4b3a878e8a
The patch sets the vectorization metadata to false for Polly's fallback loops. These are the loops executed when RTCs fail. This minimizes the multiple loop versioning carried out by Polly and subsequently by the Loop Vectorizer. --------- Co-authored-by: Michael Kruse <github@meinersbur.de>
61 lines
2.4 KiB
LLVM
61 lines
2.4 KiB
LLVM
; RUN: opt %loadNPMPolly -passes=polly-codegen -S < %s | FileCheck %s --check-prefix=SCOPES
|
|
;
|
|
; Check that we create alias scopes that indicate the accesses to A, B and C cannot alias in any way.
|
|
;
|
|
; SCOPES-LABEL: polly.stmt.for.body:
|
|
; SCOPES: %[[BIdx:[._a-zA-Z0-9]*]] = getelementptr{{.*}} ptr %B, i64 %{{.*}}
|
|
; SCOPES: load i32, ptr %[[BIdx]], align 4, !alias.scope !2, !noalias !5
|
|
; SCOPES: %[[CIdx:[._a-zA-Z0-9]*]] = getelementptr{{.*}} ptr %C, i64 %{{.*}}
|
|
; SCOPES: load float, ptr %[[CIdx]], align 4, !alias.scope !8, !noalias !9
|
|
; SCOPES: %[[AIdx:[._a-zA-Z0-9]*]] = getelementptr{{.*}} ptr %A, i64 %{{.*}}
|
|
; SCOPES: store i32 %{{[._a-zA-Z0-9]*}}, ptr %[[AIdx]], align 4, !alias.scope !10, !noalias !11
|
|
;
|
|
; SCOPES: !0 = distinct !{!0, !1}
|
|
; SCOPES: !1 = !{!"llvm.loop.vectorize.enable", i32 0}
|
|
; SCOPES: !2 = !{!3}
|
|
; SCOPES: !3 = distinct !{!3, !4, !"polly.alias.scope.MemRef_B"}
|
|
; SCOPES: !4 = distinct !{!4, !"polly.alias.scope.domain"}
|
|
; SCOPES: !5 = !{!6, !7}
|
|
; SCOPES: !6 = distinct !{!6, !4, !"polly.alias.scope.MemRef_C"}
|
|
; SCOPES: !7 = distinct !{!7, !4, !"polly.alias.scope.MemRef_A"}
|
|
; SCOPES: !8 = !{!6}
|
|
; SCOPES: !9 = !{!3, !7}
|
|
; SCOPES: !10 = !{!7}
|
|
; SCOPES: !11 = !{!3, !6}
|
|
;
|
|
; void jd(int *A, int *B, float *C) {
|
|
; for (int i = 0; i < 1024; i++)
|
|
; A[i] = B[i] + C[i];
|
|
; }
|
|
;
|
|
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
|
|
|
|
define void @jd(ptr %A, ptr %B, ptr %C) {
|
|
entry:
|
|
br label %for.cond
|
|
|
|
for.cond: ; preds = %for.inc, %entry
|
|
%indvars.iv = phi i64 [ %indvars.iv.next, %for.inc ], [ 0, %entry ]
|
|
%exitcond = icmp ne i64 %indvars.iv, 1024
|
|
br i1 %exitcond, label %for.body, label %for.end
|
|
|
|
for.body: ; preds = %for.cond
|
|
%arrayidx = getelementptr inbounds i32, ptr %B, i64 %indvars.iv
|
|
%tmp = load i32, ptr %arrayidx, align 4
|
|
%conv = sitofp i32 %tmp to float
|
|
%arrayidx2 = getelementptr inbounds float, ptr %C, i64 %indvars.iv
|
|
%tmp1 = load float, ptr %arrayidx2, align 4
|
|
%add = fadd fast float %conv, %tmp1
|
|
%conv3 = fptosi float %add to i32
|
|
%arrayidx5 = getelementptr inbounds i32, ptr %A, i64 %indvars.iv
|
|
store i32 %conv3, ptr %arrayidx5, align 4
|
|
br label %for.inc
|
|
|
|
for.inc: ; preds = %for.body
|
|
%indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
|
|
br label %for.cond
|
|
|
|
for.end: ; preds = %for.cond
|
|
ret void
|
|
}
|