| blob | 33d998ffbbc7e5eaf06823c6b12ff0d1d269500f |
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
18 ;
19 ; FPTOUI vXf64
20 ;
22 define void @fptoui_8f64_8i64() #0 {
23 ; SSE-LABEL: @fptoui_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:%.*]] = fptoui double [[A0]] to i64
33 ; SSE-NEXT: [[CVT1:%.*]] = fptoui double [[A1]] to i64
34 ; SSE-NEXT: [[CVT2:%.*]] = fptoui double [[A2]] to i64
35 ; SSE-NEXT: [[CVT3:%.*]] = fptoui double [[A3]] to i64
36 ; SSE-NEXT: [[CVT4:%.*]] = fptoui double [[A4]] to i64
37 ; SSE-NEXT: [[CVT5:%.*]] = fptoui double [[A5]] to i64
38 ; SSE-NEXT: [[CVT6:%.*]] = fptoui double [[A6]] to i64
39 ; SSE-NEXT: [[CVT7:%.*]] = fptoui 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
48 ; SSE-NEXT: ret void
49 ;
50 ; AVX256NODQ-LABEL: @fptoui_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:%.*]] = fptoui double [[A0]] to i64
60 ; AVX256NODQ-NEXT: [[CVT1:%.*]] = fptoui double [[A1]] to i64
61 ; AVX256NODQ-NEXT: [[CVT2:%.*]] = fptoui double [[A2]] to i64
62 ; AVX256NODQ-NEXT: [[CVT3:%.*]] = fptoui double [[A3]] to i64
63 ; AVX256NODQ-NEXT: [[CVT4:%.*]] = fptoui double [[A4]] to i64
64 ; AVX256NODQ-NEXT: [[CVT5:%.*]] = fptoui double [[A5]] to i64
65 ; AVX256NODQ-NEXT: [[CVT6:%.*]] = fptoui double [[A6]] to i64
66 ; AVX256NODQ-NEXT: [[CVT7:%.*]] = fptoui 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
76 ;
77 ; AVX512-LABEL: @fptoui_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:%.*]] = fptoui <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
82 ;
83 ; AVX256DQ-LABEL: @fptoui_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:%.*]] = fptoui <4 x double> [[TMP1]] to <4 x i64>
87 ; AVX256DQ-NEXT: [[TMP4:%.*]] = fptoui <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
91 ;
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 = fptoui double %a0 to i64
101 %cvt1 = fptoui double %a1 to i64
102 %cvt2 = fptoui double %a2 to i64
103 %cvt3 = fptoui double %a3 to i64
104 %cvt4 = fptoui double %a4 to i64
105 %cvt5 = fptoui double %a5 to i64
106 %cvt6 = fptoui double %a6 to i64
107 %cvt7 = fptoui 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
116 ret void
117 }
119 define void @fptoui_8f64_8i32() #0 {
120 ; SSE-LABEL: @fptoui_8f64_8i32(
121 ; SSE-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
122 ; SSE-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
123 ; SSE-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
124 ; SSE-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
125 ; SSE-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
126 ; SSE-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
127 ; SSE-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
128 ; SSE-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
129 ; SSE-NEXT: [[CVT0:%.*]] = fptoui double [[A0]] to i32
130 ; SSE-NEXT: [[CVT1:%.*]] = fptoui double [[A1]] to i32
131 ; SSE-NEXT: [[CVT2:%.*]] = fptoui double [[A2]] to i32
132 ; SSE-NEXT: [[CVT3:%.*]] = fptoui double [[A3]] to i32
133 ; SSE-NEXT: [[CVT4:%.*]] = fptoui double [[A4]] to i32
134 ; SSE-NEXT: [[CVT5:%.*]] = fptoui double [[A5]] to i32
135 ; SSE-NEXT: [[CVT6:%.*]] = fptoui double [[A6]] to i32
136 ; SSE-NEXT: [[CVT7:%.*]] = fptoui double [[A7]] to i32
137 ; SSE-NEXT: store i32 [[CVT0]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 0), align 4
138 ; SSE-NEXT: store i32 [[CVT1]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 1), align 4
139 ; SSE-NEXT: store i32 [[CVT2]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 2), align 4
140 ; SSE-NEXT: store i32 [[CVT3]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 3), align 4
141 ; SSE-NEXT: store i32 [[CVT4]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4), align 4
142 ; SSE-NEXT: store i32 [[CVT5]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 5), align 4
143 ; SSE-NEXT: store i32 [[CVT6]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 6), align 4
144 ; SSE-NEXT: store i32 [[CVT7]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 7), align 4
145 ; SSE-NEXT: ret void
146 ;
147 ; AVX256NODQ-LABEL: @fptoui_8f64_8i32(
148 ; AVX256NODQ-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
149 ; AVX256NODQ-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
150 ; AVX256NODQ-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
151 ; AVX256NODQ-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
152 ; AVX256NODQ-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
153 ; AVX256NODQ-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
154 ; AVX256NODQ-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
