caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
2417618d5ca4b151908df09d8e3a00a49f029222
clang-p2996/llvm/test/Transforms/LoopVectorize/RISCV/divrem.ll

1390 lines
86 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -loop-vectorize -scalable-vectorization=on -mtriple riscv64-linux-gnu -mattr=+v,+f -S 2>%t | FileCheck %s
; RUN: opt < %s -loop-vectorize -scalable-vectorization=off -riscv-v-vector-bits-min=-1 -mtriple riscv64-linux-gnu -mattr=+v,+f -S 2>%t | FileCheck --check-prefix=FIXED %s
; Tests specific to div/rem handling - both predicated and not
target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n64-S128"
target triple = "riscv64"
define void @vector_udiv(ptr noalias nocapture %a, i64 %v, i64 %n) {
; CHECK-LABEL: @vector_udiv(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP0]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 1 x i64> poison, i64 [[V:%.*]], i32 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 1 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = udiv <vscale x 1 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: store <vscale x 1 x i64> [[TMP5]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[DIVREM:%.*]] = udiv i64 [[ELEM]], [[V]]
; CHECK-NEXT: store i64 [[DIVREM]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @vector_udiv(
; FIXED-NEXT: entry:
; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[V:%.*]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[V]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
; FIXED-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
; FIXED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; FIXED-NEXT: [[TMP6:%.*]] = udiv <2 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; FIXED-NEXT: [[TMP7:%.*]] = udiv <2 x i64> [[WIDE_LOAD1]], [[BROADCAST_SPLAT3]]
; FIXED-NEXT: store <2 x i64> [[TMP6]], ptr [[TMP4]], align 8
; FIXED-NEXT: store <2 x i64> [[TMP7]], ptr [[TMP5]], align 8
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; FIXED-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; FIXED: scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[DIVREM:%.*]] = udiv i64 [[ELEM]], [[V]]
; FIXED-NEXT: store i64 [[DIVREM]], ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%divrem = udiv i64 %elem, %v
store i64 %divrem, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @vector_sdiv(ptr noalias nocapture %a, i64 %v, i64 %n) {
; CHECK-LABEL: @vector_sdiv(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP0]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 1 x i64> poison, i64 [[V:%.*]], i32 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 1 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = sdiv <vscale x 1 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: store <vscale x 1 x i64> [[TMP5]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[DIVREM:%.*]] = sdiv i64 [[ELEM]], [[V]]
; CHECK-NEXT: store i64 [[DIVREM]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @vector_sdiv(
; FIXED-NEXT: entry:
; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[V:%.*]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[V]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
; FIXED-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
; FIXED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; FIXED-NEXT: [[TMP6:%.*]] = sdiv <2 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; FIXED-NEXT: [[TMP7:%.*]] = sdiv <2 x i64> [[WIDE_LOAD1]], [[BROADCAST_SPLAT3]]
; FIXED-NEXT: store <2 x i64> [[TMP6]], ptr [[TMP4]], align 8
; FIXED-NEXT: store <2 x i64> [[TMP7]], ptr [[TMP5]], align 8
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; FIXED-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; FIXED: scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[DIVREM:%.*]] = sdiv i64 [[ELEM]], [[V]]
; FIXED-NEXT: store i64 [[DIVREM]], ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%divrem = sdiv i64 %elem, %v
store i64 %divrem, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @vector_urem(ptr noalias nocapture %a, i64 %v, i64 %n) {
; CHECK-LABEL: @vector_urem(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP0]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 1 x i64> poison, i64 [[V:%.*]], i32 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 1 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = urem <vscale x 1 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: store <vscale x 1 x i64> [[TMP5]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[DIVREM:%.*]] = urem i64 [[ELEM]], [[V]]
; CHECK-NEXT: store i64 [[DIVREM]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @vector_urem(
; FIXED-NEXT: entry:
; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[V:%.*]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[V]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
; FIXED-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
; FIXED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; FIXED-NEXT: [[TMP6:%.*]] = urem <2 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; FIXED-NEXT: [[TMP7:%.*]] = urem <2 x i64> [[WIDE_LOAD1]], [[BROADCAST_SPLAT3]]
; FIXED-NEXT: store <2 x i64> [[TMP6]], ptr [[TMP4]], align 8
; FIXED-NEXT: store <2 x i64> [[TMP7]], ptr [[TMP5]], align 8
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; FIXED-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; FIXED: scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[DIVREM:%.*]] = urem i64 [[ELEM]], [[V]]
; FIXED-NEXT: store i64 [[DIVREM]], ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%divrem = urem i64 %elem, %v
store i64 %divrem, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @vector_srem(ptr noalias nocapture %a, i64 %v, i64 %n) {
; CHECK-LABEL: @vector_srem(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP0]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 1 x i64> poison, i64 [[V:%.*]], i32 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 1 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = srem <vscale x 1 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: store <vscale x 1 x i64> [[TMP5]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[DIVREM:%.*]] = srem i64 [[ELEM]], [[V]]
