x86cpu.h (4247B)
1 /* Copyright (c) 2014, Cisco Systems, INC 2 Written by XiangMingZhu WeiZhou MinPeng YanWang 3 4 Redistribution and use in source and binary forms, with or without 5 modification, are permitted provided that the following conditions 6 are met: 7 8 - Redistributions of source code must retain the above copyright 9 notice, this list of conditions and the following disclaimer. 10 11 - Redistributions in binary form must reproduce the above copyright 12 notice, this list of conditions and the following disclaimer in the 13 documentation and/or other materials provided with the distribution. 14 15 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 16 ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 17 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 18 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER 19 OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, 20 EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 21 PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 22 PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 23 LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 24 NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 25 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28 #if !defined(X86CPU_H) 29 # define X86CPU_H 30 31 # if defined(OPUS_X86_MAY_HAVE_SSE) 32 # define MAY_HAVE_SSE(name) name ## _sse 33 # else 34 # define MAY_HAVE_SSE(name) name ## _c 35 # endif 36 37 # if defined(OPUS_X86_MAY_HAVE_SSE2) 38 # define MAY_HAVE_SSE2(name) name ## _sse2 39 # else 40 # define MAY_HAVE_SSE2(name) name ## _c 41 # endif 42 43 # if defined(OPUS_X86_MAY_HAVE_SSE4_1) 44 # define MAY_HAVE_SSE4_1(name) name ## _sse4_1 45 # else 46 # define MAY_HAVE_SSE4_1(name) name ## _c 47 # endif 48 49 # if defined(OPUS_X86_MAY_HAVE_AVX2) 50 # define MAY_HAVE_AVX2(name) name ## _avx2 51 # else 52 # define MAY_HAVE_AVX2(name) name ## _c 53 # endif 54 55 # if defined(OPUS_HAVE_RTCD) && \ 56 ((defined(OPUS_X86_MAY_HAVE_SSE) && !defined(OPUS_X86_PRESUME_SSE)) || \ 57 (defined(OPUS_X86_MAY_HAVE_SSE2) && !defined(OPUS_X86_PRESUME_SSE2)) || \ 58 (defined(OPUS_X86_MAY_HAVE_SSE4_1) && !defined(OPUS_X86_PRESUME_SSE4_1)) || \ 59 (defined(OPUS_X86_MAY_HAVE_AVX2) && !defined(OPUS_X86_PRESUME_AVX2))) 60 int opus_select_arch(void); 61 # endif 62 63 # if defined(OPUS_X86_MAY_HAVE_SSE2) 64 # include "opus_defines.h" 65 66 /*MOVD should not impose any alignment restrictions, but the C standard does, 67 and UBSan will report errors if we actually make unaligned accesses. 68 Use this to work around those restrictions (which should hopefully all get 69 optimized to a single MOVD instruction). 70 GCC implemented _mm_loadu_si32() since GCC 11; HOWEVER, there is a bug! 71 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99754 72 LLVM implemented _mm_loadu_si32() since Clang 8.0, however the 73 __clang_major__ version number macro is unreliable, as vendors 74 (specifically, Apple) will use different numbering schemes than upstream. 75 Clang's advice is "use feature detection", but they do not provide feature 76 detection support for specific SIMD functions. 77 We follow the approach from the SIMDe project and instead detect unrelated 78 features that should be available in the version we want (see 79 <https://github.com/simd-everywhere/simde/blob/master/simde/simde-detect-clang.h>).*/ 80 # if defined(__clang__) 81 # if __has_warning("-Wextra-semi-stmt") || \ 82 __has_builtin(__builtin_rotateleft32) 83 # define OPUS_CLANG_8 (1) 84 # endif 85 # endif 86 # if !defined(_MSC_VER) && !OPUS_GNUC_PREREQ(11,3) && !defined(OPUS_CLANG_8) 87 # include <string.h> 88 # include <emmintrin.h> 89 90 # ifdef _mm_loadu_si32 91 # undef _mm_loadu_si32 92 # endif 93 # define _mm_loadu_si32 WORKAROUND_mm_loadu_si32 94 static inline __m128i WORKAROUND_mm_loadu_si32(void const* mem_addr) { 95 int val; 96 memcpy(&val, mem_addr, sizeof(val)); 97 return _mm_cvtsi32_si128(val); 98 } 99 # elif defined(_MSC_VER) 100 /* MSVC needs this for _mm_loadu_si32 */ 101 # include <immintrin.h> 102 # endif 103 104 # define OP_CVTEPI8_EPI32_M32(x) \ 105 (_mm_cvtepi8_epi32(_mm_loadu_si32(x))) 106 107 # define OP_CVTEPI16_EPI32_M64(x) \ 108 (_mm_cvtepi16_epi32(_mm_loadl_epi64((__m128i *)(void*)(x)))) 109 110 # endif 111 112 #endif