summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--llvm/lib/Target/X86/X86InstrInfo.cpp24
-rw-r--r--llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll52
-rw-r--r--llvm/test/CodeGen/X86/stack-folding-fp-avx512vl.ll21
-rw-r--r--llvm/test/CodeGen/X86/stack-folding-int-avx512.ll52
-rw-r--r--llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll21
5 files changed, 170 insertions, 0 deletions
diff --git a/llvm/lib/Target/X86/X86InstrInfo.cpp b/llvm/lib/Target/X86/X86InstrInfo.cpp
index 88b608d3e45..4f6b0b5f26b 100644
--- a/llvm/lib/Target/X86/X86InstrInfo.cpp
+++ b/llvm/lib/Target/X86/X86InstrInfo.cpp
@@ -2104,6 +2104,10 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::VPUNPCKLWDZrr, X86::VPUNPCKLWDZrm, 0 },
{ X86::VPXORDZrr, X86::VPXORDZrm, 0 },
{ X86::VPXORQZrr, X86::VPXORQZrm, 0 },
+ { X86::VSHUFF32X4Zrri, X86::VSHUFF32X4Zrmi, 0 },
+ { X86::VSHUFF64X2Zrri, X86::VSHUFF64X2Zrmi, 0 },
+ { X86::VSHUFI64X2Zrri, X86::VSHUFI64X2Zrmi, 0 },
+ { X86::VSHUFI32X4Zrri, X86::VSHUFI32X4Zrmi, 0 },
{ X86::VSHUFPDZrri, X86::VSHUFPDZrmi, 0 },
{ X86::VSHUFPSZrri, X86::VSHUFPSZrmi, 0 },
{ X86::VSUBPDZrr, X86::VSUBPDZrm, 0 },
@@ -2380,6 +2384,10 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::VPXORDZ256rr, X86::VPXORDZ256rm, 0 },
{ X86::VPXORQZ128rr, X86::VPXORQZ128rm, 0 },
{ X86::VPXORQZ256rr, X86::VPXORQZ256rm, 0 },
+ { X86::VSHUFF32X4Z256rri, X86::VSHUFF32X4Z256rmi, 0 },
+ { X86::VSHUFF64X2Z256rri, X86::VSHUFF64X2Z256rmi, 0 },
+ { X86::VSHUFI32X4Z256rri, X86::VSHUFI32X4Z256rmi, 0 },
+ { X86::VSHUFI64X2Z256rri, X86::VSHUFI64X2Z256rmi, 0 },
{ X86::VSHUFPDZ128rri, X86::VSHUFPDZ128rmi, 0 },
{ X86::VSHUFPDZ256rri, X86::VSHUFPDZ256rmi, 0 },
{ X86::VSHUFPSZ128rri, X86::VSHUFPSZ128rmi, 0 },
@@ -2793,6 +2801,10 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::VPUNPCKLWDZrrkz, X86::VPUNPCKLWDZrmkz, 0 },
{ X86::VPXORDZrrkz, X86::VPXORDZrmkz, 0 },
{ X86::VPXORQZrrkz, X86::VPXORQZrmkz, 0 },
+ { X86::VSHUFF32X4Zrrikz, X86::VSHUFF32X4Zrmikz, 0 },
+ { X86::VSHUFF64X2Zrrikz, X86::VSHUFF64X2Zrmikz, 0 },
+ { X86::VSHUFI32X4Zrrikz, X86::VSHUFI32X4Zrmikz, 0 },
+ { X86::VSHUFI64X2Zrrikz, X86::VSHUFI64X2Zrmikz, 0 },
{ X86::VSHUFPDZrrikz, X86::VSHUFPDZrmikz, 0 },
{ X86::VSHUFPSZrrikz, X86::VSHUFPSZrmikz, 0 },
{ X86::VSUBPDZrrkz, X86::VSUBPDZrmkz, 0 },
@@ -2922,6 +2934,10 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::VPUNPCKLWDZ256rrkz, X86::VPUNPCKLWDZ256rmkz, 0 },
{ X86::VPXORDZ256rrkz, X86::VPXORDZ256rmkz, 0 },
{ X86::VPXORQZ256rrkz, X86::VPXORQZ256rmkz, 0 },
+ { X86::VSHUFF32X4Z256rrikz, X86::VSHUFF32X4Z256rmikz, 0 },
+ { X86::VSHUFF64X2Z256rrikz, X86::VSHUFF64X2Z256rmikz, 0 },
+ { X86::VSHUFI32X4Z256rrikz, X86::VSHUFI32X4Z256rmikz, 0 },
+ { X86::VSHUFI64X2Z256rrikz, X86::VSHUFI64X2Z256rmikz, 0 },
{ X86::VSHUFPDZ256rrikz, X86::VSHUFPDZ256rmikz, 0 },
{ X86::VSHUFPSZ256rrikz, X86::VSHUFPSZ256rmikz, 0 },
{ X86::VSUBPDZ256rrkz, X86::VSUBPDZ256rmkz, 0 },
@@ -3410,6 +3426,10 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::VPUNPCKLWDZrrk, X86::VPUNPCKLWDZrmk, 0 },
{ X86::VPXORDZrrk, X86::VPXORDZrmk, 0 },
{ X86::VPXORQZrrk, X86::VPXORQZrmk, 0 },
+ { X86::VSHUFF32X4Zrrik, X86::VSHUFF32X4Zrmik, 0 },
+ { X86::VSHUFF64X2Zrrik, X86::VSHUFF64X2Zrmik, 0 },
+ { X86::VSHUFI32X4Zrrik, X86::VSHUFI32X4Zrmik, 0 },
+ { X86::VSHUFI64X2Zrrik, X86::VSHUFI64X2Zrmik, 0 },