; CHECK-NEXT: store i64 [[DIVREM]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @vector_srem(
; FIXED-NEXT: entry:
; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[V:%.*]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[V]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
; FIXED-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
; FIXED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; FIXED-NEXT: [[TMP6:%.*]] = srem <2 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; FIXED-NEXT: [[TMP7:%.*]] = srem <2 x i64> [[WIDE_LOAD1]], [[BROADCAST_SPLAT3]]
; FIXED-NEXT: store <2 x i64> [[TMP6]], ptr [[TMP4]], align 8
; FIXED-NEXT: store <2 x i64> [[TMP7]], ptr [[TMP5]], align 8
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; FIXED-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; FIXED: scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[DIVREM:%.*]] = srem i64 [[ELEM]], [[V]]
; FIXED-NEXT: store i64 [[DIVREM]], ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%divrem = srem i64 %elem, %v
store i64 %divrem, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @predicated_udiv(ptr noalias nocapture %a, i64 %v, i64 %n) {
; CHECK-LABEL: @predicated_udiv(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP0]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 1 x i64> poison, i64 [[V:%.*]], i32 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 1 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <vscale x 1 x i64> [[BROADCAST_SPLAT]], zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = select <vscale x 1 x i1> [[TMP5]], <vscale x 1 x i64> [[BROADCAST_SPLAT]], <vscale x 1 x i64> shufflevector (<vscale x 1 x i64> insertelement (<vscale x 1 x i64> poison, i64 1, i32 0), <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP7:%.*]] = udiv <vscale x 1 x i64> [[WIDE_LOAD]], [[TMP6]]
; CHECK-NEXT: [[TMP8:%.*]] = xor <vscale x 1 x i1> [[TMP5]], shufflevector (<vscale x 1 x i1> insertelement (<vscale x 1 x i1> poison, i1 true, i32 0), <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[PREDPHI:%.*]] = select <vscale x 1 x i1> [[TMP5]], <vscale x 1 x i64> [[TMP7]], <vscale x 1 x i64> [[WIDE_LOAD]]
; CHECK-NEXT: store <vscale x 1 x i64> [[PREDPHI]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[C:%.*]] = icmp ne i64 [[V]], 0
; CHECK-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; CHECK: do_op:
; CHECK-NEXT: [[DIVREM:%.*]] = udiv i64 [[ELEM]], [[V]]
; CHECK-NEXT: br label [[LATCH]]
; CHECK: latch:
; CHECK-NEXT: [[PHI:%.*]] = phi i64 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; CHECK-NEXT: store i64 [[PHI]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @predicated_udiv(
; FIXED-NEXT: entry:
; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[V:%.*]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[V]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_UDIV_CONTINUE9:%.*]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
; FIXED-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
; FIXED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; FIXED-NEXT: [[TMP6:%.*]] = icmp ne <2 x i64> [[BROADCAST_SPLAT]], zeroinitializer
; FIXED-NEXT: [[TMP7:%.*]] = icmp ne <2 x i64> [[BROADCAST_SPLAT3]], zeroinitializer
; FIXED-NEXT: [[TMP8:%.*]] = extractelement <2 x i1> [[TMP6]], i32 0
; FIXED-NEXT: br i1 [[TMP8]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
; FIXED: pred.udiv.if:
; FIXED-NEXT: [[TMP9:%.*]] = extractelement <2 x i64> [[WIDE_LOAD]], i32 0
; FIXED-NEXT: [[TMP10:%.*]] = udiv i64 [[TMP9]], [[V]]
; FIXED-NEXT: [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
; FIXED-NEXT: br label [[PRED_UDIV_CONTINUE]]
; FIXED: pred.udiv.continue:
; FIXED-NEXT: [[TMP12:%.*]] = phi <2 x i64> [ poison, [[VECTOR_BODY]] ], [ [[TMP11]], [[PRED_UDIV_IF]] ]
; FIXED-NEXT: [[TMP13:%.*]] = extractelement <2 x i1> [[TMP6]], i32 1
; FIXED-NEXT: br i1 [[TMP13]], label [[PRED_UDIV_IF4:%.*]], label [[PRED_UDIV_CONTINUE5:%.*]]
; FIXED: pred.udiv.if4:
; FIXED-NEXT: [[TMP14:%.*]] = extractelement <2 x i64> [[WIDE_LOAD]], i32 1
; FIXED-NEXT: [[TMP15:%.*]] = udiv i64 [[TMP14]], [[V]]
; FIXED-NEXT: [[TMP16:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP15]], i32 1
; FIXED-NEXT: br label [[PRED_UDIV_CONTINUE5]]
; FIXED: pred.udiv.continue5:
; FIXED-NEXT: [[TMP17:%.*]] = phi <2 x i64> [ [[TMP12]], [[PRED_UDIV_CONTINUE]] ], [ [[TMP16]], [[PRED_UDIV_IF4]] ]
; FIXED-NEXT: [[TMP18:%.*]] = extractelement <2 x i1> [[TMP7]], i32 0
; FIXED-NEXT: br i1 [[TMP18]], label [[PRED_UDIV_IF6:%.*]], label [[PRED_UDIV_CONTINUE7:%.*]]
; FIXED: pred.udiv.if6:
; FIXED-NEXT: [[TMP19:%.*]] = extractelement <2 x i64> [[WIDE_LOAD1]], i32 0
; FIXED-NEXT: [[TMP20:%.*]] = udiv i64 [[TMP19]], [[V]]
; FIXED-NEXT: [[TMP21:%.*]] = insertelement <2 x i64> poison, i64 [[TMP20]], i32 0
; FIXED-NEXT: br label [[PRED_UDIV_CONTINUE7]]
; FIXED: pred.udiv.continue7:
; FIXED-NEXT: [[TMP22:%.*]] = phi <2 x i64> [ poison, [[PRED_UDIV_CONTINUE5]] ], [ [[TMP21]], [[PRED_UDIV_IF6]] ]
; FIXED-NEXT: [[TMP23:%.*]] = extractelement <2 x i1> [[TMP7]], i32 1
; FIXED-NEXT: br i1 [[TMP23]], label [[PRED_UDIV_IF8:%.*]], label [[PRED_UDIV_CONTINUE9]]
; FIXED: pred.udiv.if8:
; FIXED-NEXT: [[TMP24:%.*]] = extractelement <2 x i64> [[WIDE_LOAD1]], i32 1
; FIXED-NEXT: [[TMP25:%.*]] = udiv i64 [[TMP24]], [[V]]
; FIXED-NEXT: [[TMP26:%.*]] = insertelement <2 x i64> [[TMP22]], i64 [[TMP25]], i32 1
; FIXED-NEXT: br label [[PRED_UDIV_CONTINUE9]]
; FIXED: pred.udiv.continue9:
; FIXED-NEXT: [[TMP27:%.*]] = phi <2 x i64> [ [[TMP22]], [[PRED_UDIV_CONTINUE7]] ], [ [[TMP26]], [[PRED_UDIV_IF8]] ]
; FIXED-NEXT: [[TMP28:%.*]] = xor <2 x i1> [[TMP6]], <i1 true, i1 true>
; FIXED-NEXT: [[TMP29:%.*]] = xor <2 x i1> [[TMP7]], <i1 true, i1 true>
; FIXED-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP6]], <2 x i64> [[TMP17]], <2 x i64> [[WIDE_LOAD]]
; FIXED-NEXT: [[PREDPHI10:%.*]] = select <2 x i1> [[TMP7]], <2 x i64> [[TMP27]], <2 x i64> [[WIDE_LOAD1]]
; FIXED-NEXT: [[TMP30:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: store <2 x i64> [[PREDPHI]], ptr [[TMP30]], align 8
; FIXED-NEXT: [[TMP31:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: store <2 x i64> [[PREDPHI10]], ptr [[TMP31]], align 8