155 ; AVX256NODQ-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
156 ; AVX256NODQ-NEXT: [[CVT0:%.*]] = fptoui double [[A0]] to i32
157 ; AVX256NODQ-NEXT: [[CVT1:%.*]] = fptoui double [[A1]] to i32
158 ; AVX256NODQ-NEXT: [[CVT2:%.*]] = fptoui double [[A2]] to i32
159 ; AVX256NODQ-NEXT: [[CVT3:%.*]] = fptoui double [[A3]] to i32
160 ; AVX256NODQ-NEXT: [[CVT4:%.*]] = fptoui double [[A4]] to i32
161 ; AVX256NODQ-NEXT: [[CVT5:%.*]] = fptoui double [[A5]] to i32
162 ; AVX256NODQ-NEXT: [[CVT6:%.*]] = fptoui double [[A6]] to i32
163 ; AVX256NODQ-NEXT: [[CVT7:%.*]] = fptoui double [[A7]] to i32
164 ; AVX256NODQ-NEXT: store i32 [[CVT0]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 0), align 4
165 ; AVX256NODQ-NEXT: store i32 [[CVT1]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 1), align 4
166 ; AVX256NODQ-NEXT: store i32 [[CVT2]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 2), align 4
167 ; AVX256NODQ-NEXT: store i32 [[CVT3]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 3), align 4
168 ; AVX256NODQ-NEXT: store i32 [[CVT4]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4), align 4
169 ; AVX256NODQ-NEXT: store i32 [[CVT5]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 5), align 4
170 ; AVX256NODQ-NEXT: store i32 [[CVT6]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 6), align 4
171 ; AVX256NODQ-NEXT: store i32 [[CVT7]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 7), align 4
172 ; AVX256NODQ-NEXT: ret void
173 ;
174 ; AVX512-LABEL: @fptoui_8f64_8i32(
175 ; AVX512-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 8
176 ; AVX512-NEXT: [[TMP2:%.*]] = fptoui <8 x double> [[TMP1]] to <8 x i32>
177 ; AVX512-NEXT: store <8 x i32> [[TMP2]], <8 x i32>* bitcast ([16 x i32]* @dst32 to <8 x i32>*), align 4
178 ; AVX512-NEXT: ret void
179 ;
180 ; AVX256DQ-LABEL: @fptoui_8f64_8i32(
181 ; AVX256DQ-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 8
182 ; AVX256DQ-NEXT: [[TMP2:%.*]] = fptoui <8 x double> [[TMP1]] to <8 x i32>
183 ; AVX256DQ-NEXT: store <8 x i32> [[TMP2]], <8 x i32>* bitcast ([16 x i32]* @dst32 to <8 x i32>*), align 4
184 ; AVX256DQ-NEXT: ret void
185 ;
186 %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
187 %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
188 %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
189 %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
190 %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
191 %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
192 %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
193 %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
194 %cvt0 = fptoui double %a0 to i32
195 %cvt1 = fptoui double %a1 to i32
196 %cvt2 = fptoui double %a2 to i32
197 %cvt3 = fptoui double %a3 to i32
198 %cvt4 = fptoui double %a4 to i32
199 %cvt5 = fptoui double %a5 to i32
200 %cvt6 = fptoui double %a6 to i32
201 %cvt7 = fptoui double %a7 to i32
202 store i32 %cvt0, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 0), align 4
203 store i32 %cvt1, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 1), align 4
204 store i32 %cvt2, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 2), align 4
205 store i32 %cvt3, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 3), align 4
206 store i32 %cvt4, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4), align 4
207 store i32 %cvt5, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 5), align 4
208 store i32 %cvt6, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 6), align 4
209 store i32 %cvt7, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 7), align 4
210 ret void
211 }
213 define void @fptoui_8f64_8i16() #0 {
214 ; SSE-LABEL: @fptoui_8f64_8i16(
215 ; SSE-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
216 ; SSE-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
217 ; SSE-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
218 ; SSE-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
219 ; SSE-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
220 ; SSE-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
221 ; SSE-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
222 ; SSE-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
223 ; SSE-NEXT: [[CVT0:%.*]] = fptoui double [[A0]] to i16
224 ; SSE-NEXT: [[CVT1:%.*]] = fptoui double [[A1]] to i16
225 ; SSE-NEXT: [[CVT2:%.*]] = fptoui double [[A2]] to i16
226 ; SSE-NEXT: [[CVT3:%.*]] = fptoui double [[A3]] to i16
227 ; SSE-NEXT: [[CVT4:%.*]] = fptoui double [[A4]] to i16
228 ; SSE-NEXT: [[CVT5:%.*]] = fptoui double [[A5]] to i16
229 ; SSE-NEXT: [[CVT6:%.*]] = fptoui double [[A6]] to i16
