[CostModel][X86] Split off BtVer2 cost checks

llvm-svn: 330433

15 files changed, 1454 insertions, 101 deletions

diff --git a/llvm/test/Analysis/CostModel/X86/alternate-shuffle-cost.ll b/llvm/test/Analysis/CostModel/X86/alternate-shuffle-cost.ll
index ee2267542f5..0bc955441e2 100644
--- a/llvm/test/Analysis/CostModel/X86/alternate-shuffle-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/alternate-shuffle-cost.ll
@@ -9,7 +9,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 ; Verify the cost model for alternate shuffles.
 
@@ -22,6 +22,10 @@ define <2 x i32> @test_v2i32(<2 x i32> %a, <2 x i32> %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x i32> %a, <2 x i32> %b, <2 x i32> <i32 0, i32 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %1
 ;
+; BTVER2-LABEL: 'test_v2i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x i32> %a, <2 x i32> %b, <2 x i32> <i32 0, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %1
+;
   %1 = shufflevector <2 x i32> %a, <2 x i32> %b, <2 x i32> <i32 0, i32 3>
   ret <2 x i32> %1
 }
@@ -43,6 +47,10 @@ define <2 x float> @test_v2f32(<2 x float> %a, <2 x float> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x float> %a, <2 x float> %b, <2 x i32> <i32 0, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %1
 ;
+; BTVER2-LABEL: 'test_v2f32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x float> %a, <2 x float> %b, <2 x i32> <i32 0, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %1
+;
   %1 = shufflevector <2 x float> %a, <2 x float> %b, <2 x i32> <i32 0, i32 3>
   ret <2 x float> %1
 }
@@ -52,6 +60,10 @@ define <2 x i32> @test_v2i32_2(<2 x i32> %a, <2 x i32> %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x i32> %a, <2 x i32> %b, <2 x i32> <i32 2, i32 1>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %1
 ;
+; BTVER2-LABEL: 'test_v2i32_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x i32> %a, <2 x i32> %b, <2 x i32> <i32 2, i32 1>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %1
+;
   %1 = shufflevector <2 x i32> %a, <2 x i32> %b, <2 x i32> <i32 2, i32 1>
   ret <2 x i32> %1
 }
@@ -73,6 +85,10 @@ define <2 x float> @test_v2f32_2(<2 x float> %a, <2 x float> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x float> %a, <2 x float> %b, <2 x i32> <i32 2, i32 1>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %1
 ;
+; BTVER2-LABEL: 'test_v2f32_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x float> %a, <2 x float> %b, <2 x i32> <i32 2, i32 1>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %1
+;
   %1 = shufflevector <2 x float> %a, <2 x float> %b, <2 x i32> <i32 2, i32 1>
   ret <2 x float> %1
 }
@@ -84,6 +100,10 @@ define <2 x i64> @test_v2i64(<2 x i64> %a, <2 x i64> %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %1
 ;
+; BTVER2-LABEL: 'test_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %1
+;
   %1 = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 3>
   ret <2 x i64> %1
 }
@@ -93,6 +113,10 @@ define <2 x double> @test_v2f64(<2 x double> %a, <2 x double> %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x double> %a, <2 x double> %b, <2 x i32> <i32 0, i32 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %1
 ;
+; BTVER2-LABEL: 'test_v2f64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x double> %a, <2 x double> %b, <2 x i32> <i32 0, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %1
+;
   %1 = shufflevector <2 x double> %a, <2 x double> %b, <2 x i32> <i32 0, i32 3>
   ret <2 x double> %1
 }
@@ -102,6 +126,10 @@ define <2 x i64> @test_v2i64_2(<2 x i64> %a, <2 x i64> %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 2, i32 1>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %1
 ;
+; BTVER2-LABEL: 'test_v2i64_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 2, i32 1>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %1
+;
   %1 = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 2, i32 1>
   ret <2 x i64> %1
 }
@@ -111,6 +139,10 @@ define <2 x double> @test_v2f64_2(<2 x double> %a, <2 x double> %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x double> %a, <2 x double> %b, <2 x i32> <i32 2, i32 1>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %1
 ;
+; BTVER2-LABEL: 'test_v2f64_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <2 x double> %a, <2 x double> %b, <2 x i32> <i32 2, i32 1>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %1
+;
   %1 = shufflevector <2 x double> %a, <2 x double> %b, <2 x i32> <i32 2, i32 1>
   ret <2 x double> %1
 }
@@ -134,6 +166,10 @@ define <4 x i32> @test_v4i32(<4 x i32> %a, <4 x i32> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %1
 ;
+; BTVER2-LABEL: 'test_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %1
+;
   %1 = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
   ret <4 x i32> %1
 }
@@ -155,6 +191,10 @@ define <4 x i32> @test_v4i32_2(<4 x i32> %a, <4 x i32> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %1
 ;
+; BTVER2-LABEL: 'test_v4i32_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %1
+;
   %1 = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
   ret <4 x i32> %1
 }
@@ -176,6 +216,10 @@ define <4 x float> @test_v4f32(<4 x float> %a, <4 x float> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
 ;
+; BTVER2-LABEL: 'test_v4f32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
+;
   %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
   ret <4 x float> %1
 }
@@ -197,6 +241,10 @@ define <4 x float> @test_v4f32_2(<4 x float> %a, <4 x float> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
 ;
+; BTVER2-LABEL: 'test_v4f32_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
+;
   %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
   ret <4 x float> %1
 }
@@ -210,6 +258,10 @@ define <4 x i64> @test_v4i64(<4 x i64> %a, <4 x i64> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %1
 ;
+; BTVER2-LABEL: 'test_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %1
+;
   %1 = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
   ret <4 x i64> %1
 }
@@ -223,6 +275,10 @@ define <4 x i64> @test_v4i64_2(<4 x i64> %a, <4 x i64> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %1
 ;
+; BTVER2-LABEL: 'test_v4i64_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %1
+;
   %1 = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
   ret <4 x i64> %1
 }
@@ -236,6 +292,10 @@ define <4 x double> @test_v4f64(<4 x double> %a, <4 x double> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x double> %1
 ;
+; BTVER2-LABEL: 'test_v4f64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x double> %1
+;
   %1 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
   ret <4 x double> %1
 }
@@ -249,6 +309,10 @@ define <4 x double> @test_v4f64_2(<4 x double> %a, <4 x double> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x double> %1
 ;
+; BTVER2-LABEL: 'test_v4f64_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x double> %1
+;
   %1 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
   ret <4 x double> %1
 }
@@ -272,6 +336,10 @@ define <8 x i16> @test_v8i16(<8 x i16> %a, <8 x i16> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %1
 ;
+; BTVER2-LABEL: 'test_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %1
+;
   %1 = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
   ret <8 x i16> %1
 }
@@ -293,6 +361,10 @@ define <8 x i16> @test_v8i16_2(<8 x i16> %a, <8 x i16> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %1
 ;
+; BTVER2-LABEL: 'test_v8i16_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %1
+;
   %1 = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
   ret <8 x i16> %1
 }
@@ -314,6 +386,10 @@ define <8 x i32> @test_v8i32(<8 x i32> %a, <8 x i32> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %1
 ;
+; BTVER2-LABEL: 'test_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %1
+;
   %1 = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
   ret <8 x i32> %1
 }
@@ -335,6 +411,10 @@ define <8 x i32> @test_v8i32_2(<8 x i32> %a, <8 x i32> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %1
 ;
+; BTVER2-LABEL: 'test_v8i32_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %1
+;
   %1 = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
   ret <8 x i32> %1
 }
@@ -356,6 +436,10 @@ define <8 x float> @test_v8f32(<8 x float> %a, <8 x float> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %1
 ;
+; BTVER2-LABEL: 'test_v8f32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %1
+;
   %1 = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
   ret <8 x float> %1
 }
@@ -377,6 +461,10 @@ define <8 x float> @test_v8f32_2(<8 x float> %a, <8 x float> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %1
 ;
+; BTVER2-LABEL: 'test_v8f32_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %1
+;
   %1 = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
   ret <8 x float> %1
 }
@@ -400,6 +488,10 @@ define <16 x i8> @test_v16i8(<16 x i8> %a, <16 x i8> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %1
 ;
+; BTVER2-LABEL: 'test_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %1
+;
   %1 = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
   ret <16 x i8> %1
 }
@@ -421,6 +513,10 @@ define <16 x i8> @test_v16i8_2(<16 x i8> %a, <16 x i8> %b) {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %1
 ;
+; BTVER2-LABEL: 'test_v16i8_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %1
+;
   %1 = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
   ret <16 x i8> %1
 }
@@ -446,6 +542,10 @@ define <16 x i16> @test_v16i16(<16 x i16> %a, <16 x i16> %b) {
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %1
 ;
+; BTVER2-LABEL: 'test_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %1 = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %1
+;
   %1 = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
   ret <16 x i16> %1
 }
@@ -471,6 +571,10 @@ define <16 x i16> @test_v16i16_2(<16 x i16> %a, <16 x i16> %b) {
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %1
 ;
+; BTVER2-LABEL: 'test_v16i16_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %1 = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %1
+;
   %1 = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
   ret <16 x i16> %1
 }
@@ -496,6 +600,10 @@ define <32 x i8> @test_v32i8(<32 x i8> %a, <32 x i8> %b) {
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 33, i32 2, i32 35, i32 4, i32 37, i32 6, i32 39, i32 8, i32 41, i32 10, i32 43, i32 12, i32 45, i32 14, i32 47, i32 16, i32 49, i32 18, i32 51, i32 20, i32 53, i32 22, i32 55, i32 24, i32 57, i32 26, i32 59, i32 28, i32 61, i32 30, i32 63>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %1
 ;
+; BTVER2-LABEL: 'test_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %1 = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 33, i32 2, i32 35, i32 4, i32 37, i32 6, i32 39, i32 8, i32 41, i32 10, i32 43, i32 12, i32 45, i32 14, i32 47, i32 16, i32 49, i32 18, i32 51, i32 20, i32 53, i32 22, i32 55, i32 24, i32 57, i32 26, i32 59, i32 28, i32 61, i32 30, i32 63>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %1
+;
   %1 = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 33, i32 2, i32 35, i32 4, i32 37, i32 6, i32 39, i32 8, i32 41, i32 10, i32 43, i32 12, i32 45, i32 14, i32 47, i32 16, i32 49, i32 18, i32 51, i32 20, i32 53, i32 22, i32 55, i32 24, i32 57, i32 26, i32 59, i32 28, i32 61, i32 30, i32 63>
   ret <32 x i8> %1
 }
@@ -521,6 +629,10 @@ define <32 x i8> @test_v32i8_2(<32 x i8> %a, <32 x i8> %b) {
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 32, i32 1, i32 34, i32 3, i32 36, i32 5, i32 38, i32 7, i32 40, i32 9, i32 42, i32 11, i32 44, i32 13, i32 46, i32 15, i32 48, i32 17, i32 50, i32 19, i32 52, i32 21, i32 54, i32 23, i32 56, i32 25, i32 58, i32 27, i32 60, i32 29, i32 62, i32 31>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %1
 ;
+; BTVER2-LABEL: 'test_v32i8_2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %1 = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 32, i32 1, i32 34, i32 3, i32 36, i32 5, i32 38, i32 7, i32 40, i32 9, i32 42, i32 11, i32 44, i32 13, i32 46, i32 15, i32 48, i32 17, i32 50, i32 19, i32 52, i32 21, i32 54, i32 23, i32 56, i32 25, i32 58, i32 27, i32 60, i32 29, i32 62, i32 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %1
+;
   %1 = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 32, i32 1, i32 34, i32 3, i32 36, i32 5, i32 38, i32 7, i32 40, i32 9, i32 42, i32 11, i32 44, i32 13, i32 46, i32 15, i32 48, i32 17, i32 50, i32 19, i32 52, i32 21, i32 54, i32 23, i32 56, i32 25, i32 58, i32 27, i32 60, i32 29, i32 62, i32 31>
   ret <32 x i8> %1
 }
diff --git a/llvm/test/Analysis/CostModel/X86/arith-fp.ll b/llvm/test/Analysis/CostModel/X86/arith-fp.ll
index 4ca0648f022..a6b5cf61473 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-fp.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-fp.ll
@@ -6,9 +6,9 @@
 ; RUN: opt < %s -enable-no-nans-fp-math  -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
 ; RUN: opt < %s -enable-no-nans-fp-math  -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
 ;
-; RUN: opt < %s -enable-no-nans-fp-math  -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SLM
-; RUN: opt < %s -enable-no-nans-fp-math  -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,GLM
-; RUN: opt < %s -enable-no-nans-fp-math  -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -enable-no-nans-fp-math  -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=slm | FileCheck %s --check-prefixes=SLM
+; RUN: opt < %s -enable-no-nans-fp-math  -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=goldmont | FileCheck %s --check-prefixes=GLM
+; RUN: opt < %s -enable-no-nans-fp-math  -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.8.0"
@@ -91,6 +91,17 @@ define i32 @fadd(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fadd <8 x double> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fadd'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fadd <8 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fadd <16 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fadd <4 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fadd <8 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = fadd float undef, undef
   %V4F32 = fadd <4 x float> undef, undef
   %V8F32 = fadd <8 x float> undef, undef
@@ -182,6 +193,17 @@ define i32 @fsub(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fsub'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = fsub float undef, undef
   %V4F32 = fsub <4 x float> undef, undef
   %V8F32 = fsub <8 x float> undef, undef
@@ -273,6 +295,17 @@ define i32 @fmul(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fmul'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = fmul float undef, undef
   %V4F32 = fmul <4 x float> undef, undef
   %V8F32 = fmul <8 x float> undef, undef
@@ -364,6 +397,17 @@ define i32 @fdiv(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 260 for instruction: %V8F64 = fdiv <8 x double> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fdiv'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = fdiv float undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = fdiv <4 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = fdiv <8 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = fdiv <16 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %F64 = fdiv double undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = fdiv <2 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = fdiv <4 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = fdiv <8 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = fdiv float undef, undef
   %V4F32 = fdiv <4 x float> undef, undef
   %V8F32 = fdiv <8 x float> undef, undef
@@ -433,6 +477,17 @@ define i32 @frem(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'frem'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = frem <8 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = frem <16 x float> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = frem <4 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = frem <8 x double> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = frem float undef, undef
   %V4F32 = frem <4 x float> undef, undef
   %V8F32 = frem <8 x float> undef, undef
@@ -524,6 +579,17 @@ define i32 @fsqrt(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fsqrt'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = call float @llvm.sqrt.f32(float undef)
   %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
   %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
@@ -593,6 +659,17 @@ define i32 @fabs(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fabs'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = call float @llvm.fabs.f32(float undef)
   %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
   %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
@@ -662,6 +739,17 @@ define i32 @fcopysign(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fcopysign'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = call float @llvm.copysign.f32(float undef, float undef)
   %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
   %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
@@ -731,6 +819,17 @@ define i32 @fma(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fma'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 87 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 174 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
   %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
   %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
diff --git a/llvm/test/Analysis/CostModel/X86/arith.ll b/llvm/test/Analysis/CostModel/X86/arith.ll
index 07445b62638..b998732acbc 100644
--- a/llvm/test/Analysis/CostModel/X86/arith.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith.ll
@@ -7,9 +7,9 @@
 ; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
 ; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+avx512f,+avx512dq | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512DQ
 ;
-; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SLM
-; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,GLM
-; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=slm | FileCheck %s --check-prefixes=SLM
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=goldmont | FileCheck %s --check-prefixes=GLM
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.8.0"
@@ -167,6 +167,25 @@ define i32 @add(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = add <64 x i8> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'add'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = add <2 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = add <4 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = add <8 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = add i32 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = add <4 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = add <8 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = add <16 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = add i16 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = add <8 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = add <16 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = add <32 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = add i8 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = add <16 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = add <32 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = add <64 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = add i64 undef, undef
   %V2I64 = add <2 x i64> undef, undef
   %V4I64 = add <4 x i64> undef, undef
@@ -343,6 +362,25 @@ define i32 @sub(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = sub <64 x i8> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sub'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = sub i64 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = sub <2 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = sub <4 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = sub <8 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = sub i32 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = sub <4 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = sub <8 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = sub <16 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = sub i16 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = sub <8 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = sub <16 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = sub <32 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = sub i8 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = sub <16 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = sub <32 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = sub <64 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = sub i64 undef, undef
   %V2I64 = sub <2 x i64> undef, undef
   %V4I64 = sub <4 x i64> undef, undef
@@ -500,6 +538,25 @@ define i32 @or(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = or <64 x i8> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'or'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = or i64 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = or <2 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = or <4 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = or <8 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = or i32 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = or <4 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = or <8 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = or <16 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = or i16 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = or <8 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = or <16 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = or <32 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = or i8 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = or <16 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = or <32 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64I8 = or <64 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = or i64 undef, undef
   %V2I64 = or <2 x i64> undef, undef
   %V4I64 = or <4 x i64> undef, undef
@@ -657,6 +714,25 @@ define i32 @xor(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = xor <64 x i8> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'xor'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = xor i64 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = xor <2 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = xor <4 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = xor <8 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = xor i32 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = xor <4 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = xor <8 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = xor <16 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = xor i16 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = xor <8 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = xor <16 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = xor <32 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = xor i8 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = xor <16 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = xor <32 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64I8 = xor <64 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = xor i64 undef, undef
   %V2I64 = xor <2 x i64> undef, undef
   %V4I64 = xor <4 x i64> undef, undef
@@ -814,6 +890,25 @@ define i32 @and(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = and <64 x i8> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'and'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = and i64 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = and <2 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = and <4 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = and <8 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = and i32 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = and <4 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = and <8 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = and <16 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = and i16 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = and <8 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = and <16 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = and <32 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = and i8 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = and <16 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = and <32 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64I8 = and <64 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = and i64 undef, undef
   %V2I64 = and <2 x i64> undef, undef
   %V4I64 = and <4 x i64> undef, undef
@@ -1009,6 +1104,25 @@ define i32 @mul(i32 %arg) {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64I8 = mul <64 x i8> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'mul'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = mul <2 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = mul <4 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = mul <8 x i64> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = mul <4 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = mul <8 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = mul <16 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = mul i16 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = mul <8 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = mul <16 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = mul <32 x i16> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = mul i8 undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I8 = mul <16 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I8 = mul <32 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V64I8 = mul <64 x i8> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = mul i64 undef, undef
   %V2I64 = mul <2 x i64> undef, undef
   %V4I64 = mul <4 x i64> undef, undef
@@ -1064,6 +1178,10 @@ define void @mul_2i32() {
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %A0 = mul <2 x i32> undef, undef
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'mul_2i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %A0 = mul <2 x i32> undef, undef
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %A0 = mul <2 x i32> undef, undef
 
   ret void
diff --git a/llvm/test/Analysis/CostModel/X86/fptosi.ll b/llvm/test/Analysis/CostModel/X86/fptosi.ll
index 2b18d12d463..ff405087253 100644
--- a/llvm/test/Analysis/CostModel/X86/fptosi.ll
+++ b/llvm/test/Analysis/CostModel/X86/fptosi.ll
@@ -8,7 +8,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 define i32 @fptosi_double_i64(i32 %arg) {
 ; SSE-LABEL: 'fptosi_double_i64'
@@ -39,6 +39,13 @@ define i32 @fptosi_double_i64(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = fptosi <8 x double> undef to <8 x i64>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptosi_double_i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = fptosi double undef to i64
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = fptosi <2 x double> undef to <2 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = fptosi <4 x double> undef to <4 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = fptosi <8 x double> undef to <8 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = fptosi double undef to i64
   %V2I64 = fptosi <2 x double> undef to <2 x i64>
   %V4I64 = fptosi <4 x double> undef to <4 x i64>
@@ -68,6 +75,13 @@ define i32 @fptosi_double_i32(i32 %arg) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = fptosi <8 x double> undef to <8 x i32>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptosi_double_i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = fptosi double undef to i32
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = fptosi <2 x double> undef to <2 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = fptosi <4 x double> undef to <4 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I32 = fptosi <8 x double> undef to <8 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I32 = fptosi double undef to i32
   %V2I32 = fptosi <2 x double> undef to <2 x i32>
   %V4I32 = fptosi <4 x double> undef to <4 x i32>
@@ -104,6 +118,13 @@ define i32 @fptosi_double_i16(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = fptosi <8 x double> undef to <8 x i16>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptosi_double_i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = fptosi double undef to i16
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I16 = fptosi <2 x double> undef to <2 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I16 = fptosi <4 x double> undef to <4 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I16 = fptosi <8 x double> undef to <8 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I16 = fptosi double undef to i16
   %V2I16 = fptosi <2 x double> undef to <2 x i16>
   %V4I16 = fptosi <4 x double> undef to <4 x i16>
@@ -140,6 +161,13 @@ define i32 @fptosi_double_i8(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I8 = fptosi <8 x double> undef to <8 x i8>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptosi_double_i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = fptosi double undef to i8
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I8 = fptosi <2 x double> undef to <2 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I8 = fptosi <4 x double> undef to <4 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I8 = fptosi <8 x double> undef to <8 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I8 = fptosi double undef to i8
   %V2I8 = fptosi <2 x double> undef to <2 x i8>
   %V4I8 = fptosi <4 x double> undef to <4 x i8>
@@ -180,6 +208,14 @@ define i32 @fptosi_float_i64(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I64 = fptosi <16 x float> undef to <16 x i64>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptosi_float_i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = fptosi float undef to i64
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = fptosi <2 x float> undef to <2 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = fptosi <4 x float> undef to <4 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = fptosi <8 x float> undef to <8 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16I64 = fptosi <16 x float> undef to <16 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = fptosi float undef to i64
   %V2I64 = fptosi <2 x float> undef to <2 x i64>
   %V4I64 = fptosi <4 x float> undef to <4 x i64>
@@ -196,6 +232,13 @@ define i32 @fptosi_float_i32(i32 %arg) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = fptosi <16 x float> undef to <16 x i32>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptosi_float_i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = fptosi float undef to i32
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = fptosi <4 x float> undef to <4 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = fptosi <8 x float> undef to <8 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = fptosi <16 x float> undef to <16 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I32 = fptosi float undef to i32
   %V4I32 = fptosi <4 x float> undef to <4 x i32>
   %V8I32 = fptosi <8 x float> undef to <8 x i32>
@@ -225,6 +268,13 @@ define i32 @fptosi_float_i16(i32 %arg) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = fptosi <16 x float> undef to <16 x i16>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptosi_float_i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = fptosi float undef to i16
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I16 = fptosi <4 x float> undef to <4 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = fptosi <8 x float> undef to <8 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I16 = fptosi <16 x float> undef to <16 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I16 = fptosi float undef to i16
   %V4I16 = fptosi <4 x float> undef to <4 x i16>
   %V8I16 = fptosi <8 x float> undef to <8 x i16>
@@ -254,6 +304,13 @@ define i32 @fptosi_float_i8(i32 %arg) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = fptosi <16 x float> undef to <16 x i8>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptosi_float_i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = fptosi float undef to i8
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I8 = fptosi <4 x float> undef to <4 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I8 = fptosi <8 x float> undef to <8 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I8 = fptosi <16 x float> undef to <16 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I8 = fptosi float undef to i8
   %V4I8 = fptosi <4 x float> undef to <4 x i8>
   %V8I8 = fptosi <8 x float> undef to <8 x i8>
diff --git a/llvm/test/Analysis/CostModel/X86/fptoui.ll b/llvm/test/Analysis/CostModel/X86/fptoui.ll
index 1d2331c2c1e..ae42fd6c4d0 100644
--- a/llvm/test/Analysis/CostModel/X86/fptoui.ll
+++ b/llvm/test/Analysis/CostModel/X86/fptoui.ll
