RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-20 11:33:24 +02:00
Code Issues Projects Releases Wiki Activity

Temporarily Revert "[SLP] allow forming 2-way reduction patterns"

Browse Source 
After speaking with Sanjay - seeing a number of miscompiles and working
on tracking down a testcase. None of the follow on patches seem to
have helped so far.

This reverts commit 7ff57705ba196ce649d6034614b3b9df57e1f84f.
This commit is contained in:
Eric Christopher 2019-11-20 15:18:08 -08:00
parent 4783cb036c
commit fd20022682
4 changed files with 20 additions and 43 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
include/llvm/Transforms/Vectorize/SLPVectorizer.h
View File

@ -114,12 +114,9 @@ private:
/// Try to find horizontal reduction or otherwise vectorize a chain of binary
/// operators.
/// \p Try2WayRdx specializes the analysis to only attempt a 2-element
/// reduction.
bool vectorizeRootInstruction(PHINode *P, Value *V, BasicBlock *BB,
slpvectorizer::BoUpSLP &R,
TargetTransformInfo *TTI,
bool Try2WayRdx = false);
TargetTransformInfo *TTI);
/// Try to vectorize trees that start at insertvalue instructions.
bool vectorizeInsertValueInst(InsertValueInst *IVI, BasicBlock *BB,

37
lib/Transforms/Vectorize/SLPVectorizer.cpp
View File

@ -6653,7 +6653,7 @@ public:
/// Attempt to vectorize the tree found by
/// matchAssociativeReduction.
bool tryToReduce(BoUpSLP &V, TargetTransformInfo *TTI, bool Try2WayRdx) {
bool tryToReduce(BoUpSLP &V, TargetTransformInfo *TTI) {
if (ReducedVals.empty())
return false;
@ -6661,14 +6661,11 @@ public:
// to a nearby power-of-2. Can safely generate oversized
// vectors and rely on the backend to split them to legal sizes.
unsigned NumReducedVals = ReducedVals.size();
if (Try2WayRdx && NumReducedVals != 2)
return false;
unsigned MinRdxVals = Try2WayRdx ? 2 : 4;
if (NumReducedVals < MinRdxVals)
if (NumReducedVals < 4)
return false;
unsigned ReduxWidth = PowerOf2Floor(NumReducedVals);
unsigned MinRdxWidth = Log2_32(MinRdxVals);
Value *VectorizedTree = nullptr;
// FIXME: Fast-math-flags should be set based on the instructions in the
@ -6704,7 +6701,7 @@ public:
SmallVector<Value *, 16> IgnoreList;
for (auto &V : ReductionOps)
IgnoreList.append(V.begin(), V.end());
while (i < NumReducedVals - ReduxWidth + 1 && ReduxWidth > MinRdxWidth) {
while (i < NumReducedVals - ReduxWidth + 1 && ReduxWidth > 2) {
auto VL = makeArrayRef(&ReducedVals[i], ReduxWidth);
V.buildTree(VL, ExternallyUsedValues, IgnoreList);
Optional<ArrayRef<unsigned>> Order = V.bestOrder();
@ -7048,7 +7045,7 @@ static Value *getReductionValue(const DominatorTree *DT, PHINode *P,
/// performed.
static bool tryToVectorizeHorReductionOrInstOperands(
PHINode *P, Instruction *Root, BasicBlock *BB, BoUpSLP &R,
TargetTransformInfo *TTI, bool Try2WayRdx,
TargetTransformInfo *TTI,
const function_ref<bool(Instruction *, BoUpSLP &)> Vectorize) {
if (!ShouldVectorizeHor)
return false;
@ -7079,7 +7076,7 @@ static bool tryToVectorizeHorReductionOrInstOperands(
if (BI || SI) {
HorizontalReduction HorRdx;
if (HorRdx.matchAssociativeReduction(P, Inst)) {
if (HorRdx.tryToReduce(R, TTI, Try2WayRdx)) {
if (HorRdx.tryToReduce(R, TTI)) {
Res = true;
// Set P to nullptr to avoid re-analysis of phi node in
// matchAssociativeReduction function unless this is the root node.
@ -7122,8 +7119,7 @@ static bool tryToVectorizeHorReductionOrInstOperands(
bool SLPVectorizerPass::vectorizeRootInstruction(PHINode *P, Value *V,
BasicBlock *BB, BoUpSLP &R,
TargetTransformInfo *TTI,
bool Try2WayRdx) {
TargetTransformInfo *TTI) {
if (!V)
return false;
auto *I = dyn_cast<Instruction>(V);
@ -7136,7 +7132,7 @@ bool SLPVectorizerPass::vectorizeRootInstruction(PHINode *P, Value *V,
auto &&ExtraVectorization = [this](Instruction *I, BoUpSLP &R) -> bool {
return tryToVectorize(I, R);
};
return tryToVectorizeHorReductionOrInstOperands(P, I, BB, R, TTI, Try2WayRdx,
return tryToVectorizeHorReductionOrInstOperands(P, I, BB, R, TTI,
ExtraVectorization);
}
@ -7332,23 +7328,6 @@ bool SLPVectorizerPass::vectorizeChainsInBlock(BasicBlock *BB, BoUpSLP &R) {
PostProcessInstructions.push_back(&*it);
}
// Make a final attempt to match a 2-way reduction if nothing else worked.
// We do not try this above because it may interfere with other vectorization
// attempts.
// TODO: The constraints are copied from the above call to
// vectorizeRootInstruction(), but that might be too restrictive?
BasicBlock::iterator LastInst = --BB->end();
if (!Changed && LastInst->use_empty() &&
(LastInst->getType()->isVoidTy() || isa<CallInst>(LastInst) ||
isa<InvokeInst>(LastInst))) {
if (ShouldStartVectorizeHorAtStore || !isa<StoreInst>(LastInst)) {
for (auto *V : LastInst->operand_values()) {
Changed |= vectorizeRootInstruction(nullptr, V, BB, R, TTI,
/* Try2WayRdx */ true);
}
}
}
return Changed;
}

