RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 03:02:36 +01:00
Code Issues Projects Releases Wiki Activity

Make MDNode::intersect faster than O(n * m)

Browse Source 
It is pretty easy to get it down to O(nlogn + mlogm).  This
implementation has the added benefit of automatically deduplicating
entries between the two sets.

llvm-svn: 278837
This commit is contained in:
David Majnemer 2016-08-16 18:48:37 +00:00
parent eb18f1195e
commit b52c8324a5
3 changed files with 7 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
lib/IR/Metadata.cpp
View File

@ -875,14 +875,13 @@ MDNode *MDNode::intersect(MDNode *A, MDNode *B) {
if (!A || !B) if (!A || !B)
return nullptr; return nullptr;
SmallVector<Metadata *, 4> MDs; SmallSetVector<Metadata *, 4> MDs(A->op_begin(), A->op_end());
for (Metadata *MD : A->operands()) SmallPtrSet<Metadata *, 4> BSet(B->op_begin(), B->op_end());
if (is_contained(B->operands(), MD)) MDs.remove_if([&](Metadata *MD) { return !is_contained(BSet, MD); });
MDs.push_back(MD);
// FIXME: This preserves long-standing behaviour, but is it really the right // FIXME: This preserves long-standing behaviour, but is it really the right
// behaviour? Or was that an unintended side-effect of node uniquing? // behaviour? Or was that an unintended side-effect of node uniquing?
return getOrSelfReference(A->getContext(), MDs); return getOrSelfReference(A->getContext(), MDs.getArrayRef());
} }
MDNode *MDNode::getMostGenericAliasScope(MDNode *A, MDNode *B) { MDNode *MDNode::getMostGenericAliasScope(MDNode *A, MDNode *B) {

4
test/Transforms/LoopVersioningLICM/loopversioningLICM1.ll
View File

@ -12,11 +12,11 @@
; CHECK-NEXT: %j.113 = phi i32 [ %j.016, %for.body3.ph ], [ %inc, %for.body3 ] ; CHECK-NEXT: %j.113 = phi i32 [ %j.016, %for.body3.ph ], [ %inc, %for.body3 ]
; CHECK-NEXT: %idxprom = zext i32 %j.113 to i64 ; CHECK-NEXT: %idxprom = zext i32 %j.113 to i64
; CHECK-NEXT: %arrayidx = getelementptr inbounds i32, i32* %var1, i64 %idxprom ; CHECK-NEXT: %arrayidx = getelementptr inbounds i32, i32* %var1, i64 %idxprom
; CHECK-NEXT: store i32 %add, i32* %arrayidx, align 4, !alias.scope !6, !noalias !6 ; CHECK-NEXT: store i32 %add, i32* %arrayidx, align 4, !alias.scope !2, !noalias !2
; CHECK-NEXT: %add8 = add nsw i32 %add86, %add ; CHECK-NEXT: %add8 = add nsw i32 %add86, %add
; CHECK-NEXT: %inc = add nuw i32 %j.113, 1 ; CHECK-NEXT: %inc = add nuw i32 %j.113, 1
; CHECK-NEXT: %cmp2 = icmp ult i32 %inc, %itr ; CHECK-NEXT: %cmp2 = icmp ult i32 %inc, %itr
; CHECK-NEXT: br i1 %cmp2, label %for.body3, label %for.inc11.loopexit.loopexit5, !llvm.loop !7 ; CHECK-NEXT: br i1 %cmp2, label %for.body3, label %for.inc11.loopexit.loopexit5, !llvm.loop !5
define i32 @foo(i32* nocapture %var1, i32* nocapture readnone %var2, i32* nocapture %var3, i32 %itr) #0 { define i32 @foo(i32* nocapture %var1, i32* nocapture readnone %var2, i32* nocapture %var3, i32 %itr) #0 {
entry: entry:
%cmp14 = icmp eq i32 %itr, 0 %cmp14 = icmp eq i32 %itr, 0

2
test/Transforms/LoopVersioningLICM/loopversioningLICM2.ll
View File

@ -9,7 +9,7 @@
; ;
; CHECK: for.cond1.for.inc17_crit_edge.us.loopexit5: ; preds = %for.body3.us ; CHECK: for.cond1.for.inc17_crit_edge.us.loopexit5: ; preds = %for.body3.us
; CHECK-NEXT: %add14.us.lcssa = phi float [ %add14.us, %for.body3.us ] ; CHECK-NEXT: %add14.us.lcssa = phi float [ %add14.us, %for.body3.us ]
; CHECK-NEXT: store float %add14.us.lcssa, float* %arrayidx.us, align 4, !alias.scope !7, !noalias !8 ; CHECK-NEXT: store float %add14.us.lcssa, float* %arrayidx.us, align 4, !alias.scope !0, !noalias !0
; CHECK-NEXT: br label %for.cond1.for.inc17_crit_edge.us ; CHECK-NEXT: br label %for.cond1.for.inc17_crit_edge.us
; ;
define i32 @foo(float* nocapture %var2, float** nocapture readonly %var3, i32 %itr) #0 { define i32 @foo(float* nocapture %var2, float** nocapture readonly %var3, i32 %itr) #0 {