aboutsummaryrefslogtreecommitdiffstats
path: root/test/Transforms
diff options
context:
space:
mode:
authorNadav Rotem <nrotem@apple.com>2013-04-26 05:08:59 +0000
committerNadav Rotem <nrotem@apple.com>2013-04-26 05:08:59 +0000
commit7557e521e50180817a165f8c897220b0e2020b7b (patch)
tree41cab12892eba9cc53bf9848f3c353990d505782 /test/Transforms
parent276b8549ee177ee584699638dd17eb358e25ace1 (diff)
downloadexternal_llvm-7557e521e50180817a165f8c897220b0e2020b7b.zip
external_llvm-7557e521e50180817a165f8c897220b0e2020b7b.tar.gz
external_llvm-7557e521e50180817a165f8c897220b0e2020b7b.tar.bz2
LoopVectorizer: Calculate the number of pointers to disambiguate at runtime based on the numbers of reads and writes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@180593 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/Transforms')
-rw-r--r--test/Transforms/LoopVectorize/runtime-limit.ll84
1 files changed, 84 insertions, 0 deletions
diff --git a/test/Transforms/LoopVectorize/runtime-limit.ll b/test/Transforms/LoopVectorize/runtime-limit.ll
new file mode 100644
index 0000000..d783974
--- /dev/null
+++ b/test/Transforms/LoopVectorize/runtime-limit.ll
@@ -0,0 +1,84 @@
+; RUN: opt < %s -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S | FileCheck %s
+
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
+target triple = "x86_64-apple-macosx10.8.0"
+
+; We are vectorizing with 6 runtime checks.
+;CHECK: func1x6
+;CHECK: <4 x i32>
+;CHECK: ret
+define i32 @func1x6(i32* nocapture %out, i32* nocapture %A, i32* nocapture %B, i32* nocapture %C, i32* nocapture %D, i32* nocapture %E, i32* nocapture %F) {
+entry:
+ br label %for.body
+
+for.body: ; preds = %for.body, %entry
+ %i.016 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+ %arrayidx = getelementptr inbounds i32* %A, i64 %i.016
+ %0 = load i32* %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds i32* %B, i64 %i.016
+ %1 = load i32* %arrayidx1, align 4
+ %add = add nsw i32 %1, %0
+ %arrayidx2 = getelementptr inbounds i32* %C, i64 %i.016
+ %2 = load i32* %arrayidx2, align 4
+ %add3 = add nsw i32 %add, %2
+ %arrayidx4 = getelementptr inbounds i32* %E, i64 %i.016
+ %3 = load i32* %arrayidx4, align 4
+ %add5 = add nsw i32 %add3, %3
+ %arrayidx6 = getelementptr inbounds i32* %F, i64 %i.016
+ %4 = load i32* %arrayidx6, align 4
+ %add7 = add nsw i32 %add5, %4
+ %arrayidx8 = getelementptr inbounds i32* %out, i64 %i.016
+ store i32 %add7, i32* %arrayidx8, align 4
+ %inc = add i64 %i.016, 1
+ %exitcond = icmp eq i64 %inc, 256
+ br i1 %exitcond, label %for.end, label %for.body
+
+for.end: ; preds = %for.body
+ ret i32 undef
+}
+
+; We are not vectorizing with 12 runtime checks.
+;CHECK: func2x6
+;CHECK-NOT: <4 x i32>
+;CHECK: ret
+define i32 @func2x6(i32* nocapture %out, i32* nocapture %out2, i32* nocapture %A, i32* nocapture %B, i32* nocapture %C, i32* nocapture %D, i32* nocapture %E, i32* nocapture %F) {
+entry:
+ br label %for.body
+
+for.body: ; preds = %for.body, %entry
+ %i.037 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+ %arrayidx = getelementptr inbounds i32* %A, i64 %i.037
+ %0 = load i32* %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds i32* %B, i64 %i.037
+ %1 = load i32* %arrayidx1, align 4
+ %add = add nsw i32 %1, %0
+ %arrayidx2 = getelementptr inbounds i32* %C, i64 %i.037
+ %2 = load i32* %arrayidx2, align 4
+ %add3 = add nsw i32 %add, %2
+ %arrayidx4 = getelementptr inbounds i32* %E, i64 %i.037
+ %3 = load i32* %arrayidx4, align 4
+ %add5 = add nsw i32 %add3, %3
+ %arrayidx6 = getelementptr inbounds i32* %F, i64 %i.037
+ %4 = load i32* %arrayidx6, align 4
+ %add7 = add nsw i32 %add5, %4
+ %arrayidx8 = getelementptr inbounds i32* %out, i64 %i.037
+ store i32 %add7, i32* %arrayidx8, align 4
+ %5 = load i32* %arrayidx, align 4
+ %6 = load i32* %arrayidx1, align 4
+ %add11 = add nsw i32 %6, %5
+ %7 = load i32* %arrayidx2, align 4
+ %add13 = add nsw i32 %add11, %7
+ %8 = load i32* %arrayidx4, align 4
+ %add15 = add nsw i32 %add13, %8
+ %9 = load i32* %arrayidx6, align 4
+ %add17 = add nsw i32 %add15, %9
+ %arrayidx18 = getelementptr inbounds i32* %out2, i64 %i.037
+ store i32 %add17, i32* %arrayidx18, align 4
+ %inc = add i64 %i.037, 1
+ %exitcond = icmp eq i64 %inc, 256
+ br i1 %exitcond, label %for.end, label %for.body
+
+for.end: ; preds = %for.body
+ ret i32 undef
+}
+