diff options
| author | Craig Topper <craig.topper@intel.com> | 2018-01-07 06:24:24 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2018-01-07 06:24:24 +0000 |
| commit | 11aede13db63e7908171dbb3edf4ee6afdf9dd84 (patch) | |
| tree | f2a1102afa807e0a0b41e0f83e57c4ade85ca476 | |
| parent | 40cc8338f75174fced450bfa567cc722c9da3317 (diff) | |
| download | bcm5719-llvm-11aede13db63e7908171dbb3edf4ee6afdf9dd84.tar.gz bcm5719-llvm-11aede13db63e7908171dbb3edf4ee6afdf9dd84.zip | |
[X86] Add EVEX vcvtph2ps to the load folding tables.
llvm-svn: 321951
| -rw-r--r-- | llvm/lib/Target/X86/X86InstrInfo.cpp | 5 | ||||
| -rw-r--r-- | llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll | 9 |
2 files changed, 13 insertions, 1 deletions
diff --git a/llvm/lib/Target/X86/X86InstrInfo.cpp b/llvm/lib/Target/X86/X86InstrInfo.cpp index be918860f9a..24a63801d3d 100644 --- a/llvm/lib/Target/X86/X86InstrInfo.cpp +++ b/llvm/lib/Target/X86/X86InstrInfo.cpp @@ -1099,8 +1099,11 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI) { X86::VPSRLWZ128ri, X86::VPSRLWZ128mi, 0 }, // F16C foldable instructions - { X86::VCVTPH2PSrr, X86::VCVTPH2PSrm, 0 }, + { X86::VCVTPH2PSrr, X86::VCVTPH2PSrm, TB_NO_REVERSE }, { X86::VCVTPH2PSYrr, X86::VCVTPH2PSYrm, 0 }, + { X86::VCVTPH2PSZ128rr, X86::VCVTPH2PSZ128rm, TB_NO_REVERSE }, + { X86::VCVTPH2PSZ256rr, X86::VCVTPH2PSZ256rm, 0 }, + { X86::VCVTPH2PSZrr, X86::VCVTPH2PSZrm, 0 }, // AES foldable instructions { X86::AESIMCrr, X86::AESIMCrm, 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 5b057566abc..9057f262b31 100644 --- a/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll +++ b/llvm/test/CodeGen/X86/stack-folding-fp-avx512.ll @@ -208,6 +208,15 @@ define <8 x float> @stack_fold_cvtpd2ps(<8 x double> %a0) { ret <8 x float> %2 } +define <16 x float> @stack_fold_cvtph2ps(<16 x i16> %a0) { + ;CHECK-LABEL: stack_fold_cvtph2ps + ;CHECK: vcvtph2ps {{-?[0-9]*}}(%rsp), {{%zmm[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 = call <16 x float> @llvm.x86.avx512.mask.vcvtph2ps.512(<16 x i16> %a0, <16 x float> undef, i16 -1, i32 4) + ret <16 x float> %2 +} +declare <16 x float> @llvm.x86.avx512.mask.vcvtph2ps.512(<16 x i16>, <16 x float>, i16, i32) nounwind readonly + define <16 x i16> @stack_fold_cvtps2ph(<16 x float> %a0) { ;CHECK-LABEL: stack_fold_cvtps2ph ;CHECK: vcvtps2ph $0, {{%zmm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 32-byte Folded Spill |
