1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt < %s -mtriple=x86_64-unknown -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=SSE
3 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX256 --check-prefix=AVX256NODQ
4 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=bdver1 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX256 --check-prefix=AVX256NODQ
5 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX256 --check-prefix=AVX256NODQ
6 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skylake-avx512 -mattr=-prefer-256-bit -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX512
7 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skylake-avx512 -mattr=+prefer-256-bit -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX256 --check-prefix=AVX256DQ
9 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
11 @src64 = common global [8 x double] zeroinitializer, align 64
12 @src32 = common global [16 x float] zeroinitializer, align 64
13 @dst64 = common global [8 x i64] zeroinitializer, align 64
14 @dst32 = common global [16 x i32] zeroinitializer, align 64
15 @dst16 = common global [32 x i16] zeroinitializer, align 64
16 @dst8 = common global [64 x i8] zeroinitializer, align 64
22 define void @fptosi_8f64_8i64() #0 {
23 ; SSE-LABEL: @fptosi_8f64_8i64(
24 ; SSE-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
25 ; SSE-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
26 ; SSE-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
27 ; SSE-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
28 ; SSE-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
29 ; SSE-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
30 ; SSE-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
31 ; SSE-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
32 ; SSE-NEXT: [[CVT0:%.*]] = fptosi double [[A0]] to i64
33 ; SSE-NEXT: [[CVT1:%.*]] = fptosi double [[A1]] to i64
34 ; SSE-NEXT: [[CVT2:%.*]] = fptosi double [[A2]] to i64
35 ; SSE-NEXT: [[CVT3:%.*]] = fptosi double [[A3]] to i64
36 ; SSE-NEXT: [[CVT4:%.*]] = fptosi double [[A4]] to i64
37 ; SSE-NEXT: [[CVT5:%.*]] = fptosi double [[A5]] to i64
38 ; SSE-NEXT: [[CVT6:%.*]] = fptosi double [[A6]] to i64
39 ; SSE-NEXT: [[CVT7:%.*]] = fptosi double [[A7]] to i64
40 ; SSE-NEXT: store i64 [[CVT0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
41 ; SSE-NEXT: store i64 [[CVT1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
42 ; SSE-NEXT: store i64 [[CVT2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
43 ; SSE-NEXT: store i64 [[CVT3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
44 ; SSE-NEXT: store i64 [[CVT4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
45 ; SSE-NEXT: store i64 [[CVT5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
46 ; SSE-NEXT: store i64 [[CVT6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
47 ; SSE-NEXT: store i64 [[CVT7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
50 ; AVX256NODQ-LABEL: @fptosi_8f64_8i64(
51 ; AVX256NODQ-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
52 ; AVX256NODQ-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
53 ; AVX256NODQ-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
54 ; AVX256NODQ-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
55 ; AVX256NODQ-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
56 ; AVX256NODQ-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
57 ; AVX256NODQ-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
58 ; AVX256NODQ-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
59 ; AVX256NODQ-NEXT: [[CVT0:%.*]] = fptosi double [[A0]] to i64
60 ; AVX256NODQ-NEXT: [[CVT1:%.*]] = fptosi double [[A1]] to i64
61 ; AVX256NODQ-NEXT: [[CVT2:%.*]] = fptosi double [[A2]] to i64
62 ; AVX256NODQ-NEXT: [[CVT3:%.*]] = fptosi double [[A3]] to i64
63 ; AVX256NODQ-NEXT: [[CVT4:%.*]] = fptosi double [[A4]] to i64
64 ; AVX256NODQ-NEXT: [[CVT5:%.*]] = fptosi double [[A5]] to i64
65 ; AVX256NODQ-NEXT: [[CVT6:%.*]] = fptosi double [[A6]] to i64
66 ; AVX256NODQ-NEXT: [[CVT7:%.*]] = fptosi double [[A7]] to i64
67 ; AVX256NODQ-NEXT: store i64 [[CVT0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
68 ; AVX256NODQ-NEXT: store i64 [[CVT1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
69 ; AVX256NODQ-NEXT: store i64 [[CVT2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
70 ; AVX256NODQ-NEXT: store i64 [[CVT3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
