diff options
author | Lang Hames <lhames@gmail.com> | 2012-06-19 22:51:23 +0000 |
---|---|---|
committer | Lang Hames <lhames@gmail.com> | 2012-06-19 22:51:23 +0000 |
commit | d693cafcfb9e67ba7040cb810e4409a166421482 (patch) | |
tree | 86a20682e48b7aaf402fc95f6497eae48e8fa89a /include/llvm/Target/TargetOptions.h | |
parent | fa8becb6f9e8aa2cbe3bab79e1fc4cdf2ffbb8e5 (diff) | |
download | external_llvm-d693cafcfb9e67ba7040cb810e4409a166421482.zip external_llvm-d693cafcfb9e67ba7040cb810e4409a166421482.tar.gz external_llvm-d693cafcfb9e67ba7040cb810e4409a166421482.tar.bz2 |
Add DAG-combines for aggressive FMA formation.
This patch adds DAG combines to form FMAs from pairs of FADD + FMUL or
FSUB + FMUL. The combines are performed when:
(a) Either
AllowExcessFPPrecision option (-enable-excess-fp-precision for llc)
OR
UnsafeFPMath option (-enable-unsafe-fp-math)
are set, and
(b) TargetLoweringInfo::isFMAFasterThanMulAndAdd(VT) is true for the type of
the FADD/FSUB, and
(c) The FMUL only has one user (the FADD/FSUB).
If your target has fast FMA instructions you can make use of these combines by
overriding TargetLoweringInfo::isFMAFasterThanMulAndAdd(VT) to return true for
types supported by your FMA instruction, and adding patterns to match ISD::FMA
to your FMA instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158757 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'include/llvm/Target/TargetOptions.h')
-rw-r--r-- | include/llvm/Target/TargetOptions.h | 16 |
1 files changed, 8 insertions, 8 deletions
diff --git a/include/llvm/Target/TargetOptions.h b/include/llvm/Target/TargetOptions.h index bc60673..84287fb 100644 --- a/include/llvm/Target/TargetOptions.h +++ b/include/llvm/Target/TargetOptions.h @@ -35,7 +35,7 @@ namespace llvm { TargetOptions() : PrintMachineCode(false), NoFramePointerElim(false), NoFramePointerElimNonLeaf(false), LessPreciseFPMADOption(false), - NoExcessFPPrecision(false), UnsafeFPMath(false), NoInfsFPMath(false), + AllowExcessFPPrecision(false), UnsafeFPMath(false), NoInfsFPMath(false), NoNaNsFPMath(false), HonorSignDependentRoundingFPMathOption(false), UseSoftFloat(false), NoZerosInBSS(false), JITExceptionHandling(false), JITEmitDebugInfo(false), JITEmitDebugInfoToDisk(false), @@ -74,13 +74,13 @@ namespace llvm { unsigned LessPreciseFPMADOption : 1; bool LessPreciseFPMAD() const; - /// NoExcessFPPrecision - This flag is enabled when the - /// -disable-excess-fp-precision flag is specified on the command line. - /// When this flag is off (the default), the code generator is allowed to - /// produce results that are "more precise" than IEEE allows. This includes - /// use of FMA-like operations and use of the X86 FP registers without - /// rounding all over the place. - unsigned NoExcessFPPrecision : 1; + /// AllowExcessFPPrecision - This flag is enabled when the + /// -enable-excess-fp-precision flag is specified on the command line. This + /// flag is OFF by default. When it is turned on, the code generator is + /// allowed to produce results that are "more precise" than IEEE allows. + /// This includes use of FMA-like operations and use of the X86 FP registers + /// without rounding all over the place. + unsigned AllowExcessFPPrecision : 1; /// UnsafeFPMath - This flag is enabled when the /// -enable-unsafe-fp-math flag is specified on the command line. When |