@@ -8,7 +8,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 define i32 @fptoui_double_i64(i32 %arg) {
 ; SSE-LABEL: 'fptoui_double_i64'
@@ -39,6 +39,13 @@ define i32 @fptoui_double_i64(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = fptoui <8 x double> undef to <8 x i64>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptoui_double_i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = fptoui double undef to i64
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = fptoui <2 x double> undef to <2 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = fptoui <4 x double> undef to <4 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V8I64 = fptoui <8 x double> undef to <8 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = fptoui double undef to i64
   %V2I64 = fptoui <2 x double> undef to <2 x i64>
   %V4I64 = fptoui <4 x double> undef to <4 x i64>
@@ -75,6 +82,13 @@ define i32 @fptoui_double_i32(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = fptoui <8 x double> undef to <8 x i32>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptoui_double_i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = fptoui double undef to i32
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = fptoui <2 x double> undef to <2 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = fptoui <4 x double> undef to <4 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I32 = fptoui <8 x double> undef to <8 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I32 = fptoui double undef to i32
   %V2I32 = fptoui <2 x double> undef to <2 x i32>
   %V4I32 = fptoui <4 x double> undef to <4 x i32>
@@ -111,6 +125,13 @@ define i32 @fptoui_double_i16(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = fptoui <8 x double> undef to <8 x i16>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptoui_double_i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = fptoui double undef to i16
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I16 = fptoui <2 x double> undef to <2 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = fptoui <4 x double> undef to <4 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I16 = fptoui <8 x double> undef to <8 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I16 = fptoui double undef to i16
   %V2I16 = fptoui <2 x double> undef to <2 x i16>
   %V4I16 = fptoui <4 x double> undef to <4 x i16>
@@ -147,6 +168,13 @@ define i32 @fptoui_double_i8(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = fptoui <8 x double> undef to <8 x i8>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptoui_double_i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = fptoui double undef to i8
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I8 = fptoui <2 x double> undef to <2 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I8 = fptoui <4 x double> undef to <4 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I8 = fptoui <8 x double> undef to <8 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I8 = fptoui double undef to i8
   %V2I8 = fptoui <2 x double> undef to <2 x i8>
   %V4I8 = fptoui <4 x double> undef to <4 x i8>
@@ -187,6 +215,14 @@ define i32 @fptoui_float_i64(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I64 = fptoui <16 x float> undef to <16 x i64>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptoui_float_i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = fptoui float undef to i64
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = fptoui <2 x float> undef to <2 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = fptoui <4 x float> undef to <4 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V8I64 = fptoui <8 x float> undef to <8 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V16I64 = fptoui <16 x float> undef to <16 x i64>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I64 = fptoui float undef to i64
   %V2I64 = fptoui <2 x float> undef to <2 x i64>
   %V4I64 = fptoui <4 x float> undef to <4 x i64>
@@ -217,6 +253,13 @@ define i32 @fptoui_float_i32(i32 %arg) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = fptoui <16 x float> undef to <16 x i32>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptoui_float_i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = fptoui float undef to i32
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = fptoui <4 x float> undef to <4 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = fptoui <8 x float> undef to <8 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V16I32 = fptoui <16 x float> undef to <16 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I32 = fptoui float undef to i32
   %V4I32 = fptoui <4 x float> undef to <4 x i32>
   %V8I32 = fptoui <8 x float> undef to <8 x i32>
@@ -246,6 +289,13 @@ define i32 @fptoui_float_i16(i32 %arg) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = fptoui <16 x float> undef to <16 x i16>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptoui_float_i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = fptoui float undef to i16
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = fptoui <4 x float> undef to <4 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = fptoui <8 x float> undef to <8 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I16 = fptoui <16 x float> undef to <16 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I16 = fptoui float undef to i16
   %V4I16 = fptoui <4 x float> undef to <4 x i16>
   %V8I16 = fptoui <8 x float> undef to <8 x i16>
@@ -275,6 +325,13 @@ define i32 @fptoui_float_i8(i32 %arg) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = fptoui <16 x float> undef to <16 x i8>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'fptoui_float_i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = fptoui float undef to i8
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I8 = fptoui <4 x float> undef to <4 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I8 = fptoui <8 x float> undef to <8 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I8 = fptoui <16 x float> undef to <16 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %I8 = fptoui float undef to i8
   %V4I8 = fptoui <4 x float> undef to <4 x i8>
   %V8I8 = fptoui <8 x float> undef to <8 x i8>
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-broadcast.ll b/llvm/test/Analysis/CostModel/X86/shuffle-broadcast.ll
index 025791370ae..2a27c822f2b 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-broadcast.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-broadcast.ll
@@ -10,7 +10,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 ;
 ; Verify the cost model for broadcast shuffles.
@@ -41,6 +41,12 @@ define void @test_vXf64(<2 x double> %src128, <4 x double> %src256, <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <8 x double> %src512, <8 x double> undef, <8 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXf64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %src128, <2 x double> undef, <2 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %src256, <4 x double> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512 = shufflevector <8 x double> %src512, <8 x double> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <2 x double> %src128, <2 x double> undef, <2 x i32> zeroinitializer
   %V256 = shufflevector <4 x double> %src256, <4 x double> undef, <4 x i32> zeroinitializer
   %V512 = shufflevector <8 x double> %src512, <8 x double> undef, <8 x i32> zeroinitializer
@@ -72,6 +78,12 @@ define void @test_vXi64(<2 x i64> %src128, <4 x i64> %src256, <8 x i64> %src512)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %src128, <2 x i64> undef, <2 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %src256, <4 x i64> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <2 x i64> %src128, <2 x i64> undef, <2 x i32> zeroinitializer
   %V256 = shufflevector <4 x i64> %src256, <4 x i64> undef, <4 x i32> zeroinitializer
   %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> zeroinitializer
@@ -107,6 +119,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <16 x float> %src512, <16 x float> undef, <16 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXf32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64 = shufflevector <2 x float> %src64, <2 x float> undef, <2 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <4 x float> %src128, <4 x float> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x float> %src256, <8 x float> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512 = shufflevector <16 x float> %src512, <16 x float> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V64 = shufflevector <2 x float> %src64, <2 x float> undef, <2 x i32> zeroinitializer
   %V128 = shufflevector <4 x float> %src128, <4 x float> undef, <4 x i32> zeroinitializer
   %V256 = shufflevector <8 x float> %src256, <8 x float> undef, <8 x i32> zeroinitializer
@@ -143,6 +162,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <16 x i32> %src512, <16 x i32> undef, <16 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64 = shufflevector <2 x i32> %src64, <2 x i32> undef, <2 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <4 x i32> %src128, <4 x i32> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x i32> %src256, <8 x i32> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512 = shufflevector <16 x i32> %src512, <16 x i32> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V64 = shufflevector <2 x i32> %src64, <2 x i32> undef, <2 x i32> zeroinitializer
   %V128 = shufflevector <4 x i32> %src128, <4 x i32> undef, <4 x i32> zeroinitializer
   %V256 = shufflevector <8 x i32> %src256, <8 x i32> undef, <8 x i32> zeroinitializer
@@ -187,6 +213,12 @@ define void @test_vXi16(<8 x i16> %src128, <16 x i16> %src256, <32 x i16> %src51
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <32 x i16> %src512, <32 x i16> undef, <32 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <8 x i16> %src128, <8 x i16> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V256 = shufflevector <16 x i16> %src256, <16 x i16> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V512 = shufflevector <32 x i16> %src512, <32 x i16> undef, <32 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <8 x i16> %src128, <8 x i16> undef, <8 x i32> zeroinitializer
   %V256 = shufflevector <16 x i16> %src256, <16 x i16> undef, <16 x i32> zeroinitializer
   %V512 = shufflevector <32 x i16> %src512, <32 x i16> undef, <32 x i32> zeroinitializer
@@ -230,6 +262,12 @@ define void @test_vXi8(<16 x i8> %src128, <32 x i8> %src256, <64 x i8> %src512)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> zeroinitializer
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <16 x i8> %src128, <16 x i8> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <32 x i8> %src256, <32 x i8> undef, <32 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <16 x i8> %src128, <16 x i8> undef, <16 x i32> zeroinitializer
   %V256 = shufflevector <32 x i8> %src256, <32 x i8> undef, <32 x i32> zeroinitializer
   %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> zeroinitializer
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-reverse.ll b/llvm/test/Analysis/CostModel/X86/shuffle-reverse.ll
index 5768238798c..11f39510b02 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-reverse.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-reverse.ll
@@ -10,7 +10,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 ;
 ; Verify the cost model for reverse shuffles.
@@ -41,6 +41,12 @@ define void @test_vXf64(<2 x double> %src128, <4 x double> %src256, <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <8 x double> %src512, <8 x double> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXf64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %src128, <2 x double> undef, <2 x i32> <i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %src256, <4 x double> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %src512, <8 x double> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <2 x double> %src128, <2 x double> undef, <2 x i32> <i32 1, i32 0>
   %V256 = shufflevector <4 x double> %src256, <4 x double> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <8 x double> %src512, <8 x double> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -72,6 +78,12 @@ define void @test_vXi64(<2 x i64> %src128, <4 x i64> %src256, <8 x i64> %src512)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %src128, <2 x i64> undef, <2 x i32> <i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %src256, <4 x i64> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <2 x i64> %src128, <2 x i64> undef, <2 x i32> <i32 1, i32 0>
   %V256 = shufflevector <4 x i64> %src256, <4 x i64> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -107,6 +119,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <16 x float> %src512, <16 x float> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXf32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64 = shufflevector <2 x float> %src64, <2 x float> undef, <2 x i32> <i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <4 x float> %src128, <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x float> %src256, <8 x float> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x float> %src512, <16 x float> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V64 = shufflevector <2 x float> %src64, <2 x float> undef, <2 x i32> <i32 1, i32 0>
   %V128 = shufflevector <4 x float> %src128, <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
   %V256 = shufflevector <8 x float> %src256, <8 x float> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -143,6 +162,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <16 x i32> %src512, <16 x i32> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64 = shufflevector <2 x i32> %src64, <2 x i32> undef, <2 x i32> <i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <4 x i32> %src128, <4 x i32> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x i32> %src256, <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x i32> %src512, <16 x i32> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V64 = shufflevector <2 x i32> %src64, <2 x i32> undef, <2 x i32> <i32 1, i32 0>
   %V128 = shufflevector <4 x i32> %src128, <4 x i32> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
   %V256 = shufflevector <8 x i32> %src256, <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -199,6 +225,12 @@ define void @test_vXi16(<8 x i16> %src128, <16 x i16> %src256, <32 x i16> %src51
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <32 x i16> %src512, <32 x i16> undef, <32 x i32> <i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <8 x i16> %src128, <8 x i16> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <16 x i16> %src256, <16 x i16> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <32 x i16> %src512, <32 x i16> undef, <32 x i32> <i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <8 x i16> %src128, <8 x i16> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V256 = shufflevector <16 x i16> %src256, <16 x i16> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <32 x i16> %src512, <32 x i16> undef, <32 x i32> <i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -254,6 +286,12 @@ define void @test_vXi8(<16 x i8> %src128, <32 x i8> %src256, <64 x i8> %src512)
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> <i32 63, i32 62, i32 61, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <16 x i8> %src128, <16 x i8> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <32 x i8> %src256, <32 x i8> undef, <32 x i32> <i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> <i32 63, i32 62, i32 61, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <16 x i8> %src128, <16 x i8> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V256 = shufflevector <32 x i8> %src256, <32 x i8> undef, <32 x i32> <i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> <i32 63, i32 62, i32 61, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-single-src.ll b/llvm/test/Analysis/CostModel/X86/shuffle-single-src.ll
index f93794518bd..b90575d97c4 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-single-src.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-single-src.ll
@@ -11,7 +11,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 ;
 ; Verify the cost model for 1 src shuffles
