[llvm] r280818 - [LSV] Use the original loads' names for the extractelement instructions.
Justin Lebar via llvm-commits
llvm-commits at lists.llvm.org
Wed Sep 7 08:49:49 PDT 2016
Author: jlebar
Date: Wed Sep 7 10:49:48 2016
New Revision: 280818
URL: http://llvm.org/viewvc/llvm-project?rev=280818&view=rev
Log:
[LSV] Use the original loads' names for the extractelement instructions.
Summary:
LSV replaces multiple adjacent loads with one vectorized load and a
bunch of extractelement instructions. This patch makes the
extractelement instructions' names match those of the original loads,
for (hopefully) improved readability.
Reviewers: asbirlea, tstellarAMD
Subscribers: arsenm, mzolotukhin
Differential Revision: https://reviews.llvm.org/D23748
Modified:
llvm/trunk/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp
llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll
llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll
Modified: llvm/trunk/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp?rev=280818&r1=280817&r2=280818&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp (original)
+++ llvm/trunk/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp Wed Sep 7 10:49:48 2016
@@ -983,7 +983,8 @@ bool Vectorizer::vectorizeLoadChain(
Instruction *UI = cast<Instruction>(Use);
unsigned Idx = cast<ConstantInt>(UI->getOperand(1))->getZExtValue();
unsigned NewIdx = Idx + I * VecWidth;
- Value *V = Builder.CreateExtractElement(LI, Builder.getInt32(NewIdx));
+ Value *V = Builder.CreateExtractElement(LI, Builder.getInt32(NewIdx),
+ UI->getName());
if (V->getType() != UI->getType())
V = Builder.CreateBitCast(V, UI->getType());
@@ -1002,8 +1003,9 @@ bool Vectorizer::vectorizeLoadChain(
I->eraseFromParent();
} else {
for (unsigned I = 0, E = Chain.size(); I != E; ++I) {
- Value *V = Builder.CreateExtractElement(LI, Builder.getInt32(I));
Value *CV = Chain[I];
+ Value *V =
+ Builder.CreateExtractElement(LI, Builder.getInt32(I), CV->getName());
if (V->getType() != CV->getType()) {
V = Builder.CreateBitOrPointerCast(V, CV->getType());
}
Modified: llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll?rev=280818&r1=280817&r2=280818&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll (original)
+++ llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll Wed Sep 7 10:49:48 2016
@@ -196,11 +196,11 @@ define void @merge_global_store_4_consta
}
; CHECK-LABEL: @merge_global_store_2_adjacent_loads_i32
-; CHECK: [[LOAD:%[0-9]+]] = load <2 x i32>
-; CHECK: [[ELT0:%[0-9]+]] = extractelement <2 x i32> [[LOAD]], i32 0
-; CHECK: [[ELT1:%[0-9]+]] = extractelement <2 x i32> [[LOAD]], i32 1
-; CHECK: [[INSERT0:%[0-9]+]] = insertelement <2 x i32> undef, i32 [[ELT0]], i32 0
-; CHECK: [[INSERT1:%[0-9]+]] = insertelement <2 x i32> [[INSERT0]], i32 [[ELT1]], i32 1
+; CHECK: [[LOAD:%[^ ]+]] = load <2 x i32>