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; FIXED-NEXT: [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; FIXED: scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[C:%.*]] = icmp ne i64 [[V]], 0
; FIXED-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; FIXED: do_op:
; FIXED-NEXT: [[DIVREM:%.*]] = udiv i64 [[ELEM]], [[V]]
; FIXED-NEXT: br label [[LATCH]]
; FIXED: latch:
; FIXED-NEXT: [[PHI:%.*]] = phi i64 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; FIXED-NEXT: store i64 [[PHI]], ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%c = icmp ne i64 %v, 0
br i1 %c, label %do_op, label %latch
do_op:
%divrem = udiv i64 %elem, %v
br label %latch
latch:
%phi = phi i64 [%elem, %for.body], [%divrem, %do_op]
store i64 %phi, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @predicated_sdiv(ptr noalias nocapture %a, i64 %v, i64 %n) {
; CHECK-LABEL: @predicated_sdiv(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP0]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 1 x i64> poison, i64 [[V:%.*]], i32 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 1 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <vscale x 1 x i64> [[BROADCAST_SPLAT]], zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = select <vscale x 1 x i1> [[TMP5]], <vscale x 1 x i64> [[BROADCAST_SPLAT]], <vscale x 1 x i64> shufflevector (<vscale x 1 x i64> insertelement (<vscale x 1 x i64> poison, i64 1, i32 0), <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP7:%.*]] = sdiv <vscale x 1 x i64> [[WIDE_LOAD]], [[TMP6]]
; CHECK-NEXT: [[TMP8:%.*]] = xor <vscale x 1 x i1> [[TMP5]], shufflevector (<vscale x 1 x i1> insertelement (<vscale x 1 x i1> poison, i1 true, i32 0), <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[PREDPHI:%.*]] = select <vscale x 1 x i1> [[TMP5]], <vscale x 1 x i64> [[TMP7]], <vscale x 1 x i64> [[WIDE_LOAD]]
; CHECK-NEXT: store <vscale x 1 x i64> [[PREDPHI]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[C:%.*]] = icmp ne i64 [[V]], 0
; CHECK-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; CHECK: do_op:
; CHECK-NEXT: [[DIVREM:%.*]] = sdiv i64 [[ELEM]], [[V]]
; CHECK-NEXT: br label [[LATCH]]
; CHECK: latch:
; CHECK-NEXT: [[PHI:%.*]] = phi i64 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; CHECK-NEXT: store i64 [[PHI]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @predicated_sdiv(
; FIXED-NEXT: entry:
; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[V:%.*]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[V]], i32 0
; FIXED-NEXT: [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_SDIV_CONTINUE9:%.*]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
; FIXED-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
; FIXED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; FIXED-NEXT: [[TMP6:%.*]] = icmp ne <2 x i64> [[BROADCAST_SPLAT]], zeroinitializer
; FIXED-NEXT: [[TMP7:%.*]] = icmp ne <2 x i64> [[BROADCAST_SPLAT3]], zeroinitializer
; FIXED-NEXT: [[TMP8:%.*]] = extractelement <2 x i1> [[TMP6]], i32 0
; FIXED-NEXT: br i1 [[TMP8]], label [[PRED_SDIV_IF:%.*]], label [[PRED_SDIV_CONTINUE:%.*]]
; FIXED: pred.sdiv.if:
; FIXED-NEXT: [[TMP9:%.*]] = extractelement <2 x i64> [[WIDE_LOAD]], i32 0
; FIXED-NEXT: [[TMP10:%.*]] = sdiv i64 [[TMP9]], [[V]]
; FIXED-NEXT: [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE]]
; FIXED: pred.sdiv.continue:
; FIXED-NEXT: [[TMP12:%.*]] = phi <2 x i64> [ poison, [[VECTOR_BODY]] ], [ [[TMP11]], [[PRED_SDIV_IF]] ]
; FIXED-NEXT: [[TMP13:%.*]] = extractelement <2 x i1> [[TMP6]], i32 1
; FIXED-NEXT: br i1 [[TMP13]], label [[PRED_SDIV_IF4:%.*]], label [[PRED_SDIV_CONTINUE5:%.*]]
; FIXED: pred.sdiv.if4:
; FIXED-NEXT: [[TMP14:%.*]] = extractelement <2 x i64> [[WIDE_LOAD]], i32 1
; FIXED-NEXT: [[TMP15:%.*]] = sdiv i64 [[TMP14]], [[V]]
; FIXED-NEXT: [[TMP16:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP15]], i32 1
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE5]]
; FIXED: pred.sdiv.continue5:
; FIXED-NEXT: [[TMP17:%.*]] = phi <2 x i64> [ [[TMP12]], [[PRED_SDIV_CONTINUE]] ], [ [[TMP16]], [[PRED_SDIV_IF4]] ]
; FIXED-NEXT: [[TMP18:%.*]] = extractelement <2 x i1> [[TMP7]], i32 0
; FIXED-NEXT: br i1 [[TMP18]], label [[PRED_SDIV_IF6:%.*]], label [[PRED_SDIV_CONTINUE7:%.*]]
; FIXED: pred.sdiv.if6:
; FIXED-NEXT: [[TMP19:%.*]] = extractelement <2 x i64> [[WIDE_LOAD1]], i32 0
; FIXED-NEXT: [[TMP20:%.*]] = sdiv i64 [[TMP19]], [[V]]
; FIXED-NEXT: [[TMP21:%.*]] = insertelement <2 x i64> poison, i64 [[TMP20]], i32 0
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE7]]
; FIXED: pred.sdiv.continue7:
; FIXED-NEXT: [[TMP22:%.*]] = phi <2 x i64> [ poison, [[PRED_SDIV_CONTINUE5]] ], [ [[TMP21]], [[PRED_SDIV_IF6]] ]
; FIXED-NEXT: [[TMP23:%.*]] = extractelement <2 x i1> [[TMP7]], i32 1
; FIXED-NEXT: br i1 [[TMP23]], label [[PRED_SDIV_IF8:%.*]], label [[PRED_SDIV_CONTINUE9]]
; FIXED: pred.sdiv.if8:
; FIXED-NEXT: [[TMP24:%.*]] = extractelement <2 x i64> [[WIDE_LOAD1]], i32 1
; FIXED-NEXT: [[TMP25:%.*]] = sdiv i64 [[TMP24]], [[V]]
; FIXED-NEXT: [[TMP26:%.*]] = insertelement <2 x i64> [[TMP22]], i64 [[TMP25]], i32 1
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE9]]
; FIXED: pred.sdiv.continue9:
; FIXED-NEXT: [[TMP27:%.*]] = phi <2 x i64> [ [[TMP22]], [[PRED_SDIV_CONTINUE7]] ], [ [[TMP26]], [[PRED_SDIV_IF8]] ]
; FIXED-NEXT: [[TMP28:%.*]] = xor <2 x i1> [[TMP6]], <i1 true, i1 true>
; FIXED-NEXT: [[TMP29:%.*]] = xor <2 x i1> [[TMP7]], <i1 true, i1 true>
; FIXED-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP6]], <2 x i64> [[TMP17]], <2 x i64> [[WIDE_LOAD]]
; FIXED-NEXT: [[PREDPHI10:%.*]] = select <2 x i1> [[TMP7]], <2 x i64> [[TMP27]], <2 x i64> [[WIDE_LOAD1]]
; FIXED-NEXT: [[TMP30:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: store <2 x i64> [[PREDPHI]], ptr [[TMP30]], align 8
; FIXED-NEXT: [[TMP31:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: store <2 x i64> [[PREDPHI10]], ptr [[TMP31]], align 8
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; FIXED-NEXT: [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; FIXED: scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[C:%.*]] = icmp ne i64 [[V]], 0
; FIXED-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; FIXED: do_op:
