summaryrefslogtreecommitdiffstats
path: root/llvm/test/Transforms/SLPVectorizer/AArch64/gather-root.ll
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test/Transforms/SLPVectorizer/AArch64/gather-root.ll')
-rw-r--r--llvm/test/Transforms/SLPVectorizer/AArch64/gather-root.ll176
1 files changed, 92 insertions, 84 deletions
diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/gather-root.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/gather-root.ll
index dda8a0c3071..76f41647c62 100644
--- a/llvm/test/Transforms/SLPVectorizer/AArch64/gather-root.ll
+++ b/llvm/test/Transforms/SLPVectorizer/AArch64/gather-root.ll
@@ -31,50 +31,54 @@ define void @PR28330(i32 %n) {
;
; GATHER-LABEL: @PR28330(
; GATHER-NEXT: entry:
-; GATHER-NEXT: [[TMP0:%.*]] = load <2 x i8>, <2 x i8>* bitcast (i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 1) to <2 x i8>*), align 1
-; GATHER-NEXT: [[TMP1:%.*]] = icmp eq <2 x i8> [[TMP0]], zeroinitializer
-; GATHER-NEXT: [[TMP4:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 3), align 1
-; GATHER-NEXT: [[TMP5:%.*]] = icmp eq i8 [[TMP4]], 0
-; GATHER-NEXT: [[TMP6:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 4), align 4
-; GATHER-NEXT: [[TMP7:%.*]] = icmp eq i8 [[TMP6]], 0
-; GATHER-NEXT: [[TMP8:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 5), align 1
-; GATHER-NEXT: [[TMP9:%.*]] = icmp eq i8 [[TMP8]], 0
-; GATHER-NEXT: [[TMP10:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 6), align 2
-; GATHER-NEXT: [[TMP11:%.*]] = icmp eq i8 [[TMP10]], 0
-; GATHER-NEXT: [[TMP12:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 7), align 1
-; GATHER-NEXT: [[TMP13:%.*]] = icmp eq i8 [[TMP12]], 0
-; GATHER-NEXT: [[TMP14:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 8), align 8
-; GATHER-NEXT: [[TMP15:%.*]] = icmp eq i8 [[TMP14]], 0
+; GATHER-NEXT: [[TMP0:%.*]] = load <8 x i8>, <8 x i8>* bitcast (i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 1) to <8 x i8>*), align 1
+; GATHER-NEXT: [[TMP1:%.*]] = icmp eq <8 x i8> [[TMP0]], zeroinitializer
; GATHER-NEXT: br label [[FOR_BODY:%.*]]
; GATHER: for.body:
-; GATHER-NEXT: [[TMP17:%.*]] = phi i32 [ [[BIN_EXTRA:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
-; GATHER-NEXT: [[TMP2:%.*]] = select <2 x i1> [[TMP1]], <2 x i32> <i32 -720, i32 -720>, <2 x i32> <i32 -80, i32 -80>
-; GATHER-NEXT: [[TMP3:%.*]] = extractelement <2 x i32> [[TMP2]], i32 0
-; GATHER-NEXT: [[TMP20:%.*]] = add i32 [[TMP17]], [[TMP3]]
-; GATHER-NEXT: [[TMP4:%.*]] = extractelement <2 x i32> [[TMP2]], i32 1
