diff options
| author | Dimitry Andric <dim@FreeBSD.org> | 2017-01-14 15:37:50 +0000 |
|---|---|---|
| committer | Dimitry Andric <dim@FreeBSD.org> | 2017-01-14 15:37:50 +0000 |
| commit | 581a6d8501ff5614297da837b81ed3b6956361ea (patch) | |
| tree | 985ee91d0ca1d3e6506ac5ff7e37f5b67adfec09 /lib/Transforms/Utils/SimplifyLibCalls.cpp | |
| parent | 909545a822eef491158f831688066f0ec2866938 (diff) | |
Diffstat (limited to 'lib/Transforms/Utils/SimplifyLibCalls.cpp')
| -rw-r--r-- | lib/Transforms/Utils/SimplifyLibCalls.cpp | 23 |
1 files changed, 22 insertions, 1 deletions
diff --git a/lib/Transforms/Utils/SimplifyLibCalls.cpp b/lib/Transforms/Utils/SimplifyLibCalls.cpp index 11d54bcf4f89d..8eaeb1073a761 100644 --- a/lib/Transforms/Utils/SimplifyLibCalls.cpp +++ b/lib/Transforms/Utils/SimplifyLibCalls.cpp @@ -1074,6 +1074,24 @@ Value *LibCallSimplifier::optimizePow(CallInst *CI, IRBuilder<> &B) { if (Op2C->getValueAPF().isZero()) // pow(x, 0.0) -> 1.0 return ConstantFP::get(CI->getType(), 1.0); + if (Op2C->isExactlyValue(-0.5) && + hasUnaryFloatFn(TLI, Op2->getType(), LibFunc::sqrt, LibFunc::sqrtf, + LibFunc::sqrtl)) { + // If -ffast-math: + // pow(x, -0.5) -> 1.0 / sqrt(x) + if (CI->hasUnsafeAlgebra()) { + IRBuilder<>::FastMathFlagGuard Guard(B); + B.setFastMathFlags(CI->getFastMathFlags()); + + // Here we cannot lower to an intrinsic because C99 sqrt() and llvm.sqrt + // are not guaranteed to have the same semantics. + Value *Sqrt = emitUnaryFloatFnCall(Op1, TLI->getName(LibFunc::sqrt), B, + Callee->getAttributes()); + + return B.CreateFDiv(ConstantFP::get(CI->getType(), 1.0), Sqrt, "sqrtrecip"); + } + } + if (Op2C->isExactlyValue(0.5) && hasUnaryFloatFn(TLI, Op2->getType(), LibFunc::sqrt, LibFunc::sqrtf, LibFunc::sqrtl) && @@ -1121,6 +1139,10 @@ Value *LibCallSimplifier::optimizePow(CallInst *CI, IRBuilder<> &B) { !V.isInteger()) return nullptr; + // Propagate fast math flags. + IRBuilder<>::FastMathFlagGuard Guard(B); + B.setFastMathFlags(CI->getFastMathFlags()); + // We will memoize intermediate products of the Addition Chain. Value *InnerChain[33] = {nullptr}; InnerChain[1] = Op1; @@ -1131,7 +1153,6 @@ Value *LibCallSimplifier::optimizePow(CallInst *CI, IRBuilder<> &B) { bool ignored; V.convert(APFloat::IEEEdouble(), APFloat::rmTowardZero, &ignored); - // TODO: Should the new instructions propagate the 'fast' flag of the pow()? Value *FMul = getPow(InnerChain, V.convertToDouble(), B); // For negative exponents simply compute the reciprocal. if (Op2C->isNegative()) |