; FIXED-NEXT: [[DIVREM:%.*]] = sdiv i64 [[ELEM]], [[V]]
; FIXED-NEXT: br label [[LATCH]]
; FIXED: latch:
; FIXED-NEXT: [[PHI:%.*]] = phi i64 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; FIXED-NEXT: store i64 [[PHI]], ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%c = icmp ne i64 %v, 0
br i1 %c, label %do_op, label %latch
do_op:
%divrem = sdiv i64 %elem, %v
br label %latch
latch:
%phi = phi i64 [%elem, %for.body], [%divrem, %do_op]
store i64 %phi, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @predicated_udiv_by_constant(ptr noalias nocapture %a, i64 %n) {
; CHECK-LABEL: @predicated_udiv_by_constant(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP0]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[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: [[TMP2:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <vscale x 1 x i64> [[WIDE_LOAD]], shufflevector (<vscale x 1 x i64> insertelement (<vscale x 1 x i64> poison, i64 42, i32 0), <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP6:%.*]] = udiv <vscale x 1 x i64> [[WIDE_LOAD]], shufflevector (<vscale x 1 x i64> insertelement (<vscale x 1 x i64> poison, i64 27, i32 0), <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP7:%.*]] = xor <vscale x 1 x i1> [[TMP5]], shufflevector (<vscale x 1 x i1> insertelement (<vscale x 1 x i1> poison, i1 true, i32 0), <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[PREDPHI:%.*]] = select <vscale x 1 x i1> [[TMP5]], <vscale x 1 x i64> [[TMP6]], <vscale x 1 x i64> [[WIDE_LOAD]]
; CHECK-NEXT: store <vscale x 1 x i64> [[PREDPHI]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[C:%.*]] = icmp ne i64 [[ELEM]], 42
; CHECK-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; CHECK: do_op:
; CHECK-NEXT: [[DIVREM:%.*]] = udiv i64 [[ELEM]], 27
; CHECK-NEXT: br label [[LATCH]]
; CHECK: latch:
; CHECK-NEXT: [[PHI:%.*]] = phi i64 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; CHECK-NEXT: store i64 [[PHI]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP15:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @predicated_udiv_by_constant(
; FIXED-NEXT: entry:
; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
; FIXED-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
; FIXED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; FIXED-NEXT: [[TMP6:%.*]] = icmp ne <2 x i64> [[WIDE_LOAD]], <i64 42, i64 42>
; FIXED-NEXT: [[TMP7:%.*]] = icmp ne <2 x i64> [[WIDE_LOAD1]], <i64 42, i64 42>
; FIXED-NEXT: [[TMP8:%.*]] = udiv <2 x i64> [[WIDE_LOAD]], <i64 27, i64 27>
; FIXED-NEXT: [[TMP9:%.*]] = udiv <2 x i64> [[WIDE_LOAD1]], <i64 27, i64 27>
; FIXED-NEXT: [[TMP10:%.*]] = xor <2 x i1> [[TMP6]], <i1 true, i1 true>
; FIXED-NEXT: [[TMP11:%.*]] = xor <2 x i1> [[TMP7]], <i1 true, i1 true>
; FIXED-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP6]], <2 x i64> [[TMP8]], <2 x i64> [[WIDE_LOAD]]
; FIXED-NEXT: [[PREDPHI2:%.*]] = select <2 x i1> [[TMP7]], <2 x i64> [[TMP9]], <2 x i64> [[WIDE_LOAD1]]
; FIXED-NEXT: store <2 x i64> [[PREDPHI]], ptr [[TMP4]], align 8
; FIXED-NEXT: store <2 x i64> [[PREDPHI2]], ptr [[TMP5]], align 8
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; FIXED-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; FIXED: scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[C:%.*]] = icmp ne i64 [[ELEM]], 42
; FIXED-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; FIXED: do_op:
; FIXED-NEXT: [[DIVREM:%.*]] = udiv i64 [[ELEM]], 27
; FIXED-NEXT: br label [[LATCH]]
; FIXED: latch:
; FIXED-NEXT: [[PHI:%.*]] = phi i64 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; FIXED-NEXT: store i64 [[PHI]], ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP15:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%c = icmp ne i64 %elem, 42
br i1 %c, label %do_op, label %latch
do_op:
%divrem = udiv i64 %elem, 27
br label %latch
latch:
%phi = phi i64 [%elem, %for.body], [%divrem, %do_op]
store i64 %phi, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @predicated_sdiv_by_constant(ptr noalias nocapture %a, i64 %n) {
; CHECK-LABEL: @predicated_sdiv_by_constant(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP0]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP1]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[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: [[TMP2:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <vscale x 1 x i64> [[WIDE_LOAD]], shufflevector (<vscale x 1 x i64> insertelement (<vscale x 1 x i64> poison, i64 42, i32 0), <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP6:%.*]] = sdiv <vscale x 1 x i64> [[WIDE_LOAD]], shufflevector (<vscale x 1 x i64> insertelement (<vscale x 1 x i64> poison, i64 27, i32 0), <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP7:%.*]] = xor <vscale x 1 x i1> [[TMP5]], shufflevector (<vscale x 1 x i1> insertelement (<vscale x 1 x i1> poison, i1 true, i32 0), <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT: [[PREDPHI:%.*]] = select <vscale x 1 x i1> [[TMP5]], <vscale x 1 x i64> [[TMP6]], <vscale x 1 x i64> [[WIDE_LOAD]]
; CHECK-NEXT: store <vscale x 1 x i64> [[PREDPHI]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[C:%.*]] = icmp ne i64 [[ELEM]], 42
; CHECK-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; CHECK: do_op:
; CHECK-NEXT: [[DIVREM:%.*]] = sdiv i64 [[ELEM]], 27
; CHECK-NEXT: br label [[LATCH]]
; CHECK: latch:
; CHECK-NEXT: [[PHI:%.*]] = phi i64 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; CHECK-NEXT: store i64 [[PHI]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP17:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @predicated_sdiv_by_constant(
; FIXED-NEXT: entry:
; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
; FIXED-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
; FIXED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
; FIXED-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; FIXED-NEXT: [[TMP6:%.*]] = icmp ne <2 x i64> [[WIDE_LOAD]], <i64 42, i64 42>
; FIXED-NEXT: [[TMP7:%.*]] = icmp ne <2 x i64> [[WIDE_LOAD1]], <i64 42, i64 42>
; FIXED-NEXT: [[TMP8:%.*]] = sdiv <2 x i64> [[WIDE_LOAD]], <i64 27, i64 27>
; FIXED-NEXT: [[TMP9:%.*]] = sdiv <2 x i64> [[WIDE_LOAD1]], <i64 27, i64 27>