-; GATHER-NEXT: [[TMP22:%.*]] = add i32 [[TMP20]], [[TMP4]]
-; GATHER-NEXT: [[TMP23:%.*]] = select i1 [[TMP5]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP24:%.*]] = add i32 [[TMP22]], [[TMP23]]
-; GATHER-NEXT: [[TMP25:%.*]] = select i1 [[TMP7]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP26:%.*]] = add i32 [[TMP24]], [[TMP25]]
-; GATHER-NEXT: [[TMP27:%.*]] = select i1 [[TMP9]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP28:%.*]] = add i32 [[TMP26]], [[TMP27]]
-; GATHER-NEXT: [[TMP29:%.*]] = select i1 [[TMP11]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP30:%.*]] = add i32 [[TMP28]], [[TMP29]]
-; GATHER-NEXT: [[TMP31:%.*]] = select i1 [[TMP13]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP32:%.*]] = add i32 [[TMP30]], [[TMP31]]
-; GATHER-NEXT: [[TMP33:%.*]] = select i1 [[TMP15]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP5:%.*]] = insertelement <8 x i32> undef, i32 [[TMP3]], i32 0
-; GATHER-NEXT: [[TMP6:%.*]] = insertelement <8 x i32> [[TMP5]], i32 [[TMP4]], i32 1
-; GATHER-NEXT: [[TMP7:%.*]] = insertelement <8 x i32> [[TMP6]], i32 [[TMP23]], i32 2
-; GATHER-NEXT: [[TMP8:%.*]] = insertelement <8 x i32> [[TMP7]], i32 [[TMP25]], i32 3
-; GATHER-NEXT: [[TMP9:%.*]] = insertelement <8 x i32> [[TMP8]], i32 [[TMP27]], i32 4
-; GATHER-NEXT: [[TMP10:%.*]] = insertelement <8 x i32> [[TMP9]], i32 [[TMP29]], i32 5
-; GATHER-NEXT: [[TMP11:%.*]] = insertelement <8 x i32> [[TMP10]], i32 [[TMP31]], i32 6
-; GATHER-NEXT: [[TMP12:%.*]] = insertelement <8 x i32> [[TMP11]], i32 [[TMP33]], i32 7
-; GATHER-NEXT: [[TMP13:%.*]] = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> [[TMP12]])
-; GATHER-NEXT: [[BIN_EXTRA]] = add i32 [[TMP13]], [[TMP17]]
-; GATHER-NEXT: [[TMP34:%.*]] = add i32 [[TMP32]], [[TMP33]]
+; GATHER-NEXT: [[TMPP17:%.*]] = phi i32 [ [[OP_EXTRA:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
+; GATHER-NEXT: [[TMP2:%.*]] = extractelement <8 x i1> [[TMP1]], i32 0
+; GATHER-NEXT: [[TMP3:%.*]] = insertelement <8 x i1> undef, i1 [[TMP2]], i32 0
+; GATHER-NEXT: [[TMP4:%.*]] = extractelement <8 x i1> [[TMP1]], i32 1
+; GATHER-NEXT: [[TMP5:%.*]] = insertelement <8 x i1> [[TMP3]], i1 [[TMP4]], i32 1
+; GATHER-NEXT: [[TMP6:%.*]] = extractelement <8 x i1> [[TMP1]], i32 2
+; GATHER-NEXT: [[TMP7:%.*]] = insertelement <8 x i1> [[TMP5]], i1 [[TMP6]], i32 2
+; GATHER-NEXT: [[TMP8:%.*]] = extractelement <8 x i1> [[TMP1]], i32 3
+; GATHER-NEXT: [[TMP9:%.*]] = insertelement <8 x i1> [[TMP7]], i1 [[TMP8]], i32 3
+; GATHER-NEXT: [[TMP10:%.*]] = extractelement <8 x i1> [[TMP1]], i32 4
