9536a6286e
Update createScalarIVSteps to take an insert point as parameter. This ensures that the inserted scalar steps are in the same order as the recipes they replace (vs in reverse order as currently). This helps to reduce the diff for follow-up changes.
77 lines
3.9 KiB
LLVM
77 lines
3.9 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -S %s | FileCheck %s
|
|
|
|
define void @test(ptr %A, i32 %x) {
|
|
; CHECK-LABEL: @test(
|
|
; CHECK-NEXT: entry:
|
|
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_SCEVCHECK:%.*]]
|
|
; CHECK: vector.scevcheck:
|
|
; CHECK-NEXT: [[IDENT_CHECK:%.*]] = icmp ne i32 [[X:%.*]], 1
|
|
; CHECK-NEXT: br i1 [[IDENT_CHECK]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
|
|
; CHECK: vector.ph:
|
|
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
|
|
; CHECK: vector.body:
|
|
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 0
|
|
; CHECK-NEXT: [[TMP0:%.*]] = trunc i64 [[OFFSET_IDX]] to i32
|
|
; CHECK-NEXT: [[TMP1:%.*]] = add i32 [[TMP0]], 0
|
|
; CHECK-NEXT: [[TMP3:%.*]] = add nuw nsw i64 [[TMP2]], 1
|
|
; CHECK-NEXT: [[TMP4:%.*]] = trunc i64 [[TMP3]] to i32
|
|
; CHECK-NEXT: [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
|
|
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[TMP5]]
|
|
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[TMP6]], i32 0
|
|
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP7]], align 4
|
|
; CHECK-NEXT: [[TMP8:%.*]] = zext i32 [[TMP1]] to i64
|
|
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP8]]
|
|
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP9]], i32 0
|
|
; CHECK-NEXT: store <4 x float> [[WIDE_LOAD]], ptr [[TMP10]], align 4
|
|
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[OFFSET_IDX]], 4
|
|
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], undef
|
|
; CHECK-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
|
|
; CHECK: middle.block:
|
|
; CHECK-NEXT: br i1 undef, label [[EXIT:%.*]], label [[SCALAR_PH]]
|
|
; CHECK: scalar.ph:
|
|
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ undef, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ]
|
|
; CHECK-NEXT: br label [[LOOP:%.*]]
|
|
; CHECK: loop:
|
|
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
|
|
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
|
|
; CHECK-NEXT: [[T_IV_NEXT:%.*]] = trunc i64 [[IV_NEXT]] to i32
|
|
; CHECK-NEXT: [[MUL_IV_NEXT:%.*]] = mul i32 [[T_IV_NEXT]], [[X]]
|
|
; CHECK-NEXT: [[IDX_1:%.*]] = zext i32 [[MUL_IV_NEXT]] to i64
|
|
; CHECK-NEXT: [[ARRAYIDX1215:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IDX_1]]
|
|
; CHECK-NEXT: [[LV:%.*]] = load float, ptr [[ARRAYIDX1215]], align 4
|
|
; CHECK-NEXT: [[T_IV:%.*]] = trunc i64 [[IV]] to i32
|
|
; CHECK-NEXT: [[MUL_IV:%.*]] = mul i32 [[T_IV]], [[X]]
|
|
; CHECK-NEXT: [[IDX_2:%.*]] = zext i32 [[MUL_IV]] to i64
|
|
; CHECK-NEXT: [[ARRAYIDX1209:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IDX_2]]
|
|
; CHECK-NEXT: store float [[LV]], ptr [[ARRAYIDX1209]], align 4
|
|
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], undef
|
|
; CHECK-NEXT: br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
|
|
; CHECK: exit:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
br label %loop
|
|
|
|
loop: ; preds = %loop, %entry
|
|
%iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ]
|
|
%iv.next = add nuw nsw i64 %iv, 1
|
|
%t.iv.next = trunc i64 %iv.next to i32
|
|
%mul.iv.next = mul i32 %t.iv.next, %x
|
|
%idx.1 = zext i32 %mul.iv.next to i64
|
|
%arrayidx1215 = getelementptr inbounds float, ptr %A, i64 %idx.1
|
|
%lv = load float, ptr %arrayidx1215, align 4
|
|
|
|
%t.iv = trunc i64 %iv to i32
|
|
%mul.iv = mul i32 %t.iv, %x
|
|
%idx.2 = zext i32 %mul.iv to i64
|
|
%arrayidx1209 = getelementptr inbounds float, ptr %A, i64 %idx.2
|
|
store float %lv, ptr %arrayidx1209, align 4
|
|
%ec = icmp eq i64 %iv.next, undef
|
|
br i1 %ec, label %exit, label %loop
|
|
|
|
exit: ; preds = %loop
|
|
ret void
|
|
}
|