230 ; SSE-NEXT: [[CVT7:%.*]] = fptoui double [[A7]] to i16
231 ; SSE-NEXT: store i16 [[CVT0]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 0), align 2
232 ; SSE-NEXT: store i16 [[CVT1]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 1), align 2
233 ; SSE-NEXT: store i16 [[CVT2]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 2), align 2
234 ; SSE-NEXT: store i16 [[CVT3]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 3), align 2
235 ; SSE-NEXT: store i16 [[CVT4]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 4), align 2
236 ; SSE-NEXT: store i16 [[CVT5]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 5), align 2
237 ; SSE-NEXT: store i16 [[CVT6]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 6), align 2
238 ; SSE-NEXT: store i16 [[CVT7]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 7), align 2
239 ; SSE-NEXT: ret void
240 ;
241 ; AVX256NODQ-LABEL: @fptoui_8f64_8i16(
242 ; AVX256NODQ-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
243 ; AVX256NODQ-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
244 ; AVX256NODQ-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
245 ; AVX256NODQ-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
246 ; AVX256NODQ-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
247 ; AVX256NODQ-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
248 ; AVX256NODQ-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
249 ; AVX256NODQ-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
250 ; AVX256NODQ-NEXT: [[CVT0:%.*]] = fptoui double [[A0]] to i16
251 ; AVX256NODQ-NEXT: [[CVT1:%.*]] = fptoui double [[A1]] to i16
252 ; AVX256NODQ-NEXT: [[CVT2:%.*]] = fptoui double [[A2]] to i16
253 ; AVX256NODQ-NEXT: [[CVT3:%.*]] = fptoui double [[A3]] to i16
254 ; AVX256NODQ-NEXT: [[CVT4:%.*]] = fptoui double [[A4]] to i16
255 ; AVX256NODQ-NEXT: [[CVT5:%.*]] = fptoui double [[A5]] to i16
256 ; AVX256NODQ-NEXT: [[CVT6:%.*]] = fptoui double [[A6]] to i16
257 ; AVX256NODQ-NEXT: [[CVT7:%.*]] = fptoui double [[A7]] to i16
258 ; AVX256NODQ-NEXT: store i16 [[CVT0]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 0), align 2
259 ; AVX256NODQ-NEXT: store i16 [[CVT1]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 1), align 2
260 ; AVX256NODQ-NEXT: store i16 [[CVT2]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 2), align 2
261 ; AVX256NODQ-NEXT: store i16 [[CVT3]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 3), align 2
262 ; AVX256NODQ-NEXT: store i16 [[CVT4]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 4), align 2
263 ; AVX256NODQ-NEXT: store i16 [[CVT5]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 5), align 2
264 ; AVX256NODQ-NEXT: store i16 [[CVT6]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 6), align 2
265 ; AVX256NODQ-NEXT: store i16 [[CVT7]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 7), align 2
266 ; AVX256NODQ-NEXT: ret void
267 ;
268 ; AVX512-LABEL: @fptoui_8f64_8i16(
269 ; AVX512-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 8
270 ; AVX512-NEXT: [[TMP2:%.*]] = fptoui <8 x double> [[TMP1]] to <8 x i16>
271 ; AVX512-NEXT: store <8 x i16> [[TMP2]], <8 x i16>* bitcast ([32 x i16]* @dst16 to <8 x i16>*), align 2
272 ; AVX512-NEXT: ret void
273 ;
274 ; AVX256DQ-LABEL: @fptoui_8f64_8i16(
275 ; AVX256DQ-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 8
276 ; AVX256DQ-NEXT: [[TMP2:%.*]] = fptoui <8 x double> [[TMP1]] to <8 x i16>
277 ; AVX256DQ-NEXT: store <8 x i16> [[TMP2]], <8 x i16>* bitcast ([32 x i16]* @dst16 to <8 x i16>*), align 2
278 ; AVX256DQ-NEXT: ret void
279 ;
280 %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
281 %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
282 %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
283 %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
284 %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
285 %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
286 %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
287 %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
288 %cvt0 = fptoui double %a0 to i16
289 %cvt1 = fptoui double %a1 to i16
290 %cvt2 = fptoui double %a2 to i16
291 %cvt3 = fptoui double %a3 to i16
292 %cvt4 = fptoui double %a4 to i16
293 %cvt5 = fptoui double %a5 to i16
294 %cvt6 = fptoui double %a6 to i16
295 %cvt7 = fptoui double %a7 to i16
296 store i16 %cvt0, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 0), align 2
297 store i16 %cvt1, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 1), align 2
298 store i16 %cvt2, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 2), align 2
