OLD | NEW |
1 ; RUN: opt < %s -instcombine -S | FileCheck %s | 1 ; RUN: opt < %s -instcombine -S | FileCheck %s |
2 | 2 |
3 define i16 @test1(float %f) { | 3 define i16 @test1(float %f) { |
4 entry: | 4 entry: |
5 ; CHECK: @test1 | 5 ; CHECK: @test1 |
6 ; CHECK: fmul float | 6 ; CHECK: fmul float |
7 ; CHECK-NOT: insertelement {{.*}} 0.00 | 7 ; CHECK-NOT: insertelement {{.*}} 0.00 |
8 ; CHECK-NOT: call {{.*}} @llvm.x86.sse.mul | 8 ; CHECK-NOT: call {{.*}} @llvm.x86.sse.mul |
9 ; CHECK-NOT: call {{.*}} @llvm.x86.sse.sub | 9 ; CHECK-NOT: call {{.*}} @llvm.x86.sse.sub |
10 ; CHECK: ret | 10 ; CHECK: ret |
11 %tmp = insertelement <4 x float> undef, float %f, i32 0 ; <<4 x
float>> [#uses=1] | 11 %tmp = insertelement <4 x float> undef, float %f, i32 0 ; <<4 x
float>> [#uses=1] |
12 %tmp10 = insertelement <4 x float> %tmp, float 0.000000e+00, i32 1
; <<4 x float>> [#uses=1] | 12 %tmp10 = insertelement <4 x float> %tmp, float 0.000000e+00, i32 1
; <<4 x float>> [#uses=1] |
13 %tmp11 = insertelement <4 x float> %tmp10, float 0.000000e+00, i32 2
; <<4 x float>> [#uses=1] | 13 %tmp11 = insertelement <4 x float> %tmp10, float 0.000000e+00, i32 2
; <<4 x float>> [#uses=1] |
14 %tmp12 = insertelement <4 x float> %tmp11, float 0.000000e+00, i32 3
; <<4 x float>> [#uses=1] | 14 %tmp12 = insertelement <4 x float> %tmp11, float 0.000000e+00, i32 3
; <<4 x float>> [#uses=1] |
15 %tmp28 = tail call <4 x float> @llvm.x86.sse.sub.ss( <4 x float> %tmp12,
<4 x float> < float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float
0.000000e+00 > ) ; <<4 x float>> [#uses=1] | 15 %tmp28 = tail call <4 x float> @llvm.x86.sse.sub.ss( <4 x float> %tmp12,
<4 x float> < float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float
0.000000e+00 > ) ; <<4 x float>> [#uses=1] |
16 %tmp37 = tail call <4 x float> @llvm.x86.sse.mul.ss( <4 x float> %tmp28,
<4 x float> < float 5.000000e-01, float 0.000000e+00, float 0.000000e+00, float
0.000000e+00 > ) ; <<4 x float>> [#uses=1] | 16 %tmp37 = tail call <4 x float> @llvm.x86.sse.mul.ss( <4 x float> %tmp28,
<4 x float> < float 5.000000e-01, float 0.000000e+00, float 0.000000e+00, float
0.000000e+00 > ) ; <<4 x float>> [#uses=1] |
17 %tmp48 = tail call <4 x float> @llvm.x86.sse.min.ss( <4 x float> %tmp37,
<4 x float> < float 6.553500e+04, float 0.000000e+00, float 0.000000e+00, float
0.000000e+00 > ) ; <<4 x float>> [#uses=1] | 17 %tmp48 = tail call <4 x float> @llvm.x86.sse.min.ss( <4 x float> %tmp37,
<4 x float> < float 6.553500e+04, float 0.000000e+00, float 0.000000e+00, float
0.000000e+00 > ) ; <<4 x float>> [#uses=1] |
18 %tmp59 = tail call <4 x float> @llvm.x86.sse.max.ss( <4 x float> %tmp48,
<4 x float> zeroinitializer ) ; <<4 x float>> [#uses=1] | 18 %tmp59 = tail call <4 x float> @llvm.x86.sse.max.ss( <4 x float> %tmp48,
<4 x float> zeroinitializer ) ; <<4 x float>> [#uses=1] |
19 %tmp.upgrd.1 = tail call i32 @llvm.x86.sse.cvttss2si( <4 x float> %tmp59
) ; <i32> [#uses=1] | 19 %tmp.upgrd.1 = tail call i32 @llvm.x86.sse.cvttss2si( <4 x float> %tmp59
) ; <i32> [#uses=1] |
