[llvm] r329821 - [InstCombine] limit X - (cast(-Y) --> X + cast(Y) with hasOneUse()

via llvm-commits llvm-commits at lists.llvm.org
Thu Apr 12 10:57:37 PDT 2018


hi! this patch caused a 0.14% instruction count regression on our test suite. it is likely quite bad for us because fpext/fptrunc are free.

—escha

> On Apr 11, 2018, at 8:57 AM, Sanjay Patel via llvm-commits <llvm-commits at lists.llvm.org> wrote:
> 
> Author: spatel
> Date: Wed Apr 11 08:57:18 2018
> New Revision: 329821
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=329821&view=rev
> Log:
> [InstCombine] limit X - (cast(-Y) --> X + cast(Y) with hasOneUse()
> 
> Modified:
>    llvm/trunk/lib/Transforms/InstCombine/InstCombineAddSub.cpp
>    llvm/trunk/test/Transforms/InstCombine/fsub.ll
> 
> Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineAddSub.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineAddSub.cpp?rev=329821&r1=329820&r2=329821&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Transforms/InstCombine/InstCombineAddSub.cpp (original)
> +++ llvm/trunk/lib/Transforms/InstCombine/InstCombineAddSub.cpp Wed Apr 11 08:57:18 2018
> @@ -1731,16 +1731,16 @@ Instruction *InstCombiner::visitFSub(Bin
>   if (match(Op1, m_FNeg(m_Value(Y))))
>     return BinaryOperator::CreateFAddFMF(Op0, Y, &I);
> 
> -  if (FPTruncInst *FPTI = dyn_cast<FPTruncInst>(Op1)) {
> -    if (Value *V = dyn_castFNegVal(FPTI->getOperand(0))) {
> -      Value *NewTrunc = Builder.CreateFPTrunc(V, I.getType());
> -      return BinaryOperator::CreateFAddFMF(Op0, NewTrunc, &I);
> -    }
> -  } else if (FPExtInst *FPEI = dyn_cast<FPExtInst>(Op1)) {
> -    if (Value *V = dyn_castFNegVal(FPEI->getOperand(0))) {
> -      Value *NewExt = Builder.CreateFPExt(V, I.getType());
> -      return BinaryOperator::CreateFAddFMF(Op0, NewExt, &I);
> -    }
> +  // Similar to above, but look through a cast of the negated value:
> +  // X - (fptrunc(-Y)) --> X + fptrunc(Y)
> +  if (match(Op1, m_OneUse(m_FPTrunc(m_FNeg(m_Value(Y)))))) {
> +    Value *TruncY = Builder.CreateFPTrunc(Y, I.getType());
> +    return BinaryOperator::CreateFAddFMF(Op0, TruncY, &I);
> +  }
> +  // X - (fpext(-Y)) --> X + fpext(Y)
> +  if (match(Op1, m_OneUse(m_FPExt(m_FNeg(m_Value(Y)))))) {
> +    Value *ExtY = Builder.CreateFPExt(Y, I.getType());
> +    return BinaryOperator::CreateFAddFMF(Op0, ExtY, &I);
>   }
> 
>   // Handle specials cases for FSub with selects feeding the operation
> 
> Modified: llvm/trunk/test/Transforms/InstCombine/fsub.ll
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstCombine/fsub.ll?rev=329821&r1=329820&r2=329821&view=diff
> ==============================================================================
> --- llvm/trunk/test/Transforms/InstCombine/fsub.ll (original)
> +++ llvm/trunk/test/Transforms/InstCombine/fsub.ll Wed Apr 11 08:57:18 2018
> @@ -189,14 +189,13 @@ define double @neg_ext_op1_fast(float %a
>   ret double %t3
> }
> 
> -; FIXME: Extra use should prevent the transform.
> +; Extra use should prevent the transform.
> 
> define float @neg_ext_op1_extra_use(half %a, float %b) {
> ; CHECK-LABEL: @neg_ext_op1_extra_use(
> ; CHECK-NEXT:    [[T1:%.*]] = fsub half 0xH8000, [[A:%.*]]
> ; CHECK-NEXT:    [[T2:%.*]] = fpext half [[T1]] to float
> -; CHECK-NEXT:    [[TMP1:%.*]] = fpext half [[A]] to float
> -; CHECK-NEXT:    [[T3:%.*]] = fadd float [[TMP1]], [[B:%.*]]
> +; CHECK-NEXT:    [[T3:%.*]] = fsub float [[B:%.*]], [[T2]]
> ; CHECK-NEXT:    call void @use(float [[T2]])
> ; CHECK-NEXT:    ret float [[T3]]
> ;
> @@ -207,8 +206,8 @@ define float @neg_ext_op1_extra_use(half
>   ret float %t3
> }
> 
> -; One-use fptrunc is always hoisted above fneg, so the corresponding 
> -; multi-use bug for fptrunc isn't visible with a fold starting from 
> +; One-use fptrunc is always hoisted above fneg, so the corresponding
> +; multi-use bug for fptrunc isn't visible with a fold starting from
> ; the last fsub.
> 
> define float @neg_trunc_op1_extra_use(double %a, float %b) {
> @@ -226,15 +225,13 @@ define float @neg_trunc_op1_extra_use(do
>   ret float %t3
> }
> 
> -; FIXME: But the bug is visible when both preceding values have other uses.
> ; Extra uses should prevent the transform.
> 
> define float @neg_trunc_op1_extra_uses(double %a, float %b) {
> ; CHECK-LABEL: @neg_trunc_op1_extra_uses(
> ; CHECK-NEXT:    [[T1:%.*]] = fsub double -0.000000e+00, [[A:%.*]]
> ; CHECK-NEXT:    [[T2:%.*]] = fptrunc double [[T1]] to float
> -; CHECK-NEXT:    [[TMP1:%.*]] = fptrunc double [[A]] to float
> -; CHECK-NEXT:    [[T3:%.*]] = fadd float [[TMP1]], [[B:%.*]]
> +; CHECK-NEXT:    [[T3:%.*]] = fsub float [[B:%.*]], [[T2]]
> ; CHECK-NEXT:    call void @use2(float [[T2]], double [[T1]])
> ; CHECK-NEXT:    ret float [[T3]]
> ;
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits



More information about the llvm-commits mailing list