; FIXED-NEXT: [[TMP10:%.*]] = xor <2 x i1> [[TMP6]], <i1 true, i1 true>
; FIXED-NEXT: [[TMP11:%.*]] = xor <2 x i1> [[TMP7]], <i1 true, i1 true>
; FIXED-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP6]], <2 x i64> [[TMP8]], <2 x i64> [[WIDE_LOAD]]
; FIXED-NEXT: [[PREDPHI2:%.*]] = select <2 x i1> [[TMP7]], <2 x i64> [[TMP9]], <2 x i64> [[WIDE_LOAD1]]
; FIXED-NEXT: store <2 x i64> [[PREDPHI]], ptr [[TMP4]], align 8
; FIXED-NEXT: store <2 x i64> [[PREDPHI2]], ptr [[TMP5]], align 8
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; FIXED-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; FIXED: scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[C:%.*]] = icmp ne i64 [[ELEM]], 42
; FIXED-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; FIXED: do_op:
; FIXED-NEXT: [[DIVREM:%.*]] = sdiv i64 [[ELEM]], 27
; FIXED-NEXT: br label [[LATCH]]
; FIXED: latch:
; FIXED-NEXT: [[PHI:%.*]] = phi i64 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; FIXED-NEXT: store i64 [[PHI]], ptr [[ARRAYIDX]], align 8
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP17:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%c = icmp ne i64 %elem, 42
br i1 %c, label %do_op, label %latch
do_op:
%divrem = sdiv i64 %elem, 27
br label %latch
latch:
%phi = phi i64 [%elem, %for.body], [%divrem, %do_op]
store i64 %phi, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @predicated_sdiv_by_minus_one(ptr noalias nocapture %a, i64 %n) {
; CHECK-LABEL: @predicated_sdiv_by_minus_one(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 16
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 16
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[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: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP6:%.*]] = mul i64 [[TMP5]], 8
; CHECK-NEXT: [[TMP7:%.*]] = add i64 [[TMP6]], 0
; CHECK-NEXT: [[TMP8:%.*]] = mul i64 [[TMP7]], 1
; CHECK-NEXT: [[TMP9:%.*]] = add i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x i8>, ptr [[TMP12]], align 1
; CHECK-NEXT: [[TMP13:%.*]] = call i32 @llvm.vscale.i32()
; CHECK-NEXT: [[TMP14:%.*]] = mul i32 [[TMP13]], 8
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i32 [[TMP14]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 8 x i8>, ptr [[TMP15]], align 1
; CHECK-NEXT: [[TMP16:%.*]] = icmp ne <vscale x 8 x i8> [[WIDE_LOAD]], shufflevector (<vscale x 8 x i8> insertelement (<vscale x 8 x i8> poison, i8 -128, i32 0), <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP17:%.*]] = icmp ne <vscale x 8 x i8> [[WIDE_LOAD1]], shufflevector (<vscale x 8 x i8> insertelement (<vscale x 8 x i8> poison, i8 -128, i32 0), <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP18:%.*]] = select <vscale x 8 x i1> [[TMP16]], <vscale x 8 x i8> shufflevector (<vscale x 8 x i8> insertelement (<vscale x 8 x i8> poison, i8 -1, i32 0), <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer), <vscale x 8 x i8> shufflevector (<vscale x 8 x i8> insertelement (<vscale x 8 x i8> poison, i8 1, i32 0), <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP19:%.*]] = select <vscale x 8 x i1> [[TMP17]], <vscale x 8 x i8> shufflevector (<vscale x 8 x i8> insertelement (<vscale x 8 x i8> poison, i8 -1, i32 0), <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer), <vscale x 8 x i8> shufflevector (<vscale x 8 x i8> insertelement (<vscale x 8 x i8> poison, i8 1, i32 0), <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP20:%.*]] = sdiv <vscale x 8 x i8> [[WIDE_LOAD]], [[TMP18]]
; CHECK-NEXT: [[TMP21:%.*]] = sdiv <vscale x 8 x i8> [[WIDE_LOAD1]], [[TMP19]]
; CHECK-NEXT: [[TMP22:%.*]] = xor <vscale x 8 x i1> [[TMP16]], shufflevector (<vscale x 8 x i1> insertelement (<vscale x 8 x i1> poison, i1 true, i32 0), <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP23:%.*]] = xor <vscale x 8 x i1> [[TMP17]], shufflevector (<vscale x 8 x i1> insertelement (<vscale x 8 x i1> poison, i1 true, i32 0), <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer)
; CHECK-NEXT: [[PREDPHI:%.*]] = select <vscale x 8 x i1> [[TMP16]], <vscale x 8 x i8> [[TMP20]], <vscale x 8 x i8> [[WIDE_LOAD]]
; CHECK-NEXT: [[PREDPHI2:%.*]] = select <vscale x 8 x i1> [[TMP17]], <vscale x 8 x i8> [[TMP21]], <vscale x 8 x i8> [[WIDE_LOAD1]]
; CHECK-NEXT: store <vscale x 8 x i8> [[PREDPHI]], ptr [[TMP12]], align 1
; CHECK-NEXT: [[TMP24:%.*]] = call i32 @llvm.vscale.i32()
; CHECK-NEXT: [[TMP25:%.*]] = mul i32 [[TMP24]], 8
; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i32 [[TMP25]]
; CHECK-NEXT: store <vscale x 8 x i8> [[PREDPHI2]], ptr [[TMP26]], align 1
; CHECK-NEXT: [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP28:%.*]] = mul i64 [[TMP27]], 16
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP28]]
; CHECK-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[IV]]
; CHECK-NEXT: [[ELEM:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT: [[C:%.*]] = icmp ne i8 [[ELEM]], -128
; CHECK-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; CHECK: do_op:
; CHECK-NEXT: [[DIVREM:%.*]] = sdiv i8 [[ELEM]], -1
; CHECK-NEXT: br label [[LATCH]]
; CHECK: latch:
; CHECK-NEXT: [[PHI:%.*]] = phi i8 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; CHECK-NEXT: store i8 [[PHI]], ptr [[ARRAYIDX]], align 1
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP19:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
; FIXED-LABEL: @predicated_sdiv_by_minus_one(
; FIXED-NEXT: iter.check:
; FIXED-NEXT: br i1 false, label [[VEC_EPILOG_SCALAR_PH:%.*]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]]
; FIXED: vector.main.loop.iter.check:
; FIXED-NEXT: br i1 false, label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH:%.*]]
; FIXED: vector.ph:
; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
; FIXED: vector.body:
; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_SDIV_CONTINUE30:%.*]] ]
; FIXED-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; FIXED-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP0]]
; FIXED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i32 0
; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP2]], align 1
; FIXED-NEXT: [[TMP3:%.*]] = icmp ne <16 x i8> [[WIDE_LOAD]], <i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128>
