[llvm] 399ff08 - [LV] Precommit tests with any-of reductions and epilogue vectorization.
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Wed Apr 3 05:33:45 PDT 2024
Author: Florian Hahn
Date: 2024-04-03T13:32:32+01:00
New Revision: 399ff08e29de4f2bbcfd47f87bb1795ba3a4e091
URL: https://github.com/llvm/llvm-project/commit/399ff08e29de4f2bbcfd47f87bb1795ba3a4e091
DIFF: https://github.com/llvm/llvm-project/commit/399ff08e29de4f2bbcfd47f87bb1795ba3a4e091.diff
LOG: [LV] Precommit tests with any-of reductions and epilogue vectorization.
Test case for failures from
https://lab.llvm.org/buildbot/#/builders/74/builds/26697
caused the revert of 95fef1d in 589c7ab.
Added:
llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll
Modified:
Removed:
################################################################################
diff --git a/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll b/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll
new file mode 100644
index 00000000000000..0b872709ec6c6e
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll
@@ -0,0 +1,222 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
+; RUN: opt -p loop-vectorize -force-vector-width=4 -enable-epilogue-vectorization -epilogue-vectorization-force-VF=4 -S %s | FileCheck %s
+
+target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
+
+define i32 @any_of_reduction_epilog(ptr %src, i64 %N) {
+; CHECK-LABEL: define i32 @any_of_reduction_epilog(
+; CHECK-SAME: ptr [[SRC:%.*]], i64 [[N:%.*]]) {
+; CHECK-NEXT: iter.check:
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[N]], 1
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 4
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]]
+; CHECK: vector.main.loop.iter.check:
+; CHECK-NEXT: [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[TMP0]], 4
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK1]], label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK: vector.ph:
+; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 4
+; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
+; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP8:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i32 0
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP3]], align 1
+; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <4 x i8> [[WIDE_LOAD]], zeroinitializer
+; CHECK-NEXT: [[TMP8]] = select <4 x i1> [[TMP4]], <4 x i32> <i32 1, i32 1, i32 1, i32 1>, <4 x i32> [[VEC_PHI]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: middle.block:
+; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <4 x i32> [[TMP8]], zeroinitializer
+; CHECK-NEXT: [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
+; CHECK-NEXT: [[RDX_SELECT:%.*]] = select i1 [[TMP7]], i32 1, i32 0
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
+; CHECK: vec.epilog.iter.check:
+; CHECK-NEXT: [[N_VEC_REMAINING:%.*]] = sub i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT: [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], 4
+; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; CHECK: vec.epilog.ph:
+; CHECK-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ]
+; CHECK-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; CHECK-NEXT: [[N_MOD_VF2:%.*]] = urem i64 [[TMP0]], 4
+; CHECK-NEXT: [[N_VEC3:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF2]]
+; CHECK-NEXT: [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[BC_MERGE_RDX]], i64 0
+; CHECK-NEXT: [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <4 x i32> [[MINMAX_IDENT_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
+; CHECK: vec.epilog.vector.body:
+; CHECK-NEXT: [[INDEX5:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT8:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI6:%.*]] = phi <4 x i32> [ [[MINMAX_IDENT_SPLAT]], [[VEC_EPILOG_PH]] ], [ [[TMP17:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP10:%.*]] = add i64 [[INDEX5]], 0
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i32 0
+; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x i8>, ptr [[TMP12]], align 1
+; CHECK-NEXT: [[TMP13:%.*]] = icmp eq <4 x i8> [[WIDE_LOAD7]], zeroinitializer
+; CHECK-NEXT: [[TMP17]] = select <4 x i1> [[TMP13]], <4 x i32> <i32 1, i32 1, i32 1, i32 1>, <4 x i32> [[VEC_PHI6]]
+; CHECK-NEXT: [[INDEX_NEXT8]] = add nuw i64 [[INDEX5]], 4
+; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT8]], [[N_VEC3]]
+; CHECK-NEXT: br i1 [[TMP15]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK: vec.epilog.middle.block:
+; CHECK-NEXT: [[TMP14:%.*]] = icmp ne <4 x i32> [[TMP17]], zeroinitializer
+; CHECK-NEXT: [[TMP16:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP14]])
+; CHECK-NEXT: [[RDX_SELECT9:%.*]] = select i1 [[TMP16]], i32 1, i32 0
+; CHECK-NEXT: [[CMP_N4:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC3]]
+; CHECK-NEXT: br i1 [[CMP_N4]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
+; CHECK: vec.epilog.scalar.ph:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC3]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
+; CHECK-NEXT: [[BC_MERGE_RDX10:%.*]] = phi i32 [ 0, [[ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ], [ [[RDX_SELECT9]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; CHECK-NEXT: br label [[LOOP:%.*]]
+; CHECK: loop:
+; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[RED:%.*]] = phi i32 [ [[BC_MERGE_RDX10]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[SELECT:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[IV]]
+; CHECK-NEXT: [[LOAD:%.*]] = load i8, ptr [[GEP]], align 1
+; CHECK-NEXT: [[ICMP:%.*]] = icmp eq i8 [[LOAD]], 0
+; CHECK-NEXT: [[SELECT]] = select i1 [[ICMP]], i32 1, i32 [[RED]]
+; CHECK-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT: [[ICMP3:%.*]] = icmp eq i64 [[IV]], [[N]]
+; CHECK-NEXT: br i1 [[ICMP3]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK: exit:
+; CHECK-NEXT: [[SELECT_LCSSA:%.*]] = phi i32 [ [[SELECT]], [[LOOP]] ], [ [[RDX_SELECT]], [[MIDDLE_BLOCK]] ], [ [[RDX_SELECT9]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; CHECK-NEXT: ret i32 [[SELECT_LCSSA]]
+;
+entry:
+ br label %loop
+
+loop:
+ %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
+ %red = phi i32 [ 0, %entry ], [ %select, %loop ]
+ %gep = getelementptr inbounds i8, ptr %src, i64 %iv
+ %load = load i8, ptr %gep, align 1
+ %icmp = icmp eq i8 %load, 0
+ %select = select i1 %icmp, i32 1, i32 %red
+ %iv.next = add i64 %iv, 1
+ %icmp3 = icmp eq i64 %iv, %N
+ br i1 %icmp3, label %exit, label %loop
+
+exit:
+ ret i32 %select
+}
+
+
+define i1 @any_of_reduction_i1_epilog(i64 %N, i32 %a) {
+; CHECK-LABEL: define i1 @any_of_reduction_i1_epilog(
+; CHECK-SAME: i64 [[N:%.*]], i32 [[A:%.*]]) {
+; CHECK-NEXT: iter.check:
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[N]], 1
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 4
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]]
+; CHECK: vector.main.loop.iter.check:
+; CHECK-NEXT: [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[TMP0]], 4
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK1]], label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK: vector.ph:
+; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 4
+; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[A]], i64 0
+; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[IND_END:%.*]] = trunc i64 [[N_VEC]] to i32
+; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP1:%.*]] = icmp eq <4 x i32> [[VEC_IND]], [[BROADCAST_SPLAT]]
+; CHECK-NEXT: [[TMP3]] = select <4 x i1> [[TMP1]], <4 x i1> [[VEC_PHI]], <4 x i1> zeroinitializer
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4>
+; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK: middle.block:
+; CHECK-NEXT: [[RDX_SELECT_CMP:%.*]] = icmp ne <4 x i1> [[TMP3]], zeroinitializer
+; CHECK-NEXT: [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP]])
+; CHECK-NEXT: [[RDX_SELECT:%.*]] = select i1 [[TMP6]], i1 false, i1 false
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
+; CHECK: vec.epilog.iter.check:
+; CHECK-NEXT: [[IND_END6:%.*]] = trunc i64 [[N_VEC]] to i32
+; CHECK-NEXT: [[N_VEC_REMAINING:%.*]] = sub i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT: [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_VEC_REMAINING]], 4
+; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; CHECK: vec.epilog.ph:
+; CHECK-NEXT: [[BC_MERGE_RDX:%.*]] = phi i1 [ false, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[IND_END]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; CHECK-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
+; CHECK-NEXT: [[N_MOD_VF2:%.*]] = urem i64 [[TMP0]], 4
+; CHECK-NEXT: [[N_VEC3:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF2]]
+; CHECK-NEXT: [[IND_END5:%.*]] = trunc i64 [[N_VEC3]] to i32
+; CHECK-NEXT: [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[BC_MERGE_RDX]], i64 0
