Doomhowl-Interactive/llvm-project
0
0
Fork 0
Code Issues Pull Requests Actions 5 Packages Projects Releases Wiki Activity
Files
main
llvm-project/llvm/test/Transforms/LoopVectorize/RISCV/select-cmp-reduction.ll
Elvis Wang f5ef1edacc Reapply "[LV] Enable scalable FindLast on RISCV. (#184931)" (#190938) 
Orginal PR: https://github.com/llvm/llvm-project/pull/184931
Original PR message:
This patch enables FindLast reduction vectorization with scalable
vectors
on RISCV.

Fixed the previous crash by
https://github.com/llvm/llvm-project/pull/191166 and
https://github.com/llvm/llvm-project/pull/191517
2026-04-13 23:29:05 +00:00

355 lines
20 KiB
LLVM
Raw Permalink Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 5
; RUN: opt -p loop-vectorize -mtriple riscv64 -mattr=+v -S < %s | FileCheck %s
define i32 @select_icmp(i32 %x, i32 %y, ptr nocapture readonly %c, i64 %n) {
; CHECK-LABEL: define i32 @select_icmp(
; CHECK-SAME: i32 [[X:%.*]], i32 [[Y:%.*]], ptr readonly captures(none) [[C:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[X]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP8:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ [[N]], %[[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP14:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP6]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP14]])
; CHECK-NEXT: [[TMP7:%.*]] = icmp sge <vscale x 4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP8]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> [[TMP7]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[VEC_PHI]], i32 [[TMP14]])
; CHECK-NEXT: [[TMP9:%.*]] = zext i32 [[TMP14]] to i64
; CHECK-NEXT: [[INDEX_EVL_NEXT]] = add i64 [[TMP9]], [[INDEX]]
; CHECK-NEXT: [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP9]]
; CHECK-NEXT: [[TMP13:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT: br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[TMP10:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP8]])
; CHECK-NEXT: [[TMP11:%.*]] = freeze i1 [[TMP10]]
; CHECK-NEXT: [[RDX_SELECT:%.*]] = select i1 [[TMP11]], i32 [[Y]], i32 0
; CHECK-NEXT: br label %[[FOR_END:.*]]
; CHECK: [[FOR_END]]:
; CHECK-NEXT: ret i32 [[RDX_SELECT]]
;
entry:
br label %for.body
for.body:
%indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
%a = phi i32 [ 0, %entry], [ %cond, %for.body ]
%arrayidx = getelementptr inbounds i32, ptr %c, i64 %indvars.iv
%0 = load i32, ptr %arrayidx, align 4
%cmp1 = icmp slt i32 %0, %x
%cond = select i1 %cmp1, i32 %a, i32 %y
%indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
%exitcond.not = icmp eq i64 %indvars.iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret i32 %cond
}
define i32 @select_fcmp(float %x, i32 %y, ptr nocapture readonly %c, i64 %n) {
; CHECK-LABEL: define i32 @select_fcmp(
; CHECK-SAME: float [[X:%.*]], i32 [[Y:%.*]], ptr readonly captures(none) [[C:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[X]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x float> [[BROADCAST_SPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP8:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ [[N]], %[[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP14:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[C]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = call <vscale x 4 x float> @llvm.vp.load.nxv4f32.p0(ptr align 4 [[TMP6]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP14]])
; CHECK-NEXT: [[TMP7:%.*]] = fcmp fast uge <vscale x 4 x float> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP8]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> [[TMP7]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[VEC_PHI]], i32 [[TMP14]])
; CHECK-NEXT: [[TMP9:%.*]] = zext i32 [[TMP14]] to i64
; CHECK-NEXT: [[INDEX_EVL_NEXT]] = add i64 [[TMP9]], [[INDEX]]
; CHECK-NEXT: [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP9]]