+; GATHER-NEXT: [[TMP11:%.*]] = insertelement <8 x i1> [[TMP9]], i1 [[TMP10]], i32 4
+; GATHER-NEXT: [[TMP12:%.*]] = extractelement <8 x i1> [[TMP1]], i32 5
+; GATHER-NEXT: [[TMP13:%.*]] = insertelement <8 x i1> [[TMP11]], i1 [[TMP12]], i32 5
+; GATHER-NEXT: [[TMP14:%.*]] = extractelement <8 x i1> [[TMP1]], i32 6
+; GATHER-NEXT: [[TMP15:%.*]] = insertelement <8 x i1> [[TMP13]], i1 [[TMP14]], i32 6
+; GATHER-NEXT: [[TMP16:%.*]] = extractelement <8 x i1> [[TMP1]], i32 7
+; GATHER-NEXT: [[TMP17:%.*]] = insertelement <8 x i1> [[TMP15]], i1 [[TMP16]], i32 7
+; GATHER-NEXT: [[TMP18:%.*]] = select <8 x i1> [[TMP17]], <8 x i32> <i32 -720, i32 -720, i32 -720, i32 -720, i32 -720, i32 -720, i32 -720, i32 -720>, <8 x i32> <i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32 -80>
+; GATHER-NEXT: [[TMP19:%.*]] = extractelement <8 x i32> [[TMP18]], i32 0
+; GATHER-NEXT: [[TMPP20:%.*]] = add i32 [[TMPP17]], [[TMP19]]
+; GATHER-NEXT: [[TMP20:%.*]] = extractelement <8 x i32> [[TMP18]], i32 1
+; GATHER-NEXT: [[TMPP22:%.*]] = add i32 [[TMPP20]], [[TMP20]]
+; GATHER-NEXT: [[TMP21:%.*]] = extractelement <8 x i32> [[TMP18]], i32 2
+; GATHER-NEXT: [[TMPP24:%.*]] = add i32 [[TMPP22]], [[TMP21]]
+; GATHER-NEXT: [[TMP22:%.*]] = extractelement <8 x i32> [[TMP18]], i32 3
+; GATHER-NEXT: [[TMPP26:%.*]] = add i32 [[TMPP24]], [[TMP22]]
+; GATHER-NEXT: [[TMP23:%.*]] = extractelement <8 x i32> [[TMP18]], i32 4
+; GATHER-NEXT: [[TMPP28:%.*]] = add i32 [[TMPP26]], [[TMP23]]
+; GATHER-NEXT: [[TMP24:%.*]] = extractelement <8 x i32> [[TMP18]], i32 5
+; GATHER-NEXT: [[TMPP30:%.*]] = add i32 [[TMPP28]], [[TMP24]]
+; GATHER-NEXT: [[TMP25:%.*]] = extractelement <8 x i32> [[TMP18]], i32 6
+; GATHER-NEXT: [[TMPP32:%.*]] = add i32 [[TMPP30]], [[TMP25]]
+; GATHER-NEXT: [[TMP26:%.*]] = insertelement <8 x i32> undef, i32 [[TMP19]], i32 0
+; GATHER-NEXT: [[TMP27:%.*]] = insertelement <8 x i32> [[TMP26]], i32 [[TMP20]], i32 1
+; GATHER-NEXT: [[TMP28:%.*]] = insertelement <8 x i32> [[TMP27]], i32 [[TMP21]], i32 2
+; GATHER-NEXT: [[TMP29:%.*]] = insertelement <8 x i32> [[TMP28]], i32 [[TMP22]], i32 3
+; GATHER-NEXT: [[TMP30:%.*]] = insertelement <8 x i32> [[TMP29]], i32 [[TMP23]], i32 4
+; GATHER-NEXT: [[TMP31:%.*]] = insertelement <8 x i32> [[TMP30]], i32 [[TMP24]], i32 5
+; GATHER-NEXT: [[TMP32:%.*]] = insertelement <8 x i32> [[TMP31]], i32 [[TMP25]], i32 6
+; GATHER-NEXT: [[TMP33:%.*]] = extractelement <8 x i32> [[TMP18]], i32 7
+; GATHER-NEXT: [[TMP34:%.*]] = insertelement <8 x i32> [[TMP32]], i32 [[TMP33]], i32 7