@@ -53,6 +53,13 @@ define void @test_vXf64(<2 x double> %src128, <4 x double> %src256, <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1024 = shufflevector <16 x double> %src1024, <16 x double> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 13, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXf64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %src128, <2 x double> undef, <2 x i32> <i32 1, i32 1>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V256 = shufflevector <4 x double> %src256, <4 x double> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <8 x double> %src512, <8 x double> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V1024 = shufflevector <16 x double> %src1024, <16 x double> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 13, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <2 x double> %src128, <2 x double> undef, <2 x i32> <i32 1, i32 1>
   %V256 = shufflevector <4 x double> %src256, <4 x double> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 0>
   %V512 = shufflevector <8 x double> %src512, <8 x double> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -91,6 +98,12 @@ define void @test_vXi64(<2 x i64> %src128, <4 x i64> %src256, <8 x i64> %src512)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %src128, <2 x i64> undef, <2 x i32> <i32 1, i32 1>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V256 = shufflevector <4 x i64> %src256, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <2 x i64> %src128, <2 x i64> undef, <2 x i32> <i32 1, i32 1>
   %V256 = shufflevector <4 x i64> %src256, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 0>
   %V512 = shufflevector <8 x i64> %src512, <8 x i64> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -128,6 +141,12 @@ define void @test_vXf32(<4 x float> %src128, <8 x float> %src256, <16 x float> %
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <16 x float> %src512, <16 x float> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 13, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXf32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <4 x float> %src128, <4 x float> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x float> %src256, <8 x float> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x float> %src512, <16 x float> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 13, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <4 x float> %src128, <4 x float> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 0>
   %V256 = shufflevector <8 x float> %src256, <8 x float> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <16 x float> %src512, <16 x float> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 13, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -170,6 +189,13 @@ define void @test_vXi32(<4 x i32> %src128, <8 x i32> %src256, <16 x i32> %src512
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1024 = shufflevector <32 x i32> %src1024, <32 x i32> undef, <32 x i32> <i32 31, i32 30, i32 20, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 11, i32 9, i32 8, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <4 x i32> %src128, <4 x i32> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x i32> %src256, <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 5, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x i32> %src512, <16 x i32> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 13, i32 10, i32 9, i32 8, i32 8, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V1024 = shufflevector <32 x i32> %src1024, <32 x i32> undef, <32 x i32> <i32 31, i32 30, i32 20, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 11, i32 9, i32 8, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <4 x i32> %src128, <4 x i32> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 0>
   %V256 = shufflevector <8 x i32> %src256, <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 5, i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <16 x i32> %src512, <16 x i32> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 13, i32 10, i32 9, i32 8, i32 8, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -241,6 +267,13 @@ define void @test_vXi16(<8 x i16> %src128, <16 x i16> %src256, <32 x i16> %src51
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1024 = shufflevector <64 x i16> %src1024, <64 x i16> undef, <64 x i32> <i32 63, i32 62, i32 61, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <8 x i16> %src128, <8 x i16> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V256 = shufflevector <16 x i16> %src256, <16 x i16> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 13, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V512 = shufflevector <32 x i16> %src512, <32 x i16> undef, <32 x i32> <i32 31, i32 30, i32 20, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 11, i32 9, i32 8, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 180 for instruction: %V1024 = shufflevector <64 x i16> %src1024, <64 x i16> undef, <64 x i32> <i32 63, i32 62, i32 61, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <8 x i16> %src128, <8 x i16> undef, <8 x i32> <i32 7, i32 6, i32 6, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V256 = shufflevector <16 x i16> %src256, <16 x i16> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 13, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <32 x i16> %src512, <32 x i16> undef, <32 x i32> <i32 31, i32 30, i32 20, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 11, i32 9, i32 8, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -303,6 +336,12 @@ define void @test_vXi8(<16 x i8> %src128, <32 x i8> %src256, <64 x i8> %src512)
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> <i32 63, i32 62, i32 61, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <16 x i8> %src128, <16 x i8> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 11, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V256 = shufflevector <32 x i8> %src256, <32 x i8> undef, <32 x i32> <i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 8, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> <i32 63, i32 62, i32 61, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <16 x i8> %src128, <16 x i8> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 11, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V256 = shufflevector <32 x i8> %src256, <32 x i8> undef, <32 x i32> <i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 8, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <64 x i8> %src512, <64 x i8> undef, <64 x i32> <i32 63, i32 62, i32 61, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-two-src.ll b/llvm/test/Analysis/CostModel/X86/shuffle-two-src.ll
index a6529a84c72..c8bad46fe3e 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-two-src.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-two-src.ll
@@ -11,7 +11,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 ;
 ; Verify the cost model for 2 src shuffles
@@ -53,6 +53,13 @@ define void @test_vXf64(<2 x double> %src128, <4 x double> %src256, <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V1024 = shufflevector <16 x double> %src1024, <16 x double> %src1024_1, <16 x i32> <i32 30, i32 14, i32 13, i32 12, i32 13, i32 10, i32 18, i32 8, i32 8, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXf64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %src128, <2 x double> %src128_1, <2 x i32> <i32 3, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <4 x double> %src256, <4 x double> %src256_1, <4 x i32> <i32 3, i32 3, i32 7, i32 6>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512 = shufflevector <8 x double> %src512, <8 x double> %src512_1, <8 x i32> <i32 7, i32 6, i32 12, i32 4, i32 3, i32 2, i32 1, i32 15>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V1024 = shufflevector <16 x double> %src1024, <16 x double> %src1024_1, <16 x i32> <i32 30, i32 14, i32 13, i32 12, i32 13, i32 10, i32 18, i32 8, i32 8, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <2 x double> %src128, <2 x double> %src128_1, <2 x i32> <i32 3, i32 0>
   %V256 = shufflevector <4 x double> %src256, <4 x double> %src256_1, <4 x i32> <i32 3, i32 3, i32 7, i32 6>
   %V512 = shufflevector <8 x double> %src512, <8 x double> %src512_1, <8 x i32> <i32 7, i32 6, i32 12, i32 4, i32 3, i32 2, i32 1, i32 15>
@@ -96,6 +103,13 @@ define void @test_vXi64(<2 x i64> %src128, <4 x i64> %src256, <8 x i64> %src512,
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V1024 = shufflevector <16 x i64> %src1024, <16 x i64> %src1024_1, <16 x i32> <i32 30, i32 14, i32 13, i32 12, i32 13, i32 10, i32 18, i32 8, i32 8, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %src128, <2 x i64> %src128_1, <2 x i32> <i32 3, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <4 x i64> %src256, <4 x i64> %src256_1, <4 x i32> <i32 3, i32 3, i32 7, i32 6>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512 = shufflevector <8 x i64> %src512, <8 x i64> %src512_1, <8 x i32> <i32 7, i32 6, i32 12, i32 4, i32 3, i32 2, i32 1, i32 15>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V1024 = shufflevector <16 x i64> %src1024, <16 x i64> %src1024_1, <16 x i32> <i32 30, i32 14, i32 13, i32 12, i32 13, i32 10, i32 18, i32 8, i32 8, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <2 x i64> %src128, <2 x i64> %src128_1, <2 x i32> <i32 3, i32 0>
   %V256 = shufflevector <4 x i64> %src256, <4 x i64> %src256_1, <4 x i32> <i32 3, i32 3, i32 7, i32 6>
   %V512 = shufflevector <8 x i64> %src512, <8 x i64> %src512_1, <8 x i32> <i32 7, i32 6, i32 12, i32 4, i32 3, i32 2, i32 1, i32 15>
@@ -139,6 +153,13 @@ define void @test_vXf32(<4 x float> %src128, <8 x float> %src256, <16 x float> %
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V1024 = shufflevector <32 x float> %src1024, <32 x float> %src1024_1, <32 x i32> <i32 31, i32 33, i32 20, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 48, i32 13, i32 12, i32 11, i32 11, i32 9, i32 45, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXf32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %src128, <4 x float> %src128_1, <4 x i32> <i32 3, i32 6, i32 1, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x float> %src256, <8 x float> %src256_1, <8 x i32> <i32 7, i32 6, i32 8, i32 4, i32 3, i32 2, i32 12, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512 = shufflevector <16 x float> %src512, <16 x float> %src512_1, <16 x i32> <i32 15, i32 17, i32 13, i32 20, i32 11, i32 10, i32 8, i32 8, i32 7, i32 22, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V1024 = shufflevector <32 x float> %src1024, <32 x float> %src1024_1, <32 x i32> <i32 31, i32 33, i32 20, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 48, i32 13, i32 12, i32 11, i32 11, i32 9, i32 45, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <4 x float> %src128, <4 x float> %src128_1, <4 x i32> <i32 3, i32 6, i32 1, i32 5>
   %V256 = shufflevector <8 x float> %src256, <8 x float> %src256_1, <8 x i32> <i32 7, i32 6, i32 8, i32 4, i32 3, i32 2, i32 12, i32 0>
   %V512 = shufflevector <16 x float> %src512, <16 x float> %src512_1, <16 x i32> <i32 15, i32 17, i32 13, i32 20, i32 11, i32 10, i32 8, i32 8, i32 7, i32 22, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -182,6 +203,13 @@ define void @test_vXi32(<4 x i32> %src128, <8 x i32> %src256, <16 x i32> %src512
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V1024 = shufflevector <32 x i32> %src1024, <32 x i32> %src1024_1, <32 x i32> <i32 31, i32 33, i32 20, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 48, i32 13, i32 12, i32 11, i32 11, i32 9, i32 45, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %src128, <4 x i32> %src128_1, <4 x i32> <i32 3, i32 6, i32 1, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x i32> %src256, <8 x i32> %src256_1, <8 x i32> <i32 7, i32 6, i32 8, i32 4, i32 3, i32 2, i32 12, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512 = shufflevector <16 x i32> %src512, <16 x i32> %src512_1, <16 x i32> <i32 15, i32 17, i32 13, i32 20, i32 11, i32 10, i32 8, i32 8, i32 7, i32 22, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V1024 = shufflevector <32 x i32> %src1024, <32 x i32> %src1024_1, <32 x i32> <i32 31, i32 33, i32 20, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 48, i32 13, i32 12, i32 11, i32 11, i32 9, i32 45, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <4 x i32> %src128, <4 x i32> %src128_1, <4 x i32> <i32 3, i32 6, i32 1, i32 5>
   %V256 = shufflevector <8 x i32> %src256, <8 x i32> %src256_1, <8 x i32> <i32 7, i32 6, i32 8, i32 4, i32 3, i32 2, i32 12, i32 0>
   %V512 = shufflevector <16 x i32> %src512, <16 x i32> %src512_1, <16 x i32> <i32 15, i32 17, i32 13, i32 20, i32 11, i32 10, i32 8, i32 8, i32 7, i32 22, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -253,6 +281,13 @@ define void @test_vXi16(<8 x i16> %src128, <16 x i16> %src256, <32 x i16> %src51
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V1024 = shufflevector <64 x i16> %src1024, <64 x i16> %src1024_1, <64 x i32> <i32 63, i32 62, i32 71, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 66, i32 2, i32 1, i32 0>
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %src128, <8 x i16> %src128_1, <8 x i32> <i32 7, i32 6, i32 6, i32 8, i32 9, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V256 = shufflevector <16 x i16> %src256, <16 x i16> %src256_1, <16 x i32> <i32 15, i32 14, i32 13, i32 20, i32 21, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V512 = shufflevector <32 x i16> %src512, <32 x i16> %src512_1, <32 x i32> <i32 31, i32 30, i32 45, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 38, i32 11, i32 11, i32 9, i32 8, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 420 for instruction: %V1024 = shufflevector <64 x i16> %src1024, <64 x i16> %src1024_1, <64 x i32> <i32 63, i32 62, i32 71, i32 60, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 66, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <8 x i16> %src128, <8 x i16> %src128_1, <8 x i32> <i32 7, i32 6, i32 6, i32 8, i32 9, i32 2, i32 1, i32 0>
   %V256 = shufflevector <16 x i16> %src256, <16 x i16> %src256_1, <16 x i32> <i32 15, i32 14, i32 13, i32 20, i32 21, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <32 x i16> %src512, <32 x i16> %src512_1, <32 x i32> <i32 31, i32 30, i32 45, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 38, i32 11, i32 11, i32 9, i32 8, i32 7, i32 11, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
@@ -315,6 +350,12 @@ define void @test_vXi8(<16 x i8> %src128, <32 x i8> %src256, <64 x i8> %src512,
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512 = shufflevector <64 x i8> %src512, <64 x i8> %src512_1, <64 x i32> <i32 63, i32 100, i32 61, i32 96, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; AVX512VBMI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
+; BTVER2-LABEL: 'test_vXi8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %src128, <16 x i8> %src128_1, <16 x i32> <i32 29, i32 14, i32 28, i32 12, i32 11, i32 10, i32 11, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V256 = shufflevector <32 x i8> %src256, <32 x i8> %src256_1, <32 x i32> <i32 31, i32 30, i32 45, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 8, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V512 = shufflevector <64 x i8> %src512, <64 x i8> %src512_1, <64 x i32> <i32 63, i32 100, i32 61, i32 96, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
   %V128 = shufflevector <16 x i8> %src128, <16 x i8> %src128_1, <16 x i32> <i32 29, i32 14, i32 28, i32 12, i32 11, i32 10, i32 11, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V256 = shufflevector <32 x i8> %src256, <32 x i8> %src256_1, <32 x i32> <i32 31, i32 30, i32 45, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 22, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 8, i32 8, i32 7, i32 6, i32 8, i32 4, i32 3, i32 2, i32 1, i32 0>
   %V512 = shufflevector <64 x i8> %src512, <64 x i8> %src512_1, <64 x i32> <i32 63, i32 100, i32 61, i32 96, i32 59, i32 58, i32 57, i32 56, i32 55, i32 54, i32 53, i32 52, i32 51, i32 50, i32 49, i32 48, i32 47, i32 46, i32 45, i32 44, i32 43, i32 42, i32 41, i32 40, i32 39, i32 38, i32 37, i32 36, i32 35, i32 34, i32 33, i32 32, i32 31, i32 30, i32 29, i32 28, i32 27, i32 26, i32 25, i32 24, i32 23, i32 20, i32 21, i32 20, i32 19, i32 18, i32 17, i32 16, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
diff --git a/llvm/test/Analysis/CostModel/X86/sitofp.ll b/llvm/test/Analysis/CostModel/X86/sitofp.ll
index 2e545f548e7..c39988a4fc5 100644
--- a/llvm/test/Analysis/CostModel/X86/sitofp.ll
+++ b/llvm/test/Analysis/CostModel/X86/sitofp.ll
@@ -8,7 +8,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 define i32 @sitofp_i8_double() {
 ; SSE-LABEL: 'sitofp_i8_double'
@@ -32,6 +32,13 @@ define i32 @sitofp_i8_double() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v8i8_v8f64 = sitofp <8 x i8> undef to <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sitofp_i8_double'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i8_f64 = sitofp i8 undef to double
