shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity
llvm-project/llvm/test/Transforms/LoopVectorize/uniform-blend.ll
Ramkumar Ramachandra ef023cae38
Reland [VPlan] Expand WidenInt inductions with nuw/nsw (#168354) 
Changes: The previous patch had to be reverted to a mismatching-OpType
assert in cse. The reduced-test has now been added corresponding to a
RVV pointer-induction, and the pointer-induction case has been updated
to use createOverflowingBinaryOp.

While at it, record VPIRFlags in VPWidenInductionRecipe.
2025-11-17 13:44:25 +00:00

261 lines
12 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -S %s | FileCheck %s
@dst = external global [32 x i16], align 1
define void @blend_uniform_iv_trunc(i1 %c) {
; CHECK-LABEL: define void @blend_uniform_iv_trunc(
; CHECK-SAME: i1 [[C:%.*]]) {
; 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_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = trunc i64 [[INDEX]] to i16
; CHECK-NEXT: [[TMP6:%.*]] = select i1 [[C]], i16 [[TMP0]], i16 poison
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i16 [[TMP6]]
; CHECK-NEXT: store <4 x i16> zeroinitializer, ptr [[TMP7]], align 2
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 32
; CHECK-NEXT: br i1 [[TMP4]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br label %[[EXIT:.*]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
entry:
br label %loop.header
loop.header: ; preds = %loop.latch, %entry
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
%iv.trunc.2 = trunc i64 %iv to i16
br i1 %c, label %loop.next, label %loop.latch
loop.next: ; preds = %loop.header
br label %loop.latch
loop.latch: ; preds = %loop.next, %loop.header
%blend = phi i16 [ poison, %loop.header ], [ %iv.trunc.2, %loop.next ]
%dst.ptr = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i16 %blend
store i16 0, ptr %dst.ptr
%iv.next = add nuw nsw i64 %iv, 1
%cmp439 = icmp ult i64 %iv, 31
br i1 %cmp439, label %loop.header, label %exit
exit: ; preds = %loop.latch
ret void
}
define void @blend_uniform_iv(i1 %c) {
; CHECK-LABEL: define void @blend_uniform_iv(
; CHECK-SAME: i1 [[C:%.*]]) {
; 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_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP6:%.*]] = select i1 [[C]], i64 [[INDEX]], i64 poison
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i64 [[TMP6]]
; CHECK-NEXT: store <4 x i16> zeroinitializer, ptr [[TMP7]], align 2
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 32
; CHECK-NEXT: br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br label %[[EXIT:.*]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
entry:
br label %loop.header
loop.header: ; preds = %loop.latch, %entry
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
br i1 %c, label %loop.next, label %loop.latch
loop.next: ; preds = %loop.header
br label %loop.latch
loop.latch: ; preds = %loop.next, %loop.header
%blend = phi i64 [ poison, %loop.header ], [ %iv, %loop.next ]
%dst.ptr = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i64 %blend
store i16 0, ptr %dst.ptr
%iv.next = add nuw nsw i64 %iv, 1
%cmp439 = icmp ult i64 %iv, 31
br i1 %cmp439, label %loop.header, label %exit
exit: ; preds = %loop.latch
ret void
}
define void @blend_chain_iv(i1 %c) {
; CHECK-LABEL: define void @blend_chain_iv(
; CHECK-SAME: i1 [[C:%.*]]) {
; 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_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[PREDPHI1:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[PREDPHI2:%.*]] = select i1 [[C]], <4 x i64> [[PREDPHI1]], <4 x i64> poison
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i64> [[PREDPHI2]], i32 0
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x i64> [[PREDPHI2]], i32 1
; CHECK-NEXT: [[TMP5:%.*]] = extractelement <4 x i64> [[PREDPHI2]], i32 2
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i64> [[PREDPHI2]], i32 3
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i64 [[TMP1]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i64 [[TMP3]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i64 [[TMP5]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i64 [[TMP7]]
; CHECK-NEXT: store i16 0, ptr [[TMP2]], align 2
; CHECK-NEXT: store i16 0, ptr [[TMP4]], align 2
; CHECK-NEXT: store i16 0, ptr [[TMP6]], align 2
; CHECK-NEXT: store i16 0, ptr [[TMP8]], align 2
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <4 x i64> [[PREDPHI1]], splat (i64 4)
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], 32
; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br label %[[EXIT:.*]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
entry:
br label %loop.header
loop.header: ; preds = %loop.latch, %entry
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
br i1 %c, label %loop.next, label %loop.latch
loop.next: ; preds = %loop.header
br i1 %c, label %loop.next.2, label %loop.next.3
loop.next.2:
br label %loop.next.3
loop.next.3:
%blend.1 = phi i64 [ poison, %loop.next ], [ %iv, %loop.next.2 ]
br label %loop.latch
loop.latch: ; preds = %loop.next, %loop.header
%blend = phi i64 [ poison, %loop.header ], [ %blend.1, %loop.next.3 ]
%dst.ptr = getelementptr inbounds [32 x i16], ptr @dst, i16 0, i64 %blend
store i16 0, ptr %dst.ptr
%iv.next = add nuw nsw i64 %iv, 1
%cmp439 = icmp ult i64 %iv, 31
br i1 %cmp439, label %loop.header, label %exit
exit: ; preds = %loop.latch
ret void
}
define void @redundant_branch_and_blends_without_mask(ptr %A) {
; CHECK-LABEL: define void @redundant_branch_and_blends_without_mask(
; CHECK-SAME: ptr [[A:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 0
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 1
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 2
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 3
; CHECK-NEXT: [[TMP35:%.*]] = insertelement <4 x ptr> poison, ptr [[TMP5]], i32 0
; CHECK-NEXT: [[TMP36:%.*]] = insertelement <4 x ptr> [[TMP35]], ptr [[TMP6]], i32 1
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x ptr> [[TMP36]], ptr [[TMP7]], i32 2
; CHECK-NEXT: [[TMP38:%.*]] = insertelement <4 x ptr> [[TMP37]], ptr [[TMP8]], i32 3
; CHECK-NEXT: br i1 true, label %[[PRED_LOAD_IF:.*]], label %[[PRED_LOAD_CONTINUE:.*]]
; CHECK: [[PRED_LOAD_IF]]:
; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[TMP10]], i32 0
; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE]]
; CHECK: [[PRED_LOAD_CONTINUE]]:
; CHECK-NEXT: [[TMP12:%.*]] = phi <4 x i32> [ poison, %[[VECTOR_BODY]] ], [ [[TMP11]], %[[PRED_LOAD_IF]] ]
; CHECK-NEXT: br i1 true, label %[[PRED_LOAD_IF1:.*]], label %[[PRED_LOAD_CONTINUE2:.*]]
; CHECK: [[PRED_LOAD_IF1]]:
; CHECK-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <4 x i32> [[TMP12]], i32 [[TMP14]], i32 1
; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE2]]
; CHECK: [[PRED_LOAD_CONTINUE2]]:
; CHECK-NEXT: [[TMP16:%.*]] = phi <4 x i32> [ [[TMP12]], %[[PRED_LOAD_CONTINUE]] ], [ [[TMP15]], %[[PRED_LOAD_IF1]] ]
; CHECK-NEXT: br i1 false, label %[[PRED_LOAD_IF3:.*]], label %[[PRED_LOAD_CONTINUE4:.*]]
; CHECK: [[PRED_LOAD_IF3]]:
; CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP7]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = insertelement <4 x i32> [[TMP16]], i32 [[TMP18]], i32 2
; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE4]]
; CHECK: [[PRED_LOAD_CONTINUE4]]:
; CHECK-NEXT: [[TMP20:%.*]] = phi <4 x i32> [ [[TMP16]], %[[PRED_LOAD_CONTINUE2]] ], [ [[TMP19]], %[[PRED_LOAD_IF3]] ]
; CHECK-NEXT: br i1 false, label %[[PRED_LOAD_IF5:.*]], label %[[PRED_LOAD_CONTINUE6:.*]]
; CHECK: [[PRED_LOAD_IF5]]:
; CHECK-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x i32> [[TMP20]], i32 [[TMP22]], i32 3
; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE6]]
; CHECK: [[PRED_LOAD_CONTINUE6]]:
; CHECK-NEXT: [[TMP24:%.*]] = phi <4 x i32> [ [[TMP20]], %[[PRED_LOAD_CONTINUE4]] ], [ [[TMP23]], %[[PRED_LOAD_IF5]] ]
; CHECK-NEXT: [[TMP25:%.*]] = add <4 x i32> [[TMP24]], splat (i32 10)
; CHECK-NEXT: [[TMP26:%.*]] = add <4 x i32> [[TMP24]], [[TMP25]]
; CHECK-NEXT: br i1 true, label %[[PRED_STORE_IF:.*]], label %[[PRED_STORE_CONTINUE:.*]]
; CHECK: [[PRED_STORE_IF]]:
; CHECK-NEXT: [[TMP28:%.*]] = extractelement <4 x i32> [[TMP26]], i32 0
; CHECK-NEXT: store i32 [[TMP28]], ptr [[TMP5]], align 4
; CHECK-NEXT: br label %[[PRED_STORE_CONTINUE]]
; CHECK: [[PRED_STORE_CONTINUE]]:
; CHECK-NEXT: br i1 true, label %[[PRED_STORE_IF7:.*]], label %[[PRED_STORE_CONTINUE8:.*]]
; CHECK: [[PRED_STORE_IF7]]:
; CHECK-NEXT: [[TMP30:%.*]] = extractelement <4 x i32> [[TMP26]], i32 1
; CHECK-NEXT: store i32 [[TMP30]], ptr [[TMP6]], align 4
; CHECK-NEXT: br label %[[PRED_STORE_CONTINUE8]]
; CHECK: [[PRED_STORE_CONTINUE8]]:
; CHECK-NEXT: br i1 false, label %[[PRED_STORE_IF9:.*]], label %[[PRED_STORE_CONTINUE10:.*]]
; CHECK: [[PRED_STORE_IF9]]:
; CHECK-NEXT: [[TMP32:%.*]] = extractelement <4 x i32> [[TMP26]], i32 2
; CHECK-NEXT: store i32 [[TMP32]], ptr [[TMP7]], align 4
; CHECK-NEXT: br label %[[PRED_STORE_CONTINUE10]]
; CHECK: [[PRED_STORE_CONTINUE10]]:
; CHECK-NEXT: br i1 false, label %[[PRED_STORE_IF11:.*]], label %[[PRED_STORE_CONTINUE12:.*]]
; CHECK: [[PRED_STORE_IF11]]:
; CHECK-NEXT: [[TMP34:%.*]] = extractelement <4 x i32> [[TMP26]], i32 3
; CHECK-NEXT: store i32 [[TMP34]], ptr [[TMP8]], align 4
; CHECK-NEXT: br label %[[PRED_STORE_CONTINUE12]]
; CHECK: [[PRED_STORE_CONTINUE12]]:
; CHECK-NEXT: br label %[[MIDDLE_BLOCK:.*]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br label %[[EXIT:.*]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
entry:
br label %loop.header
loop.header:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
%gep.iv = getelementptr inbounds i32, ptr %A, i64 %iv
%l = load i32, ptr %gep.iv
%add = add i32 %l, 10
br label %loop.latch
loop.latch:
%p.1 = phi i32 [ %l, %loop.header ]
%p.2 = phi i32 [ %add, %loop.header ]
%res = add i32 %p.1, %p.2
store i32 %res, ptr %gep.iv
%iv.next = add i64 %iv, 1
%ec = icmp eq i64 %iv, 1
br i1 %ec, label %exit, label %loop.header
exit:
ret void
}
;.
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META1]], [[META2]]}
; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META1]], [[META2]]}
;.
Reference in New Issue View Git Blame Copy Permalink