71 ; AVX256NODQ-NEXT: store i64 [[CVT4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
72 ; AVX256NODQ-NEXT: store i64 [[CVT5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
73 ; AVX256NODQ-NEXT: store i64 [[CVT6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
74 ; AVX256NODQ-NEXT: store i64 [[CVT7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
75 ; AVX256NODQ-NEXT: ret void
77 ; AVX512-LABEL: @fptosi_8f64_8i64(
78 ; AVX512-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 8
79 ; AVX512-NEXT: [[TMP2:%.*]] = fptosi <8 x double> [[TMP1]] to <8 x i64>
80 ; AVX512-NEXT: store <8 x i64> [[TMP2]], <8 x i64>* bitcast ([8 x i64]* @dst64 to <8 x i64>*), align 8
81 ; AVX512-NEXT: ret void
83 ; AVX256DQ-LABEL: @fptosi_8f64_8i64(
84 ; AVX256DQ-NEXT: [[TMP1:%.*]] = load <4 x double>, <4 x double>* bitcast ([8 x double]* @src64 to <4 x double>*), align 8
85 ; AVX256DQ-NEXT: [[TMP2:%.*]] = load <4 x double>, <4 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4) to <4 x double>*), align 8
86 ; AVX256DQ-NEXT: [[TMP3:%.*]] = fptosi <4 x double> [[TMP1]] to <4 x i64>
87 ; AVX256DQ-NEXT: [[TMP4:%.*]] = fptosi <4 x double> [[TMP2]] to <4 x i64>
88 ; AVX256DQ-NEXT: store <4 x i64> [[TMP3]], <4 x i64>* bitcast ([8 x i64]* @dst64 to <4 x i64>*), align 8
89 ; AVX256DQ-NEXT: store <4 x i64> [[TMP4]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4) to <4 x i64>*), align 8
90 ; AVX256DQ-NEXT: ret void
92 %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
93 %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
94 %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
95 %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
96 %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
97 %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
98 %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
99 %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
100 %cvt0 = fptosi double %a0 to i64
101 %cvt1 = fptosi double %a1 to i64
102 %cvt2 = fptosi double %a2 to i64
103 %cvt3 = fptosi double %a3 to i64
104 %cvt4 = fptosi double %a4 to i64
105 %cvt5 = fptosi double %a5 to i64
106 %cvt6 = fptosi double %a6 to i64
107 %cvt7 = fptosi double %a7 to i64
108 store i64 %cvt0, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
109 store i64 %cvt1, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
110 store i64 %cvt2, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
111 store i64 %cvt3, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
112 store i64 %cvt4, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
113 store i64 %cvt5, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
114 store i64 %cvt6, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
115 store i64 %cvt7, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
119 define void @fptosi_8f64_8i32() #0 {
120 ; SSE-LABEL: @fptosi_8f64_8i32(
121 ; SSE-NEXT: [[TMP1:%.*]] = load <4 x double>, <4 x double>* bitcast ([8 x double]* @src64 to <4 x double>*), align 8
122 ; SSE-NEXT: [[TMP2:%.*]] = load <4 x double>, <4 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4) to <4 x double>*), align 8
123 ; SSE-NEXT: [[TMP3:%.*]] = fptosi <4 x double> [[TMP1]] to <4 x i32>
124 ; SSE-NEXT: [[TMP4:%.*]] = fptosi <4 x double> [[TMP2]] to <4 x i32>
125 ; SSE-NEXT: store <4 x i32> [[TMP3]], <4 x i32>* bitcast ([16 x i32]* @dst32 to <4 x i32>*), align 4
126 ; SSE-NEXT: store <4 x i32> [[TMP4]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4) to <4 x i32>*), align 4
129 ; AVX-LABEL: @fptosi_8f64_8i32(
130 ; AVX-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 8
131 ; AVX-NEXT: [[TMP2:%.*]] = fptosi <8 x double> [[TMP1]] to <8 x i32>
132 ; AVX-NEXT: store <8 x i32> [[TMP2]], <8 x i32>* bitcast ([16 x i32]* @dst32 to <8 x i32>*), align 4
135 %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
136 %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
137 %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
138 %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
139 %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
140 %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
141 %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
142 %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
143 %cvt0 = fptosi double %a0 to i32
144 %cvt1 = fptosi double %a1 to i32
145 %cvt2 = fptosi double %a2 to i32
146 %cvt3 = fptosi double %a3 to i32
147 %cvt4 = fptosi double %a4 to i32
148 %cvt5 = fptosi double %a5 to i32
