6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-07-09 10:29:25 +00:00
Code Issues Projects Releases Wiki Activity

SLPVectorizer: compare entire intrinsic for SLP compatibility.

Browse Source 
Some Intrinsics are overloaded to the extent that return type equality (all
that's been checked up to now) does not guarantee that the arguments are the
same. In these cases SLP vectorizer should not recurse into the operands, which
can be achieved by comparing them as "Function *" rather than simply the ID.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@205424 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Tim Northover 2014-04-02 14:39:02 +00:00
parent 3844cadc9a
commit 24e78e0125
3 changed files with 23 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
lib/Transforms/Vectorize/SLPVectorizer.cpp
View File

@ -955,11 +955,11 @@ void BoUpSLP::buildTree_rec(ArrayRef<Value *> VL, unsigned Depth) {
return;
}
Intrinsic::ID ID = II->getIntrinsicID();
Function *Int = II->getCalledFunction();
for (unsigned i = 1, e = VL.size(); i != e; ++i) {
IntrinsicInst *II2 = dyn_cast<IntrinsicInst>(VL[i]);
if (!II2 || II2->getIntrinsicID() != ID) {
if (!II2 || II2->getCalledFunction() != Int) {
newTreeEntry(VL, false);
DEBUG(dbgs() << "SLP: mismatched calls:" << *II << "!=" << *VL[i]
<< "\n");

3
test/Transforms/SLPVectorizer/ARM64/lit.local.cfg Normal file
View File

@ -0,0 +1,3 @@
targets = set(config.root.targets_to_build.split())
if not 'ARM64' in targets:
config.unsupported = True

18
test/Transforms/SLPVectorizer/ARM64/mismatched-intrinsics.ll Normal file
View File

@ -0,0 +1,18 @@
; RUN: opt -S -slp-vectorizer %s | FileCheck %s
target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
target triple = "arm64-apple-ios5.0.0"
define i64 @mismatched_intrinsics(<4 x i32> %in1, <2 x i32> %in2) nounwind {
; CHECK-LABEL: @mismatched_intrinsics
; CHECK: call i64 @llvm.arm64.neon.saddlv.i64.v4i32
; CHECK: call i64 @llvm.arm64.neon.saddlv.i64.v2i32
%vaddlvq_s32.i = tail call i64 @llvm.arm64.neon.saddlv.i64.v4i32(<4 x i32> %in1) #2
%vaddlv_s32.i = tail call i64 @llvm.arm64.neon.saddlv.i64.v2i32(<2 x i32> %in2) #2
%tst = icmp sgt i64 %vaddlvq_s32.i, %vaddlv_s32.i
%equal = sext i1 %tst to i64
ret i64 %equal
}
declare i64 @llvm.arm64.neon.saddlv.i64.v4i32(<4 x i32> %in1)
declare i64 @llvm.arm64.neon.saddlv.i64.v2i32(<2 x i32> %in1)