ca2396e673
Summary: For better vectorization result we should take into consideration the cost of the user insertelement instructions when we try to vectorize sequences that build the whole vector. I.e. if we have the following scalar code: ``` <Scalar code> insertelement <ScalarCode>, ... ``` we should consider the cost of the last `insertelement ` instructions as the cost of the scalar code. Reviewers: RKSimon, spatel, hfinkel, mkuper Subscribers: javed.absar, llvm-commits Differential Revision: https://reviews.llvm.org/D42657 llvm-svn: 324893
63 lines
3.1 KiB
LLVM
63 lines
3.1 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt -slp-vectorizer < %s -S -o - -mtriple=x86_64-apple-macosx10.10.0 -mcpu=core2 | FileCheck %s
|
|
|
|
define <4 x i32> @sign_extend_v_v(<4 x i16> %lhs) {
|
|
; CHECK-LABEL: @sign_extend_v_v(
|
|
; CHECK-NEXT: entry:
|
|
; CHECK-NEXT: [[TMP0:%.*]] = sext <4 x i16> [[LHS:%.*]] to <4 x i32>
|
|
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i32> [[TMP0]], i32 0
|
|
; CHECK-NEXT: [[VECINIT:%.*]] = insertelement <4 x i32> undef, i32 [[TMP1]], i32 0
|
|
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i32> [[TMP0]], i32 1
|
|
; CHECK-NEXT: [[VECINIT3:%.*]] = insertelement <4 x i32> [[VECINIT]], i32 [[TMP2]], i32 1
|
|
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x i32> [[TMP0]], i32 2
|
|
; CHECK-NEXT: [[VECINIT6:%.*]] = insertelement <4 x i32> [[VECINIT3]], i32 [[TMP3]], i32 2
|
|
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x i32> [[TMP0]], i32 3
|
|
; CHECK-NEXT: [[VECINIT9:%.*]] = insertelement <4 x i32> [[VECINIT6]], i32 [[TMP4]], i32 3
|
|
; CHECK-NEXT: ret <4 x i32> [[VECINIT9]]
|
|
;
|
|
entry:
|
|
%vecext = extractelement <4 x i16> %lhs, i32 0
|
|
%conv = sext i16 %vecext to i32
|
|
%vecinit = insertelement <4 x i32> undef, i32 %conv, i32 0
|
|
%vecext1 = extractelement <4 x i16> %lhs, i32 1
|
|
%conv2 = sext i16 %vecext1 to i32
|
|
%vecinit3 = insertelement <4 x i32> %vecinit, i32 %conv2, i32 1
|
|
%vecext4 = extractelement <4 x i16> %lhs, i32 2
|
|
%conv5 = sext i16 %vecext4 to i32
|
|
%vecinit6 = insertelement <4 x i32> %vecinit3, i32 %conv5, i32 2
|
|
%vecext7 = extractelement <4 x i16> %lhs, i32 3
|
|
%conv8 = sext i16 %vecext7 to i32
|
|
%vecinit9 = insertelement <4 x i32> %vecinit6, i32 %conv8, i32 3
|
|
ret <4 x i32> %vecinit9
|
|
}
|
|
|
|
define <4 x i16> @truncate_v_v(<4 x i32> %lhs) {
|
|
; CHECK-LABEL: @truncate_v_v(
|
|
; CHECK-NEXT: entry:
|
|
; CHECK-NEXT: [[TMP0:%.*]] = trunc <4 x i32> [[LHS:%.*]] to <4 x i16>
|
|
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i16> [[TMP0]], i32 0
|
|
; CHECK-NEXT: [[VECINIT:%.*]] = insertelement <4 x i16> undef, i16 [[TMP1]], i32 0
|
|
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i16> [[TMP0]], i32 1
|
|
; CHECK-NEXT: [[VECINIT3:%.*]] = insertelement <4 x i16> [[VECINIT]], i16 [[TMP2]], i32 1
|
|
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x i16> [[TMP0]], i32 2
|
|
; CHECK-NEXT: [[VECINIT6:%.*]] = insertelement <4 x i16> [[VECINIT3]], i16 [[TMP3]], i32 2
|
|
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x i16> [[TMP0]], i32 3
|
|
; CHECK-NEXT: [[VECINIT9:%.*]] = insertelement <4 x i16> [[VECINIT6]], i16 [[TMP4]], i32 3
|
|
; CHECK-NEXT: ret <4 x i16> [[VECINIT9]]
|
|
;
|
|
entry:
|
|
%vecext = extractelement <4 x i32> %lhs, i32 0
|
|
%conv = trunc i32 %vecext to i16
|
|
%vecinit = insertelement <4 x i16> undef, i16 %conv, i32 0
|
|
%vecext1 = extractelement <4 x i32> %lhs, i32 1
|
|
%conv2 = trunc i32 %vecext1 to i16
|
|
%vecinit3 = insertelement <4 x i16> %vecinit, i16 %conv2, i32 1
|
|
%vecext4 = extractelement <4 x i32> %lhs, i32 2
|
|
%conv5 = trunc i32 %vecext4 to i16
|
|
%vecinit6 = insertelement <4 x i16> %vecinit3, i16 %conv5, i32 2
|
|
%vecext7 = extractelement <4 x i32> %lhs, i32 3
|
|
%conv8 = trunc i32 %vecext7 to i16
|
|
%vecinit9 = insertelement <4 x i16> %vecinit6, i16 %conv8, i32 3
|
|
ret <4 x i16> %vecinit9
|
|
}
|