+; CHECK: [[ELT0:%[^ ]+]] = extractelement <2 x i32> [[LOAD]], i32 0
+; CHECK: [[ELT1:%[^ ]+]] = extractelement <2 x i32> [[LOAD]], i32 1
+; CHECK: [[INSERT0:%[^ ]+]] = insertelement <2 x i32> undef, i32 [[ELT0]], i32 0
+; CHECK: [[INSERT1:%[^ ]+]] = insertelement <2 x i32> [[INSERT0]], i32 [[ELT1]], i32 1
; CHECK: store <2 x i32> [[INSERT1]]
define void @merge_global_store_2_adjacent_loads_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %in) #0 {
%out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
@@ -235,11 +235,11 @@ define void @merge_global_store_2_adjace
}
; CHECK-LABEL: @merge_global_store_2_adjacent_loads_shuffle_i32
-; CHECK: [[LOAD:%[0-9]+]] = load <2 x i32>
-; CHECK: [[ELT0:%[0-9]+]] = extractelement <2 x i32> [[LOAD]], i32 0
-; CHECK: [[ELT1:%[0-9]+]] = extractelement <2 x i32> [[LOAD]], i32 1
-; CHECK: [[INSERT0:%[0-9]+]] = insertelement <2 x i32> undef, i32 [[ELT1]], i32 0
-; CHECK: [[INSERT1:%[0-9]+]] = insertelement <2 x i32> [[INSERT0]], i32 [[ELT0]], i32 1
+; CHECK: [[LOAD:%[^ ]+]] = load <2 x i32>
+; CHECK: [[ELT0:%[^ ]+]] = extractelement <2 x i32> [[LOAD]], i32 0
+; CHECK: [[ELT1:%[^ ]+]] = extractelement <2 x i32> [[LOAD]], i32 1
+; CHECK: [[INSERT0:%[^ ]+]] = insertelement <2 x i32> undef, i32 [[ELT1]], i32 0
+; CHECK: [[INSERT1:%[^ ]+]] = insertelement <2 x i32> [[INSERT0]], i32 [[ELT0]], i32 1
; CHECK: store <2 x i32> [[INSERT1]]
define void @merge_global_store_2_adjacent_loads_shuffle_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %in) #0 {
%out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
Modified: llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll?rev=280818&r1=280817&r2=280818&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll (original)
+++ llvm/trunk/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll Wed Sep 7 10:49:48 2016
@@ -6,8 +6,8 @@ declare i32 @llvm.amdgcn.workitem.id.x()
; CHECK-LABEL: @merge_v2p1i8(
; CHECK: load <2 x i64>
-; CHECK: inttoptr i64 %{{[0-9]+}} to i8 addrspace(1)*
-; CHECK: inttoptr i64 %{{[0-9]+}} to i8 addrspace(1)*
+; CHECK: inttoptr i64 %{{[^ ]+}} to i8 addrspace(1)*
+; CHECK: inttoptr i64 %{{[^ ]+}} to i8 addrspace(1)*
; CHECK: store <2 x i64> zeroinitializer
define void @merge_v2p1i8(i8 addrspace(1)* addrspace(1)* nocapture %a, i8 addrspace(1)* addrspace(1)* nocapture readonly %b) #0 {
entry:
@@ -25,8 +25,8 @@ entry:
; CHECK-LABEL: @merge_v2p3i8(
; CHECK: load <2 x i32>
-; CHECK: inttoptr i32 %{{[0-9]+}} to i8 addrspace(3)*
-; CHECK: inttoptr i32 %{{[0-9]+}} to i8 addrspace(3)*
+; CHECK: inttoptr i32 %{{[^ ]+}} to i8 addrspace(3)*
+; CHECK: inttoptr i32 %{{[^ ]+}} to i8 addrspace(3)*
; CHECK: store <2 x i32> zeroinitializer
define void @merge_v2p3i8(i8 addrspace(3)* addrspace(3)* nocapture %a, i8 addrspace(3)* addrspace(3)* nocapture readonly %b) #0 {
entry:
@@ -44,7 +44,7 @@ entry:
; CHECK-LABEL: @merge_load_i64_ptr64(
; CHECK: load <2 x i64>