149 %cvt6 = fptosi double %a6 to i32
150 %cvt7 = fptosi double %a7 to i32
151 store i32 %cvt0, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 0), align 4
152 store i32 %cvt1, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 1), align 4
153 store i32 %cvt2, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 2), align 4
154 store i32 %cvt3, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 3), align 4
155 store i32 %cvt4, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4), align 4
156 store i32 %cvt5, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 5), align 4
157 store i32 %cvt6, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 6), align 4
158 store i32 %cvt7, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 7), align 4
162 define void @fptosi_8f64_8i16() #0 {
163 ; SSE-LABEL: @fptosi_8f64_8i16(
164 ; SSE-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
165 ; SSE-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
166 ; SSE-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
167 ; SSE-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
168 ; SSE-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
169 ; SSE-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
170 ; SSE-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
171 ; SSE-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
172 ; SSE-NEXT: [[CVT0:%.*]] = fptosi double [[A0]] to i16
173 ; SSE-NEXT: [[CVT1:%.*]] = fptosi double [[A1]] to i16
174 ; SSE-NEXT: [[CVT2:%.*]] = fptosi double [[A2]] to i16
175 ; SSE-NEXT: [[CVT3:%.*]] = fptosi double [[A3]] to i16
176 ; SSE-NEXT: [[CVT4:%.*]] = fptosi double [[A4]] to i16
177 ; SSE-NEXT: [[CVT5:%.*]] = fptosi double [[A5]] to i16
178 ; SSE-NEXT: [[CVT6:%.*]] = fptosi double [[A6]] to i16
179 ; SSE-NEXT: [[CVT7:%.*]] = fptosi double [[A7]] to i16
180 ; SSE-NEXT: store i16 [[CVT0]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 0), align 2
181 ; SSE-NEXT: store i16 [[CVT1]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 1), align 2
182 ; SSE-NEXT: store i16 [[CVT2]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 2), align 2
183 ; SSE-NEXT: store i16 [[CVT3]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 3), align 2
184 ; SSE-NEXT: store i16 [[CVT4]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 4), align 2
185 ; SSE-NEXT: store i16 [[CVT5]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 5), align 2
186 ; SSE-NEXT: store i16 [[CVT6]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 6), align 2
187 ; SSE-NEXT: store i16 [[CVT7]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 7), align 2
190 ; AVX-LABEL: @fptosi_8f64_8i16(
191 ; AVX-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 8
192 ; AVX-NEXT: [[TMP2:%.*]] = fptosi <8 x double> [[TMP1]] to <8 x i16>
193 ; AVX-NEXT: store <8 x i16> [[TMP2]], <8 x i16>* bitcast ([32 x i16]* @dst16 to <8 x i16>*), align 2
196 %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
197 %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
198 %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
199 %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
200 %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
201 %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
202 %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
203 %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
204 %cvt0 = fptosi double %a0 to i16
205 %cvt1 = fptosi double %a1 to i16
206 %cvt2 = fptosi double %a2 to i16
207 %cvt3 = fptosi double %a3 to i16
208 %cvt4 = fptosi double %a4 to i16
209 %cvt5 = fptosi double %a5 to i16
210 %cvt6 = fptosi double %a6 to i16
211 %cvt7 = fptosi double %a7 to i16
212 store i16 %cvt0, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 0), align 2
213 store i16 %cvt1, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 1), align 2
214 store i16 %cvt2, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 2), align 2
215 store i16 %cvt3, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 3), align 2
216 store i16 %cvt4, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 4), align 2
217 store i16 %cvt5, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 5), align 2
218 store i16 %cvt6, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 6), align 2
219 store i16 %cvt7, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 7), align 2
223 define void @fptosi_8f64_8i8() #0 {
224 ; SSE-LABEL: @fptosi_8f64_8i8(
225 ; SSE-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