{ X86::VSHUFPDZrrik, X86::VSHUFPDZrmik, 0 },
{ X86::VSHUFPSZrrik, X86::VSHUFPSZrmik, 0 },
{ X86::VSUBPDZrrk, X86::VSUBPDZrmk, 0 },
@@ -3555,6 +3575,10 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::VPUNPCKLWDZ256rrk, X86::VPUNPCKLWDZ256rmk, 0 },
{ X86::VPXORDZ256rrk, X86::VPXORDZ256rmk, 0 },
{ X86::VPXORQZ256rrk, X86::VPXORQZ256rmk, 0 },
+ { X86::VSHUFF32X4Z256rrik, X86::VSHUFF32X4Z256rmik, 0 },
+ { X86::VSHUFF64X2Z256rrik, X86::VSHUFF64X2Z256rmik, 0 },
+ { X86::VSHUFI32X4Z256rrik, X86::VSHUFI32X4Z256rmik, 0 },
+ { X86::VSHUFI64X2Z256rrik, X86::VSHUFI64X2Z256rmik, 0 },
{ X86::VSHUFPDZ256rrik, X86::VSHUFPDZ256rmik, 0 },
{ X86::VSHUFPSZ256rrik, X86::VSHUFPSZ256rmik, 0 },
{ X86::VSUBPDZ256rrk, X86::VSUBPDZ256rmk, 0 },
diff --git a/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll b/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll
index 9057f262b31..bb09d04f7a5 100644
--- a/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll
@@ -400,6 +400,58 @@ define <16 x float> @stack_fold_orps_zmm(<16 x float> %a0, <16 x float> %a1) #0
ret <16 x float> %6
}
+define <8 x double> @stack_fold_shuff64x2(<8 x double> %a, <8 x double> %b) {
+ ;CHECK-LABEL: stack_fold_shuff64x2
+ ;CHECK: vshuff64x2 $24, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <8 x double> %a, <8 x double> %b, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9>
+ ret <8 x double> %2
+}
+
+define <8 x double> @stack_fold_shuff64x2_mask(<8 x double> %a, <8 x double> %b, i8 %mask, <8 x double>* %passthru) {
+ ;CHECK-LABEL: stack_fold_shuff64x2_mask
+ ;CHECK: vshuff64x2 $24, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{{%k[1-7]}}} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <8 x double> %a, <8 x double> %b, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9>
+ %3 = bitcast i8 %mask to <8 x i1>
+ ; load needed to keep the operation from being scheduled above the asm block
+ %4 = load <8 x double>, <8 x double>* %passthru
+ %5 = select <8 x i1> %3, <8 x double> %2, <8 x double> %4
+ ret <8 x double> %5
+}
+
+define <8 x double> @stack_fold_shuff64x2_maskz(<8 x double> %a, <8 x double> %b, i8 %mask, <8 x double>* %passthru) {
+ ;CHECK-LABEL: stack_fold_shuff64x2_maskz
+ ;CHECK: vshuff64x2 $24, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{{%k[1-7]}}} {z} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <8 x double> %a, <8 x double> %b, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9>
+ %3 = bitcast i8 %mask to <8 x i1>
+ %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> zeroinitializer
+ ret <8 x double> %4
+}
+
+define <16 x float> @stack_fold_shuff32x4_mask(<16 x float> %a, <16 x float> %b, i16 %mask, <16 x float>* %passthru) {
+ ;CHECK-LABEL: stack_fold_shuff32x4_mask
+ ;CHECK: vshuff32x4 $20, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{{%k[1-7]}}} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 16, i32 17, i32 18, i32 19>
+ %3 = bitcast i16 %mask to <16 x i1>
+ ; load needed to keep the operation from being scheduled above the asm block
+ %4 = load <16 x float>, <16 x float>* %passthru
+ %5 = select <16 x i1> %3, <16 x float> %2, <16 x float> %4
+ ret <16 x float> %5
+}
+
+define <16 x float> @stack_fold_shuff32x4_maskz(<16 x float> %a, <16 x float> %b, i16 %mask) {