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %cvt_v2i8_v2f64 = sitofp <2 x i8> undef to <2 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %cvt_v4i8_v4f64 = sitofp <4 x i8> undef to <4 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %cvt_v8i8_v8f64 = sitofp <8 x i8> undef to <8 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i8_f64 = sitofp i8 undef to double
   %cvt_v2i8_v2f64 = sitofp <2 x i8> undef to <2 x double>
   %cvt_v4i8_v4f64 = sitofp <4 x i8> undef to <4 x double>
@@ -61,6 +68,13 @@ define i32 @sitofp_i16_double() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v8i16_v8f64 = sitofp <8 x i16> undef to <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sitofp_i16_double'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i16_f64 = sitofp i16 undef to double
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %cvt_v2i16_v2f64 = sitofp <2 x i16> undef to <2 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %cvt_v4i16_v4f64 = sitofp <4 x i16> undef to <4 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %cvt_v8i16_v8f64 = sitofp <8 x i16> undef to <8 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i16_f64 = sitofp i16 undef to double
   %cvt_v2i16_v2f64 = sitofp <2 x i16> undef to <2 x double>
   %cvt_v4i16_v4f64 = sitofp <4 x i16> undef to <4 x double>
@@ -90,6 +104,13 @@ define i32 @sitofp_i32_double() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v8i32_v8f64 = sitofp <8 x i32> undef to <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sitofp_i32_double'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i32_f64 = sitofp i32 undef to double
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %cvt_v2i32_v2f64 = sitofp <2 x i32> undef to <2 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v4i32_v4f64 = sitofp <4 x i32> undef to <4 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %cvt_v8i32_v8f64 = sitofp <8 x i32> undef to <8 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i32_f64 = sitofp i32 undef to double
   %cvt_v2i32_v2f64 = sitofp <2 x i32> undef to <2 x double>
   %cvt_v4i32_v4f64 = sitofp <4 x i32> undef to <4 x double>
@@ -126,6 +147,13 @@ define i32 @sitofp_i64_double() {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v8i64_v8f64 = sitofp <8 x i64> undef to <8 x double>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sitofp_i64_double'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i64_f64 = sitofp i64 undef to double
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %cvt_v2i64_v2f64 = sitofp <2 x i64> undef to <2 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %cvt_v4i64_v4f64 = sitofp <4 x i64> undef to <4 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %cvt_v8i64_v8f64 = sitofp <8 x i64> undef to <8 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i64_f64 = sitofp i64 undef to double
   %cvt_v2i64_v2f64 = sitofp <2 x i64> undef to <2 x double>
   %cvt_v4i64_v4f64 = sitofp <4 x i64> undef to <4 x double>
@@ -155,6 +183,13 @@ define i32 @sitofp_i8_float() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v16i8_v16f32 = sitofp <16 x i8> undef to <16 x float>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sitofp_i8_float'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i8_f32 = sitofp i8 undef to float
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %cvt_v4i8_v4f32 = sitofp <4 x i8> undef to <4 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %cvt_v8i8_v8f32 = sitofp <8 x i8> undef to <8 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %cvt_v16i8_v16f32 = sitofp <16 x i8> undef to <16 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i8_f32 = sitofp i8 undef to float
   %cvt_v4i8_v4f32 = sitofp <4 x i8> undef to <4 x float>
   %cvt_v8i8_v8f32 = sitofp <8 x i8> undef to <8 x float>
@@ -184,6 +219,13 @@ define i32 @sitofp_i16_float() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v16i16_v16f32 = sitofp <16 x i16> undef to <16 x float>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sitofp_i16_float'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i16_f32 = sitofp i16 undef to float
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %cvt_v4i16_v4f32 = sitofp <4 x i16> undef to <4 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %cvt_v8i16_v8f32 = sitofp <8 x i16> undef to <8 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %cvt_v16i16_v16f32 = sitofp <16 x i16> undef to <16 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i16_f32 = sitofp i16 undef to float
   %cvt_v4i16_v4f32 = sitofp <4 x i16> undef to <4 x float>
   %cvt_v8i16_v8f32 = sitofp <8 x i16> undef to <8 x float>
@@ -213,6 +255,13 @@ define i32 @sitofp_i32_float() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v16i32_v16f32 = sitofp <16 x i32> undef to <16 x float>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sitofp_i32_float'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i32_f32 = sitofp i32 undef to float
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v4i32_v4f32 = sitofp <4 x i32> undef to <4 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v8i32_v8f32 = sitofp <8 x i32> undef to <8 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %cvt_v16i32_v16f32 = sitofp <16 x i32> undef to <16 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i32_f32 = sitofp i32 undef to float
   %cvt_v4i32_v4f32 = sitofp <4 x i32> undef to <4 x float>
   %cvt_v8i32_v8f32 = sitofp <8 x i32> undef to <8 x float>
@@ -253,6 +302,14 @@ define i32 @sitofp_i64_float() {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %cvt_v16i64_v16f32 = sitofp <16 x i64> undef to <16 x float>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'sitofp_i64_float'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i64_f32 = sitofp i64 undef to float
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %cvt_v2i64_v2f32 = sitofp <2 x i64> undef to <2 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %cvt_v4i64_v4f32 = sitofp <4 x i64> undef to <4 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %cvt_v8i64_v8f32 = sitofp <8 x i64> undef to <8 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %cvt_v16i64_v16f32 = sitofp <16 x i64> undef to <16 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i64_f32 = sitofp i64 undef to float
   %cvt_v2i64_v2f32 = sitofp <2 x i64> undef to <2 x float>
   %cvt_v4i64_v4f32 = sitofp <4 x i64> undef to <4 x float>
diff --git a/llvm/test/Analysis/CostModel/X86/trunc.ll b/llvm/test/Analysis/CostModel/X86/trunc.ll
index bf425820334..2e0dc8e5daf 100644
--- a/llvm/test/Analysis/CostModel/X86/trunc.ll
+++ b/llvm/test/Analysis/CostModel/X86/trunc.ll
@@ -9,7 +9,7 @@
 ;
 ; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 define i32 @trunc_vXi32() {
 ; SSE-LABEL: 'trunc_vXi32'
@@ -36,6 +36,12 @@ define i32 @trunc_vXi32() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i32>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'trunc_vXi32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i32>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %V2i64 = trunc <2 x i64> undef to <2 x i32>
   %V4i64 = trunc <4 x i64> undef to <4 x i32>
   %V8i64 = trunc <8 x i64> undef to <8 x i32>
@@ -97,6 +103,15 @@ define i32 @trunc_vXi16() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'trunc_vXi16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i16>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %V2i64 = trunc <2 x i64> undef to <2 x i16>
   %V4i64 = trunc <4 x i64> undef to <4 x i16>
   %V8i64 = trunc <8 x i64> undef to <8 x i16>
@@ -212,6 +227,21 @@ define i32 @trunc_vXi8() {
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V32i16 = trunc <32 x i16> undef to <32 x i8>
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'trunc_vXi8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i64 = trunc <2 x i64> undef to <2 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4i64 = trunc <4 x i64> undef to <4 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V8i64 = trunc <8 x i64> undef to <8 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i32 = trunc <2 x i32> undef to <2 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4i32 = trunc <4 x i32> undef to <4 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8i32 = trunc <8 x i32> undef to <8 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16i32 = trunc <16 x i32> undef to <16 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V2i16 = trunc <2 x i16> undef to <2 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4i16 = trunc <4 x i16> undef to <4 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8i16 = trunc <8 x i16> undef to <8 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16i16 = trunc <16 x i16> undef to <16 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V32i16 = trunc <32 x i16> undef to <32 x i8>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %V2i64 = trunc <2 x i64> undef to <2 x i8>
   %V4i64 = trunc <4 x i64> undef to <4 x i8>
   %V8i64 = trunc <8 x i64> undef to <8 x i8>
diff --git a/llvm/test/Analysis/CostModel/X86/uitofp.ll b/llvm/test/Analysis/CostModel/X86/uitofp.ll
index 3920764ce57..9b8bd082923 100644
--- a/llvm/test/Analysis/CostModel/X86/uitofp.ll
+++ b/llvm/test/Analysis/CostModel/X86/uitofp.ll
@@ -8,7 +8,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-apple-darwin -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 define i32 @uitofp_i8_double() {
 ; SSE-LABEL: 'uitofp_i8_double'
@@ -32,6 +32,13 @@ define i32 @uitofp_i8_double() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v8i8_v8f64 = uitofp <8 x i8> undef to <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'uitofp_i8_double'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i8_f64 = uitofp i8 undef to double
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %cvt_v2i8_v2f64 = uitofp <2 x i8> undef to <2 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v4i8_v4f64 = uitofp <4 x i8> undef to <4 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %cvt_v8i8_v8f64 = uitofp <8 x i8> undef to <8 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i8_f64 = uitofp i8 undef to double
   %cvt_v2i8_v2f64 = uitofp <2 x i8> undef to <2 x double>
   %cvt_v4i8_v4f64 = uitofp <4 x i8> undef to <4 x double>
@@ -61,6 +68,13 @@ define i32 @uitofp_i16_double() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v8i16_v8f64 = uitofp <8 x i16> undef to <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'uitofp_i16_double'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i16_f64 = uitofp i16 undef to double
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %cvt_v2i16_v2f64 = uitofp <2 x i16> undef to <2 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v4i16_v4f64 = uitofp <4 x i16> undef to <4 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %cvt_v8i16_v8f64 = uitofp <8 x i16> undef to <8 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i16_f64 = uitofp i16 undef to double
   %cvt_v2i16_v2f64 = uitofp <2 x i16> undef to <2 x double>
   %cvt_v4i16_v4f64 = uitofp <4 x i16> undef to <4 x double>
@@ -90,6 +104,13 @@ define i32 @uitofp_i32_double() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v8i32_v8f64 = uitofp <8 x i32> undef to <8 x double>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'uitofp_i32_double'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i32_f64 = uitofp i32 undef to double
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %cvt_v2i32_v2f64 = uitofp <2 x i32> undef to <2 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %cvt_v4i32_v4f64 = uitofp <4 x i32> undef to <4 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %cvt_v8i32_v8f64 = uitofp <8 x i32> undef to <8 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i32_f64 = uitofp i32 undef to double
   %cvt_v2i32_v2f64 = uitofp <2 x i32> undef to <2 x double>
   %cvt_v4i32_v4f64 = uitofp <4 x i32> undef to <4 x double>
@@ -126,6 +147,13 @@ define i32 @uitofp_i64_double() {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v8i64_v8f64 = uitofp <8 x i64> undef to <8 x double>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'uitofp_i64_double'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i64_f64 = uitofp i64 undef to double
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %cvt_v2i64_v2f64 = uitofp <2 x i64> undef to <2 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %cvt_v4i64_v4f64 = uitofp <4 x i64> undef to <4 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %cvt_v8i64_v8f64 = uitofp <8 x i64> undef to <8 x double>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i64_f64 = uitofp i64 undef to double
   %cvt_v2i64_v2f64 = uitofp <2 x i64> undef to <2 x double>
   %cvt_v4i64_v4f64 = uitofp <4 x i64> undef to <4 x double>
@@ -155,6 +183,13 @@ define i32 @uitofp_i8_float() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v16i8_v16f32 = uitofp <16 x i8> undef to <16 x float>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'uitofp_i8_float'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i8_f32 = uitofp i8 undef to float
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v4i8_v4f32 = uitofp <4 x i8> undef to <4 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %cvt_v8i8_v8f32 = uitofp <8 x i8> undef to <8 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %cvt_v16i8_v16f32 = uitofp <16 x i8> undef to <16 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i8_f32 = uitofp i8 undef to float
   %cvt_v4i8_v4f32 = uitofp <4 x i8> undef to <4 x float>
   %cvt_v8i8_v8f32 = uitofp <8 x i8> undef to <8 x float>
@@ -184,6 +219,13 @@ define i32 @uitofp_i16_float() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v16i16_v16f32 = uitofp <16 x i16> undef to <16 x float>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'uitofp_i16_float'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i16_f32 = uitofp i16 undef to float
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %cvt_v4i16_v4f32 = uitofp <4 x i16> undef to <4 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %cvt_v8i16_v8f32 = uitofp <8 x i16> undef to <8 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %cvt_v16i16_v16f32 = uitofp <16 x i16> undef to <16 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i16_f32 = uitofp i16 undef to float
   %cvt_v4i16_v4f32 = uitofp <4 x i16> undef to <4 x float>
   %cvt_v8i16_v8f32 = uitofp <8 x i16> undef to <8 x float>
@@ -220,6 +262,13 @@ define i32 @uitofp_i32_float() {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_v16i32_v16f32 = uitofp <16 x i32> undef to <16 x float>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'uitofp_i32_float'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i32_f32 = uitofp i32 undef to float
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %cvt_v4i32_v4f32 = uitofp <4 x i32> undef to <4 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %cvt_v8i32_v8f32 = uitofp <8 x i32> undef to <8 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %cvt_v16i32_v16f32 = uitofp <16 x i32> undef to <16 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i32_f32 = uitofp i32 undef to float
   %cvt_v4i32_v4f32 = uitofp <4 x i32> undef to <4 x float>
   %cvt_v8i32_v8f32 = uitofp <8 x i32> undef to <8 x float>
@@ -260,6 +309,14 @@ define i32 @uitofp_i64_float() {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %cvt_v16i64_v16f32 = uitofp <16 x i64> undef to <16 x float>
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
+; BTVER2-LABEL: 'uitofp_i64_float'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cvt_i64_f32 = uitofp i64 undef to float
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %cvt_v2i64_v2f32 = uitofp <2 x i64> undef to <2 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %cvt_v4i64_v4f32 = uitofp <4 x i64> undef to <4 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %cvt_v8i64_v8f32 = uitofp <8 x i64> undef to <8 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %cvt_v16i64_v16f32 = uitofp <16 x i64> undef to <16 x float>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
   %cvt_i64_f32 = uitofp i64 undef to float
   %cvt_v2i64_v2f32 = uitofp <2 x i64> undef to <2 x float>
   %cvt_v4i64_v4f32 = uitofp <4 x i64> undef to <4 x float>
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
index 2c328ba7aa4..1d1aafabe92 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
@@ -14,7 +14,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 ; Verify the cost of vector arithmetic shift right instructions.
