1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt < %s -instcombine -S | FileCheck %s
3 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
5 define <2 x double> @test_round_sd(<2 x double> %a, <2 x double> %b) {
6 ; CHECK-LABEL: @test_round_sd(
7 ; CHECK-NEXT: [[TMP1:%.*]] = tail call <2 x double> @llvm.x86.sse41.round.sd(<2 x double> [[A:%.*]], <2 x double> [[B:%.*]], i32 10)
8 ; CHECK-NEXT: ret <2 x double> [[TMP1]]
10 %1 = insertelement <2 x double> %a, double 1.000000e+00, i32 0
11 %2 = insertelement <2 x double> %b, double 2.000000e+00, i32 1
12 %3 = tail call <2 x double> @llvm.x86.sse41.round.sd(<2 x double> %1, <2 x double> %2, i32 10)
16 define double @test_round_sd_0(double %a, double %b) {
17 ; CHECK-LABEL: @test_round_sd_0(
18 ; CHECK-NEXT: [[TMP1:%.*]] = insertelement <2 x double> undef, double [[B:%.*]], i32 0
19 ; CHECK-NEXT: [[TMP2:%.*]] = tail call <2 x double> @llvm.x86.sse41.round.sd(<2 x double> undef, <2 x double> [[TMP1]], i32 10)
20 ; CHECK-NEXT: [[TMP3:%.*]] = extractelement <2 x double> [[TMP2]], i32 0
21 ; CHECK-NEXT: ret double [[TMP3]]
23 %1 = insertelement <2 x double> undef, double %a, i32 0
24 %2 = insertelement <2 x double> %1, double 1.000000e+00, i32 1
25 %3 = insertelement <2 x double> undef, double %b, i32 0
26 %4 = insertelement <2 x double> %3, double 2.000000e+00, i32 1
27 %5 = tail call <2 x double> @llvm.x86.sse41.round.sd(<2 x double> %2, <2 x double> %4, i32 10)
28 %6 = extractelement <2 x double> %5, i32 0
32 define double @test_round_sd_1(double %a, double %b) {
33 ; CHECK-LABEL: @test_round_sd_1(
34 ; CHECK-NEXT: ret double 1.000000e+00
36 %1 = insertelement <2 x double> undef, double %a, i32 0
37 %2 = insertelement <2 x double> %1, double 1.000000e+00, i32 1
38 %3 = insertelement <2 x double> undef, double %b, i32 0
39 %4 = insertelement <2 x double> %3, double 2.000000e+00, i32 1
40 %5 = tail call <2 x double> @llvm.x86.sse41.round.sd(<2 x double> %2, <2 x double> %4, i32 10)
41 %6 = extractelement <2 x double> %5, i32 1
45 define <4 x float> @test_round_ss(<4 x float> %a, <4 x float> %b) {
46 ; CHECK-LABEL: @test_round_ss(
47 ; CHECK-NEXT: [[TMP1:%.*]] = tail call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> <float undef, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, <4 x float> [[B:%.*]], i32 10)
48 ; CHECK-NEXT: ret <4 x float> [[TMP1]]
50 %1 = insertelement <4 x float> %a, float 1.000000e+00, i32 1
51 %2 = insertelement <4 x float> %1, float 2.000000e+00, i32 2
52 %3 = insertelement <4 x float> %2, float 3.000000e+00, i32 3
53 %4 = insertelement <4 x float> %b, float 1.000000e+00, i32 1
54 %5 = insertelement <4 x float> %4, float 2.000000e+00, i32 2
55 %6 = insertelement <4 x float> %5, float 3.000000e+00, i32 3
56 %7 = tail call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %3, <4 x float> %6, i32 10)
60 define float @test_round_ss_0(float %a, float %b) {
61 ; CHECK-LABEL: @test_round_ss_0(
62 ; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x float> undef, float [[B:%.*]], i32 0
63 ; CHECK-NEXT: [[TMP2:%.*]] = tail call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> undef, <4 x float> [[TMP1]], i32 10)
64 ; CHECK-NEXT: [[R:%.*]] = extractelement <4 x float> [[TMP2]], i32 0
65 ; CHECK-NEXT: ret float [[R]]
67 %1 = insertelement <4 x float> undef, float %a, i32 0
68 %2 = insertelement <4 x float> %1, float 1.000000e+00, i32 1
69 %3 = insertelement <4 x float> %2, float 2.000000e+00, i32 2
70 %4 = insertelement <4 x float> %3, float 3.000000e+00, i32 3
71 %5 = insertelement <4 x float> undef, float %b, i32 0
72 %6 = insertelement <4 x float> %5, float 4.000000e+00, i32 1
73 %7 = insertelement <4 x float> %6, float 5.000000e+00, i32 2
74 %8 = insertelement <4 x float> %7, float 6.000000e+00, i32 3
75 %9 = tail call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %4, <4 x float> %8, i32 10)
76 %r = extractelement <4 x float> %9, i32 0
80 define float @test_round_ss_2(float %a, float %b) {
81 ; CHECK-LABEL: @test_round_ss_2(
82 ; CHECK-NEXT: ret float 2.000000e+00
84 %1 = insertelement <4 x float> undef, float %a, i32 0
85 %2 = insertelement <4 x float> %1, float 1.000000e+00, i32 1
86 %3 = insertelement <4 x float> %2, float 2.000000e+00, i32 2
87 %4 = insertelement <4 x float> %3, float 3.000000e+00, i32 3
88 %5 = insertelement <4 x float> undef, float %b, i32 0
89 %6 = insertelement <4 x float> %5, float 4.000000e+00, i32 1
90 %7 = insertelement <4 x float> %6, float 5.000000e+00, i32 2
91 %8 = insertelement <4 x float> %7, float 6.000000e+00, i32 3
92 %9 = tail call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %4, <4 x float> %8, i32 10)
93 %r = extractelement <4 x float> %9, i32 2
97 declare <2 x double> @llvm.x86.sse41.round.sd(<2 x double>, <2 x double>, i32) nounwind readnone
98 declare <4 x float> @llvm.x86.sse41.round.ss(<4 x float>, <4 x float>, i32) nounwind readnone