299 store i16 %cvt3, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 3), align 2
300 store i16 %cvt4, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 4), align 2
301 store i16 %cvt5, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 5), align 2
302 store i16 %cvt6, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 6), align 2
303 store i16 %cvt7, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 7), align 2
304 ret void
305 }
307 define void @fptoui_8f64_8i8() #0 {
308 ; CHECK-LABEL: @fptoui_8f64_8i8(
309 ; CHECK-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
310 ; CHECK-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
311 ; CHECK-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
312 ; CHECK-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
313 ; CHECK-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
314 ; CHECK-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
315 ; CHECK-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
316 ; CHECK-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
317 ; CHECK-NEXT: [[CVT0:%.*]] = fptoui double [[A0]] to i8
318 ; CHECK-NEXT: [[CVT1:%.*]] = fptoui double [[A1]] to i8
319 ; CHECK-NEXT: [[CVT2:%.*]] = fptoui double [[A2]] to i8
320 ; CHECK-NEXT: [[CVT3:%.*]] = fptoui double [[A3]] to i8
321 ; CHECK-NEXT: [[CVT4:%.*]] = fptoui double [[A4]] to i8
322 ; CHECK-NEXT: [[CVT5:%.*]] = fptoui double [[A5]] to i8
323 ; CHECK-NEXT: [[CVT6:%.*]] = fptoui double [[A6]] to i8
324 ; CHECK-NEXT: [[CVT7:%.*]] = fptoui double [[A7]] to i8
325 ; CHECK-NEXT: store i8 [[CVT0]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 0), align 1
326 ; CHECK-NEXT: store i8 [[CVT1]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 1), align 1
327 ; CHECK-NEXT: store i8 [[CVT2]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 2), align 1
328 ; CHECK-NEXT: store i8 [[CVT3]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 3), align 1
329 ; CHECK-NEXT: store i8 [[CVT4]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 4), align 1
330 ; CHECK-NEXT: store i8 [[CVT5]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 5), align 1
331 ; CHECK-NEXT: store i8 [[CVT6]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 6), align 1
332 ; CHECK-NEXT: store i8 [[CVT7]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 7), align 1
333 ; CHECK-NEXT: ret void
334 ;
335 %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
336 %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
337 %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
338 %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
339 %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 8
340 %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 8
341 %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 8
342 %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 8
343 %cvt0 = fptoui double %a0 to i8
344 %cvt1 = fptoui double %a1 to i8
345 %cvt2 = fptoui double %a2 to i8
346 %cvt3 = fptoui double %a3 to i8
347 %cvt4 = fptoui double %a4 to i8
348 %cvt5 = fptoui double %a5 to i8
349 %cvt6 = fptoui double %a6 to i8
350 %cvt7 = fptoui double %a7 to i8
351 store i8 %cvt0, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 0), align 1
352 store i8 %cvt1, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 1), align 1
353 store i8 %cvt2, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 2), align 1
354 store i8 %cvt3, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 3), align 1
355 store i8 %cvt4, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 4), align 1
356 store i8 %cvt5, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 5), align 1
357 store i8 %cvt6, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 6), align 1
358 store i8 %cvt7, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 7), align 1
359 ret void
360 }
362 ;
363 ; FPTOUI vXf32
364 ;
366 define void @fptoui_8f32_8i64() #0 {
367 ; SSE-LABEL: @fptoui_8f32_8i64(
368 ; SSE-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
369 ; SSE-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
370 ; SSE-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
371 ; SSE-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
372 ; SSE-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
373 ; SSE-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
374 ; SSE-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
375 ; SSE-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
376 ; SSE-NEXT: [[CVT0:%.*]] = fptoui float [[A0]] to i64
377 ; SSE-NEXT: [[CVT1:%.*]] = fptoui float [[A1]] to i64
378 ; SSE-NEXT: [[CVT2:%.*]] = fptoui float [[A2]] to i64
