Showing the changes to codec/common/src/cpu.cpp (1 of 40 files). Show all
Unified · Split
| @@ -0,0 +1,293 @@ | ||
| 1 | +/*! | |
| 2 | + * \copy | |
| 3 | + * Copyright (c) 2009-2013, Cisco Systems | |
| 4 | + * All rights reserved. | |
| 5 | + * | |
| 6 | + * Redistribution and use in source and binary forms, with or without | |
| 7 | + * modification, are permitted provided that the following conditions | |
| 8 | + * are met: | |
| 9 | + * | |
| 10 | + * * Redistributions of source code must retain the above copyright | |
| 11 | + * notice, this list of conditions and the following disclaimer. | |
| 12 | + * | |
| 13 | + * * Redistributions in binary form must reproduce the above copyright | |
| 14 | + * notice, this list of conditions and the following disclaimer in | |
| 15 | + * the documentation and/or other materials provided with the | |
| 16 | + * distribution. | |
| 17 | + * | |
| 18 | + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS | |
| 19 | + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT | |
| 20 | + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS | |
| 21 | + * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE | |
| 22 | + * COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, | |
| 23 | + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, | |
| 24 | + * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; | |
| 25 | + * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER | |
| 26 | + * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT | |
| 27 | + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN | |
| 28 | + * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE | |
| 29 | + * POSSIBILITY OF SUCH DAMAGE. | |
| 30 | + * | |
| 31 | + * | |
| 32 | + * \file cpu.cpp | |
| 33 | + * | |
| 34 | + * \brief CPU compatibility detection | |
| 35 | + * | |
| 36 | + * \date 04/29/2009 Created | |
| 37 | + * | |
| 38 | + ************************************************************************************* | |
| 39 | + */ | |
| 40 | +#include <string.h> | |
| 41 | +#include <stdio.h> | |
| 42 | +#ifdef ANDROID_NDK | |
| 43 | +#include <cpu-features.h> | |
| 44 | +#endif | |
| 45 | +#include "cpu.h" | |
| 46 | +#include "cpu_core.h" | |
| 47 | + | |
| 48 | + | |
| 49 | + | |
| 50 | +#define CPU_Vendor_AMD "AuthenticAMD" | |
| 51 | +#define CPU_Vendor_INTEL "GenuineIntel" | |
| 52 | +#define CPU_Vendor_CYRIX "CyrixInstead" | |
| 53 | + | |
| 54 | +#if defined(X86_ASM) | |
| 55 | + | |
| 56 | +uint32_t WelsCPUFeatureDetect (int32_t* pNumberOfLogicProcessors) { | |
| 57 | + uint32_t uiCPU = 0; | |
| 58 | + uint32_t uiFeatureA = 0, uiFeatureB = 0, uiFeatureC = 0, uiFeatureD = 0; | |
| 59 | + int32_t CacheLineSize = 0; | |
| 60 | + int8_t chVendorName[16] = { 0 }; | |
| 61 | + uint32_t uiMaxCpuidLevel = 0; | |
| 62 | + | |
| 63 | + if (!WelsCPUIdVerify()) { | |
| 64 | + /* cpuid is not supported in cpu */ | |
| 65 | + return 0; | |
| 66 | + } | |
| 67 | + | |
| 68 | + WelsCPUId (0, &uiFeatureA, (uint32_t*)&chVendorName[0], (uint32_t*)&chVendorName[8], (uint32_t*)&chVendorName[4]); | |
| 69 | + uiMaxCpuidLevel = uiFeatureA; | |
| 70 | + if (uiMaxCpuidLevel == 0) { | |
| 71 | + /* maximum input value for basic cpuid information */ | |
| 72 | + return 0; | |
| 73 | + } | |
| 74 | + | |
| 75 | + WelsCPUId (1, &uiFeatureA, &uiFeatureB, &uiFeatureC, &uiFeatureD); | |
| 76 | + if ((uiFeatureD & 0x00800000) == 0) { | |
| 77 | + /* Basic MMX technology is not support in cpu, mean nothing for us so return here */ | |
| 78 | + return 0; | |
| 79 | + } | |
| 80 | + | |
| 81 | + uiCPU = WELS_CPU_MMX; | |
| 82 | + if (uiFeatureD & 0x02000000) { | |
| 83 | + /* SSE technology is identical to AMD MMX extensions */ | |
| 84 | + uiCPU |= WELS_CPU_MMXEXT | WELS_CPU_SSE; | |
| 85 | + } | |
| 86 | + if (uiFeatureD & 0x04000000) { | |
| 87 | + /* SSE2 support here */ | |
| 88 | + uiCPU |= WELS_CPU_SSE2; | |
| 89 | + } | |