; CHECK-NEXT: [[TMP13:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT: br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[TMP10:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP8]])
; CHECK-NEXT: [[TMP11:%.*]] = freeze i1 [[TMP10]]
; CHECK-NEXT: [[RDX_SELECT:%.*]] = select i1 [[TMP11]], i32 [[Y]], i32 0
; CHECK-NEXT: br label %[[FOR_END:.*]]
; CHECK: [[FOR_END]]:
; CHECK-NEXT: ret i32 [[RDX_SELECT]]
;
entry:
br label %for.body
for.body:
%indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
%a = phi i32 [ 0, %entry], [ %cond, %for.body ]
%arrayidx = getelementptr inbounds float, ptr %c, i64 %indvars.iv
%0 = load float, ptr %arrayidx, align 4
%cmp1 = fcmp fast olt float %0, %x
%cond = select i1 %cmp1, i32 %a, i32 %y
%indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
%exitcond.not = icmp eq i64 %indvars.iv.next, %n
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret i32 %cond
}
define i32 @select_const_i32_from_icmp(ptr nocapture readonly %v, i64 %n) {
; CHECK-LABEL: define i32 @select_const_i32_from_icmp(
; CHECK-SAME: ptr readonly captures(none) [[V:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP8:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ [[N]], %[[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP21:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[V]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP6]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP21]])
; CHECK-NEXT: [[TMP7:%.*]] = icmp ne <vscale x 4 x i32> [[WIDE_LOAD]], splat (i32 3)
; CHECK-NEXT: [[TMP8]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> [[TMP7]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[VEC_PHI]], i32 [[TMP21]])
; CHECK-NEXT: [[TMP9:%.*]] = zext i32 [[TMP21]] to i64
; CHECK-NEXT: [[INDEX_EVL_NEXT]] = add i64 [[TMP9]], [[INDEX]]
; CHECK-NEXT: [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP9]]
; CHECK-NEXT: [[TMP20:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT: br i1 [[TMP20]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[TMP10:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP8]])
; CHECK-NEXT: [[TMP11:%.*]] = freeze i1 [[TMP10]]
; CHECK-NEXT: [[RDX_SELECT:%.*]] = select i1 [[TMP11]], i32 7, i32 3
; CHECK-NEXT: br label %[[EXIT:.*]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret i32 [[RDX_SELECT]]
;
entry:
br label %for.body
for.body:
%0 = phi i64 [ 0, %entry ], [ %6, %for.body ]
%1 = phi i32 [ 3, %entry ], [ %5, %for.body ]
%2 = getelementptr inbounds i32, ptr %v, i64 %0
%3 = load i32, ptr %2, align 4
%4 = icmp eq i32 %3, 3
%5 = select i1 %4, i32 %1, i32 7
%6 = add nuw nsw i64 %0, 1
%7 = icmp eq i64 %6, %n
br i1 %7, label %exit, label %for.body
exit:
ret i32 %5
}
define i32 @select_i32_from_icmp(ptr nocapture readonly %v, i32 %a, i32 %b, i64 %n) {
; CHECK-LABEL: define i32 @select_i32_from_icmp(
; CHECK-SAME: ptr readonly captures(none) [[V:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP8:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ [[N]], %[[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP21:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[V]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP6]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP21]])
; CHECK-NEXT: [[TMP7:%.*]] = icmp ne <vscale x 4 x i32> [[WIDE_LOAD]], splat (i32 3)
; CHECK-NEXT: [[TMP8]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> [[TMP7]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[VEC_PHI]], i32 [[TMP21]])
; CHECK-NEXT: [[TMP9:%.*]] = zext i32 [[TMP21]] to i64
; CHECK-NEXT: [[INDEX_EVL_NEXT]] = add i64 [[TMP9]], [[INDEX]]
; CHECK-NEXT: [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP9]]
; CHECK-NEXT: [[TMP20:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT: br i1 [[TMP20]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[TMP10:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP8]])
; CHECK-NEXT: [[TMP11:%.*]] = freeze i1 [[TMP10]]