226 ; SSE-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
227 ; SSE-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
228 ; SSE-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
229 ; SSE-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
230 ; SSE-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
231 ; SSE-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
232 ; SSE-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
233 ; SSE-NEXT: [[CVT0:%.*]] = fptosi double [[A0]] to i8
234 ; SSE-NEXT: [[CVT1:%.*]] = fptosi double [[A1]] to i8
235 ; SSE-NEXT: [[CVT2:%.*]] = fptosi double [[A2]] to i8
236 ; SSE-NEXT: [[CVT3:%.*]] = fptosi double [[A3]] to i8
237 ; SSE-NEXT: [[CVT4:%.*]] = fptosi double [[A4]] to i8
238 ; SSE-NEXT: [[CVT5:%.*]] = fptosi double [[A5]] to i8
239 ; SSE-NEXT: [[CVT6:%.*]] = fptosi double [[A6]] to i8
240 ; SSE-NEXT: [[CVT7:%.*]] = fptosi double [[A7]] to i8
241 ; SSE-NEXT: store i8 [[CVT0]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 0), align 1
242 ; SSE-NEXT: store i8 [[CVT1]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 1), align 1
243 ; SSE-NEXT: store i8 [[CVT2]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 2), align 1
244 ; SSE-NEXT: store i8 [[CVT3]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 3), align 1
245 ; SSE-NEXT: store i8 [[CVT4]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 4), align 1
246 ; SSE-NEXT: store i8 [[CVT5]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 5), align 1
247 ; SSE-NEXT: store i8 [[CVT6]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 6), align 1
248 ; SSE-NEXT: store i8 [[CVT7]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 7), align 1
251 ; AVX-LABEL: @fptosi_8f64_8i8(
252 ; AVX-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 8
253 ; AVX-NEXT: [[TMP2:%.*]] = fptosi <8 x double> [[TMP1]] to <8 x i8>
254 ; AVX-NEXT: store <8 x i8> [[TMP2]], <8 x i8>* bitcast ([64 x i8]* @dst8 to <8 x i8>*), align 1
257 %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
258 %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
259 %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
260 %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
261 %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
262 %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
263 %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
264 %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
265 %cvt0 = fptosi double %a0 to i8
266 %cvt1 = fptosi double %a1 to i8
267 %cvt2 = fptosi double %a2 to i8
268 %cvt3 = fptosi double %a3 to i8
269 %cvt4 = fptosi double %a4 to i8
270 %cvt5 = fptosi double %a5 to i8
271 %cvt6 = fptosi double %a6 to i8
272 %cvt7 = fptosi double %a7 to i8
273 store i8 %cvt0, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 0), align 1
274 store i8 %cvt1, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 1), align 1
275 store i8 %cvt2, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 2), align 1
276 store i8 %cvt3, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 3), align 1
277 store i8 %cvt4, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 4), align 1
278 store i8 %cvt5, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 5), align 1
279 store i8 %cvt6, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 6), align 1
280 store i8 %cvt7, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 7), align 1
288 define void @fptosi_8f32_8i64() #0 {
289 ; SSE-LABEL: @fptosi_8f32_8i64(
290 ; SSE-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
291 ; SSE-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
292 ; SSE-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
293 ; SSE-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
294 ; SSE-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
295 ; SSE-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
296 ; SSE-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
297 ; SSE-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
298 ; SSE-NEXT: [[CVT0:%.*]] = fptosi float [[A0]] to i64
299 ; SSE-NEXT: [[CVT1:%.*]] = fptosi float [[A1]] to i64
300 ; SSE-NEXT: [[CVT2:%.*]] = fptosi float [[A2]] to i64
301 ; SSE-NEXT: [[CVT3:%.*]] = fptosi float [[A3]] to i64
302 ; SSE-NEXT: [[CVT4:%.*]] = fptosi float [[A4]] to i64
303 ; SSE-NEXT: [[CVT5:%.*]] = fptosi float [[A5]] to i64
304 ; SSE-NEXT: [[CVT6:%.*]] = fptosi float [[A6]] to i64
