RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2024-11-28 06:00:28 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[InstCombine] allow (icmp sgt smin(PosA, B), 0) fold for vectors

Browse Source 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@281624 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Sanjay Patel 2016-09-15 16:23:20 +00:00
parent b794567962
commit f71deb6fd3
2 changed files with 21 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
lib/Transforms/InstCombine/InstCombineCompares.cpp
View File

@ -1376,11 +1376,10 @@ static Instruction *ProcessUGT_ADDCST_ADD(ICmpInst &I, Value *A, Value *B,
// Fold icmp Pred X, C.
Instruction *InstCombiner::foldICmpWithConstant(ICmpInst &Cmp) {
CmpInst::Predicate Pred = Cmp.getPredicate();
Value *X = Cmp.getOperand(0), *C = Cmp.getOperand(1);
Value *X = Cmp.getOperand(0);
// FIXME: Use m_APInt to allow folds for splat constants.
ConstantInt *CI = dyn_cast<ConstantInt>(C);
if (!CI)
const APInt *C;
if (!match(Cmp.getOperand(1), m_APInt(C)))
return nullptr;
Value *A = nullptr, *B = nullptr;
@ -1400,22 +1399,27 @@ Instruction *InstCombiner::foldICmpWithConstant(ICmpInst &Cmp) {
ConstantInt *CI2; // I = icmp ugt (add (add A, B), CI2), CI
if (Pred == ICmpInst::ICMP_UGT &&
match(X, m_Add(m_Add(m_Value(A), m_Value(B)), m_ConstantInt(CI2))))
if (Instruction *Res = ProcessUGT_ADDCST_ADD(Cmp, A, B, CI2, CI, *this))
if (Instruction *Res = ProcessUGT_ADDCST_ADD(
Cmp, A, B, CI2, cast<ConstantInt>(Cmp.getOperand(1)), *this))
return Res;
}
// (icmp sgt smin(PosA, B) 0) -> (icmp sgt B 0)
if (CI->isZero() && Pred == ICmpInst::ICMP_SGT)
if (auto *SI = dyn_cast<SelectInst>(X)) {
SelectPatternResult SPR = matchSelectPattern(SI, A, B);
if (*C == 0 && Pred == ICmpInst::ICMP_SGT) {
SelectPatternResult SPR = matchSelectPattern(X, A, B);
if (SPR.Flavor == SPF_SMIN) {
if (isKnownPositive(A, DL))
return new ICmpInst(Pred, B, CI);
return new ICmpInst(Pred, B, Cmp.getOperand(1));
if (isKnownPositive(B, DL))
return new ICmpInst(Pred, A, CI);
return new ICmpInst(Pred, A, Cmp.getOperand(1));
}
}
// FIXME: Use m_APInt to allow folds for splat constants.
ConstantInt *CI = dyn_cast<ConstantInt>(Cmp.getOperand(1));
if (!CI)
return nullptr;
// The following transforms are only worth it if the only user of the subtract
// is the icmp.
if (X->hasOneUse()) {

14
test/Transforms/InstCombine/min-positive.ll
View File

@ -19,11 +19,7 @@ define i1 @smin(i32 %other) {
define <2 x i1> @smin_vec(<2 x i32> %x, <2 x i32> %other) {
; CHECK-LABEL: @smin_vec(
; CHECK-NEXT: [[NOTNEG:%.*]] = and <2 x i32> %x, <i32 7, i32 7>
; CHECK-NEXT: [[POSITIVE:%.*]] = or <2 x i32> [[NOTNEG]], <i32 1, i32 1>
; CHECK-NEXT: [[CMP:%.*]] = icmp slt <2 x i32> [[POSITIVE]], %other
; CHECK-NEXT: [[SEL:%.*]] = select <2 x i1> [[CMP]], <2 x i32> [[POSITIVE]], <2 x i32> %other
; CHECK-NEXT: [[TEST:%.*]] = icmp sgt <2 x i32> [[SEL]], zeroinitializer
; CHECK-NEXT: [[TEST:%.*]] = icmp sgt <2 x i32> %other, zeroinitializer
; CHECK-NEXT: ret <2 x i1> [[TEST]]
;
%notneg = and <2 x i32> %x, <i32 7, i32 7>
@ -48,16 +44,12 @@ define i1 @smin_commute(i32 %other) {
define <2 x i1> @smin_commute_vec(<2 x i32> %x, <2 x i32> %other) {
; CHECK-LABEL: @smin_commute_vec(
; CHECK-NEXT: [[NOTNEG:%.*]] = and <2 x i32> %x, <i32 7, i32 7>
; CHECK-NEXT: [[POSITIVE:%.*]] = or <2 x i32> [[NOTNEG]], <i32 1, i32 1>
; CHECK-NEXT: [[CMP:%.*]] = icmp slt <2 x i32> [[POSITIVE]], %other
; CHECK-NEXT: [[SEL:%.*]] = select <2 x i1> [[CMP]], <2 x i32> %other, <2 x i32> [[POSITIVE]]
; CHECK-NEXT: [[TEST:%.*]] = icmp sgt <2 x i32> [[SEL]], zeroinitializer
; CHECK-NEXT: [[TEST:%.*]] = icmp sgt <2 x i32> %other, zeroinitializer
; CHECK-NEXT: ret <2 x i1> [[TEST]]
;
%notneg = and <2 x i32> %x, <i32 7, i32 7>
%positive = or <2 x i32> %notneg, <i32 1, i32 1>
%cmp = icmp slt <2 x i32> %positive, %other
%cmp = icmp slt <2 x i32> %other, %positive
%sel = select <2 x i1> %cmp, <2 x i32> %other, <2 x i32> %positive
%test = icmp sgt <2 x i32> %sel, zeroinitializer
ret <2 x i1> %test