; FIXED-NEXT: [[TMP4:%.*]] = extractelement <16 x i1> [[TMP3]], i32 0
; FIXED-NEXT: br i1 [[TMP4]], label [[PRED_SDIV_IF:%.*]], label [[PRED_SDIV_CONTINUE:%.*]]
; FIXED: pred.sdiv.if:
; FIXED-NEXT: [[TMP5:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 0
; FIXED-NEXT: [[TMP6:%.*]] = sdiv i8 [[TMP5]], -1
; FIXED-NEXT: [[TMP7:%.*]] = insertelement <16 x i8> poison, i8 [[TMP6]], i32 0
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE]]
; FIXED: pred.sdiv.continue:
; FIXED-NEXT: [[TMP8:%.*]] = phi <16 x i8> [ poison, [[VECTOR_BODY]] ], [ [[TMP7]], [[PRED_SDIV_IF]] ]
; FIXED-NEXT: [[TMP9:%.*]] = extractelement <16 x i1> [[TMP3]], i32 1
; FIXED-NEXT: br i1 [[TMP9]], label [[PRED_SDIV_IF1:%.*]], label [[PRED_SDIV_CONTINUE2:%.*]]
; FIXED: pred.sdiv.if1:
; FIXED-NEXT: [[TMP10:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 1
; FIXED-NEXT: [[TMP11:%.*]] = sdiv i8 [[TMP10]], -1
; FIXED-NEXT: [[TMP12:%.*]] = insertelement <16 x i8> [[TMP8]], i8 [[TMP11]], i32 1
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE2]]
; FIXED: pred.sdiv.continue2:
; FIXED-NEXT: [[TMP13:%.*]] = phi <16 x i8> [ [[TMP8]], [[PRED_SDIV_CONTINUE]] ], [ [[TMP12]], [[PRED_SDIV_IF1]] ]
; FIXED-NEXT: [[TMP14:%.*]] = extractelement <16 x i1> [[TMP3]], i32 2
; FIXED-NEXT: br i1 [[TMP14]], label [[PRED_SDIV_IF3:%.*]], label [[PRED_SDIV_CONTINUE4:%.*]]
; FIXED: pred.sdiv.if3:
; FIXED-NEXT: [[TMP15:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 2
; FIXED-NEXT: [[TMP16:%.*]] = sdiv i8 [[TMP15]], -1
; FIXED-NEXT: [[TMP17:%.*]] = insertelement <16 x i8> [[TMP13]], i8 [[TMP16]], i32 2
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE4]]
; FIXED: pred.sdiv.continue4:
; FIXED-NEXT: [[TMP18:%.*]] = phi <16 x i8> [ [[TMP13]], [[PRED_SDIV_CONTINUE2]] ], [ [[TMP17]], [[PRED_SDIV_IF3]] ]
; FIXED-NEXT: [[TMP19:%.*]] = extractelement <16 x i1> [[TMP3]], i32 3
; FIXED-NEXT: br i1 [[TMP19]], label [[PRED_SDIV_IF5:%.*]], label [[PRED_SDIV_CONTINUE6:%.*]]
; FIXED: pred.sdiv.if5:
; FIXED-NEXT: [[TMP20:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 3
; FIXED-NEXT: [[TMP21:%.*]] = sdiv i8 [[TMP20]], -1
; FIXED-NEXT: [[TMP22:%.*]] = insertelement <16 x i8> [[TMP18]], i8 [[TMP21]], i32 3
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE6]]
; FIXED: pred.sdiv.continue6:
; FIXED-NEXT: [[TMP23:%.*]] = phi <16 x i8> [ [[TMP18]], [[PRED_SDIV_CONTINUE4]] ], [ [[TMP22]], [[PRED_SDIV_IF5]] ]
; FIXED-NEXT: [[TMP24:%.*]] = extractelement <16 x i1> [[TMP3]], i32 4
; FIXED-NEXT: br i1 [[TMP24]], label [[PRED_SDIV_IF7:%.*]], label [[PRED_SDIV_CONTINUE8:%.*]]
; FIXED: pred.sdiv.if7:
; FIXED-NEXT: [[TMP25:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 4
; FIXED-NEXT: [[TMP26:%.*]] = sdiv i8 [[TMP25]], -1
; FIXED-NEXT: [[TMP27:%.*]] = insertelement <16 x i8> [[TMP23]], i8 [[TMP26]], i32 4
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE8]]
; FIXED: pred.sdiv.continue8:
; FIXED-NEXT: [[TMP28:%.*]] = phi <16 x i8> [ [[TMP23]], [[PRED_SDIV_CONTINUE6]] ], [ [[TMP27]], [[PRED_SDIV_IF7]] ]
; FIXED-NEXT: [[TMP29:%.*]] = extractelement <16 x i1> [[TMP3]], i32 5
; FIXED-NEXT: br i1 [[TMP29]], label [[PRED_SDIV_IF9:%.*]], label [[PRED_SDIV_CONTINUE10:%.*]]
; FIXED: pred.sdiv.if9:
; FIXED-NEXT: [[TMP30:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 5
; FIXED-NEXT: [[TMP31:%.*]] = sdiv i8 [[TMP30]], -1
; FIXED-NEXT: [[TMP32:%.*]] = insertelement <16 x i8> [[TMP28]], i8 [[TMP31]], i32 5
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE10]]
; FIXED: pred.sdiv.continue10:
; FIXED-NEXT: [[TMP33:%.*]] = phi <16 x i8> [ [[TMP28]], [[PRED_SDIV_CONTINUE8]] ], [ [[TMP32]], [[PRED_SDIV_IF9]] ]
; FIXED-NEXT: [[TMP34:%.*]] = extractelement <16 x i1> [[TMP3]], i32 6
; FIXED-NEXT: br i1 [[TMP34]], label [[PRED_SDIV_IF11:%.*]], label [[PRED_SDIV_CONTINUE12:%.*]]
; FIXED: pred.sdiv.if11:
; FIXED-NEXT: [[TMP35:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 6
; FIXED-NEXT: [[TMP36:%.*]] = sdiv i8 [[TMP35]], -1
; FIXED-NEXT: [[TMP37:%.*]] = insertelement <16 x i8> [[TMP33]], i8 [[TMP36]], i32 6
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE12]]
; FIXED: pred.sdiv.continue12:
; FIXED-NEXT: [[TMP38:%.*]] = phi <16 x i8> [ [[TMP33]], [[PRED_SDIV_CONTINUE10]] ], [ [[TMP37]], [[PRED_SDIV_IF11]] ]
; FIXED-NEXT: [[TMP39:%.*]] = extractelement <16 x i1> [[TMP3]], i32 7
; FIXED-NEXT: br i1 [[TMP39]], label [[PRED_SDIV_IF13:%.*]], label [[PRED_SDIV_CONTINUE14:%.*]]
; FIXED: pred.sdiv.if13:
; FIXED-NEXT: [[TMP40:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 7
; FIXED-NEXT: [[TMP41:%.*]] = sdiv i8 [[TMP40]], -1
; FIXED-NEXT: [[TMP42:%.*]] = insertelement <16 x i8> [[TMP38]], i8 [[TMP41]], i32 7
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE14]]
; FIXED: pred.sdiv.continue14:
; FIXED-NEXT: [[TMP43:%.*]] = phi <16 x i8> [ [[TMP38]], [[PRED_SDIV_CONTINUE12]] ], [ [[TMP42]], [[PRED_SDIV_IF13]] ]
; FIXED-NEXT: [[TMP44:%.*]] = extractelement <16 x i1> [[TMP3]], i32 8
; FIXED-NEXT: br i1 [[TMP44]], label [[PRED_SDIV_IF15:%.*]], label [[PRED_SDIV_CONTINUE16:%.*]]
; FIXED: pred.sdiv.if15:
; FIXED-NEXT: [[TMP45:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 8
; FIXED-NEXT: [[TMP46:%.*]] = sdiv i8 [[TMP45]], -1
; FIXED-NEXT: [[TMP47:%.*]] = insertelement <16 x i8> [[TMP43]], i8 [[TMP46]], i32 8
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE16]]
; FIXED: pred.sdiv.continue16:
; FIXED-NEXT: [[TMP48:%.*]] = phi <16 x i8> [ [[TMP43]], [[PRED_SDIV_CONTINUE14]] ], [ [[TMP47]], [[PRED_SDIV_IF15]] ]
; FIXED-NEXT: [[TMP49:%.*]] = extractelement <16 x i1> [[TMP3]], i32 9
; FIXED-NEXT: br i1 [[TMP49]], label [[PRED_SDIV_IF17:%.*]], label [[PRED_SDIV_CONTINUE18:%.*]]
; FIXED: pred.sdiv.if17:
; FIXED-NEXT: [[TMP50:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 9
; FIXED-NEXT: [[TMP51:%.*]] = sdiv i8 [[TMP50]], -1
; FIXED-NEXT: [[TMP52:%.*]] = insertelement <16 x i8> [[TMP48]], i8 [[TMP51]], i32 9
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE18]]
; FIXED: pred.sdiv.continue18:
; FIXED-NEXT: [[TMP53:%.*]] = phi <16 x i8> [ [[TMP48]], [[PRED_SDIV_CONTINUE16]] ], [ [[TMP52]], [[PRED_SDIV_IF17]] ]
; FIXED-NEXT: [[TMP54:%.*]] = extractelement <16 x i1> [[TMP3]], i32 10