+ ;CHECK-LABEL: stack_fold_shuff32x4_maskz
+ ;CHECK: vshuff32x4 $20, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{{%k[1-7]}}} {z} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 16, i32 17, i32 18, i32 19>
+ %3 = bitcast i16 %mask to <16 x i1>
+ %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> zeroinitializer
+ ret <16 x float> %4
+}
+
define <8 x double> @stack_fold_subpd_zmm(<8 x double> %a0, <8 x double> %a1) {
;CHECK-LABEL: stack_fold_subpd_zmm
;CHECK: vsubpd {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{.*#+}} 64-byte Folded Reload
diff --git a/llvm/test/CodeGen/X86/stack-folding-fp-avx512vl.ll b/llvm/test/CodeGen/X86/stack-folding-fp-avx512vl.ll
index 717e942fff1..2607c4f0c6e 100644
--- a/llvm/test/CodeGen/X86/stack-folding-fp-avx512vl.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-fp-avx512vl.ll
@@ -452,6 +452,27 @@ define <8 x float> @stack_fold_orps_ymm(<8 x float> %a0, <8 x float> %a1) {
ret <8 x float> %6
}
+define <4 x double> @stack_fold_shuff64x2_maskz(<4 x double> %a, <4 x double> %b, i8 %mask) {
+ ;CHECK-LABEL: stack_fold_shuff64x2_maskz
+ ;CHECK: vshuff64x2 $1, {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{{%k[1-7]}}} {z} {{.*#+}} 32-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+ %3 = bitcast i8 %mask to <8 x i1>
+ %4 = shufflevector <8 x i1> %3, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+ %5 = select <4 x i1> %4, <4 x double> %2, <4 x double> zeroinitializer
+ ret <4 x double> %5
+}
+
+define <8 x float> @stack_fold_shuff32x4_maskz(<8 x float> %a, <8 x float> %b, i8 %mask) {
+ ;CHECK-LABEL: stack_fold_shuff32x4_maskz
+ ;CHECK: vshuff32x4 $1, {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{{%k[1-7]}}} {z} {{.*#+}} 32-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
+ %3 = bitcast i8 %mask to <8 x i1>
+ %4 = select <8 x i1> %3, <8 x float> %2, <8 x float> zeroinitializer
+ ret <8 x float> %4
+}
+
define <4 x float> @stack_fold_shufps(<4 x float> %a0, <4 x float> %a1) {
;CHECK-LABEL: stack_fold_shufps
;CHECK: vshufps $200, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
diff --git a/llvm/test/CodeGen/X86/stack-folding-int-avx512.ll b/llvm/test/CodeGen/X86/stack-folding-int-avx512.ll
index 284021ac504..444eec3c783 100644
--- a/llvm/test/CodeGen/X86/stack-folding-int-avx512.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-int-avx512.ll
@@ -1742,6 +1742,58 @@ define <32 x i16> @stack_fold_psubw(<32 x i16> %a0, <32 x i16> %a1) {
ret <32 x i16> %2
}
+define <8 x i64> @stack_fold_shufi64x2(<8 x i64> %a, <8 x i64> %b) {
+ ;CHECK-LABEL: stack_fold_shufi64x2
+ ;CHECK: vshufi64x2 $24, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <8 x i64> %a, <8 x i64> %b, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9>
+ ret <8 x i64> %2
+}
+
+define <8 x i64> @stack_fold_shufi64x2_mask(<8 x i64> %a, <8 x i64> %b, i8 %mask, <8 x i64>* %passthru) {
+ ;CHECK-LABEL: stack_fold_shufi64x2_mask
+ ;CHECK: vshufi64x2 $24, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{{%k[1-7]}}} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <8 x i64> %a, <8 x i64> %b, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9>
+ %3 = bitcast i8 %mask to <8 x i1>
+ ; load needed to keep the operation from being scheduled above the asm block
+ %4 = load <8 x i64>, <8 x i64>* %passthru