 
@@ -43,6 +43,10 @@ define <2 x i64> @var_shift_v2i64(<2 x i64> %a, <2 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <2 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <2 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = ashr <2 x i64> %a, %b
   ret <2 x i64> %shift
 }
@@ -68,6 +72,10 @@ define <4 x i64> @var_shift_v4i64(<4 x i64> %a, <4 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = ashr <4 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = ashr <4 x i64> %a, %b
   ret <4 x i64> %shift
 }
@@ -93,6 +101,10 @@ define <8 x i64> @var_shift_v8i64(<8 x i64> %a, <8 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = ashr <8 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = ashr <8 x i64> %a, %b
   ret <8 x i64> %shift
 }
@@ -126,9 +138,10 @@ define <4 x i32> @var_shift_v4i32(<4 x i32> %a, <4 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
-; XOPAVX-LABEL: 'var_shift_v4i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, %b
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+; BTVER2-LABEL: 'var_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <4 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = ashr <4 x i32> %a, %b
   ret <4 x i32> %shift
 }
@@ -162,9 +175,10 @@ define <8 x i32> @var_shift_v8i32(<8 x i32> %a, <8 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
-; XOPAVX-LABEL: 'var_shift_v8i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <8 x i32> %a, %b
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; BTVER2-LABEL: 'var_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = ashr <8 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = ashr <8 x i32> %a, %b
   ret <8 x i32> %shift
 }
@@ -198,9 +212,10 @@ define <16 x i32> @var_shift_v16i32(<16 x i32> %a, <16 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <16 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
-; XOPAVX-LABEL: 'var_shift_v16i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <16 x i32> %a, %b
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; BTVER2-LABEL: 'var_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = ashr <16 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = ashr <16 x i32> %a, %b
   ret <16 x i32> %shift
 }
@@ -238,6 +253,10 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %shift = ashr <8 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = ashr <8 x i16> %a, %b
   ret <8 x i16> %shift
 }
@@ -279,6 +298,10 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <16 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %shift = ashr <16 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = ashr <16 x i16> %a, %b
   ret <16 x i16> %shift
 }
@@ -320,6 +343,10 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <32 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %shift = ashr <32 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = ashr <32 x i16> %a, %b
   ret <32 x i16> %shift
 }
@@ -345,6 +372,10 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = ashr <16 x i8> %a, %b
   ret <16 x i8> %shift
 }
@@ -374,6 +405,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <32 x i8> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %shift = ashr <32 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = ashr <32 x i8> %a, %b
   ret <32 x i8> %shift
 }
@@ -415,6 +450,10 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <64 x i8> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %shift = ashr <64 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = ashr <64 x i8> %a, %b
   ret <64 x i8> %shift
 }
@@ -448,6 +487,12 @@ define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <2 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <2 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %insert = insertelement <2 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
   %shift = ashr <2 x i64> %a, %splat
@@ -491,11 +536,12 @@ define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
-; XOPAVX-LABEL: 'splatvar_shift_v4i64'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <4 x i64> %a, %splat
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+; BTVER2-LABEL: 'splatvar_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %insert = insertelement <4 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
   %shift = ashr <4 x i64> %a, %splat
@@ -539,11 +585,12 @@ define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
-; XOPAVX-LABEL: 'splatvar_shift_v8i64'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <8 x i64> %a, %splat
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+; BTVER2-LABEL: 'splatvar_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %insert = insertelement <8 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
   %shift = ashr <8 x i64> %a, %splat
@@ -557,6 +604,12 @@ define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %splat
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
   %shift = ashr <4 x i32> %a, %splat
@@ -600,11 +653,12 @@ define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
-; XOPAVX-LABEL: 'splatvar_shift_v8i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <8 x i32> %a, %splat
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; BTVER2-LABEL: 'splatvar_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %insert = insertelement <8 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
   %shift = ashr <8 x i32> %a, %splat
@@ -648,11 +702,12 @@ define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <16 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
-; XOPAVX-LABEL: 'splatvar_shift_v16i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <16 x i32> %a, %splat
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; BTVER2-LABEL: 'splatvar_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %insert = insertelement <16 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
   %shift = ashr <16 x i32> %a, %splat
@@ -690,6 +745,12 @@ define <8 x i16> @splatvar_shift_v8i16(<8 x i16> %a, i16 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i16> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i16> %insert, <8 x i16> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %insert = insertelement <8 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <8 x i16> %insert, <8 x i16> undef, <8 x i32> zeroinitializer
   %shift = ashr <8 x i16> %a, %splat
@@ -739,11 +800,12 @@ define <16 x i16> @splatvar_shift_v16i16(<16 x i16> %a, i16 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <16 x i16> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
-; XOPAVX-LABEL: 'splatvar_shift_v16i16'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i16> undef, i16 %b, i32 0
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <16 x i16> %a, %splat
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+; BTVER2-LABEL: 'splatvar_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <16 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %insert = insertelement <16 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer
   %shift = ashr <16 x i16> %a, %splat
@@ -811,11 +873,12 @@ define <32 x i16> @splatvar_shift_v32i16(<32 x i16> %a, i16 %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <32 x i16> %a, %splat
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
-; XOPAVX-LABEL: 'splatvar_shift_v32i16'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i16> undef, i16 %b, i32 0
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <32 x i16> %a, %splat
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+; BTVER2-LABEL: 'splatvar_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <32 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %insert = insertelement <32 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer
   %shift = ashr <32 x i16> %a, %splat
@@ -853,6 +916,12 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %insert = insertelement <16 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
   %shift = ashr <16 x i8> %a, %splat
@@ -902,11 +971,12 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <32 x i8> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
-; XOPAVX-LABEL: 'splatvar_shift_v32i8'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <32 x i8> %a, %splat
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+; BTVER2-LABEL: 'splatvar_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %shift = ashr <32 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %insert = insertelement <32 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
   %shift = ashr <32 x i8> %a, %splat
@@ -974,11 +1044,12 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <64 x i8> %a, %splat
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
-; XOPAVX-LABEL: 'splatvar_shift_v64i8'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <64 x i8> %a, %splat
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+; BTVER2-LABEL: 'splatvar_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %shift = ashr <64 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %insert = insertelement <64 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
   %shift = ashr <64 x i8> %a, %splat
@@ -1010,6 +1081,10 @@ define <2 x i64> @constant_shift_v2i64(<2 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <2 x i64> %a, <i64 1, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <2 x i64> %a, <i64 1, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = ashr <2 x i64> %a, <i64 1, i64 7>
   ret <2 x i64> %shift
 }
@@ -1035,6 +1110,10 @@ define <4 x i64> @constant_shift_v4i64(<4 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = ashr <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = ashr <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
   ret <4 x i64> %shift
 }
@@ -1060,6 +1139,10 @@ define <8 x i64> @constant_shift_v8i64(<8 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = ashr <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = ashr <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
   ret <8 x i64> %shift
 }
@@ -1093,9 +1176,10 @@ define <4 x i32> @constant_shift_v4i32(<4 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
-; XOPAVX-LABEL: 'constant_shift_v4i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = ashr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+; BTVER2-LABEL: 'constant_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = ashr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
   ret <4 x i32> %shift
 }
@@ -1129,9 +1213,10 @@ define <8 x i32> @constant_shift_v8i32(<8 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
-; XOPAVX-LABEL: 'constant_shift_v8i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; BTVER2-LABEL: 'constant_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = ashr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = ashr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   ret <8 x i32> %shift
 }
@@ -1165,9 +1250,10 @@ define <16 x i32> @constant_shift_v16i32(<16 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
-; XOPAVX-LABEL: 'constant_shift_v16i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; BTVER2-LABEL: 'constant_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = ashr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = ashr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   ret <16 x i32> %shift
 }
@@ -1205,6 +1291,10 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %shift = ashr <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = ashr <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <8 x i16> %shift
 }
@@ -1246,6 +1336,10 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %shift = ashr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = ashr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <16 x i16> %shift
 }
@@ -1287,6 +1381,10 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %shift = ashr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = ashr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <32 x i16> %shift
 }
@@ -1312,6 +1410,10 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = ashr <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <16 x i8> %shift
 }
@@ -1341,6 +1443,10 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %shift = ashr <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = ashr <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <32 x i8> %shift
 }
@@ -1382,6 +1488,10 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %shift = ashr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = ashr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <64 x i8> %shift
 }
@@ -1407,6 +1517,10 @@ define <2 x i64> @splatconstant_shift_v2i64(<2 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <2 x i64> %a, <i64 7, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <2 x i64> %a, <i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = ashr <2 x i64> %a, <i64 7, i64 7>
   ret <2 x i64> %shift
 }
@@ -1436,9 +1550,10 @@ define <4 x i64> @splatconstant_shift_v4i64(<4 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
-; XOPAVX-LABEL: 'splatconstant_shift_v4i64'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+; BTVER2-LABEL: 'splatconstant_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = ashr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
   ret <4 x i64> %shift
 }
@@ -1468,9 +1583,10 @@ define <8 x i64> @splatconstant_shift_v8i64(<8 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
-; XOPAVX-LABEL: 'splatconstant_shift_v8i64'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+; BTVER2-LABEL: 'splatconstant_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = ashr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
   ret <8 x i64> %shift
 }
@@ -1480,6 +1596,10 @@ define <4 x i32> @splatconstant_shift_v4i32(<4 x i32> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = ashr <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
   ret <4 x i32> %shift
 }
@@ -1509,9 +1629,10 @@ define <8 x i32> @splatconstant_shift_v8i32(<8 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
-; XOPAVX-LABEL: 'splatconstant_shift_v8i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+; BTVER2-LABEL: 'splatconstant_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = ashr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
   ret <8 x i32> %shift
 }
@@ -1541,9 +1662,10 @@ define <16 x i32> @splatconstant_shift_v16i32(<16 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
-; XOPAVX-LABEL: 'splatconstant_shift_v16i32'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+; BTVER2-LABEL: 'splatconstant_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = ashr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
   ret <16 x i32> %shift
 }
@@ -1553,6 +1675,10 @@ define <8 x i16> @splatconstant_shift_v8i16(<8 x i16> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = ashr <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <8 x i16> %shift
 }
@@ -1582,9 +1708,10 @@ define <16 x i16> @splatconstant_shift_v16i16(<16 x i16> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
-; XOPAVX-LABEL: 'splatconstant_shift_v16i16'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+; BTVER2-LABEL: 'splatconstant_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = ashr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <16 x i16> %shift
 }
@@ -1626,9 +1753,10 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = ashr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
-; XOPAVX-LABEL: 'splatconstant_shift_v32i16'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+; BTVER2-LABEL: 'splatconstant_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = ashr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = ashr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <32 x i16> %shift
 }
@@ -1650,6 +1778,10 @@ define <16 x i8> @splatconstant_shift_v16i8(<16 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = ashr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <16 x i8> %shift
 }
@@ -1679,9 +1811,10 @@ define <32 x i8> @splatconstant_shift_v32i8(<32 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
-; XOPAVX-LABEL: 'splatconstant_shift_v32i8'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = ashr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+; BTVER2-LABEL: 'splatconstant_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = ashr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = ashr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <32 x i8> %shift
 }
@@ -1723,9 +1856,10 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = ashr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
-; XOPAVX-LABEL: 'splatconstant_shift_v64i8'
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = ashr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
-; XOPAVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+; BTVER2-LABEL: 'splatconstant_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = ashr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = ashr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <64 x i8> %shift
 }
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
index 23a9c3ac36d..8ebe6f82165 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
@@ -14,7 +14,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 ; Verify the cost of vector logical shift right instructions.