; FIXED-NEXT: br i1 [[TMP54]], label [[PRED_SDIV_IF19:%.*]], label [[PRED_SDIV_CONTINUE20:%.*]]
; FIXED: pred.sdiv.if19:
; FIXED-NEXT: [[TMP55:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 10
; FIXED-NEXT: [[TMP56:%.*]] = sdiv i8 [[TMP55]], -1
; FIXED-NEXT: [[TMP57:%.*]] = insertelement <16 x i8> [[TMP53]], i8 [[TMP56]], i32 10
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE20]]
; FIXED: pred.sdiv.continue20:
; FIXED-NEXT: [[TMP58:%.*]] = phi <16 x i8> [ [[TMP53]], [[PRED_SDIV_CONTINUE18]] ], [ [[TMP57]], [[PRED_SDIV_IF19]] ]
; FIXED-NEXT: [[TMP59:%.*]] = extractelement <16 x i1> [[TMP3]], i32 11
; FIXED-NEXT: br i1 [[TMP59]], label [[PRED_SDIV_IF21:%.*]], label [[PRED_SDIV_CONTINUE22:%.*]]
; FIXED: pred.sdiv.if21:
; FIXED-NEXT: [[TMP60:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 11
; FIXED-NEXT: [[TMP61:%.*]] = sdiv i8 [[TMP60]], -1
; FIXED-NEXT: [[TMP62:%.*]] = insertelement <16 x i8> [[TMP58]], i8 [[TMP61]], i32 11
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE22]]
; FIXED: pred.sdiv.continue22:
; FIXED-NEXT: [[TMP63:%.*]] = phi <16 x i8> [ [[TMP58]], [[PRED_SDIV_CONTINUE20]] ], [ [[TMP62]], [[PRED_SDIV_IF21]] ]
; FIXED-NEXT: [[TMP64:%.*]] = extractelement <16 x i1> [[TMP3]], i32 12
; FIXED-NEXT: br i1 [[TMP64]], label [[PRED_SDIV_IF23:%.*]], label [[PRED_SDIV_CONTINUE24:%.*]]
; FIXED: pred.sdiv.if23:
; FIXED-NEXT: [[TMP65:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 12
; FIXED-NEXT: [[TMP66:%.*]] = sdiv i8 [[TMP65]], -1
; FIXED-NEXT: [[TMP67:%.*]] = insertelement <16 x i8> [[TMP63]], i8 [[TMP66]], i32 12
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE24]]
; FIXED: pred.sdiv.continue24:
; FIXED-NEXT: [[TMP68:%.*]] = phi <16 x i8> [ [[TMP63]], [[PRED_SDIV_CONTINUE22]] ], [ [[TMP67]], [[PRED_SDIV_IF23]] ]
; FIXED-NEXT: [[TMP69:%.*]] = extractelement <16 x i1> [[TMP3]], i32 13
; FIXED-NEXT: br i1 [[TMP69]], label [[PRED_SDIV_IF25:%.*]], label [[PRED_SDIV_CONTINUE26:%.*]]
; FIXED: pred.sdiv.if25:
; FIXED-NEXT: [[TMP70:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 13
; FIXED-NEXT: [[TMP71:%.*]] = sdiv i8 [[TMP70]], -1
; FIXED-NEXT: [[TMP72:%.*]] = insertelement <16 x i8> [[TMP68]], i8 [[TMP71]], i32 13
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE26]]
; FIXED: pred.sdiv.continue26:
; FIXED-NEXT: [[TMP73:%.*]] = phi <16 x i8> [ [[TMP68]], [[PRED_SDIV_CONTINUE24]] ], [ [[TMP72]], [[PRED_SDIV_IF25]] ]
; FIXED-NEXT: [[TMP74:%.*]] = extractelement <16 x i1> [[TMP3]], i32 14
; FIXED-NEXT: br i1 [[TMP74]], label [[PRED_SDIV_IF27:%.*]], label [[PRED_SDIV_CONTINUE28:%.*]]
; FIXED: pred.sdiv.if27:
; FIXED-NEXT: [[TMP75:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 14
; FIXED-NEXT: [[TMP76:%.*]] = sdiv i8 [[TMP75]], -1
; FIXED-NEXT: [[TMP77:%.*]] = insertelement <16 x i8> [[TMP73]], i8 [[TMP76]], i32 14
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE28]]
; FIXED: pred.sdiv.continue28:
; FIXED-NEXT: [[TMP78:%.*]] = phi <16 x i8> [ [[TMP73]], [[PRED_SDIV_CONTINUE26]] ], [ [[TMP77]], [[PRED_SDIV_IF27]] ]
; FIXED-NEXT: [[TMP79:%.*]] = extractelement <16 x i1> [[TMP3]], i32 15
; FIXED-NEXT: br i1 [[TMP79]], label [[PRED_SDIV_IF29:%.*]], label [[PRED_SDIV_CONTINUE30]]
; FIXED: pred.sdiv.if29:
; FIXED-NEXT: [[TMP80:%.*]] = extractelement <16 x i8> [[WIDE_LOAD]], i32 15
; FIXED-NEXT: [[TMP81:%.*]] = sdiv i8 [[TMP80]], -1
; FIXED-NEXT: [[TMP82:%.*]] = insertelement <16 x i8> [[TMP78]], i8 [[TMP81]], i32 15
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE30]]
; FIXED: pred.sdiv.continue30:
; FIXED-NEXT: [[TMP83:%.*]] = phi <16 x i8> [ [[TMP78]], [[PRED_SDIV_CONTINUE28]] ], [ [[TMP82]], [[PRED_SDIV_IF29]] ]
; FIXED-NEXT: [[TMP84:%.*]] = xor <16 x i1> [[TMP3]], <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>
; FIXED-NEXT: [[PREDPHI:%.*]] = select <16 x i1> [[TMP3]], <16 x i8> [[TMP83]], <16 x i8> [[WIDE_LOAD]]
; FIXED-NEXT: [[TMP85:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i32 0
; FIXED-NEXT: store <16 x i8> [[PREDPHI]], ptr [[TMP85]], align 1
; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; FIXED-NEXT: [[TMP86:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; FIXED-NEXT: br i1 [[TMP86]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
; FIXED: middle.block:
; FIXED-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
; FIXED: vec.epilog.iter.check:
; FIXED-NEXT: br i1 true, label [[VEC_EPILOG_SCALAR_PH]], label [[VEC_EPILOG_PH]]
; FIXED: vec.epilog.ph:
; FIXED-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ 1024, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
; FIXED-NEXT: br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
; FIXED: vec.epilog.vector.body:
; FIXED-NEXT: [[OFFSET_IDX:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT51:%.*]], [[PRED_SDIV_CONTINUE49:%.*]] ]
; FIXED-NEXT: [[TMP87:%.*]] = add i64 [[OFFSET_IDX]], 0
; FIXED-NEXT: [[TMP88:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP87]]
; FIXED-NEXT: [[TMP89:%.*]] = getelementptr inbounds i8, ptr [[TMP88]], i32 0
; FIXED-NEXT: [[WIDE_LOAD33:%.*]] = load <8 x i8>, ptr [[TMP89]], align 1
; FIXED-NEXT: [[TMP90:%.*]] = icmp ne <8 x i8> [[WIDE_LOAD33]], <i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128>
; FIXED-NEXT: [[TMP91:%.*]] = extractelement <8 x i1> [[TMP90]], i32 0
; FIXED-NEXT: br i1 [[TMP91]], label [[PRED_SDIV_IF34:%.*]], label [[PRED_SDIV_CONTINUE35:%.*]]
; FIXED: pred.sdiv.if34:
; FIXED-NEXT: [[TMP92:%.*]] = extractelement <8 x i8> [[WIDE_LOAD33]], i32 0
; FIXED-NEXT: [[TMP93:%.*]] = sdiv i8 [[TMP92]], -1
; FIXED-NEXT: [[TMP94:%.*]] = insertelement <8 x i8> poison, i8 [[TMP93]], i32 0
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE35]]
; FIXED: pred.sdiv.continue35:
; FIXED-NEXT: [[TMP95:%.*]] = phi <8 x i8> [ poison, [[VEC_EPILOG_VECTOR_BODY]] ], [ [[TMP94]], [[PRED_SDIV_IF34]] ]
; FIXED-NEXT: [[TMP96:%.*]] = extractelement <8 x i1> [[TMP90]], i32 1
; FIXED-NEXT: br i1 [[TMP96]], label [[PRED_SDIV_IF36:%.*]], label [[PRED_SDIV_CONTINUE37:%.*]]
; FIXED: pred.sdiv.if36:
; FIXED-NEXT: [[TMP97:%.*]] = extractelement <8 x i8> [[WIDE_LOAD33]], i32 1
; FIXED-NEXT: [[TMP98:%.*]] = sdiv i8 [[TMP97]], -1
; FIXED-NEXT: [[TMP99:%.*]] = insertelement <8 x i8> [[TMP95]], i8 [[TMP98]], i32 1
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE37]]