+ %5 = select <8 x i1> %3, <8 x i64> %2, <8 x i64> %4
+ ret <8 x i64> %5
+}
+
+define <8 x i64> @stack_fold_shufi64x2_maskz(<8 x i64> %a, <8 x i64> %b, i8 %mask, <8 x i64>* %passthru) {
+ ;CHECK-LABEL: stack_fold_shufi64x2_maskz
+ ;CHECK: vshufi64x2 $24, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{{%k[1-7]}}} {z} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <8 x i64> %a, <8 x i64> %b, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9>
+ %3 = bitcast i8 %mask to <8 x i1>
+ %4 = select <8 x i1> %3, <8 x i64> %2, <8 x i64> zeroinitializer
+ ret <8 x i64> %4
+}
+
+define <16 x i32> @stack_fold_shufi32x4_mask(<16 x i32> %a, <16 x i32> %b, i16 %mask, <16 x i32>* %passthru) {
+ ;CHECK-LABEL: stack_fold_shufi32x4_mask
+ ;CHECK: vshufi32x4 $20, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{{%k[1-7]}}} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 16, i32 17, i32 18, i32 19>
+ %3 = bitcast i16 %mask to <16 x i1>
+ ; load needed to keep the operation from being scheduled above the asm block
+ %4 = load <16 x i32>, <16 x i32>* %passthru
+ %5 = select <16 x i1> %3, <16 x i32> %2, <16 x i32> %4
+ ret <16 x i32> %5
+}
+
+define <16 x i32> @stack_fold_shufi32x4_maskz(<16 x i32> %a, <16 x i32> %b, i16 %mask) {
+ ;CHECK-LABEL: stack_fold_shufi32x4_maskz
+ ;CHECK: vshufi32x4 $20, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{{%k[1-7]}}} {z} {{.*#+}} 64-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 16, i32 17, i32 18, i32 19>
+ %3 = bitcast i16 %mask to <16 x i1>
+ %4 = select <16 x i1> %3, <16 x i32> %2, <16 x i32> zeroinitializer
+ ret <16 x i32> %4
+}
+
define <16 x i32> @stack_fold_ternlogd(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2) {
;CHECK-LABEL: stack_fold_ternlogd
;CHECK: vpternlogd $33, {{-?[0-9]*}}(%rsp), {{%zmm[0-9][0-9]*}}, {{%zmm[0-9][0-9]*}} {{.*#+}} 64-byte Folded Reload
diff --git a/llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll b/llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll
index 4bfad5154ef..7c7c4abe2e1 100644
--- a/llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll
@@ -2449,3 +2449,24 @@ define <32 x i8> @stack_fold_punpckhbw_maskz_ymm(<32 x i8> %a0, <32 x i8> %a1, i
%4 = select <32 x i1> %3, <32 x i8> %2, <32 x i8> zeroinitializer
ret <32 x i8> %4
}
+
+define <4 x i64> @stack_fold_shufi64x2_maskz(<4 x i64> %a, <4 x i64> %b, i8 %mask) {
+ ;CHECK-LABEL: stack_fold_shufi64x2_maskz
+ ;CHECK: vshufi64x2 $1, {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{{%k[1-7]}}} {z} {{.*#+}} 32-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+ %3 = bitcast i8 %mask to <8 x i1>
+ %4 = shufflevector <8 x i1> %3, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+ %5 = select <4 x i1> %4, <4 x i64> %2, <4 x i64> zeroinitializer
+ ret <4 x i64> %5
+}
+
+define <8 x i32> @stack_fold_shufi32x4_maskz(<8 x i32> %a, <8 x i32> %b, i8 %mask) {
+ ;CHECK-LABEL: stack_fold_shufi32x4_maskz
+ ;CHECK: vshufi32x4 $1, {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{{%k[1-7]}}} {z} {{.*#+}} 32-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
+ %2 = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
+ %3 = bitcast i8 %mask to <8 x i1>
+ %4 = select <8 x i1> %3, <8 x i32> %2, <8 x i32> zeroinitializer
+ ret <8 x i32> %4
+}
OpenPOWER on IntegriCloud