20 %tmp69 = trunc i32 %tmp.upgrd.1 to i16 ; <i16> [#uses=1] | 20 %tmp69 = trunc i32 %tmp.upgrd.1 to i16 ; <i16> [#uses=1] |
21 ret i16 %tmp69 | 21 ret i16 %tmp69 |
22 } | 22 } |
23 | 23 |
24 define i32 @test2(float %f) { | |
25 ; CHECK: @test2 | |
26 ; CHECK-NOT: insertelement | |
27 ; CHECK-NOT: extractelement | |
28 ; CHECK: ret | |
29 %tmp5 = fmul float %f, %f | |
30 %tmp9 = insertelement <4 x float> undef, float %tmp5, i32 0 | |
31 %tmp10 = insertelement <4 x float> %tmp9, float 0.000000e+00, i32 1 | |
32 %tmp11 = insertelement <4 x float> %tmp10, float 0.000000e+00, i32 2 | |
33 %tmp12 = insertelement <4 x float> %tmp11, float 0.000000e+00, i32 3 | |
34 %tmp19 = bitcast <4 x float> %tmp12 to <4 x i32> | |
35 %tmp21 = extractelement <4 x i32> %tmp19, i32 0 | |
36 ret i32 %tmp21 | |
37 } | |
38 | |
39 define i64 @test3(float %f, double %d) { | 24 define i64 @test3(float %f, double %d) { |
40 ; CHECK: @test3 | 25 ; CHECK: @test3 |
41 ; CHECK-NOT: insertelement {{.*}} 0.00 | 26 ; CHECK-NOT: insertelement {{.*}} 0.00 |
42 ; CHECK: ret | 27 ; CHECK: ret |
43 entry: | 28 entry: |
44 %v00 = insertelement <4 x float> undef, float %f, i32 0 | 29 %v00 = insertelement <4 x float> undef, float %f, i32 0 |
45 %v01 = insertelement <4 x float> %v00, float 0.000000e+00, i32 1 | 30 %v01 = insertelement <4 x float> %v00, float 0.000000e+00, i32 1 |
46 %v02 = insertelement <4 x float> %v01, float 0.000000e+00, i32 2 | 31 %v02 = insertelement <4 x float> %v01, float 0.000000e+00, i32 2 |
47 %v03 = insertelement <4 x float> %v02, float 0.000000e+00, i32 3 | 32 %v03 = insertelement <4 x float> %v02, float 0.000000e+00, i32 3 |
48 %tmp0 = tail call i32 @llvm.x86.sse.cvtss2si(<4 x float> %v03) | 33 %tmp0 = tail call i32 @llvm.x86.sse.cvtss2si(<4 x float> %v03) |
(...skipping 28 matching lines...) Expand all Loading... |
77 %tmp9 = add i32 %tmp4, %tmp6 | 62 %tmp9 = add i32 %tmp4, %tmp6 |
78 %tmp10 = add i32 %tmp8, %tmp9 | 63 %tmp10 = add i32 %tmp8, %tmp9 |
79 %tmp11 = sext i32 %tmp10 to i64 | 64 %tmp11 = sext i32 %tmp10 to i64 |
80 %tmp12 = add i64 %tmp1, %tmp3 | 65 %tmp12 = add i64 %tmp1, %tmp3 |
81 %tmp13 = add i64 %tmp5, %tmp7 | 66 %tmp13 = add i64 %tmp5, %tmp7 |
82 %tmp14 = add i64 %tmp12, %tmp13 | 67 %tmp14 = add i64 %tmp12, %tmp13 |
83 %tmp15 = add i64 %tmp11, %tmp14 | 68 %tmp15 = add i64 %tmp11, %tmp14 |
84 ret i64 %tmp15 | 69 ret i64 %tmp15 |
85 } | 70 } |
86 | 71 |
87 define void @get_image() nounwind { | |
88 ; CHECK: @get_image | |
89 ; CHECK-NOT: extractelement | |
90 ; CHECK: unreachable | |
91 entry: | |
92 %0 = call i32 @fgetc(i8* null) nounwind ; <i32> [#uses=1] | |
93 %1 = trunc i32 %0 to i8 ; <i8> [#uses=1] | |
94 %tmp2 = insertelement <100 x i8> zeroinitializer, i8 %1, i32 1 ; <<10
0 x i8>> [#uses=1] | |
95 %tmp1 = extractelement <100 x i8> %tmp2, i32 0 ; <i8> [#uses=1] | |
96 %2 = icmp eq i8 %tmp1, 80 ; <i1> [#uses=1] | |
97 br i1 %2, label %bb2, label %bb3 | |
98 | |
99 bb2: ; preds = %entry | |
100 br label %bb3 | |
101 | |
102 bb3: ; preds = %bb2, %entry | |
103 unreachable | |