; CHECK-NEXT: [[RDX_SELECT:%.*]] = select i1 [[TMP11]], i32 [[B]], i32 [[A]]
; CHECK-NEXT: br label %[[EXIT:.*]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret i32 [[RDX_SELECT]]
;
entry:
br label %for.body
for.body:
%0 = phi i64 [ 0, %entry ], [ %6, %for.body ]
%1 = phi i32 [ %a, %entry ], [ %5, %for.body ]
%2 = getelementptr inbounds i32, ptr %v, i64 %0
%3 = load i32, ptr %2, align 4
%4 = icmp eq i32 %3, 3
%5 = select i1 %4, i32 %1, i32 %b
%6 = add nuw nsw i64 %0, 1
%7 = icmp eq i64 %6, %n
br i1 %7, label %exit, label %for.body
exit:
ret i32 %5
}
define i32 @select_const_i32_from_fcmp(ptr nocapture readonly %v, i64 %n) {
; CHECK-LABEL: define i32 @select_const_i32_from_fcmp(
; CHECK-SAME: ptr readonly captures(none) [[V:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP8:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ [[N]], %[[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP21:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[V]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = call <vscale x 4 x float> @llvm.vp.load.nxv4f32.p0(ptr align 4 [[TMP6]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP21]])
; CHECK-NEXT: [[TMP7:%.*]] = fcmp fast one <vscale x 4 x float> [[WIDE_LOAD]], splat (float 3.000000e+00)
; CHECK-NEXT: [[TMP8]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> [[TMP7]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[VEC_PHI]], i32 [[TMP21]])
; CHECK-NEXT: [[TMP9:%.*]] = zext i32 [[TMP21]] to i64
; CHECK-NEXT: [[INDEX_EVL_NEXT]] = add i64 [[TMP9]], [[INDEX]]
; CHECK-NEXT: [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP9]]
; CHECK-NEXT: [[TMP20:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT: br i1 [[TMP20]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[TMP10:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP8]])
; CHECK-NEXT: [[TMP11:%.*]] = freeze i1 [[TMP10]]
; CHECK-NEXT: [[RDX_SELECT:%.*]] = select i1 [[TMP11]], i32 1, i32 2
; CHECK-NEXT: br label %[[EXIT:.*]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret i32 [[RDX_SELECT]]
;
entry:
br label %for.body
for.body:
%0 = phi i64 [ 0, %entry ], [ %6, %for.body ]
%1 = phi i32 [ 2, %entry ], [ %5, %for.body ]
%2 = getelementptr inbounds float, ptr %v, i64 %0
%3 = load float, ptr %2, align 4
%4 = fcmp fast ueq float %3, 3.0
%5 = select i1 %4, i32 %1, i32 1
%6 = add nuw nsw i64 %0, 1
%7 = icmp eq i64 %6, %n
br i1 %7, label %exit, label %for.body
exit:
ret i32 %5
}
define float @select_const_f32_from_icmp(ptr nocapture readonly %v, i64 %n) {
; CHECK-LABEL: define float @select_const_f32_from_icmp(
; CHECK-SAME: ptr readonly captures(none) [[V:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[TMP0:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[CURRENT_ITERATION_NEXT:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x float> [ splat (float 3.000000e+00), %[[VECTOR_PH]] ], [ [[TMP10:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP9:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ [[N]], %[[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[V]], i64 [[TMP0]]
; CHECK-NEXT: [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP2]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP1]])
; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <vscale x 4 x i32> [[VP_OP_LOAD]], splat (i32 3)
; CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[TMP5]], <vscale x 4 x i1> zeroinitializer, i32 [[TMP1]])
; CHECK-NEXT: [[TMP7:%.*]] = freeze <vscale x 4 x i1> [[TMP6]]
; CHECK-NEXT: [[TMP8:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[TMP7]])
; CHECK-NEXT: [[TMP9]] = select i1 [[TMP8]], <vscale x 4 x i1> [[TMP6]], <vscale x 4 x i1> [[TMP4]]
; CHECK-NEXT: [[TMP10]] = select i1 [[TMP8]], <vscale x 4 x float> splat (float 7.000000e+00), <vscale x 4 x float> [[VEC_PHI]]
; CHECK-NEXT: [[TMP11:%.*]] = zext i32 [[TMP1]] to i64
