maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[ARM][MVE] Tail-predication: use unsigned SCEV ranges for tripcount 

Loop tripcount expressions have a positive range, so use unsigned SCEV ranges
for them.

Differential Revision: https://reviews.llvm.org/D87608
This commit is contained in:
Sjoerd Meijer 2020-09-15 13:10:30 +01:00
parent aa8be5aeea
commit 635b87511e
1 changed files with 5 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
llvm/lib/Target/ARM/MVETailPredication.cpp
View File

@ -457,13 +457,10 @@ bool MVETailPredication::IsSafeActiveMask(IntrinsicInst *ActiveLaneMask,
// upperbound(TC) <= UINT_MAX - VectorWidth // upperbound(TC) <= UINT_MAX - VectorWidth
// //
unsigned SizeInBits = TripCount->getType()->getScalarSizeInBits(); unsigned SizeInBits = TripCount->getType()->getScalarSizeInBits();
auto Diff = APInt(SizeInBits, ~0) - APInt(SizeInBits, VectorWidth); auto MaxMinusVW = APInt(SizeInBits, ~0) - APInt(SizeInBits, VectorWidth);
uint64_t MaxMinusVW = Diff.getZExtValue(); APInt UpperboundTC = SE->getUnsignedRangeMax(TC);
// FIXME: since ranges can be negative we work with signed ranges here, but
// we shouldn't extract the zext'ed values for them.
uint64_t UpperboundTC = SE->getSignedRange(TC).getUpper().getZExtValue();
if (UpperboundTC > MaxMinusVW && !ForceTailPredication) { if (UpperboundTC.ugt(MaxMinusVW) && !ForceTailPredication) {
LLVM_DEBUG(dbgs() << "ARM TP: Overflow possible in tripcount rounding:\n"; LLVM_DEBUG(dbgs() << "ARM TP: Overflow possible in tripcount rounding:\n";
dbgs() << "upperbound(TC) <= UINT_MAX - VectorWidth\n"; dbgs() << "upperbound(TC) <= UINT_MAX - VectorWidth\n";
dbgs() << UpperboundTC << " <= " << MaxMinusVW << " == false\n";); dbgs() << UpperboundTC << " <= " << MaxMinusVW << " == false\n";);
@ -501,8 +498,8 @@ bool MVETailPredication::IsSafeActiveMask(IntrinsicInst *ActiveLaneMask,
auto *Ceil = SE->getUDivExpr(ECPlusVWMinus1, auto *Ceil = SE->getUDivExpr(ECPlusVWMinus1,
SE->getSCEV(ConstantInt::get(TripCount->getType(), VectorWidth))); SE->getSCEV(ConstantInt::get(TripCount->getType(), VectorWidth)));
ConstantRange RangeCeil = SE->getSignedRange(Ceil) ; ConstantRange RangeCeil = SE->getUnsignedRange(Ceil) ;
ConstantRange RangeTC = SE->getSignedRange(TC) ; ConstantRange RangeTC = SE->getUnsignedRange(TC) ;
if (!RangeTC.isSingleElement()) { if (!RangeTC.isSingleElement()) {
auto ZeroRange = auto ZeroRange =
ConstantRange(APInt(TripCount->getType()->getScalarSizeInBits(), 0)); ConstantRange(APInt(TripCount->getType()->getScalarSizeInBits(), 0));