379 ; SSE-NEXT: [[CVT3:%.*]] = fptoui float [[A3]] to i64
380 ; SSE-NEXT: [[CVT4:%.*]] = fptoui float [[A4]] to i64
381 ; SSE-NEXT: [[CVT5:%.*]] = fptoui float [[A5]] to i64
382 ; SSE-NEXT: [[CVT6:%.*]] = fptoui float [[A6]] to i64
383 ; SSE-NEXT: [[CVT7:%.*]] = fptoui float [[A7]] to i64
384 ; SSE-NEXT: store i64 [[CVT0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
385 ; SSE-NEXT: store i64 [[CVT1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
386 ; SSE-NEXT: store i64 [[CVT2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
387 ; SSE-NEXT: store i64 [[CVT3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
388 ; SSE-NEXT: store i64 [[CVT4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
389 ; SSE-NEXT: store i64 [[CVT5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
390 ; SSE-NEXT: store i64 [[CVT6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
391 ; SSE-NEXT: store i64 [[CVT7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
392 ; SSE-NEXT: ret void
393 ;
394 ; AVX256NODQ-LABEL: @fptoui_8f32_8i64(
395 ; AVX256NODQ-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
396 ; AVX256NODQ-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
397 ; AVX256NODQ-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
398 ; AVX256NODQ-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
399 ; AVX256NODQ-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
400 ; AVX256NODQ-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
401 ; AVX256NODQ-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
402 ; AVX256NODQ-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
403 ; AVX256NODQ-NEXT: [[CVT0:%.*]] = fptoui float [[A0]] to i64
404 ; AVX256NODQ-NEXT: [[CVT1:%.*]] = fptoui float [[A1]] to i64
405 ; AVX256NODQ-NEXT: [[CVT2:%.*]] = fptoui float [[A2]] to i64
406 ; AVX256NODQ-NEXT: [[CVT3:%.*]] = fptoui float [[A3]] to i64
407 ; AVX256NODQ-NEXT: [[CVT4:%.*]] = fptoui float [[A4]] to i64
408 ; AVX256NODQ-NEXT: [[CVT5:%.*]] = fptoui float [[A5]] to i64
409 ; AVX256NODQ-NEXT: [[CVT6:%.*]] = fptoui float [[A6]] to i64
410 ; AVX256NODQ-NEXT: [[CVT7:%.*]] = fptoui float [[A7]] to i64
411 ; AVX256NODQ-NEXT: store i64 [[CVT0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
412 ; AVX256NODQ-NEXT: store i64 [[CVT1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
413 ; AVX256NODQ-NEXT: store i64 [[CVT2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
414 ; AVX256NODQ-NEXT: store i64 [[CVT3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
415 ; AVX256NODQ-NEXT: store i64 [[CVT4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
416 ; AVX256NODQ-NEXT: store i64 [[CVT5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
417 ; AVX256NODQ-NEXT: store i64 [[CVT6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
418 ; AVX256NODQ-NEXT: store i64 [[CVT7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
419 ; AVX256NODQ-NEXT: ret void
420 ;
421 ; AVX512-LABEL: @fptoui_8f32_8i64(
422 ; AVX512-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
423 ; AVX512-NEXT: [[TMP2:%.*]] = fptoui <8 x float> [[TMP1]] to <8 x i64>
424 ; AVX512-NEXT: store <8 x i64> [[TMP2]], <8 x i64>* bitcast ([8 x i64]* @dst64 to <8 x i64>*), align 8
425 ; AVX512-NEXT: ret void
426 ;
427 ; AVX256DQ-LABEL: @fptoui_8f32_8i64(
428 ; AVX256DQ-NEXT: [[TMP1:%.*]] = load <4 x float>, <4 x float>* bitcast ([16 x float]* @src32 to <4 x float>*), align 4
429 ; 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
430 ; AVX256DQ-NEXT: [[TMP3:%.*]] = fptoui <4 x float> [[TMP1]] to <4 x i64>
431 ; AVX256DQ-NEXT: [[TMP4:%.*]] = fptoui <4 x float> [[TMP2]] to <4 x i64>
432 ; AVX256DQ-NEXT: store <4 x i64> [[TMP3]], <4 x i64>* bitcast ([8 x i64]* @dst64 to <4 x i64>*), align 8
433 ; 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
434 ; AVX256DQ-NEXT: ret void
435 ;
436 %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
437 %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
438 %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
439 %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
440 %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
441 %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
442 %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
443 %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
444 %cvt0 = fptoui float %a0 to i64
445 %cvt1 = fptoui float %a1 to i64
446 %cvt2 = fptoui float %a2 to i64
447 %cvt3 = fptoui float %a3 to i64
448 %cvt4 = fptoui float %a4 to i64
449 %cvt5 = fptoui float %a5 to i64
450 %cvt6 = fptoui float %a6 to i64
451 %cvt7 = fptoui float %a7 to i64
