1 /* SPDX-License-Identifier: MIT
2 *
3 * Permission is hereby granted, free of charge, to any person
4 * obtaining a copy of this software and associated documentation
5 * files (the "Software"), to deal in the Software without
6 * restriction, including without limitation the rights to use, copy,
7 * modify, merge, publish, distribute, sublicense, and/or sell copies
8 * of the Software, and to permit persons to whom the Software is
9 * furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice shall be
12 * included in all copies or substantial portions of the Software.
13 *
14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
15 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
16 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
17 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
18 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
19 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
20 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 * SOFTWARE.
22 *
23 * Copyright:
24 * 2020 Evan Nemerson <evan@nemerson.com>
25 * 2020 Hidayat Khan <huk2209@gmail.com>
26 * 2020 Christopher Moore <moore@free.fr>
27 */
28
29 #if !defined(SIMDE_X86_AVX512_SLLI_H)
30 #define SIMDE_X86_AVX512_SLLI_H
31
32 #include "types.h"
33 #include "../avx2.h"
34 #include "mov.h"
35 #include "setzero.h"
36
37 HEDLEY_DIAGNOSTIC_PUSH
38 SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
39 SIMDE_BEGIN_DECLS_
40
41 SIMDE_FUNCTION_ATTRIBUTES
42 simde__m512i
simde_mm512_slli_epi16(simde__m512i a,const unsigned int imm8)43 simde_mm512_slli_epi16 (simde__m512i a, const unsigned int imm8)
44 SIMDE_REQUIRE_RANGE(imm8, 0, 255) {
45 #if defined(SIMDE_X86_AVX512BW_NATIVE) && (defined(HEDLEY_GCC_VERSION) && ((__GNUC__ == 5 && __GNUC_MINOR__ == 5) || (__GNUC__ == 6 && __GNUC_MINOR__ >= 4)))
46 simde__m512i r;
47
48 SIMDE_CONSTIFY_16_(_mm512_slli_epi16, r, simde_mm512_setzero_si512(), imm8, a);
49
50 return r;
51 #elif defined(SIMDE_X86_AVX512BW_NATIVE)
52 return SIMDE_BUG_IGNORE_SIGN_CONVERSION(_mm512_slli_epi16(a, imm8));
53 #else
54 simde__m512i_private
55 r_,
56 a_ = simde__m512i_to_private(a);
57
58 #if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
59 if(imm8 < 16)
60 r_.i16 = HEDLEY_STATIC_CAST(__typeof__(r_.i16), (a_.i16 << HEDLEY_STATIC_CAST(int16_t, imm8)));
61 else
62 return simde_mm512_setzero_si512();
63 #else
64 SIMDE_VECTORIZE
65 for (size_t i = 0 ; i < (sizeof(r_.i16) / sizeof(r_.i16[0])) ; i++) {
66 r_.i16[i] = (imm8 < 16) ? HEDLEY_STATIC_CAST(int16_t, a_.i16[i] << (imm8 & 0xff)) : 0;
67 }
68 #endif
69
70 return simde__m512i_from_private(r_);
71 #endif
72 }
73 #if defined(SIMDE_X86_AVX512BW_ENABLE_NATIVE_ALIASES)
74 #undef _mm512_slli_epi16
75 #define _mm512_slli_epi16(a, imm8) simde_mm512_slli_epi16(a, imm8)
76 #endif
77
78 SIMDE_FUNCTION_ATTRIBUTES
79 simde__m512i
simde_mm512_slli_epi32(simde__m512i a,unsigned int imm8)80 simde_mm512_slli_epi32 (simde__m512i a, unsigned int imm8) {
81 /* I guess the restriction was added in 6.4, back-ported to 5.5, then
82 * removed (fixed) in 7? */
83 #if defined(SIMDE_X86_AVX512F_NATIVE) && (defined(HEDLEY_GCC_VERSION) && ((__GNUC__ == 5 && __GNUC_MINOR__ == 5) || (__GNUC__ == 6 && __GNUC_MINOR__ >= 4)))
84 simde__m512i r;
85
86 SIMDE_CONSTIFY_32_(_mm512_slli_epi32, r, simde_mm512_setzero_si512(), imm8, a);
87
88 return r;
89 #elif defined(SIMDE_X86_AVX512F_NATIVE)
90 return SIMDE_BUG_IGNORE_SIGN_CONVERSION(_mm512_slli_epi32(a, imm8));
91 #else