305 ; SSE-NEXT: [[CVT7:%.*]] = fptosi float [[A7]] to i64
306 ; SSE-NEXT: store i64 [[CVT0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
307 ; SSE-NEXT: store i64 [[CVT1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
308 ; SSE-NEXT: store i64 [[CVT2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
309 ; SSE-NEXT: store i64 [[CVT3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
310 ; SSE-NEXT: store i64 [[CVT4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
311 ; SSE-NEXT: store i64 [[CVT5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
312 ; SSE-NEXT: store i64 [[CVT6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
313 ; SSE-NEXT: store i64 [[CVT7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
316 ; AVX256NODQ-LABEL: @fptosi_8f32_8i64(
317 ; AVX256NODQ-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
318 ; AVX256NODQ-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
319 ; AVX256NODQ-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
320 ; AVX256NODQ-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
321 ; AVX256NODQ-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
322 ; AVX256NODQ-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
323 ; AVX256NODQ-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
324 ; AVX256NODQ-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
325 ; AVX256NODQ-NEXT: [[CVT0:%.*]] = fptosi float [[A0]] to i64
326 ; AVX256NODQ-NEXT: [[CVT1:%.*]] = fptosi float [[A1]] to i64
327 ; AVX256NODQ-NEXT: [[CVT2:%.*]] = fptosi float [[A2]] to i64
328 ; AVX256NODQ-NEXT: [[CVT3:%.*]] = fptosi float [[A3]] to i64
329 ; AVX256NODQ-NEXT: [[CVT4:%.*]] = fptosi float [[A4]] to i64
330 ; AVX256NODQ-NEXT: [[CVT5:%.*]] = fptosi float [[A5]] to i64
331 ; AVX256NODQ-NEXT: [[CVT6:%.*]] = fptosi float [[A6]] to i64
332 ; AVX256NODQ-NEXT: [[CVT7:%.*]] = fptosi float [[A7]] to i64
333 ; AVX256NODQ-NEXT: store i64 [[CVT0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
334 ; AVX256NODQ-NEXT: store i64 [[CVT1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
335 ; AVX256NODQ-NEXT: store i64 [[CVT2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
336 ; AVX256NODQ-NEXT: store i64 [[CVT3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
337 ; AVX256NODQ-NEXT: store i64 [[CVT4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
338 ; AVX256NODQ-NEXT: store i64 [[CVT5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
339 ; AVX256NODQ-NEXT: store i64 [[CVT6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
340 ; AVX256NODQ-NEXT: store i64 [[CVT7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
341 ; AVX256NODQ-NEXT: ret void
343 ; AVX512-LABEL: @fptosi_8f32_8i64(
344 ; AVX512-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
345 ; AVX512-NEXT: [[TMP2:%.*]] = fptosi <8 x float> [[TMP1]] to <8 x i64>
346 ; AVX512-NEXT: store <8 x i64> [[TMP2]], <8 x i64>* bitcast ([8 x i64]* @dst64 to <8 x i64>*), align 8
347 ; AVX512-NEXT: ret void
349 ; AVX256DQ-LABEL: @fptosi_8f32_8i64(
350 ; AVX256DQ-NEXT: [[TMP1:%.*]] = load <4 x float>, <4 x float>* bitcast ([16 x float]* @src32 to <4 x float>*), align 4
351 ; AVX256DQ-NEXT: [[TMP2:%.*]] = load <4 x float>, <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4) to <4 x float>*), align 4
352 ; AVX256DQ-NEXT: [[TMP3:%.*]] = fptosi <4 x float> [[TMP1]] to <4 x i64>
353 ; AVX256DQ-NEXT: [[TMP4:%.*]] = fptosi <4 x float> [[TMP2]] to <4 x i64>
354 ; AVX256DQ-NEXT: store <4 x i64> [[TMP3]], <4 x i64>* bitcast ([8 x i64]* @dst64 to <4 x i64>*), align 8
355 ; AVX256DQ-NEXT: store <4 x i64> [[TMP4]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4) to <4 x i64>*), align 8
356 ; AVX256DQ-NEXT: ret void
358 %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
359 %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
360 %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
361 %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
362 %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
363 %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
364 %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
365 %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
366 %cvt0 = fptosi float %a0 to i64
367 %cvt1 = fptosi float %a1 to i64
368 %cvt2 = fptosi float %a2 to i64
369 %cvt3 = fptosi float %a3 to i64
370 %cvt4 = fptosi float %a4 to i64
371 %cvt5 = fptosi float %a5 to i64
372 %cvt6 = fptosi float %a6 to i64
373 %cvt7 = fptosi float %a7 to i64