452 store i64 %cvt0, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 0), align 8
453 store i64 %cvt1, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 1), align 8
454 store i64 %cvt2, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 2), align 8
455 store i64 %cvt3, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 3), align 8
456 store i64 %cvt4, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 4), align 8
457 store i64 %cvt5, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 5), align 8
458 store i64 %cvt6, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 6), align 8
459 store i64 %cvt7, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @dst64, i32 0, i64 7), align 8
460 ret void
461 }
463 define void @fptoui_8f32_8i32() #0 {
464 ; SSE-LABEL: @fptoui_8f32_8i32(
465 ; SSE-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
466 ; SSE-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
467 ; SSE-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
468 ; SSE-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
469 ; SSE-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
470 ; SSE-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
471 ; SSE-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
472 ; SSE-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
473 ; SSE-NEXT: [[CVT0:%.*]] = fptoui float [[A0]] to i32
474 ; SSE-NEXT: [[CVT1:%.*]] = fptoui float [[A1]] to i32
475 ; SSE-NEXT: [[CVT2:%.*]] = fptoui float [[A2]] to i32
476 ; SSE-NEXT: [[CVT3:%.*]] = fptoui float [[A3]] to i32
477 ; SSE-NEXT: [[CVT4:%.*]] = fptoui float [[A4]] to i32
478 ; SSE-NEXT: [[CVT5:%.*]] = fptoui float [[A5]] to i32
479 ; SSE-NEXT: [[CVT6:%.*]] = fptoui float [[A6]] to i32
480 ; SSE-NEXT: [[CVT7:%.*]] = fptoui float [[A7]] to i32
481 ; SSE-NEXT: store i32 [[CVT0]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 0), align 4
482 ; SSE-NEXT: store i32 [[CVT1]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 1), align 4
483 ; SSE-NEXT: store i32 [[CVT2]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 2), align 4
484 ; SSE-NEXT: store i32 [[CVT3]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 3), align 4
485 ; SSE-NEXT: store i32 [[CVT4]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4), align 4
486 ; SSE-NEXT: store i32 [[CVT5]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 5), align 4
487 ; SSE-NEXT: store i32 [[CVT6]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 6), align 4
488 ; SSE-NEXT: store i32 [[CVT7]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 7), align 4
489 ; SSE-NEXT: ret void
490 ;
491 ; AVX256NODQ-LABEL: @fptoui_8f32_8i32(
492 ; AVX256NODQ-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
493 ; AVX256NODQ-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
494 ; AVX256NODQ-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
495 ; AVX256NODQ-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
496 ; AVX256NODQ-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
497 ; AVX256NODQ-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
498 ; AVX256NODQ-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
499 ; AVX256NODQ-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
500 ; AVX256NODQ-NEXT: [[CVT0:%.*]] = fptoui float [[A0]] to i32
501 ; AVX256NODQ-NEXT: [[CVT1:%.*]] = fptoui float [[A1]] to i32
502 ; AVX256NODQ-NEXT: [[CVT2:%.*]] = fptoui float [[A2]] to i32
503 ; AVX256NODQ-NEXT: [[CVT3:%.*]] = fptoui float [[A3]] to i32
504 ; AVX256NODQ-NEXT: [[CVT4:%.*]] = fptoui float [[A4]] to i32
505 ; AVX256NODQ-NEXT: [[CVT5:%.*]] = fptoui float [[A5]] to i32
506 ; AVX256NODQ-NEXT: [[CVT6:%.*]] = fptoui float [[A6]] to i32
507 ; AVX256NODQ-NEXT: [[CVT7:%.*]] = fptoui float [[A7]] to i32
508 ; AVX256NODQ-NEXT: store i32 [[CVT0]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 0), align 4
509 ; AVX256NODQ-NEXT: store i32 [[CVT1]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 1), align 4
510 ; AVX256NODQ-NEXT: store i32 [[CVT2]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 2), align 4
511 ; AVX256NODQ-NEXT: store i32 [[CVT3]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 3), align 4
512 ; AVX256NODQ-NEXT: store i32 [[CVT4]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4), align 4
513 ; AVX256NODQ-NEXT: store i32 [[CVT5]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 5), align 4
514 ; AVX256NODQ-NEXT: store i32 [[CVT6]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 6), align 4
515 ; AVX256NODQ-NEXT: store i32 [[CVT7]], i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 7), align 4
516 ; AVX256NODQ-NEXT: ret void
517 ;