92 simde__m512i_private
93 r_,
94 a_ = simde__m512i_to_private(a);
95
96 /* The Intel Intrinsics Guide says that only the 8 LSBits of imm8 are
97 * used. In this case we should do "imm8 &= 0xff". However in
98 * practice all bits are used. */
99 if (imm8 > 31) {
100 simde_memset(&r_, 0, sizeof(r_));
101 } else {
102 #if defined(SIMDE_X86_AVX2_NATIVE)
103 r_.m256i[0] = simde_mm256_slli_epi32(a_.m256i[0], HEDLEY_STATIC_CAST(int, imm8));
104 r_.m256i[1] = simde_mm256_slli_epi32(a_.m256i[1], HEDLEY_STATIC_CAST(int, imm8));
105 #elif defined(SIMDE_X86_SSE2_NATIVE)
106 r_.m128i[0] = simde_mm_slli_epi32(a_.m128i[0], HEDLEY_STATIC_CAST(int, imm8));
107 r_.m128i[1] = simde_mm_slli_epi32(a_.m128i[1], HEDLEY_STATIC_CAST(int, imm8));
108 r_.m128i[2] = simde_mm_slli_epi32(a_.m128i[2], HEDLEY_STATIC_CAST(int, imm8));
109 r_.m128i[3] = simde_mm_slli_epi32(a_.m128i[3], HEDLEY_STATIC_CAST(int, imm8));
110 #elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
111 r_.u32 = a_.u32 << imm8;
112 #else
113 SIMDE_VECTORIZE
114 for (size_t i = 0 ; i < (sizeof(r_.u32) / sizeof(r_.u32[0])) ; i++) {
115 r_.u32[i] = a_.u32[i] << imm8;
116 }
117 #endif
118 }
119
120 return simde__m512i_from_private(r_);
121 #endif
122 }
123 #if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
124 #undef _mm512_slli_epi32
125 #define _mm512_slli_epi32(a, imm8) simde_mm512_slli_epi32(a, imm8)
126 #endif
127
128 SIMDE_FUNCTION_ATTRIBUTES
129 simde__m512i
simde_mm512_slli_epi64(simde__m512i a,unsigned int imm8)130 simde_mm512_slli_epi64 (simde__m512i a, unsigned int imm8) {
131 #if defined(SIMDE_X86_AVX512F_NATIVE) && (defined(HEDLEY_GCC_VERSION) && ((__GNUC__ == 5 && __GNUC_MINOR__ == 5) || (__GNUC__ == 6 && __GNUC_MINOR__ >= 4)))
132 simde__m512i r;
133
134 SIMDE_CONSTIFY_64_(_mm512_slli_epi64, r, simde_mm512_setzero_si512(), imm8, a);
135
136 return r;
137 #elif defined(SIMDE_X86_AVX512F_NATIVE)
138 return SIMDE_BUG_IGNORE_SIGN_CONVERSION(_mm512_slli_epi64(a, imm8));
139 #else
140 simde__m512i_private
141 r_,
142 a_ = simde__m512i_to_private(a);
143
144 /* The Intel Intrinsics Guide says that only the 8 LSBits of imm8 are
145 * used. In this case we should do "imm8 &= 0xff". However in
146 * practice all bits are used. */
147 if (imm8 > 63) {
148 simde_memset(&r_, 0, sizeof(r_));
149 } else {
150 #if defined(SIMDE_X86_AVX2_NATIVE)
151 r_.m256i[0] = simde_mm256_slli_epi64(a_.m256i[0], HEDLEY_STATIC_CAST(int, imm8));
152 r_.m256i[1] = simde_mm256_slli_epi64(a_.m256i[1], HEDLEY_STATIC_CAST(int, imm8));
153 #elif defined(SIMDE_X86_SSE2_NATIVE)
154 r_.m128i[0] = simde_mm_slli_epi64(a_.m128i[0], HEDLEY_STATIC_CAST(int, imm8));
155 r_.m128i[1] = simde_mm_slli_epi64(a_.m128i[1], HEDLEY_STATIC_CAST(int, imm8));
156 r_.m128i[2] = simde_mm_slli_epi64(a_.m128i[2], HEDLEY_STATIC_CAST(int, imm8));
157 r_.m128i[3] = simde_mm_slli_epi64(a_.m128i[3], HEDLEY_STATIC_CAST(int, imm8));
158 #elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
159 r_.u64 = a_.u64 << imm8;
160 #else
161 SIMDE_VECTORIZE
162 for (size_t i = 0 ; i < (sizeof(r_.u64) / sizeof(r_.u64[0])) ; i++) {
163 r_.u64[i] = a_.u64[i] << imm8;
164 }
165 #endif
166 }
167
168 return simde__m512i_from_private(r_);
169 #endif
170 }
171 #if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
172 #undef _mm512_slli_epi64
173 #define _mm512_slli_epi64(a, imm8) simde_mm512_slli_epi64(a, imm8)
174 #endif
175
176 SIMDE_END_DECLS_
177 HEDLEY_DIAGNOSTIC_POP
178
179 #endif /* !defined(SIMDE_X86_AVX512_SLLI_H) */
180