374 store i64 %cvt0, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
375 store i64 %cvt1, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
376 store i64 %cvt2, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
377 store i64 %cvt3, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
378 store i64 %cvt4, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
379 store i64 %cvt5, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
380 store i64 %cvt6, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
381 store i64 %cvt7, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
385 define void @fptosi_8f32_8i32() #0 {
386 ; SSE-LABEL: @fptosi_8f32_8i32(
387 ; SSE-NEXT: [[TMP1:%.*]] = load <4 x float>, <4 x float>* bitcast ([16 x float]* @src32 to <4 x float>*), align 4
388 ; SSE-NEXT: [[TMP2:%.*]] = load <4 x float>, <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4) to <4 x float>*), align 4
389 ; SSE-NEXT: [[TMP3:%.*]] = fptosi <4 x float> [[TMP1]] to <4 x i32>
390 ; SSE-NEXT: [[TMP4:%.*]] = fptosi <4 x float> [[TMP2]] to <4 x i32>
391 ; SSE-NEXT: store <4 x i32> [[TMP3]], <4 x i32>* bitcast ([16 x i32]* @dst32 to <4 x i32>*), align 4
392 ; SSE-NEXT: store <4 x i32> [[TMP4]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4) to <4 x i32>*), align 4
395 ; AVX-LABEL: @fptosi_8f32_8i32(
396 ; AVX-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
397 ; AVX-NEXT: [[TMP2:%.*]] = fptosi <8 x float> [[TMP1]] to <8 x i32>
398 ; AVX-NEXT: store <8 x i32> [[TMP2]], <8 x i32>* bitcast ([16 x i32]* @dst32 to <8 x i32>*), align 4
401 %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
402 %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
403 %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
404 %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
405 %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
406 %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
407 %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
408 %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
409 %cvt0 = fptosi float %a0 to i32
410 %cvt1 = fptosi float %a1 to i32
411 %cvt2 = fptosi float %a2 to i32
412 %cvt3 = fptosi float %a3 to i32
413 %cvt4 = fptosi float %a4 to i32
414 %cvt5 = fptosi float %a5 to i32
415 %cvt6 = fptosi float %a6 to i32
416 %cvt7 = fptosi float %a7 to i32
417 store i32 %cvt0, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 0), align 4
418 store i32 %cvt1, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 1), align 4
419 store i32 %cvt2, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 2), align 4
420 store i32 %cvt3, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 3), align 4
421 store i32 %cvt4, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4), align 4
422 store i32 %cvt5, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 5), align 4
423 store i32 %cvt6, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 6), align 4
424 store i32 %cvt7, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 7), align 4
428 define void @fptosi_8f32_8i16() #0 {
429 ; CHECK-LABEL: @fptosi_8f32_8i16(
430 ; CHECK-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
431 ; CHECK-NEXT: [[TMP2:%.*]] = fptosi <8 x float> [[TMP1]] to <8 x i16>
432 ; CHECK-NEXT: store <8 x i16> [[TMP2]], <8 x i16>* bitcast ([32 x i16]* @dst16 to <8 x i16>*), align 2
433 ; CHECK-NEXT: ret void
435 %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
436 %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
437 %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
438 %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
439 %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
440 %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
441 %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
442 %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
443 %cvt0 = fptosi float %a0 to i16
444 %cvt1 = fptosi float %a1 to i16
445 %cvt2 = fptosi float %a2 to i16
446 %cvt3 = fptosi float %a3 to i16
447 %cvt4 = fptosi float %a4 to i16
448 %cvt5 = fptosi float %a5 to i16
449 %cvt6 = fptosi float %a6 to i16
450 %cvt7 = fptosi float %a7 to i16
451 store i16 %cvt0, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 0), align 2
452 store i16 %cvt1, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 1), align 2
453 store i16 %cvt2, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 2), align 2
454 store i16 %cvt3, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 3), align 2