518 ; AVX512-LABEL: @fptoui_8f32_8i32(
519 ; AVX512-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
520 ; AVX512-NEXT: [[TMP2:%.*]] = fptoui <8 x float> [[TMP1]] to <8 x i32>
521 ; AVX512-NEXT: store <8 x i32> [[TMP2]], <8 x i32>* bitcast ([16 x i32]* @dst32 to <8 x i32>*), align 4
522 ; AVX512-NEXT: ret void
523 ;
524 ; AVX256DQ-LABEL: @fptoui_8f32_8i32(
525 ; AVX256DQ-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
526 ; AVX256DQ-NEXT: [[TMP2:%.*]] = fptoui <8 x float> [[TMP1]] to <8 x i32>
527 ; AVX256DQ-NEXT: store <8 x i32> [[TMP2]], <8 x i32>* bitcast ([16 x i32]* @dst32 to <8 x i32>*), align 4
528 ; AVX256DQ-NEXT: ret void
529 ;
530 %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
531 %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
532 %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
533 %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
534 %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
535 %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
536 %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
537 %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
538 %cvt0 = fptoui float %a0 to i32
539 %cvt1 = fptoui float %a1 to i32
540 %cvt2 = fptoui float %a2 to i32
541 %cvt3 = fptoui float %a3 to i32
542 %cvt4 = fptoui float %a4 to i32
543 %cvt5 = fptoui float %a5 to i32
544 %cvt6 = fptoui float %a6 to i32
545 %cvt7 = fptoui float %a7 to i32
546 store i32 %cvt0, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 0), align 4
547 store i32 %cvt1, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 1), align 4
548 store i32 %cvt2, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 2), align 4
549 store i32 %cvt3, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 3), align 4
550 store i32 %cvt4, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 4), align 4
551 store i32 %cvt5, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 5), align 4
552 store i32 %cvt6, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 6), align 4
553 store i32 %cvt7, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @dst32, i32 0, i64 7), align 4
554 ret void
555 }
557 define void @fptoui_8f32_8i16() #0 {
558 ; SSE-LABEL: @fptoui_8f32_8i16(
559 ; SSE-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
560 ; SSE-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
561 ; SSE-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
562 ; SSE-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
563 ; SSE-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
564 ; SSE-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
565 ; SSE-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
566 ; SSE-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
567 ; SSE-NEXT: [[CVT0:%.*]] = fptoui float [[A0]] to i16
568 ; SSE-NEXT: [[CVT1:%.*]] = fptoui float [[A1]] to i16
569 ; SSE-NEXT: [[CVT2:%.*]] = fptoui float [[A2]] to i16
570 ; SSE-NEXT: [[CVT3:%.*]] = fptoui float [[A3]] to i16
571 ; SSE-NEXT: [[CVT4:%.*]] = fptoui float [[A4]] to i16
572 ; SSE-NEXT: [[CVT5:%.*]] = fptoui float [[A5]] to i16
573 ; SSE-NEXT: [[CVT6:%.*]] = fptoui float [[A6]] to i16
574 ; SSE-NEXT: [[CVT7:%.*]] = fptoui float [[A7]] to i16
575 ; SSE-NEXT: store i16 [[CVT0]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 0), align 2
576 ; SSE-NEXT: store i16 [[CVT1]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 1), align 2
577 ; SSE-NEXT: store i16 [[CVT2]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 2), align 2
578 ; SSE-NEXT: store i16 [[CVT3]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 3), align 2
579 ; SSE-NEXT: store i16 [[CVT4]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 4), align 2
580 ; SSE-NEXT: store i16 [[CVT5]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 5), align 2
581 ; SSE-NEXT: store i16 [[CVT6]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 6), align 2
582 ; SSE-NEXT: store i16 [[CVT7]], i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 7), align 2
583 ; SSE-NEXT: ret void
584 ;
585 ; AVX-LABEL: @fptoui_8f32_8i16(
586 ; AVX-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
587 ; AVX-NEXT: [[TMP2:%.*]] = fptoui <8 x float> [[TMP1]] to <8 x i16>
588 ; AVX-NEXT: store <8 x i16> [[TMP2]], <8 x i16>* bitcast ([32 x i16]* @dst16 to <8 x i16>*), align 2
589 ; AVX-NEXT: ret void
590 ;
591 %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
592 %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
593 %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
594 %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
595 %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