 
@@ -47,6 +47,10 @@ define <2 x i64> @var_shift_v2i64(<2 x i64> %a, <2 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <2 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = lshr <2 x i64> %a, %b
   ret <2 x i64> %shift
 }
@@ -76,6 +80,10 @@ define <4 x i64> @var_shift_v4i64(<4 x i64> %a, <4 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = lshr <4 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = lshr <4 x i64> %a, %b
   ret <4 x i64> %shift
 }
@@ -105,6 +113,10 @@ define <8 x i64> @var_shift_v8i64(<8 x i64> %a, <8 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = lshr <8 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = lshr <8 x i64> %a, %b
   ret <8 x i64> %shift
 }
@@ -138,6 +150,10 @@ define <4 x i32> @var_shift_v4i32(<4 x i32> %a, <4 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <4 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = lshr <4 x i32> %a, %b
   ret <4 x i32> %shift
 }
@@ -171,6 +187,10 @@ define <8 x i32> @var_shift_v8i32(<8 x i32> %a, <8 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = lshr <8 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = lshr <8 x i32> %a, %b
   ret <8 x i32> %shift
 }
@@ -204,6 +224,10 @@ define <16 x i32> @var_shift_v16i32(<16 x i32> %a, <16 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <16 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %shift = lshr <16 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = lshr <16 x i32> %a, %b
   ret <16 x i32> %shift
 }
@@ -241,6 +265,10 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %shift = lshr <8 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = lshr <8 x i16> %a, %b
   ret <8 x i16> %shift
 }
@@ -282,6 +310,10 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <16 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %shift = lshr <16 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = lshr <16 x i16> %a, %b
   ret <16 x i16> %shift
 }
@@ -323,6 +355,10 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <32 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %shift = lshr <32 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = lshr <32 x i16> %a, %b
   ret <32 x i16> %shift
 }
@@ -348,6 +384,10 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = lshr <16 x i8> %a, %b
   ret <16 x i8> %shift
 }
@@ -377,6 +417,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <32 x i8> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = lshr <32 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = lshr <32 x i8> %a, %b
   ret <32 x i8> %shift
 }
@@ -418,6 +462,10 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <64 x i8> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = lshr <64 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = lshr <64 x i8> %a, %b
   ret <64 x i8> %shift
 }
@@ -433,6 +481,12 @@ define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %splat
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %insert = insertelement <2 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
   %shift = lshr <2 x i64> %a, %splat
@@ -476,6 +530,12 @@ define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %insert = insertelement <4 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
   %shift = lshr <4 x i64> %a, %splat
@@ -519,6 +579,12 @@ define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %insert = insertelement <8 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
   %shift = lshr <8 x i64> %a, %splat
@@ -532,6 +598,12 @@ define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %splat
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
   %shift = lshr <4 x i32> %a, %splat
@@ -575,6 +647,12 @@ define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %insert = insertelement <8 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
   %shift = lshr <8 x i32> %a, %splat
@@ -618,6 +696,12 @@ define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <16 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %insert = insertelement <16 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
   %shift = lshr <16 x i32> %a, %splat
@@ -655,6 +739,12 @@ define <8 x i16> @splatvar_shift_v8i16(<8 x i16> %a, i16 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i16> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i16> %insert, <8 x i16> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %insert = insertelement <8 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <8 x i16> %insert, <8 x i16> undef, <8 x i32> zeroinitializer
   %shift = lshr <8 x i16> %a, %splat
@@ -704,6 +794,12 @@ define <16 x i16> @splatvar_shift_v16i16(<16 x i16> %a, i16 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <16 x i16> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <16 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %insert = insertelement <16 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer
   %shift = lshr <16 x i16> %a, %splat
@@ -771,6 +867,12 @@ define <32 x i16> @splatvar_shift_v32i16(<32 x i16> %a, i16 %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <32 x i16> %a, %splat
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <32 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %insert = insertelement <32 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer
   %shift = lshr <32 x i16> %a, %splat
@@ -808,6 +910,12 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %insert = insertelement <16 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
   %shift = lshr <16 x i8> %a, %splat
@@ -857,6 +965,12 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <32 x i8> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = lshr <32 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %insert = insertelement <32 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
   %shift = lshr <32 x i8> %a, %splat
@@ -924,6 +1038,12 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <64 x i8> %a, %splat
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = lshr <64 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %insert = insertelement <64 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
   %shift = lshr <64 x i8> %a, %splat
@@ -959,6 +1079,10 @@ define <2 x i64> @constant_shift_v2i64(<2 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, <i64 1, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <2 x i64> %a, <i64 1, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = lshr <2 x i64> %a, <i64 1, i64 7>
   ret <2 x i64> %shift
 }
@@ -988,6 +1112,10 @@ define <4 x i64> @constant_shift_v4i64(<4 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = lshr <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = lshr <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
   ret <4 x i64> %shift
 }
@@ -1017,6 +1145,10 @@ define <8 x i64> @constant_shift_v8i64(<8 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = lshr <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = lshr <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
   ret <8 x i64> %shift
 }
@@ -1050,6 +1182,10 @@ define <4 x i32> @constant_shift_v4i32(<4 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = lshr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
   ret <4 x i32> %shift
 }
@@ -1083,6 +1219,10 @@ define <8 x i32> @constant_shift_v8i32(<8 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = lshr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = lshr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   ret <8 x i32> %shift
 }
@@ -1116,6 +1256,10 @@ define <16 x i32> @constant_shift_v16i32(<16 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %shift = lshr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = lshr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   ret <16 x i32> %shift
 }
@@ -1153,6 +1297,10 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %shift = lshr <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = lshr <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <8 x i16> %shift
 }
@@ -1194,6 +1342,10 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %shift = lshr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = lshr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <16 x i16> %shift
 }
@@ -1235,6 +1387,10 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %shift = lshr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = lshr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <32 x i16> %shift
 }
@@ -1260,6 +1416,10 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = lshr <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <16 x i8> %shift
 }
@@ -1289,6 +1449,10 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %shift = lshr <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = lshr <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <32 x i8> %shift
 }
@@ -1330,6 +1494,10 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %shift = lshr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = lshr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <64 x i8> %shift
 }
@@ -1343,6 +1511,10 @@ define <2 x i64> @splatconstant_shift_v2i64(<2 x i64> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, <i64 7, i64 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <2 x i64> %a, <i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = lshr <2 x i64> %a, <i64 7, i64 7>
   ret <2 x i64> %shift
 }
@@ -1372,6 +1544,10 @@ define <4 x i64> @splatconstant_shift_v4i64(<4 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = lshr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
   ret <4 x i64> %shift
 }
@@ -1401,6 +1577,10 @@ define <8 x i64> @splatconstant_shift_v8i64(<8 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = lshr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
   ret <8 x i64> %shift
 }
@@ -1410,6 +1590,10 @@ define <4 x i32> @splatconstant_shift_v4i32(<4 x i32> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = lshr <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
   ret <4 x i32> %shift
 }
@@ -1439,6 +1623,10 @@ define <8 x i32> @splatconstant_shift_v8i32(<8 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = lshr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
   ret <8 x i32> %shift
 }
@@ -1468,6 +1656,10 @@ define <16 x i32> @splatconstant_shift_v16i32(<16 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = lshr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
   ret <16 x i32> %shift
 }
@@ -1477,6 +1669,10 @@ define <8 x i16> @splatconstant_shift_v8i16(<8 x i16> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = lshr <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <8 x i16> %shift
 }
@@ -1506,6 +1702,10 @@ define <16 x i16> @splatconstant_shift_v16i16(<16 x i16> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = lshr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = lshr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <16 x i16> %shift
 }
@@ -1547,6 +1747,10 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = lshr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = lshr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = lshr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <32 x i16> %shift
 }
@@ -1556,6 +1760,10 @@ define <16 x i8> @splatconstant_shift_v16i8(<16 x i8> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = lshr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <16 x i8> %shift
 }
@@ -1585,6 +1793,10 @@ define <32 x i8> @splatconstant_shift_v32i8(<32 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = lshr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = lshr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <32 x i8> %shift
 }
@@ -1626,6 +1838,10 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = lshr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = lshr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <64 x i8> %shift
 }
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
index 4be35794c06..c24037884b0 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
@@ -14,7 +14,7 @@
 ;
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
-; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -cost-model -analyze -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
 
 ; Verify the cost of vector shift left instructions.
