forked from OSchip/llvm-project
a476117e3a
Summary:
while investigating performance degradation of imagick benchmark
there were found inefficient pattern for UINT_TO_FP conversion.
That pattern causes RAW hazard in assembly code. Specifically,
uitofp IR operator results in poor assembler :
st %i0, [%fp - 952]
ldd [%fp - 952], %f0
it stores 32-bit integer register into memory location and then
loads 64-bit floating point data from that location.
That is exactly RAW hazard case. To optimize that case it is
possible to use SPISD::ITOF and SPISD::XTOF for conversion from
integer to floating point data type and to use ISD::BITCAST to
copy from integer register into floating point register.
The fix is to write custom UINT_TO_FP pattern using SPISD::ITOF,
SPISD::XTOF, ISD::BITCAST.
Patch by Alexey Lapshin
Reviewers: fedor.sergeev, jyknight, dcederman, lero_chris
Reviewed By: jyknight
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D36875
llvm-svn: 318704
|
||
|---|---|---|
| .. | ||
| 32abi.ll | ||
| 64abi.ll | ||
| 64bit.ll | ||
| 64cond.ll | ||
| 64spill.ll | ||
| 2006-01-22-BitConvertLegalize.ll | ||
| 2007-05-09-JumpTables.ll | ||
| 2007-07-05-LiveIntervalAssert.ll | ||
| 2008-10-10-InlineAsmMemoryOperand.ll | ||
| 2008-10-10-InlineAsmRegOperand.ll | ||
| 2009-08-28-PIC.ll | ||
| 2009-08-28-WeakLinkage.ll | ||
| 2011-01-11-CC.ll | ||
| 2011-01-11-Call.ll | ||
| 2011-01-11-FrameAddr.ll | ||
| 2011-01-19-DelaySlot.ll | ||
| 2011-01-21-ByValArgs.ll | ||
| 2011-01-22-SRet.ll | ||
| 2011-12-03-TailDuplication.ll | ||
| 2012-05-01-LowerArguments.ll | ||
| 2013-05-17-CallFrame.ll | ||
| DbgValueOtherTargets.test | ||
| LeonCASAInstructionUT.ll | ||
| LeonDetectRoundChangePassUT.ll | ||
| LeonFixAllFDIVSQRTPassUT.ll | ||
| LeonInsertNOPLoadPassUT.ll | ||
| LeonItinerariesUT.ll | ||
| LeonReplaceSDIVPassUT.ll | ||
| LeonSMACUMACInstructionUT.ll | ||
| analyze-branch.ll | ||
| atomics.ll | ||
| basictest.ll | ||
| blockaddr.ll | ||
| constpool.ll | ||
| constructor.ll | ||
| ctpop.ll | ||
| disable-fsmuld-fmuls.ll | ||
| empty-functions.ll | ||
| exception.ll | ||
| fail-alloca-align.ll | ||
| float-constants.ll | ||
| float.ll | ||
| fp128.ll | ||
| func-addr.ll | ||
| globals.ll | ||
| inlineasm-v9.ll | ||
| inlineasm.ll | ||
| leafproc.ll | ||
| lit.local.cfg | ||
| mature-mc-support.ll | ||
| missing-sret.ll | ||
| missinglabel.ll | ||
| mult-alt-generic-sparc.ll | ||
| multiple-div.ll | ||
| obj-relocs.ll | ||
| parts.ll | ||
| private.ll | ||
| register-clobber.ll | ||
| rem.ll | ||
| reserved-regs.ll | ||
| select-mask.ll | ||
| setjmp.ll | ||
| sjlj.ll | ||
| soft-float.ll | ||
| soft-mul-div.ll | ||
| spill.ll | ||
| spillsize.ll | ||
| sret-secondary.ll | ||
| stack-align.ll | ||
| stack-protector.ll | ||
| thread-pointer.ll | ||
| tls.ll | ||
| trap.ll | ||
| varargs.ll | ||
| vector-call.ll | ||
| vector-extract-elt.ll | ||
| zerostructcall.ll | ||