596 %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
597 %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
598 %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
599 %cvt0 = fptoui float %a0 to i16
600 %cvt1 = fptoui float %a1 to i16
601 %cvt2 = fptoui float %a2 to i16
602 %cvt3 = fptoui float %a3 to i16
603 %cvt4 = fptoui float %a4 to i16
604 %cvt5 = fptoui float %a5 to i16
605 %cvt6 = fptoui float %a6 to i16
606 %cvt7 = fptoui float %a7 to i16
607 store i16 %cvt0, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 0), align 2
608 store i16 %cvt1, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 1), align 2
609 store i16 %cvt2, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 2), align 2
610 store i16 %cvt3, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 3), align 2
611 store i16 %cvt4, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 4), align 2
612 store i16 %cvt5, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 5), align 2
613 store i16 %cvt6, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 6), align 2
614 store i16 %cvt7, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @dst16, i32 0, i64 7), align 2
615 ret void
616 }
618 define void @fptoui_8f32_8i8() #0 {
619 ; CHECK-LABEL: @fptoui_8f32_8i8(
620 ; CHECK-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
621 ; CHECK-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
622 ; CHECK-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
623 ; CHECK-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
624 ; CHECK-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
625 ; CHECK-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
626 ; CHECK-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
627 ; CHECK-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
628 ; CHECK-NEXT: [[CVT0:%.*]] = fptoui float [[A0]] to i8
629 ; CHECK-NEXT: [[CVT1:%.*]] = fptoui float [[A1]] to i8
630 ; CHECK-NEXT: [[CVT2:%.*]] = fptoui float [[A2]] to i8
631 ; CHECK-NEXT: [[CVT3:%.*]] = fptoui float [[A3]] to i8
632 ; CHECK-NEXT: [[CVT4:%.*]] = fptoui float [[A4]] to i8
633 ; CHECK-NEXT: [[CVT5:%.*]] = fptoui float [[A5]] to i8
634 ; CHECK-NEXT: [[CVT6:%.*]] = fptoui float [[A6]] to i8
635 ; CHECK-NEXT: [[CVT7:%.*]] = fptoui float [[A7]] to i8
636 ; CHECK-NEXT: store i8 [[CVT0]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 0), align 1
637 ; CHECK-NEXT: store i8 [[CVT1]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 1), align 1
638 ; CHECK-NEXT: store i8 [[CVT2]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 2), align 1
639 ; CHECK-NEXT: store i8 [[CVT3]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 3), align 1
640 ; CHECK-NEXT: store i8 [[CVT4]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 4), align 1
641 ; CHECK-NEXT: store i8 [[CVT5]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 5), align 1
642 ; CHECK-NEXT: store i8 [[CVT6]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 6), align 1
643 ; CHECK-NEXT: store i8 [[CVT7]], i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 7), align 1
644 ; CHECK-NEXT: ret void
645 ;
646 %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
647 %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
648 %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
649 %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
650 %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
651 %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
652 %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
653 %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
654 %cvt0 = fptoui float %a0 to i8
655 %cvt1 = fptoui float %a1 to i8
656 %cvt2 = fptoui float %a2 to i8
657 %cvt3 = fptoui float %a3 to i8
658 %cvt4 = fptoui float %a4 to i8
659 %cvt5 = fptoui float %a5 to i8
660 %cvt6 = fptoui float %a6 to i8
661 %cvt7 = fptoui float %a7 to i8
662 store i8 %cvt0, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 0), align 1
663 store i8 %cvt1, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 1), align 1
664 store i8 %cvt2, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 2), align 1
665 store i8 %cvt3, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 3), align 1
666 store i8 %cvt4, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 4), align 1
667 store i8 %cvt5, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 5), align 1
668 store i8 %cvt6, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 6), align 1
669 store i8 %cvt7, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @dst8, i32 0, i64 7), align 1
670 ret void
671 }
673 attributes #0 = { nounwind }