[X86] Add AVX512 versions of VCVTPD2PS to load folding tables.

llvm-svn: 315801

3 files changed, 27 insertions, 0 deletions

diff --git a/llvm/lib/Target/X86/X86InstrInfo.cpp b/llvm/lib/Target/X86/X86InstrInfo.cpp
index b9f78d3ad43..ae4d123e4cf 100644
--- a/llvm/lib/Target/X86/X86InstrInfo.cpp
+++ b/llvm/lib/Target/X86/X86InstrInfo.cpp
@@ -930,6 +930,7 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
     { X86::VBROADCASTSSZr,   X86::VBROADCASTSSZm,     TB_NO_REVERSE },
     { X86::VBROADCASTSDZr,   X86::VBROADCASTSDZm,     TB_NO_REVERSE },
     { X86::VCVTDQ2PDZrr,     X86::VCVTDQ2PDZrm,       0 },
+    { X86::VCVTPD2PSZrr,     X86::VCVTPD2PSZrm,       0 },
     { X86::VCVTUDQ2PDZrr,    X86::VCVTUDQ2PDZrm,      0 },
     { X86::VMOV64toPQIZrr,   X86::VMOVQI2PQIZrm,      0 },
     { X86::VMOV64toSDZrr,    X86::VMOV64toSDZrm,      0 },
@@ -991,6 +992,7 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
     { X86::VBROADCASTSSZ256r,    X86::VBROADCASTSSZ256m,    TB_NO_REVERSE },
     { X86::VBROADCASTSDZ256r,    X86::VBROADCASTSDZ256m,    TB_NO_REVERSE },
     { X86::VCVTDQ2PDZ256rr,      X86::VCVTDQ2PDZ256rm,      0 },
+    { X86::VCVTPD2PSZ256rr,      X86::VCVTPD2PSZ256rm,      0 },
     { X86::VCVTUDQ2PDZ256rr,     X86::VCVTUDQ2PDZ256rm,     0 },
     { X86::VMOVAPDZ256rr,        X86::VMOVAPDZ256rm,        TB_ALIGN_32 },
     { X86::VMOVAPSZ256rr,        X86::VMOVAPSZ256rm,        TB_ALIGN_32 },
@@ -1044,6 +1046,7 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
     // AVX-512 foldable instructions (128-bit versions)
     { X86::VBROADCASTSSZ128r,    X86::VBROADCASTSSZ128m,    TB_NO_REVERSE },
     { X86::VCVTDQ2PDZ128rr,      X86::VCVTDQ2PDZ128rm,      TB_NO_REVERSE },
+    { X86::VCVTPD2PSZ128rr,      X86::VCVTPD2PSZ128rm,      0 },
     { X86::VCVTUDQ2PDZ128rr,     X86::VCVTUDQ2PDZ128rm,     TB_NO_REVERSE },
     { X86::VMOVAPDZ128rr,        X86::VMOVAPDZ128rm,        TB_ALIGN_16 },
     { X86::VMOVAPSZ128rr,        X86::VMOVAPSZ128rm,        TB_ALIGN_16 },
diff --git a/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll b/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll
index e22e1a28e95..7bd46029f0e 100644
--- a/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll
@@ -200,6 +200,14 @@ define <8 x double> @stack_fold_cvtudq2pd(<8 x i32> %a0) {
   ret <8 x double> %2
 }
 
+define <8 x float> @stack_fold_cvtpd2ps(<8 x double> %a0) {
+  ;CHECK-LABEL: stack_fold_cvtpd2ps
+  ;CHECK:   vcvtpd2ps {{-?[0-9]*}}(%rsp), {{%ymm[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 = fptrunc <8 x double> %a0 to <8 x float>
+  ret <8 x float> %2
+}
+
 define <4 x float> @stack_fold_insertps(<4 x float> %a0, <4 x float> %a1) {
   ;CHECK-LABEL: stack_fold_insertps
   ;CHECK:       vinsertps $17, {{-?[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-fp-avx512vl.ll b/llvm/test/CodeGen/X86/stack-folding-fp-avx512vl.ll
index 331af4b6799..717e942fff1 100644
--- a/llvm/test/CodeGen/X86/stack-folding-fp-avx512vl.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-fp-avx512vl.ll
@@ -250,6 +250,22 @@ define <4 x double> @stack_fold_cvtudq2pd_ymm(<4 x i32> %a0) {
   ret <4 x double> %2
 }
 
+define <2 x float> @stack_fold_cvtpd2ps(<2 x double> %a0) {
+  ;CHECK-LABEL: stack_fold_cvtpd2ps
+  ;CHECK:   vcvtpd2psx {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-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 = fptrunc <2 x double> %a0 to <2 x float>
+  ret <2 x float> %2
+}
+
+define <4 x float> @stack_fold_cvtpd2ps_ymm(<4 x double> %a0) {
+  ;CHECK-LABEL: stack_fold_cvtpd2ps_ymm
+  ;CHECK:   vcvtpd2psy {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 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 = fptrunc <4 x double> %a0 to <4 x float>
+  ret <4 x float> %2
+}
+
 define <2 x double> @stack_fold_maxpd(<2 x double> %a0, <2 x double> %a1) #0 {
   ;CHECK-LABEL: stack_fold_maxpd
   ;CHECK:       vmaxpd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload