7d6e8f2a96
If we vectorize a e.g. store, we leave around a bunch of getelementptrs for the individual scalar stores which we removed. We can go ahead and delete them as well. This is purely for test output quality and readability. It should have no effect in any sane pipeline. Differential Revision: https://reviews.llvm.org/D122493
61 lines
2.3 KiB
LLVM
61 lines
2.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt -slp-vectorizer -slp-vectorize-hor -slp-vectorize-hor-store -S < %s -mtriple=x86_64-unknown-linux-gnu | FileCheck %s
|
|
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128-ni:1"
|
|
|
|
define void @mainTest(i32* %ptr) #0 {
|
|
; CHECK-LABEL: @mainTest(
|
|
; CHECK-NEXT: entry:
|
|
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32* [[PTR:%.*]], null
|
|
; CHECK-NEXT: br i1 [[CMP]], label [[LOOP:%.*]], label [[BAIL_OUT:%.*]]
|
|
; CHECK: loop:
|
|
; CHECK-NEXT: [[DUMMY_PHI:%.*]] = phi i32 [ 1, [[ENTRY:%.*]] ], [ [[OP_EXTRA3:%.*]], [[LOOP]] ]
|
|
; CHECK-NEXT: [[TMP0:%.*]] = bitcast i32* [[PTR]] to <4 x i32>*
|
|
; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* [[TMP0]], align 4
|
|
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
|
|
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x i32> [[TMP1]], i32 2
|
|
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x i32> [[TMP1]], i32 1
|
|
; CHECK-NEXT: [[TMP5:%.*]] = mul <4 x i32> [[TMP1]], [[TMP1]]
|
|
; CHECK-NEXT: [[TMP6:%.*]] = sext i32 [[TMP3]] to i64
|
|
; CHECK-NEXT: [[TMP7:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP5]])
|
|
; CHECK-NEXT: [[OP_EXTRA:%.*]] = add i32 [[TMP7]], 1
|
|
; CHECK-NEXT: [[OP_EXTRA1:%.*]] = add i32 [[OP_EXTRA]], [[TMP4]]
|
|
; CHECK-NEXT: [[OP_EXTRA2:%.*]] = add i32 [[OP_EXTRA1]], [[TMP3]]
|
|
; CHECK-NEXT: [[OP_EXTRA3]] = add i32 [[OP_EXTRA2]], [[TMP2]]
|
|
; CHECK-NEXT: br label [[LOOP]]
|
|
; CHECK: bail_out:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
%cmp = icmp eq i32* %ptr, null
|
|
br i1 %cmp, label %loop, label %bail_out
|
|
|
|
loop:
|
|
%dummy_phi = phi i32 [ 1, %entry ], [ %18, %loop ]
|
|
%0 = load i32, i32 * %ptr , align 4
|
|
%1 = mul i32 %0, %0
|
|
%2 = add i32 1, %1
|
|
%3 = getelementptr inbounds i32, i32 * %ptr, i64 1
|
|
%4 = load i32, i32 * %3 , align 4
|
|
%5 = mul i32 %4, %4
|
|
%6 = add i32 %2, %4
|
|
%7 = add i32 %6, %5
|
|
%8 = getelementptr inbounds i32, i32 *%ptr, i64 2
|
|
%9 = load i32, i32 * %8 , align 4
|
|
%10 = mul i32 %9, %9
|
|
%11 = add i32 %7, %9
|
|
%12 = add i32 %11, %10
|
|
%13 = sext i32 %9 to i64
|
|
%14 = getelementptr inbounds i32, i32 *%ptr, i64 3
|
|
%15 = load i32, i32 * %14 , align 4
|
|
%16 = mul i32 %15, %15
|
|
%17 = add i32 %12, %15
|
|
%18 = add i32 %17, %16
|
|
br label %loop
|
|
|
|
bail_out:
|
|
ret void
|
|
}
|
|
|
|
attributes #0 = { "target-cpu"="westmere" }
|
|
|