104 } | |
105 | |
106 ; PR4340 | |
107 define void @vac(<4 x float>* nocapture %a) nounwind { | |
108 ; CHECK: @vac | |
109 ; CHECK-NOT: load | |
110 ; CHECK: ret | |
111 entry: | |
112 %tmp1 = load <4 x float>* %a ; <<4 x float>> [#uses=1] | |
113 %vecins = insertelement <4 x float> %tmp1, float 0.000000e+00, i32 0
; <<4 x float>> [#uses=1] | |
114 %vecins4 = insertelement <4 x float> %vecins, float 0.000000e+00, i32 1;
<<4 x float>> [#uses=1] | |
115 %vecins6 = insertelement <4 x float> %vecins4, float 0.000000e+00, i32 2
; <<4 x float>> [#uses=1] | |
116 %vecins8 = insertelement <4 x float> %vecins6, float 0.000000e+00, i32 3
; <<4 x float>> [#uses=1] | |
117 store <4 x float> %vecins8, <4 x float>* %a | |
118 ret void | |
119 } | |
120 | |
121 declare i32 @fgetc(i8*) | |
122 | |
123 declare <4 x float> @llvm.x86.sse.sub.ss(<4 x float>, <4 x float>) | 72 declare <4 x float> @llvm.x86.sse.sub.ss(<4 x float>, <4 x float>) |
124 | 73 |
125 declare <4 x float> @llvm.x86.sse.mul.ss(<4 x float>, <4 x float>) | 74 declare <4 x float> @llvm.x86.sse.mul.ss(<4 x float>, <4 x float>) |
126 | 75 |
127 declare <4 x float> @llvm.x86.sse.min.ss(<4 x float>, <4 x float>) | 76 declare <4 x float> @llvm.x86.sse.min.ss(<4 x float>, <4 x float>) |
128 | 77 |
129 declare <4 x float> @llvm.x86.sse.max.ss(<4 x float>, <4 x float>) | 78 declare <4 x float> @llvm.x86.sse.max.ss(<4 x float>, <4 x float>) |
130 | 79 |
131 declare i32 @llvm.x86.sse.cvtss2si(<4 x float>) | 80 declare i32 @llvm.x86.sse.cvtss2si(<4 x float>) |
132 declare i64 @llvm.x86.sse.cvtss2si64(<4 x float>) | 81 declare i64 @llvm.x86.sse.cvtss2si64(<4 x float>) |
(...skipping 12 matching lines...) Expand all Loading... |
145 ; CHECK: %tmp2 = shufflevector | 94 ; CHECK: %tmp2 = shufflevector |
146 %tmp2 = shufflevector <4 x i16> %tmp, <4 x i16> undef, <8 x i32> <i32 0,
i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef> | 95 %tmp2 = shufflevector <4 x i16> %tmp, <4 x i16> undef, <8 x i32> <i32 0,
i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef> |
147 ; pmovzxwd ignores the upper 64-bits of its input; -instcombine should remove th
is shuffle: | 96 ; pmovzxwd ignores the upper 64-bits of its input; -instcombine should remove th
is shuffle: |
148 ; CHECK-NOT: shufflevector | 97 ; CHECK-NOT: shufflevector |
149 %tmp3 = shufflevector <8 x i16> %tmp1, <8 x i16> %tmp2, <8 x i32> <i32 8
, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7> | 98 %tmp3 = shufflevector <8 x i16> %tmp1, <8 x i16> %tmp2, <8 x i32> <i32 8
, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7> |
150 ; CHECK-NEXT: pmovzxwd | 99 ; CHECK-NEXT: pmovzxwd |
151 %0 = call <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16> %tmp3) | 100 %0 = call <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16> %tmp3) |
152 ret <4 x i32> %0 | 101 ret <4 x i32> %0 |
153 } | 102 } |
154 declare <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16>) nounwind readnone | 103 declare <4 x i32> @llvm.x86.sse41.pmovzxwd(<8 x i16>) nounwind readnone |
OLD | NEW |