-; CHECK: [[ELT1:%[0-9]+]] = extractelement <2 x i64> %{{[0-9]+}}, i32 1
+; CHECK: [[ELT1:%[^ ]+]] = extractelement <2 x i64> %{{[^ ]+}}, i32 1
; CHECK: inttoptr i64 [[ELT1]] to i8 addrspace(1)*
define void @merge_load_i64_ptr64(i64 addrspace(1)* nocapture %a) #0 {
entry:
@@ -59,7 +59,7 @@ entry:
; CHECK-LABEL: @merge_load_ptr64_i64(
; CHECK: load <2 x i64>
-; CHECK: [[ELT0:%[0-9]+]] = extractelement <2 x i64> %{{[0-9]+}}, i32 0
+; CHECK: [[ELT0:%[^ ]+]] = extractelement <2 x i64> %{{[^ ]+}}, i32 0
; CHECK: inttoptr i64 [[ELT0]] to i8 addrspace(1)*
define void @merge_load_ptr64_i64(i64 addrspace(1)* nocapture %a) #0 {
entry:
@@ -73,7 +73,7 @@ entry:
}
; CHECK-LABEL: @merge_store_ptr64_i64(
-; CHECK: [[ELT0:%[0-9]+]] = ptrtoint i8 addrspace(1)* %ptr0 to i64
+; CHECK: [[ELT0:%[^ ]+]] = ptrtoint i8 addrspace(1)* %ptr0 to i64
; CHECK: insertelement <2 x i64> undef, i64 [[ELT0]], i32 0
; CHECK: store <2 x i64>
define void @merge_store_ptr64_i64(i64 addrspace(1)* nocapture %a, i8 addrspace(1)* %ptr0, i64 %val1) #0 {
@@ -89,8 +89,8 @@ entry:
}
; CHECK-LABEL: @merge_store_i64_ptr64(
-; CHECK: [[ELT1:%[0-9]+]] = ptrtoint i8 addrspace(1)* %ptr1 to i64
-; CHECK: insertelement <2 x i64> %{{[0-9]+}}, i64 [[ELT1]], i32 1
+; CHECK: [[ELT1:%[^ ]+]] = ptrtoint i8 addrspace(1)* %ptr1 to i64
+; CHECK: insertelement <2 x i64> %{{[^ ]+}}, i64 [[ELT1]], i32 1
; CHECK: store <2 x i64>
define void @merge_store_i64_ptr64(i8 addrspace(1)* addrspace(1)* nocapture %a, i64 %val0, i8 addrspace(1)* %ptr1) #0 {
entry:
@@ -105,7 +105,7 @@ entry:
; CHECK-LABEL: @merge_load_i32_ptr32(
; CHECK: load <2 x i32>
-; CHECK: [[ELT1:%[0-9]+]] = extractelement <2 x i32> %{{[0-9]+}}, i32 1
+; CHECK: [[ELT1:%[^ ]+]] = extractelement <2 x i32> %{{[^ ]+}}, i32 1
; CHECK: inttoptr i32 [[ELT1]] to i8 addrspace(3)*
define void @merge_load_i32_ptr32(i32 addrspace(3)* nocapture %a) #0 {
entry:
@@ -120,7 +120,7 @@ entry:
; CHECK-LABEL: @merge_load_ptr32_i32(
; CHECK: load <2 x i32>
-; CHECK: [[ELT0:%[0-9]+]] = extractelement <2 x i32> %{{[0-9]+}}, i32 0
+; CHECK: [[ELT0:%[^ ]+]] = extractelement <2 x i32> %{{[^ ]+}}, i32 0
; CHECK: inttoptr i32 [[ELT0]] to i8 addrspace(3)*
define void @merge_load_ptr32_i32(i32 addrspace(3)* nocapture %a) #0 {
entry:
@@ -134,7 +134,7 @@ entry:
}
; CHECK-LABEL: @merge_store_ptr32_i32(
-; CHECK: [[ELT0:%[0-9]+]] = ptrtoint i8 addrspace(3)* %ptr0 to i32
+; CHECK: [[ELT0:%[^ ]+]] = ptrtoint i8 addrspace(3)* %ptr0 to i32
; CHECK: insertelement <2 x i32> undef, i32 [[ELT0]], i32 0
; CHECK: store <2 x i32>
define void @merge_store_ptr32_i32(i32 addrspace(3)* nocapture %a, i8 addrspace(3)* %ptr0, i32 %val1) #0 {
@@ -149,8 +149,8 @@ entry:
}
; CHECK-LABEL: @merge_store_i32_ptr32(
-; CHECK: [[ELT1:%[0-9]+]] = ptrtoint i8 addrspace(3)* %ptr1 to i32
-; CHECK: insertelement <2 x i32> %{{[0-9]+}}, i32 [[ELT1]], i32 1
+; CHECK: [[ELT1:%[^ ]+]] = ptrtoint i8 addrspace(3)* %ptr1 to i32