; CHECK-NEXT: [[CURRENT_ITERATION_NEXT]] = add i64 [[TMP11]], [[TMP0]]
; CHECK-NEXT: [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP11]]
; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT: br i1 [[TMP12]], label %[[SCALAR_PH:.*]], label %[[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: [[SCALAR_PH]]:
; CHECK-NEXT: [[DOTLCSSA:%.*]] = call float @llvm.experimental.vector.extract.last.active.nxv4f32(<vscale x 4 x float> [[TMP10]], <vscale x 4 x i1> [[TMP9]], float 3.000000e+00)
; CHECK-NEXT: br label %[[FOR_BODY1:.*]]
; CHECK: [[FOR_BODY1]]:
; CHECK-NEXT: ret float [[DOTLCSSA]]
;
entry:
br label %for.body
for.body:
%0 = phi i64 [ 0, %entry ], [ %6, %for.body ]
%1 = phi fast float [ 3.0, %entry ], [ %5, %for.body ]
%2 = getelementptr inbounds i32, ptr %v, i64 %0
%3 = load i32, ptr %2, align 4
%4 = icmp eq i32 %3, 3
%5 = select fast i1 %4, float %1, float 7.0
%6 = add nuw nsw i64 %0, 1
%7 = icmp eq i64 %6, %n
br i1 %7, label %exit, label %for.body
exit:
ret float %5
}
define i32 @pred_select_const_i32_from_icmp(ptr noalias nocapture readonly %src1, ptr noalias nocapture readonly %src2, i64 %n) {
; CHECK-LABEL: define i32 @pred_select_const_i32_from_icmp(
; CHECK-SAME: ptr noalias readonly captures(none) [[SRC1:%.*]], ptr noalias readonly captures(none) [[SRC2:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i1> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[PREDPHI:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ [[N]], %[[VECTOR_PH]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP17:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[SRC1]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP6]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP17]])
; CHECK-NEXT: [[TMP7:%.*]] = icmp sgt <vscale x 4 x i32> [[WIDE_LOAD]], splat (i32 35)
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP8]], <vscale x 4 x i1> [[TMP7]], i32 [[TMP17]])
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq <vscale x 4 x i32> [[WIDE_MASKED_LOAD]], splat (i32 2)
; CHECK-NEXT: [[TMP10:%.*]] = or <vscale x 4 x i1> [[VEC_PHI]], [[TMP9]]
; CHECK-NEXT: [[PREDPHI1:%.*]] = select <vscale x 4 x i1> [[TMP7]], <vscale x 4 x i1> [[TMP10]], <vscale x 4 x i1> [[VEC_PHI]]
; CHECK-NEXT: [[PREDPHI]] = call <vscale x 4 x i1> @llvm.vp.merge.nxv4i1(<vscale x 4 x i1> splat (i1 true), <vscale x 4 x i1> [[PREDPHI1]], <vscale x 4 x i1> [[VEC_PHI]], i32 [[TMP17]])
; CHECK-NEXT: [[TMP21:%.*]] = zext i32 [[TMP17]] to i64
; CHECK-NEXT: [[INDEX_EVL_NEXT]] = add i64 [[TMP21]], [[INDEX]]
; CHECK-NEXT: [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP21]]
; CHECK-NEXT: [[TMP16:%.*]] = icmp eq i64 [[AVL_NEXT]], 0
; CHECK-NEXT: br i1 [[TMP16]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[TMP12:%.*]] = call i1 @llvm.vector.reduce.or.nxv4i1(<vscale x 4 x i1> [[PREDPHI]])
; CHECK-NEXT: [[TMP13:%.*]] = freeze i1 [[TMP12]]
; CHECK-NEXT: [[RDX_SELECT:%.*]] = select i1 [[TMP13]], i32 1, i32 0
; CHECK-NEXT: br label %[[FOR_END_LOOPEXIT:.*]]
; CHECK: [[FOR_END_LOOPEXIT]]:
; CHECK-NEXT: ret i32 [[RDX_SELECT]]
;
entry:
br label %for.body
for.body:
%i.013 = phi i64 [ %inc, %for.inc ], [ 0, %entry ]
%r.012 = phi i32 [ %r.1, %for.inc ], [ 0, %entry ]
%arrayidx = getelementptr inbounds i32, ptr %src1, i64 %i.013
%0 = load i32, ptr %arrayidx, align 4
%cmp1 = icmp sgt i32 %0, 35
br i1 %cmp1, label %if.then, label %for.inc
if.then:
%arrayidx2 = getelementptr inbounds i32, ptr %src2, i64 %i.013
%1 = load i32, ptr %arrayidx2, align 4
%cmp3 = icmp eq i32 %1, 2
%spec.select = select i1 %cmp3, i32 1, i32 %r.012
br label %for.inc
for.inc:
%r.1 = phi i32 [ %r.012, %for.body ], [ %spec.select, %if.then ]
%inc = add nuw nsw i64 %i.013, 1
%exitcond.not = icmp eq i64 %inc, %n
br i1 %exitcond.not, label %for.end.loopexit, label %for.body
for.end.loopexit:
%r.1.lcssa = phi i32 [ %r.1, %for.inc ]
ret i32 %r.1.lcssa
}
Reference in New Issue View Git Blame Copy Permalink