diff options
| author | Craig Topper <craig.topper@gmail.com> | 2016-12-27 00:23:16 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@gmail.com> | 2016-12-27 00:23:16 +0000 |
| commit | 020b228155582eb7133a6c7be2ca1b4d9ebc61ec (patch) | |
| tree | 249b1082d8f46ce2f46141b84f1228846ef17f0e /llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp | |
| parent | 6d1b83ef8702fedfc38ec7397a8cc77844527371 (diff) | |
| download | bcm5719-llvm-020b228155582eb7133a6c7be2ca1b4d9ebc61ec.tar.gz bcm5719-llvm-020b228155582eb7133a6c7be2ca1b4d9ebc61ec.zip | |
[AVX-512][InstCombine] Teach InstCombine to turn packed add/sub/mul/div with rounding intrinsics into normal IR operations if the rounding mode is CUR_DIRECTION.
llvm-svn: 290559
Diffstat (limited to 'llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp')
| -rw-r--r-- | llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp | 44 |
1 files changed, 44 insertions, 0 deletions
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp index 8f665152703..348d1e1530f 100644 --- a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp +++ b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp @@ -1789,6 +1789,50 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) { break; } + case Intrinsic::x86_avx512_mask_add_ps_512: + case Intrinsic::x86_avx512_mask_div_ps_512: + case Intrinsic::x86_avx512_mask_mul_ps_512: + case Intrinsic::x86_avx512_mask_sub_ps_512: + case Intrinsic::x86_avx512_mask_add_pd_512: + case Intrinsic::x86_avx512_mask_div_pd_512: + case Intrinsic::x86_avx512_mask_mul_pd_512: + case Intrinsic::x86_avx512_mask_sub_pd_512: + // If the rounding mode is CUR_DIRECTION(4) we can turn these into regular + // IR operations. + if (auto *R = dyn_cast<ConstantInt>(II->getArgOperand(4))) { + if (R->getValue() == 4) { + Value *Arg0 = II->getArgOperand(0); + Value *Arg1 = II->getArgOperand(1); + + Value *V; + switch (II->getIntrinsicID()) { + default: llvm_unreachable("Case stmts out of sync!"); + case Intrinsic::x86_avx512_mask_add_ps_512: + case Intrinsic::x86_avx512_mask_add_pd_512: + V = Builder->CreateFAdd(Arg0, Arg1); + break; + case Intrinsic::x86_avx512_mask_sub_ps_512: + case Intrinsic::x86_avx512_mask_sub_pd_512: + V = Builder->CreateFSub(Arg0, Arg1); + break; + case Intrinsic::x86_avx512_mask_mul_ps_512: + case Intrinsic::x86_avx512_mask_mul_pd_512: + V = Builder->CreateFMul(Arg0, Arg1); + break; + case Intrinsic::x86_avx512_mask_div_ps_512: + case Intrinsic::x86_avx512_mask_div_pd_512: + V = Builder->CreateFDiv(Arg0, Arg1); + break; + } + + // Create a select for the masking. + V = emitX86MaskSelect(II->getArgOperand(3), V, II->getArgOperand(2), + *Builder); + return replaceInstUsesWith(*II, V); + } + } + break; + case Intrinsic::x86_avx512_mask_add_ss_round: case Intrinsic::x86_avx512_mask_div_ss_round: case Intrinsic::x86_avx512_mask_mul_ss_round: |