455 store i16 %cvt4, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 4), align 2
456 store i16 %cvt5, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 5), align 2
457 store i16 %cvt6, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 6), align 2
458 store i16 %cvt7, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 7), align 2
462 define void @fptosi_8f32_8i8() #0 {
463 ; CHECK-LABEL: @fptosi_8f32_8i8(
464 ; CHECK-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
465 ; CHECK-NEXT: [[TMP2:%.*]] = fptosi <8 x float> [[TMP1]] to <8 x i8>
466 ; CHECK-NEXT: store <8 x i8> [[TMP2]], <8 x i8>* bitcast ([64 x i8]* @dst8 to <8 x i8>*), align 1
467 ; CHECK-NEXT: ret void
469 %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
470 %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
471 %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
472 %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
473 %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
474 %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
475 %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
476 %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
477 %cvt0 = fptosi float %a0 to i8
478 %cvt1 = fptosi float %a1 to i8
479 %cvt2 = fptosi float %a2 to i8
480 %cvt3 = fptosi float %a3 to i8
481 %cvt4 = fptosi float %a4 to i8
482 %cvt5 = fptosi float %a5 to i8
483 %cvt6 = fptosi float %a6 to i8
484 %cvt7 = fptosi float %a7 to i8
485 store i8 %cvt0, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 0), align 1
486 store i8 %cvt1, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 1), align 1
487 store i8 %cvt2, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 2), align 1
488 store i8 %cvt3, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 3), align 1
489 store i8 %cvt4, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 4), align 1
490 store i8 %cvt5, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 5), align 1
491 store i8 %cvt6, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 6), align 1
492 store i8 %cvt7, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 7), align 1
500 define <4 x i32> @fptosi_4xf64_4i32(double %a0, double %a1, double %a2, double %a3) #0 {
501 ; CHECK-LABEL: @fptosi_4xf64_4i32(
502 ; CHECK-NEXT: [[CVT0:%.*]] = fptosi double [[A0:%.*]] to i32
503 ; CHECK-NEXT: [[CVT1:%.*]] = fptosi double [[A1:%.*]] to i32
504 ; CHECK-NEXT: [[CVT2:%.*]] = fptosi double [[A2:%.*]] to i32
505 ; CHECK-NEXT: [[CVT3:%.*]] = fptosi double [[A3:%.*]] to i32
506 ; CHECK-NEXT: [[RES0:%.*]] = insertelement <4 x i32> undef, i32 [[CVT0]], i32 0
507 ; CHECK-NEXT: [[RES1:%.*]] = insertelement <4 x i32> [[RES0]], i32 [[CVT1]], i32 1
508 ; CHECK-NEXT: [[RES2:%.*]] = insertelement <4 x i32> [[RES1]], i32 [[CVT2]], i32 2
509 ; CHECK-NEXT: [[RES3:%.*]] = insertelement <4 x i32> [[RES2]], i32 [[CVT3]], i32 3
510 ; CHECK-NEXT: ret <4 x i32> [[RES3]]
512 %cvt0 = fptosi double %a0 to i32
513 %cvt1 = fptosi double %a1 to i32
514 %cvt2 = fptosi double %a2 to i32
515 %cvt3 = fptosi double %a3 to i32
516 %res0 = insertelement <4 x i32> undef, i32 %cvt0, i32 0
517 %res1 = insertelement <4 x i32> %res0, i32 %cvt1, i32 1
518 %res2 = insertelement <4 x i32> %res1, i32 %cvt2, i32 2
519 %res3 = insertelement <4 x i32> %res2, i32 %cvt3, i32 3
523 define <4 x i32> @fptosi_4xf32_4i32(float %a0, float %a1, float %a2, float %a3) #0 {
524 ; CHECK-LABEL: @fptosi_4xf32_4i32(
525 ; CHECK-NEXT: [[CVT0:%.*]] = fptosi float [[A0:%.*]] to i32
526 ; CHECK-NEXT: [[CVT1:%.*]] = fptosi float [[A1:%.*]] to i32
527 ; CHECK-NEXT: [[CVT2:%.*]] = fptosi float [[A2:%.*]] to i32
528 ; CHECK-NEXT: [[CVT3:%.*]] = fptosi float [[A3:%.*]] to i32
529 ; CHECK-NEXT: [[RES0:%.*]] = insertelement <4 x i32> undef, i32 [[CVT0]], i32 0
530 ; CHECK-NEXT: [[RES1:%.*]] = insertelement <4 x i32> [[RES0]], i32 [[CVT1]], i32 1
531 ; CHECK-NEXT: [[RES2:%.*]] = insertelement <4 x i32> [[RES1]], i32 [[CVT2]], i32 2
532 ; CHECK-NEXT: [[RES3:%.*]] = insertelement <4 x i32> [[RES2]], i32 [[CVT3]], i32 3
533 ; CHECK-NEXT: ret <4 x i32> [[RES3]]
535 %cvt0 = fptosi float %a0 to i32
536 %cvt1 = fptosi float %a1 to i32
537 %cvt2 = fptosi float %a2 to i32
538 %cvt3 = fptosi float %a3 to i32
539 %res0 = insertelement <4 x i32> undef, i32 %cvt0, i32 0
540 %res1 = insertelement <4 x i32> %res0, i32 %cvt1, i32 1
541 %res2 = insertelement <4 x i32> %res1, i32 %cvt2, i32 2
542 %res3 = insertelement <4 x i32> %res2, i32 %cvt3, i32 3
546 attributes #0 = { nounwind }