+; GATHER-NEXT: [[TMP35:%.*]] = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> [[TMP34]])
+; GATHER-NEXT: [[OP_EXTRA]] = add i32 [[TMP35]], [[TMPP17]]
+; GATHER-NEXT: [[TMP34:%.*]] = add i32 [[TMPP32]], [[TMP33]]
; GATHER-NEXT: br label [[FOR_BODY]]
;
; MAX-COST-LABEL: @PR28330(
@@ -179,50 +183,54 @@ define void @PR32038(i32 %n) {
;
; GATHER-LABEL: @PR32038(
; GATHER-NEXT: entry:
-; GATHER-NEXT: [[TMP0:%.*]] = load <2 x i8>, <2 x i8>* bitcast (i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 1) to <2 x i8>*), align 1
-; GATHER-NEXT: [[TMP1:%.*]] = icmp eq <2 x i8> [[TMP0]], zeroinitializer
-; GATHER-NEXT: [[TMP4:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 3), align 1
-; GATHER-NEXT: [[TMP5:%.*]] = icmp eq i8 [[TMP4]], 0
-; GATHER-NEXT: [[TMP6:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 4), align 4
-; GATHER-NEXT: [[TMP7:%.*]] = icmp eq i8 [[TMP6]], 0
-; GATHER-NEXT: [[TMP8:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 5), align 1
-; GATHER-NEXT: [[TMP9:%.*]] = icmp eq i8 [[TMP8]], 0
-; GATHER-NEXT: [[TMP10:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 6), align 2
-; GATHER-NEXT: [[TMP11:%.*]] = icmp eq i8 [[TMP10]], 0
-; GATHER-NEXT: [[TMP12:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 7), align 1
-; GATHER-NEXT: [[TMP13:%.*]] = icmp eq i8 [[TMP12]], 0
-; GATHER-NEXT: [[TMP14:%.*]] = load i8, i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 8), align 8
-; GATHER-NEXT: [[TMP15:%.*]] = icmp eq i8 [[TMP14]], 0
+; GATHER-NEXT: [[TMP0:%.*]] = load <8 x i8>, <8 x i8>* bitcast (i8* getelementptr inbounds ([80 x i8], [80 x i8]* @a, i64 0, i64 1) to <8 x i8>*), align 1
+; GATHER-NEXT: [[TMP1:%.*]] = icmp eq <8 x i8> [[TMP0]], zeroinitializer
; GATHER-NEXT: br label [[FOR_BODY:%.*]]
; GATHER: for.body:
-; GATHER-NEXT: [[TMP17:%.*]] = phi i32 [ [[BIN_EXTRA:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
-; GATHER-NEXT: [[TMP2:%.*]] = select <2 x i1> [[TMP1]], <2 x i32> <i32 -720, i32 -720>, <2 x i32> <i32 -80, i32 -80>
-; GATHER-NEXT: [[TMP3:%.*]] = extractelement <2 x i32> [[TMP2]], i32 0
-; GATHER-NEXT: [[TMP20:%.*]] = add i32 -5, [[TMP3]]
-; GATHER-NEXT: [[TMP4:%.*]] = extractelement <2 x i32> [[TMP2]], i32 1
-; GATHER-NEXT: [[TMP22:%.*]] = add i32 [[TMP20]], [[TMP4]]
-; GATHER-NEXT: [[TMP23:%.*]] = select i1 [[TMP5]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP24:%.*]] = add i32 [[TMP22]], [[TMP23]]
-; GATHER-NEXT: [[TMP25:%.*]] = select i1 [[TMP7]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP26:%.*]] = add i32 [[TMP24]], [[TMP25]]