+; CHECK-NEXT: [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <4 x i1> [[MINMAX_IDENT_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[BC_RESUME_VAL]], i64 0
+; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[INDUCTION:%.*]] = add <4 x i32> [[DOTSPLAT]], <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT13:%.*]] = insertelement <4 x i32> poison, i32 [[A]], i64 0
+; CHECK-NEXT: [[BROADCAST_SPLAT14:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT13]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
+; CHECK: vec.epilog.vector.body:
+; CHECK-NEXT: [[INDEX9:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT15:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI10:%.*]] = phi <4 x i1> [ [[MINMAX_IDENT_SPLAT]], [[VEC_EPILOG_PH]] ], [ [[TMP10:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND11:%.*]] = phi <4 x i32> [ [[INDUCTION]], [[VEC_EPILOG_PH]] ], [ [[VEC_IND_NEXT12:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP8:%.*]] = icmp eq <4 x i32> [[VEC_IND11]], [[BROADCAST_SPLAT14]]
+; CHECK-NEXT: [[TMP10]] = select <4 x i1> [[TMP8]], <4 x i1> [[VEC_PHI10]], <4 x i1> zeroinitializer
+; CHECK-NEXT: [[INDEX_NEXT15]] = add nuw i64 [[INDEX9]], 4
+; CHECK-NEXT: [[VEC_IND_NEXT12]] = add <4 x i32> [[VEC_IND11]], <i32 4, i32 4, i32 4, i32 4>
+; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC3]]
+; CHECK-NEXT: br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK: vec.epilog.middle.block:
+; CHECK-NEXT: [[RDX_SELECT_CMP16:%.*]] = icmp ne <4 x i1> [[TMP10]], zeroinitializer
+; CHECK-NEXT: [[TMP13:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[RDX_SELECT_CMP16]])
+; CHECK-NEXT: [[RDX_SELECT16:%.*]] = select i1 [[TMP13]], i1 false, i1 false
+; CHECK-NEXT: [[CMP_N8:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC3]]
+; CHECK-NEXT: br i1 [[CMP_N8]], label [[EXIT]], label [[VEC_EPILOG_SCALAR_PH]]
+; CHECK: vec.epilog.scalar.ph:
+; CHECK-NEXT: [[BC_RESUME_VAL4:%.*]] = phi i64 [ [[N_VEC3]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
+; CHECK-NEXT: [[BC_RESUME_VAL7:%.*]] = phi i32 [ [[IND_END5]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[IND_END6]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK]] ]
+; CHECK-NEXT: [[BC_MERGE_RDX17:%.*]] = phi i1 [ false, [[ITER_CHECK]] ], [ [[RDX_SELECT]], [[VEC_EPILOG_ITER_CHECK]] ], [ [[RDX_SELECT16]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; CHECK-NEXT: br label [[LOOP:%.*]]
+; CHECK: loop:
+; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL4]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[RED_I1:%.*]] = phi i1 [ [[BC_MERGE_RDX17]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[SEL:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[IV_2:%.*]] = phi i32 [ [[BC_RESUME_VAL7]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_2_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT: [[CMP_1:%.*]] = icmp eq i32 [[IV_2]], [[A]]
+; CHECK-NEXT: [[SEL]] = select i1 [[CMP_1]], i1 [[RED_I1]], i1 false
+; CHECK-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT: [[IV_2_NEXT]] = add i32 [[IV_2]], 1
+; CHECK-NEXT: [[CMP_2:%.*]] = icmp eq i64 [[IV]], [[N]]
+; CHECK-NEXT: br i1 [[CMP_2]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
+; CHECK: exit:
+; CHECK-NEXT: [[SEL_LCSSA:%.*]] = phi i1 [ [[SEL]], [[LOOP]] ], [ [[RDX_SELECT]], [[MIDDLE_BLOCK]] ], [ [[RDX_SELECT16]], [[VEC_EPILOG_MIDDLE_BLOCK]] ]
+; CHECK-NEXT: ret i1 [[SEL_LCSSA]]
+;
+entry:
+ br label %loop
+
+loop:
+ %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
+ %red.i1 = phi i1 [ false, %entry ], [ %sel, %loop ]
+ %iv.2 = phi i32 [ 0, %entry ], [ %iv.2.next, %loop ]
+ %cmp.1 = icmp eq i32 %iv.2, %a
+ %sel = select i1 %cmp.1, i1 %red.i1, i1 false
+ %iv.next = add i64 %iv, 1
+ %iv.2.next = add i32 %iv.2, 1
+ %cmp.2 = icmp eq i64 %iv, %N
+ br i1 %cmp.2, label %exit, label %loop
+
+exit:
+ ret i1 %sel
+
+; uselistorder directives
+ uselistorder i1 %sel, { 1, 0 }
+}
+;.
+; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
+; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
+; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
+; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META1]], [[META2]]}
+; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META2]], [[META1]]}
+; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META1]], [[META2]]}
+; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META1]], [[META2]]}
+; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META2]], [[META1]]}
+;.
More information about the llvm-commits
mailing list