2
test/Feature/weak_constant.ll
View File

@ -1,5 +1,5 @@
; RUN: opt < %s -O3 -S > %t
; RUN: grep undef %t | count 2
; RUN: grep undef %t | count 1
; RUN: grep 5 %t | count 1
; RUN: grep 7 %t | count 1
; RUN: grep 9 %t | count 1

19
test/Transforms/SLPVectorizer/X86/reduction2.ll
View File

@ -54,10 +54,10 @@ define double @foo(double* nocapture %D) {
define i1 @two_wide_fcmp_reduction(<2 x double> %a0) {
; CHECK-LABEL: @two_wide_fcmp_reduction(
; CHECK-NEXT: [[A:%.*]] = fcmp ogt <2 x double> [[A0:%.*]], <double 1.000000e+00, double 1.000000e+00>
; CHECK-NEXT: [[RDX_SHUF:%.*]] = shufflevector <2 x i1> [[A]], <2 x i1> undef, <2 x i32> <i32 1, i32 undef>
; CHECK-NEXT: [[BIN_RDX:%.*]] = and <2 x i1> [[A]], [[RDX_SHUF]]
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x i1> [[BIN_RDX]], i32 0
; CHECK-NEXT: ret i1 [[TMP1]]
; CHECK-NEXT: [[B:%.*]] = extractelement <2 x i1> [[A]], i32 0
; CHECK-NEXT: [[C:%.*]] = extractelement <2 x i1> [[A]], i32 1
; CHECK-NEXT: [[D:%.*]] = and i1 [[B]], [[C]]
; CHECK-NEXT: ret i1 [[D]]
;
%a = fcmp ogt <2 x double> %a0, <double 1.0, double 1.0>
%b = extractelement <2 x i1> %a, i32 0
@ -96,11 +96,12 @@ define i1 @fcmp_lt_gt(double %a, double %b, double %c) {
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <2 x double> undef, double [[MUL]], i32 0
; CHECK-NEXT: [[TMP6:%.*]] = insertelement <2 x double> [[TMP5]], double [[MUL]], i32 1
; CHECK-NEXT: [[TMP7:%.*]] = fdiv <2 x double> [[TMP4]], [[TMP6]]
; CHECK-NEXT: [[TMP8:%.*]] = fcmp olt <2 x double> [[TMP7]], <double 0x3EB0C6F7A0B5ED8D, double 0x3EB0C6F7A0B5ED8D>
; CHECK-NEXT: [[RDX_SHUF:%.*]] = shufflevector <2 x i1> [[TMP8]], <2 x i1> undef, <2 x i32> <i32 1, i32 undef>
; CHECK-NEXT: [[BIN_RDX:%.*]] = and <2 x i1> [[TMP8]], [[RDX_SHUF]]
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <2 x i1> [[BIN_RDX]], i32 0
; CHECK-NEXT: br i1 [[TMP9]], label [[CLEANUP:%.*]], label [[LOR_LHS_FALSE:%.*]]
; CHECK-NEXT: [[TMP8:%.*]] = extractelement <2 x double> [[TMP7]], i32 1
; CHECK-NEXT: [[CMP:%.*]] = fcmp olt double [[TMP8]], 0x3EB0C6F7A0B5ED8D
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <2 x double> [[TMP7]], i32 0
; CHECK-NEXT: [[CMP4:%.*]] = fcmp olt double [[TMP9]], 0x3EB0C6F7A0B5ED8D
; CHECK-NEXT: [[OR_COND:%.*]] = and i1 [[CMP]], [[CMP4]]
; CHECK-NEXT: br i1 [[OR_COND]], label [[CLEANUP:%.*]], label [[LOR_LHS_FALSE:%.*]]
; CHECK: lor.lhs.false:
; CHECK-NEXT: [[TMP10:%.*]] = fcmp ule <2 x double> [[TMP7]], <double 1.000000e+00, double 1.000000e+00>
; CHECK-NEXT: [[TMP11:%.*]] = extractelement <2 x i1> [[TMP10]], i32 0