-; GATHER-NEXT: [[TMP27:%.*]] = select i1 [[TMP9]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP28:%.*]] = add i32 [[TMP26]], [[TMP27]]
-; GATHER-NEXT: [[TMP29:%.*]] = select i1 [[TMP11]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP30:%.*]] = add i32 [[TMP28]], [[TMP29]]
-; GATHER-NEXT: [[TMP31:%.*]] = select i1 [[TMP13]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP32:%.*]] = add i32 [[TMP30]], [[TMP31]]
-; GATHER-NEXT: [[TMP33:%.*]] = select i1 [[TMP15]], i32 -720, i32 -80
-; GATHER-NEXT: [[TMP5:%.*]] = insertelement <8 x i32> undef, i32 [[TMP3]], i32 0
-; GATHER-NEXT: [[TMP6:%.*]] = insertelement <8 x i32> [[TMP5]], i32 [[TMP4]], i32 1
-; GATHER-NEXT: [[TMP7:%.*]] = insertelement <8 x i32> [[TMP6]], i32 [[TMP23]], i32 2
-; GATHER-NEXT: [[TMP8:%.*]] = insertelement <8 x i32> [[TMP7]], i32 [[TMP25]], i32 3
-; GATHER-NEXT: [[TMP9:%.*]] = insertelement <8 x i32> [[TMP8]], i32 [[TMP27]], i32 4
-; GATHER-NEXT: [[TMP10:%.*]] = insertelement <8 x i32> [[TMP9]], i32 [[TMP29]], i32 5
-; GATHER-NEXT: [[TMP11:%.*]] = insertelement <8 x i32> [[TMP10]], i32 [[TMP31]], i32 6
-; GATHER-NEXT: [[TMP12:%.*]] = insertelement <8 x i32> [[TMP11]], i32 [[TMP33]], i32 7
-; GATHER-NEXT: [[TMP13:%.*]] = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> [[TMP12]])
-; GATHER-NEXT: [[BIN_EXTRA]] = add i32 [[TMP13]], -5
-; GATHER-NEXT: [[TMP34:%.*]] = add i32 [[TMP32]], [[TMP33]]
+; GATHER-NEXT: [[TMP17:%.*]] = phi i32 [ [[OP_EXTRA:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
+; GATHER-NEXT: [[TMP2:%.*]] = extractelement <8 x i1> [[TMP1]], i32 0
+; GATHER-NEXT: [[TMP3:%.*]] = insertelement <8 x i1> undef, i1 [[TMP2]], i32 0
+; GATHER-NEXT: [[TMP4:%.*]] = extractelement <8 x i1> [[TMP1]], i32 1
+; GATHER-NEXT: [[TMP5:%.*]] = insertelement <8 x i1> [[TMP3]], i1 [[TMP4]], i32 1
+; GATHER-NEXT: [[TMP6:%.*]] = extractelement <8 x i1> [[TMP1]], i32 2
+; GATHER-NEXT: [[TMP7:%.*]] = insertelement <8 x i1> [[TMP5]], i1 [[TMP6]], i32 2
+; GATHER-NEXT: [[TMP8:%.*]] = extractelement <8 x i1> [[TMP1]], i32 3
+; GATHER-NEXT: [[TMP9:%.*]] = insertelement <8 x i1> [[TMP7]], i1 [[TMP8]], i32 3
+; GATHER-NEXT: [[TMP10:%.*]] = extractelement <8 x i1> [[TMP1]], i32 4
+; GATHER-NEXT: [[TMP11:%.*]] = insertelement <8 x i1> [[TMP9]], i1 [[TMP10]], i32 4
+; GATHER-NEXT: [[TMP12:%.*]] = extractelement <8 x i1> [[TMP1]], i32 5
+; GATHER-NEXT: [[TMP13:%.*]] = insertelement <8 x i1> [[TMP11]], i1 [[TMP12]], i32 5
+; GATHER-NEXT: [[TMP14:%.*]] = extractelement <8 x i1> [[TMP1]], i32 6