+; CHECK: insertelement <2 x i32> %{{[^ ]+}}, i32 [[ELT1]], i32 1
; CHECK: store <2 x i32>
define void @merge_store_i32_ptr32(i8 addrspace(3)* addrspace(3)* nocapture %a, i32 %val0, i8 addrspace(3)* %ptr1) #0 {
entry:
@@ -241,9 +241,9 @@ entry:
; CHECK-LABEL: @merge_load_ptr64_f64(
; CHECK: load <2 x i64>
-; CHECK: [[ELT0:%[0-9]+]] = extractelement <2 x i64> %{{[0-9]+}}, i32 0
-; CHECK: [[ELT0_INT:%[0-9]+]] = inttoptr i64 [[ELT0]] to i8 addrspace(1)*
-; CHECK: [[ELT1_INT:%[0-9]+]] = extractelement <2 x i64> %{{[0-9]+}}, i32 1
+; CHECK: [[ELT0:%[^ ]+]] = extractelement <2 x i64> %{{[^ ]+}}, i32 0
+; CHECK: [[ELT0_INT:%[^ ]+]] = inttoptr i64 [[ELT0]] to i8 addrspace(1)*
+; CHECK: [[ELT1_INT:%[^ ]+]] = extractelement <2 x i64> %{{[^ ]+}}, i32 1
; CHECK: bitcast i64 [[ELT1_INT]] to double
define void @merge_load_ptr64_f64(double addrspace(1)* nocapture %a) #0 {
entry:
@@ -258,9 +258,9 @@ entry:
; CHECK-LABEL: @merge_load_f64_ptr64(
; CHECK: load <2 x i64>
-; CHECK: [[ELT0:%[0-9]+]] = extractelement <2 x i64> %{{[0-9]+}}, i32 0
+; CHECK: [[ELT0:%[^ ]+]] = extractelement <2 x i64> %{{[^ ]+}}, i32 0
; CHECK: bitcast i64 [[ELT0]] to double
-; CHECK: [[ELT1:%[0-9]+]] = extractelement <2 x i64> %{{[0-9]+}}, i32 1
+; CHECK: [[ELT1:%[^ ]+]] = extractelement <2 x i64> %{{[^ ]+}}, i32 1
; CHECK: inttoptr i64 [[ELT1]] to i8 addrspace(1)*
define void @merge_load_f64_ptr64(double addrspace(1)* nocapture %a) #0 {
entry:
@@ -274,10 +274,10 @@ entry:
}
; CHECK-LABEL: @merge_store_ptr64_f64(
-; CHECK: [[ELT0_INT:%[0-9]+]] = ptrtoint i8 addrspace(1)* %ptr0 to i64
+; CHECK: [[ELT0_INT:%[^ ]+]] = ptrtoint i8 addrspace(1)* %ptr0 to i64
; CHECK: insertelement <2 x i64> undef, i64 [[ELT0_INT]], i32 0
-; CHECK: [[ELT1_INT:%[0-9]+]] = bitcast double %val1 to i64
-; CHECK: insertelement <2 x i64> %{{[0-9]+}}, i64 [[ELT1_INT]], i32 1
+; CHECK: [[ELT1_INT:%[^ ]+]] = bitcast double %val1 to i64
+; CHECK: insertelement <2 x i64> %{{[^ ]+}}, i64 [[ELT1_INT]], i32 1
; CHECK: store <2 x i64>
define void @merge_store_ptr64_f64(double addrspace(1)* nocapture %a, i8 addrspace(1)* %ptr0, double %val1) #0 {
entry:
@@ -291,10 +291,10 @@ entry:
}
; CHECK-LABEL: @merge_store_f64_ptr64(
-; CHECK: [[ELT0_INT:%[0-9]+]] = bitcast double %val0 to i64
+; CHECK: [[ELT0_INT:%[^ ]+]] = bitcast double %val0 to i64
; CHECK: insertelement <2 x i64> undef, i64 [[ELT0_INT]], i32 0
-; CHECK: [[ELT1_INT:%[0-9]+]] = ptrtoint i8 addrspace(1)* %ptr1 to i64
-; CHECK: insertelement <2 x i64> %{{[0-9]+}}, i64 [[ELT1_INT]], i32 1
+; CHECK: [[ELT1_INT:%[^ ]+]] = ptrtoint i8 addrspace(1)* %ptr1 to i64
+; CHECK: insertelement <2 x i64> %{{[^ ]+}}, i64 [[ELT1_INT]], i32 1
; CHECK: store <2 x i64>
define void @merge_store_f64_ptr64(i8 addrspace(1)* addrspace(1)* nocapture %a, double %val0, i8 addrspace(1)* %ptr1) #0 {
entry:
More information about the llvm-commits
mailing list