; FIXED: pred.sdiv.continue37:
; FIXED-NEXT: [[TMP100:%.*]] = phi <8 x i8> [ [[TMP95]], [[PRED_SDIV_CONTINUE35]] ], [ [[TMP99]], [[PRED_SDIV_IF36]] ]
; FIXED-NEXT: [[TMP101:%.*]] = extractelement <8 x i1> [[TMP90]], i32 2
; FIXED-NEXT: br i1 [[TMP101]], label [[PRED_SDIV_IF38:%.*]], label [[PRED_SDIV_CONTINUE39:%.*]]
; FIXED: pred.sdiv.if38:
; FIXED-NEXT: [[TMP102:%.*]] = extractelement <8 x i8> [[WIDE_LOAD33]], i32 2
; FIXED-NEXT: [[TMP103:%.*]] = sdiv i8 [[TMP102]], -1
; FIXED-NEXT: [[TMP104:%.*]] = insertelement <8 x i8> [[TMP100]], i8 [[TMP103]], i32 2
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE39]]
; FIXED: pred.sdiv.continue39:
; FIXED-NEXT: [[TMP105:%.*]] = phi <8 x i8> [ [[TMP100]], [[PRED_SDIV_CONTINUE37]] ], [ [[TMP104]], [[PRED_SDIV_IF38]] ]
; FIXED-NEXT: [[TMP106:%.*]] = extractelement <8 x i1> [[TMP90]], i32 3
; FIXED-NEXT: br i1 [[TMP106]], label [[PRED_SDIV_IF40:%.*]], label [[PRED_SDIV_CONTINUE41:%.*]]
; FIXED: pred.sdiv.if40:
; FIXED-NEXT: [[TMP107:%.*]] = extractelement <8 x i8> [[WIDE_LOAD33]], i32 3
; FIXED-NEXT: [[TMP108:%.*]] = sdiv i8 [[TMP107]], -1
; FIXED-NEXT: [[TMP109:%.*]] = insertelement <8 x i8> [[TMP105]], i8 [[TMP108]], i32 3
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE41]]
; FIXED: pred.sdiv.continue41:
; FIXED-NEXT: [[TMP110:%.*]] = phi <8 x i8> [ [[TMP105]], [[PRED_SDIV_CONTINUE39]] ], [ [[TMP109]], [[PRED_SDIV_IF40]] ]
; FIXED-NEXT: [[TMP111:%.*]] = extractelement <8 x i1> [[TMP90]], i32 4
; FIXED-NEXT: br i1 [[TMP111]], label [[PRED_SDIV_IF42:%.*]], label [[PRED_SDIV_CONTINUE43:%.*]]
; FIXED: pred.sdiv.if42:
; FIXED-NEXT: [[TMP112:%.*]] = extractelement <8 x i8> [[WIDE_LOAD33]], i32 4
; FIXED-NEXT: [[TMP113:%.*]] = sdiv i8 [[TMP112]], -1
; FIXED-NEXT: [[TMP114:%.*]] = insertelement <8 x i8> [[TMP110]], i8 [[TMP113]], i32 4
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE43]]
; FIXED: pred.sdiv.continue43:
; FIXED-NEXT: [[TMP115:%.*]] = phi <8 x i8> [ [[TMP110]], [[PRED_SDIV_CONTINUE41]] ], [ [[TMP114]], [[PRED_SDIV_IF42]] ]
; FIXED-NEXT: [[TMP116:%.*]] = extractelement <8 x i1> [[TMP90]], i32 5
; FIXED-NEXT: br i1 [[TMP116]], label [[PRED_SDIV_IF44:%.*]], label [[PRED_SDIV_CONTINUE45:%.*]]
; FIXED: pred.sdiv.if44:
; FIXED-NEXT: [[TMP117:%.*]] = extractelement <8 x i8> [[WIDE_LOAD33]], i32 5
; FIXED-NEXT: [[TMP118:%.*]] = sdiv i8 [[TMP117]], -1
; FIXED-NEXT: [[TMP119:%.*]] = insertelement <8 x i8> [[TMP115]], i8 [[TMP118]], i32 5
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE45]]
; FIXED: pred.sdiv.continue45:
; FIXED-NEXT: [[TMP120:%.*]] = phi <8 x i8> [ [[TMP115]], [[PRED_SDIV_CONTINUE43]] ], [ [[TMP119]], [[PRED_SDIV_IF44]] ]
; FIXED-NEXT: [[TMP121:%.*]] = extractelement <8 x i1> [[TMP90]], i32 6
; FIXED-NEXT: br i1 [[TMP121]], label [[PRED_SDIV_IF46:%.*]], label [[PRED_SDIV_CONTINUE47:%.*]]
; FIXED: pred.sdiv.if46:
; FIXED-NEXT: [[TMP122:%.*]] = extractelement <8 x i8> [[WIDE_LOAD33]], i32 6
; FIXED-NEXT: [[TMP123:%.*]] = sdiv i8 [[TMP122]], -1
; FIXED-NEXT: [[TMP124:%.*]] = insertelement <8 x i8> [[TMP120]], i8 [[TMP123]], i32 6
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE47]]
; FIXED: pred.sdiv.continue47:
; FIXED-NEXT: [[TMP125:%.*]] = phi <8 x i8> [ [[TMP120]], [[PRED_SDIV_CONTINUE45]] ], [ [[TMP124]], [[PRED_SDIV_IF46]] ]
; FIXED-NEXT: [[TMP126:%.*]] = extractelement <8 x i1> [[TMP90]], i32 7
; FIXED-NEXT: br i1 [[TMP126]], label [[PRED_SDIV_IF48:%.*]], label [[PRED_SDIV_CONTINUE49]]
; FIXED: pred.sdiv.if48:
; FIXED-NEXT: [[TMP127:%.*]] = extractelement <8 x i8> [[WIDE_LOAD33]], i32 7
; FIXED-NEXT: [[TMP128:%.*]] = sdiv i8 [[TMP127]], -1
; FIXED-NEXT: [[TMP129:%.*]] = insertelement <8 x i8> [[TMP125]], i8 [[TMP128]], i32 7
; FIXED-NEXT: br label [[PRED_SDIV_CONTINUE49]]
; FIXED: pred.sdiv.continue49:
; FIXED-NEXT: [[TMP130:%.*]] = phi <8 x i8> [ [[TMP125]], [[PRED_SDIV_CONTINUE47]] ], [ [[TMP129]], [[PRED_SDIV_IF48]] ]
; FIXED-NEXT: [[TMP131:%.*]] = xor <8 x i1> [[TMP90]], <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>
; FIXED-NEXT: [[PREDPHI50:%.*]] = select <8 x i1> [[TMP90]], <8 x i8> [[TMP130]], <8 x i8> [[WIDE_LOAD33]]
; FIXED-NEXT: [[TMP132:%.*]] = getelementptr inbounds i8, ptr [[TMP88]], i32 0
; FIXED-NEXT: store <8 x i8> [[PREDPHI50]], ptr [[TMP132]], align 1
; FIXED-NEXT: [[INDEX_NEXT51]] = add nuw i64 [[OFFSET_IDX]], 8
; FIXED-NEXT: [[TMP133:%.*]] = icmp eq i64 [[INDEX_NEXT51]], 1024
; FIXED-NEXT: br i1 [[TMP133]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP19:![0-9]+]]
; FIXED: vec.epilog.middle.block:
; FIXED-NEXT: [[CMP_N31:%.*]] = icmp eq i64 1024, 1024
; FIXED-NEXT: br i1 [[CMP_N31]], label [[FOR_END]], label [[VEC_EPILOG_SCALAR_PH]]
; FIXED: vec.epilog.scalar.ph:
; FIXED-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1024, [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ 1024, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
; FIXED-NEXT: br label [[FOR_BODY:%.*]]
; FIXED: for.body:
; FIXED-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LATCH:%.*]] ]
; FIXED-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[IV]]
; FIXED-NEXT: [[ELEM:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; FIXED-NEXT: [[C:%.*]] = icmp ne i8 [[ELEM]], -128
; FIXED-NEXT: br i1 [[C]], label [[DO_OP:%.*]], label [[LATCH]]
; FIXED: do_op:
; FIXED-NEXT: [[DIVREM:%.*]] = sdiv i8 [[ELEM]], -1
; FIXED-NEXT: br label [[LATCH]]
; FIXED: latch:
; FIXED-NEXT: [[PHI:%.*]] = phi i8 [ [[ELEM]], [[FOR_BODY]] ], [ [[DIVREM]], [[DO_OP]] ]
; FIXED-NEXT: store i8 [[PHI]], ptr [[ARRAYIDX]], align 1
; FIXED-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; FIXED-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; FIXED-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
; FIXED: for.end:
; FIXED-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
%arrayidx = getelementptr inbounds i8, ptr %a, i64 %iv
%elem = load i8, ptr %arrayidx
%c = icmp ne i8 %elem, 128
br i1 %c, label %do_op, label %latch
do_op:
%divrem = sdiv i8 %elem, -1 ;; UB if %elem = INT_MIN
br label %latch
latch:
%phi = phi i8 [%elem, %for.body], [%divrem, %do_op]
store i8 %phi, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
Reference in New Issue View Git Blame Copy Permalink