+; GATHER-NEXT: [[TMP15:%.*]] = insertelement <8 x i1> [[TMP13]], i1 [[TMP14]], i32 6
+; GATHER-NEXT: [[TMP16:%.*]] = extractelement <8 x i1> [[TMP1]], i32 7
+; GATHER-NEXT: [[TMP17:%.*]] = insertelement <8 x i1> [[TMP15]], i1 [[TMP16]], i32 7
+; GATHER-NEXT: [[TMP18:%.*]] = select <8 x i1> [[TMP17]], <8 x i32> <i32 -720, i32 -720, i32 -720, i32 -720, i32 -720, i32 -720, i32 -720, i32 -720>, <8 x i32> <i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32 -80>
+; GATHER-NEXT: [[TMP19:%.*]] = extractelement <8 x i32> [[TMP18]], i32 0
+; GATHER-NEXT: [[TMPP20:%.*]] = add i32 -5, [[TMP19]]
+; GATHER-NEXT: [[TMP20:%.*]] = extractelement <8 x i32> [[TMP18]], i32 1
+; GATHER-NEXT: [[TMPP22:%.*]] = add i32 [[TMPP20]], [[TMP20]]
+; GATHER-NEXT: [[TMP21:%.*]] = extractelement <8 x i32> [[TMP18]], i32 2
+; GATHER-NEXT: [[TMPP24:%.*]] = add i32 [[TMPP22]], [[TMP21]]
+; GATHER-NEXT: [[TMP22:%.*]] = extractelement <8 x i32> [[TMP18]], i32 3
+; GATHER-NEXT: [[TMPP26:%.*]] = add i32 [[TMPP24]], [[TMP22]]
+; GATHER-NEXT: [[TMP23:%.*]] = extractelement <8 x i32> [[TMP18]], i32 4
+; GATHER-NEXT: [[TMPP28:%.*]] = add i32 [[TMPP26]], [[TMP23]]
+; GATHER-NEXT: [[TMP24:%.*]] = extractelement <8 x i32> [[TMP18]], i32 5
+; GATHER-NEXT: [[TMPP30:%.*]] = add i32 [[TMPP28]], [[TMP24]]
+; GATHER-NEXT: [[TMP25:%.*]] = extractelement <8 x i32> [[TMP18]], i32 6
+; GATHER-NEXT: [[TMPP32:%.*]] = add i32 [[TMPP30]], [[TMP25]]
+; GATHER-NEXT: [[TMP26:%.*]] = insertelement <8 x i32> undef, i32 [[TMP19]], i32 0
+; GATHER-NEXT: [[TMP27:%.*]] = insertelement <8 x i32> [[TMP26]], i32 [[TMP20]], i32 1
+; GATHER-NEXT: [[TMP28:%.*]] = insertelement <8 x i32> [[TMP27]], i32 [[TMP21]], i32 2
+; GATHER-NEXT: [[TMP29:%.*]] = insertelement <8 x i32> [[TMP28]], i32 [[TMP22]], i32 3
+; GATHER-NEXT: [[TMP30:%.*]] = insertelement <8 x i32> [[TMP29]], i32 [[TMP23]], i32 4
+; GATHER-NEXT: [[TMP31:%.*]] = insertelement <8 x i32> [[TMP30]], i32 [[TMP24]], i32 5
+; GATHER-NEXT: [[TMP32:%.*]] = insertelement <8 x i32> [[TMP31]], i32 [[TMP25]], i32 6
+; GATHER-NEXT: [[TMP33:%.*]] = extractelement <8 x i32> [[TMP18]], i32 7
+; GATHER-NEXT: [[TMP34:%.*]] = insertelement <8 x i32> [[TMP32]], i32 [[TMP33]], i32 7
+; GATHER-NEXT: [[TMP35:%.*]] = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> [[TMP34]])
+; GATHER-NEXT: [[OP_EXTRA]] = add i32 [[TMP35]], -5
+; GATHER-NEXT: [[TMP34:%.*]] = add i32 [[TMPP32]], [[TMP33]]
; GATHER-NEXT: br label [[FOR_BODY]]
;
; MAX-COST-LABEL: @PR32038(
OpenPOWER on IntegriCloud