 
@@ -43,6 +43,10 @@ define <2 x i64> @var_shift_v2i64(<2 x i64> %a, <2 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <2 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = shl <2 x i64> %a, %b
   ret <2 x i64> %shift
 }
@@ -72,6 +76,10 @@ define <4 x i64> @var_shift_v4i64(<4 x i64> %a, <4 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = shl <4 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = shl <4 x i64> %a, %b
   ret <4 x i64> %shift
 }
@@ -101,6 +109,10 @@ define <8 x i64> @var_shift_v8i64(<8 x i64> %a, <8 x i64> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i64> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = shl <8 x i64> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = shl <8 x i64> %a, %b
   ret <8 x i64> %shift
 }
@@ -130,6 +142,10 @@ define <4 x i32> @var_shift_v4i32(<4 x i32> %a, <4 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = shl <4 x i32> %a, %b
   ret <4 x i32> %shift
 }
@@ -163,6 +179,10 @@ define <8 x i32> @var_shift_v8i32(<8 x i32> %a, <8 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = shl <8 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = shl <8 x i32> %a, %b
   ret <8 x i32> %shift
 }
@@ -196,6 +216,10 @@ define <16 x i32> @var_shift_v16i32(<16 x i32> %a, <16 x i32> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <16 x i32> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = shl <16 x i32> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = shl <16 x i32> %a, %b
   ret <16 x i32> %shift
 }
@@ -233,6 +257,10 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %shift = shl <8 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = shl <8 x i16> %a, %b
   ret <8 x i16> %shift
 }
@@ -274,6 +302,10 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <16 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %shift = shl <16 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = shl <16 x i16> %a, %b
   ret <16 x i16> %shift
 }
@@ -315,6 +347,10 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <32 x i16> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %shift = shl <32 x i16> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = shl <32 x i16> %a, %b
   ret <32 x i16> %shift
 }
@@ -340,6 +376,10 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = shl <16 x i8> %a, %b
   ret <16 x i8> %shift
 }
@@ -369,6 +409,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <32 x i8> %a, %b
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = shl <32 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = shl <32 x i8> %a, %b
   ret <32 x i8> %shift
 }
@@ -410,6 +454,10 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <64 x i8> %a, %b
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'var_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %shift = shl <64 x i8> %a, %b
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = shl <64 x i8> %a, %b
   ret <64 x i8> %shift
 }
@@ -425,6 +473,12 @@ define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, i64 %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <2 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %insert = insertelement <2 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <2 x i64> %insert, <2 x i64> undef, <2 x i32> zeroinitializer
   %shift = shl <2 x i64> %a, %splat
@@ -468,6 +522,12 @@ define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %insert = insertelement <4 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer
   %shift = shl <4 x i64> %a, %splat
@@ -511,6 +571,12 @@ define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i64> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i64> undef, i64 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %insert = insertelement <8 x i64> undef, i64 %b, i32 0
   %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer
   %shift = shl <8 x i64> %a, %splat
@@ -524,6 +590,12 @@ define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, i32 %b) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <4 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %insert = insertelement <4 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <4 x i32> %insert, <4 x i32> undef, <4 x i32> zeroinitializer
   %shift = shl <4 x i32> %a, %splat
@@ -567,6 +639,12 @@ define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %insert = insertelement <8 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer
   %shift = shl <8 x i32> %a, %splat
@@ -610,6 +688,12 @@ define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <16 x i32> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i32> undef, i32 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %insert = insertelement <16 x i32> undef, i32 %b, i32 0
   %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer
   %shift = shl <16 x i32> %a, %splat
@@ -647,6 +731,12 @@ define <8 x i16> @splatvar_shift_v8i16(<8 x i16> %a, i16 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i16> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <8 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <8 x i16> %insert, <8 x i16> undef, <8 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %insert = insertelement <8 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <8 x i16> %insert, <8 x i16> undef, <8 x i32> zeroinitializer
   %shift = shl <8 x i16> %a, %splat
@@ -696,6 +786,12 @@ define <16 x i16> @splatvar_shift_v16i16(<16 x i16> %a, i16 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <16 x i16> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <16 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %insert = insertelement <16 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer
   %shift = shl <16 x i16> %a, %splat
@@ -763,6 +859,12 @@ define <32 x i16> @splatvar_shift_v32i16(<32 x i16> %a, i16 %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <32 x i16> %a, %splat
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i16> undef, i16 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <32 x i16> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %insert = insertelement <32 x i16> undef, i16 %b, i32 0
   %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer
   %shift = shl <32 x i16> %a, %splat
@@ -800,6 +902,12 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %insert = insertelement <16 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
   %shift = shl <16 x i8> %a, %splat
@@ -849,6 +957,12 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <32 x i8> %a, %splat
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = shl <32 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %insert = insertelement <32 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
   %shift = shl <32 x i8> %a, %splat
@@ -916,6 +1030,12 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <64 x i8> %a, %splat
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatvar_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %shift = shl <64 x i8> %a, %splat
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %insert = insertelement <64 x i8> undef, i8 %b, i32 0
   %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
   %shift = shl <64 x i8> %a, %splat
@@ -947,6 +1067,10 @@ define <2 x i64> @constant_shift_v2i64(<2 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, <i64 1, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <2 x i64> %a, <i64 1, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = shl <2 x i64> %a, <i64 1, i64 7>
   ret <2 x i64> %shift
 }
@@ -976,6 +1100,10 @@ define <4 x i64> @constant_shift_v4i64(<4 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shift = shl <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = shl <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31>
   ret <4 x i64> %shift
 }
@@ -1005,6 +1133,10 @@ define <8 x i64> @constant_shift_v8i64(<8 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shift = shl <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = shl <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>
   ret <8 x i64> %shift
 }
@@ -1034,6 +1166,10 @@ define <4 x i32> @constant_shift_v4i32(<4 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = shl <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
   ret <4 x i32> %shift
 }
@@ -1067,6 +1203,10 @@ define <8 x i32> @constant_shift_v8i32(<8 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = shl <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   ret <8 x i32> %shift
 }
@@ -1100,6 +1240,10 @@ define <16 x i32> @constant_shift_v16i32(<16 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = shl <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   ret <16 x i32> %shift
 }
@@ -1109,6 +1253,10 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = shl <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <8 x i16> %shift
 }
@@ -1138,6 +1286,10 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = shl <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <16 x i16> %shift
 }
@@ -1179,6 +1331,10 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = shl <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
   ret <32 x i16> %shift
 }
@@ -1204,6 +1360,10 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = shl <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <16 x i8> %shift
 }
@@ -1233,6 +1393,10 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %shift = shl <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = shl <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <32 x i8> %shift
 }
@@ -1274,6 +1438,10 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'constant_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %shift = shl <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = shl <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <64 x i8> %shift
 }
@@ -1287,6 +1455,10 @@ define <2 x i64> @splatconstant_shift_v2i64(<2 x i64> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, <i64 7, i64 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v2i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <2 x i64> %a, <i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shift
+;
   %shift = shl <2 x i64> %a, <i64 7, i64 7>
   ret <2 x i64> %shift
 }
@@ -1316,6 +1488,10 @@ define <4 x i64> @splatconstant_shift_v4i64(<4 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v4i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shift
+;
   %shift = shl <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7>
   ret <4 x i64> %shift
 }
@@ -1345,6 +1521,10 @@ define <8 x i64> @splatconstant_shift_v8i64(<8 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v8i64'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shift
+;
   %shift = shl <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
   ret <8 x i64> %shift
 }
@@ -1354,6 +1534,10 @@ define <4 x i32> @splatconstant_shift_v4i32(<4 x i32> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v4i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
+;
   %shift = shl <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
   ret <4 x i32> %shift
 }
@@ -1383,6 +1567,10 @@ define <8 x i32> @splatconstant_shift_v8i32(<8 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v8i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
+;
   %shift = shl <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
   ret <8 x i32> %shift
 }
@@ -1412,6 +1600,10 @@ define <16 x i32> @splatconstant_shift_v16i32(<16 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v16i32'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
+;
   %shift = shl <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
   ret <16 x i32> %shift
 }
@@ -1421,6 +1613,10 @@ define <8 x i16> @splatconstant_shift_v8i16(<8 x i16> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v8i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
+;
   %shift = shl <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <8 x i16> %shift
 }
@@ -1450,6 +1646,10 @@ define <16 x i16> @splatconstant_shift_v16i16(<16 x i16> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v16i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shift = shl <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
+;
   %shift = shl <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <16 x i16> %shift
 }
@@ -1491,6 +1691,10 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shift = shl <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v32i16'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shift = shl <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
+;
   %shift = shl <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
   ret <32 x i16> %shift
 }
@@ -1512,6 +1716,10 @@ define <16 x i8> @splatconstant_shift_v16i8(<16 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v16i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
+;
   %shift = shl <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <16 x i8> %shift
 }
@@ -1541,6 +1749,10 @@ define <32 x i8> @splatconstant_shift_v32i8(<32 x i8> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v32i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %shift = shl <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
+;
   %shift = shl <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <32 x i8> %shift
 }
@@ -1582,6 +1794,10 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shift = shl <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
 ;
+; BTVER2-LABEL: 'splatconstant_shift_v64i8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %shift = shl <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
+;
   %shift = shl <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
   ret <64 x i8> %shift
 }
@@ -1598,6 +1814,10 @@ define <8 x i16> @test1(<8 x i16> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <8 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shl
 ;
+; BTVER2-LABEL: 'test1'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <8 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shl
+;
   %shl = shl <8 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
   ret <8 x i16> %shl
 }
@@ -1607,6 +1827,10 @@ define <8 x i16> @test2(<8 x i16> %a) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <8 x i16> %a, <i16 0, i16 undef, i16 0, i16 0, i16 1, i16 undef, i16 -1, i16 1>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shl
 ;
+; BTVER2-LABEL: 'test2'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <8 x i16> %a, <i16 0, i16 undef, i16 0, i16 0, i16 1, i16 undef, i16 -1, i16 1>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shl
+;
   %shl = shl <8 x i16> %a, <i16 0, i16 undef, i16 0, i16 0, i16 1, i16 undef, i16 -1, i16 1>
   ret <8 x i16> %shl
 }
@@ -1641,6 +1865,10 @@ define <4 x i32> @test3(<4 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <4 x i32> %a, <i32 1, i32 -1, i32 2, i32 -3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shl
 ;
+; BTVER2-LABEL: 'test3'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shl = shl <4 x i32> %a, <i32 1, i32 -1, i32 2, i32 -3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shl
+;
   %shl = shl <4 x i32> %a, <i32 1, i32 -1, i32 2, i32 -3>
   ret <4 x i32> %shl
 }
@@ -1670,6 +1898,10 @@ define <4 x i32> @test4(<4 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <4 x i32> %a, <i32 0, i32 0, i32 1, i32 1>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shl
 ;
+; BTVER2-LABEL: 'test4'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %shl = shl <4 x i32> %a, <i32 0, i32 0, i32 1, i32 1>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shl
+;
   %shl = shl <4 x i32> %a, <i32 0, i32 0, i32 1, i32 1>
   ret <4 x i32> %shl
 }
@@ -1700,6 +1932,10 @@ define <2 x i64> @test5(<2 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <2 x i64> %a, <i64 2, i64 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shl
 ;
+; BTVER2-LABEL: 'test5'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shl = shl <2 x i64> %a, <i64 2, i64 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %shl
+;
   %shl = shl <2 x i64> %a, <i64 2, i64 3>
   ret <2 x i64> %shl
 }
@@ -1741,6 +1977,10 @@ define <16 x i16> @test6(<16 x i16> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <16 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shl
 ;
+; BTVER2-LABEL: 'test6'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shl = shl <16 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shl
+;
   %shl = shl <16 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
   ret <16 x i16> %shl
 }
@@ -1778,6 +2018,10 @@ define <8 x i32> @test7(<8 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <8 x i32> %a, <i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shl
 ;
+; BTVER2-LABEL: 'test7'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %shl = shl <8 x i32> %a, <i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shl
+;
   %shl = shl <8 x i32> %a, <i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3>
   ret <8 x i32> %shl
 }
@@ -1812,6 +2056,10 @@ define <4 x i64> @test8(<4 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <4 x i64> %a, <i64 1, i64 2, i64 3, i64 4>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shl
 ;
+; BTVER2-LABEL: 'test8'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %shl = shl <4 x i64> %a, <i64 1, i64 2, i64 3, i64 4>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %shl
+;
   %shl = shl <4 x i64> %a, <i64 1, i64 2, i64 3, i64 4>
   ret <4 x i64> %shl
 }
@@ -1855,6 +2103,10 @@ define <32 x i16> @test9(<32 x i16> %a) {
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <32 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
 ; AVX512BWVL-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shl
 ;
+; BTVER2-LABEL: 'test9'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shl = shl <32 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shl
+;
   %shl = shl <32 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
   ret <32 x i16> %shl
 }
@@ -1890,6 +2142,10 @@ define <16 x i32> @test10(<16 x i32> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <16 x i32> %a, <i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shl
 ;
+; BTVER2-LABEL: 'test10'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %shl = shl <16 x i32> %a, <i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shl
+;
   %shl = shl <16 x i32> %a, <i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3>
   ret <16 x i32> %shl
 }
@@ -1924,6 +2180,10 @@ define <8 x i64> @test11(<8 x i64> %a) {
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %shl = shl <8 x i64> %a, <i64 1, i64 1, i64 2, i64 3, i64 1, i64 1, i64 2, i64 3>
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shl
 ;
+; BTVER2-LABEL: 'test11'
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %shl = shl <8 x i64> %a, <i64 1, i64 1, i64 2, i64 3, i64 1, i64 1, i64 2, i64 3>
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i64> %shl
+;
   %shl = shl <8 x i64> %a, <i64 1, i64 1, i64 2, i64 3, i64 1, i64 1, i64 2, i64 3>
   ret <8 x i64> %shl
 }