| 90 | + if (uiFeatureD & 0x00000001) { | |
| 91 | + /* x87 FPU on-chip checking */ | |
| 92 | + uiCPU |= WELS_CPU_FPU; | |
| 93 | + } | |
| 94 | + if (uiFeatureD & 0x00008000) { | |
| 95 | + /* CMOV instruction checking */ | |
| 96 | + uiCPU |= WELS_CPU_CMOV; | |
| 97 | + } | |
| 98 | + if ((!strcmp ((const char*)chVendorName, CPU_Vendor_INTEL)) || | |
| 99 | + (!strcmp((const char*)chVendorName, CPU_Vendor_AMD)) ) { // confirmed_safe_unsafe_usage | |
| 100 | + if (uiFeatureD & 0x10000000) { | |
| 101 | + /* Multi-Threading checking: contains of multiple logic processors */ | |
| 102 | + uiCPU |= WELS_CPU_HTT; | |
| 103 | + } | |
| 104 | + } | |
| 105 | + | |
| 106 | + if (uiFeatureC & 0x00000001) { | |
| 107 | + /* SSE3 support here */ | |
| 108 | + uiCPU |= WELS_CPU_SSE3; | |
| 109 | + } | |
| 110 | + if (uiFeatureC & 0x00000200) { | |
| 111 | + /* SSSE3 support here */ | |
| 112 | + uiCPU |= WELS_CPU_SSSE3; | |
| 113 | + } | |
| 114 | + if (uiFeatureC & 0x00080000) { | |
| 115 | + /* SSE4.1 support here, 45nm Penryn processor */ | |
| 116 | + uiCPU |= WELS_CPU_SSE41; | |
| 117 | + } | |
| 118 | + if (uiFeatureC & 0x00100000) { | |
| 119 | + /* SSE4.2 support here, next generation Nehalem processor */ | |
| 120 | + uiCPU |= WELS_CPU_SSE42; | |
| 121 | + } | |
| 122 | + if (WelsCPUSupportAVX (uiFeatureA, uiFeatureC)) { | |
| 123 | + /* AVX supported */ | |
| 124 | + uiCPU |= WELS_CPU_AVX; | |
| 125 | + } | |
| 126 | + if (WelsCPUSupportFMA (uiFeatureA, uiFeatureC)) { | |
| 127 | + /* AVX FMA supported */ | |
| 128 | + uiCPU |= WELS_CPU_FMA; | |
| 129 | + } | |
| 130 | + if (uiFeatureC & 0x02000000) { | |
| 131 | + /* AES checking */ | |
| 132 | + uiCPU |= WELS_CPU_AES; | |
| 133 | + } | |
| 134 | + if (uiFeatureC & 0x00400000) { | |
| 135 | + /* MOVBE checking */ | |
| 136 | + uiCPU |= WELS_CPU_MOVBE; | |
| 137 | + } | |
| 138 | + | |
| 139 | + if( pNumberOfLogicProcessors != NULL ){ | |
| 140 | + if( uiCPU & WELS_CPU_HTT){ | |
| 141 | + *pNumberOfLogicProcessors = (uiFeatureB & 0x00ff0000) >> 16; // feature bits: 23-16 on returned EBX | |
| 142 | + } else { | |
| 143 | + *pNumberOfLogicProcessors = 0; | |
| 144 | + } | |
| 145 | + if( !strcmp((const char*)chVendorName, CPU_Vendor_INTEL) ){ | |
| 146 | + if( uiMaxCpuidLevel >= 4 ){ | |
| 147 | + uiFeatureC = 0; | |
| 148 | + WelsCPUId(0x4, &uiFeatureA, &uiFeatureB, &uiFeatureC, &uiFeatureD); | |
| 149 | + if( uiFeatureA != 0 ){ | |
| 150 | + *pNumberOfLogicProcessors = ((uiFeatureA&0xfc000000)>>26) + 1; | |
| 151 | + } | |
| 152 | + } | |
| 153 | + } | |
| 154 | + } | |
| 155 | + | |
| 156 | + WelsCPUId (0x80000000, &uiFeatureA, &uiFeatureB, &uiFeatureC, &uiFeatureD); | |
| 157 | + | |
| 158 | + if ((!strcmp ((const char*)chVendorName, CPU_Vendor_AMD)) | |
| 159 | + && (uiFeatureA >= 0x80000001)) { // confirmed_safe_unsafe_usage | |
| 160 | + WelsCPUId (0x80000001, &uiFeatureA, &uiFeatureB, &uiFeatureC, &uiFeatureD); | |
| 161 | + if (uiFeatureD & 0x00400000) { | |
| 162 | + uiCPU |= WELS_CPU_MMXEXT; | |
| 163 | + } | |
| 164 | + if (uiFeatureD & 0x80000000) { | |
| 165 | + uiCPU |= WELS_CPU_3DNOW; | |
| 166 | + } | |
| 167 | + } | |
| 168 | + | |
| 169 | + if (!strcmp ((const char*)chVendorName, CPU_Vendor_INTEL)) { // confirmed_safe_unsafe_usage | |
| 170 | + int32_t family, model; | |
| 171 | + | |
| 172 | + WelsCPUId (1, &uiFeatureA, &uiFeatureB, &uiFeatureC, &uiFeatureD); | |
| 173 | + family = ((uiFeatureA >> 8) & 0xf) + ((uiFeatureA >> 20) & 0xff); | |
| 174 | + model = ((uiFeatureA >> 4) & 0xf) + ((uiFeatureA >> 12) & 0xf0); | |
| 175 | + | |
| 176 | + if ((family == 6) && (model == 9 || model == 13 || model == 14)) { | |
| 177 | + uiCPU &= ~ (WELS_CPU_SSE2 | WELS_CPU_SSE3); | |
| 178 | + } | |
| 179 | + } | |
| 180 | + | |
| 181 | + // get cache line size | |
| 182 | + if ((!strcmp ((const char*)chVendorName, CPU_Vendor_INTEL)) | |
| 183 | + || ! (strcmp ((const char*)chVendorName, CPU_Vendor_CYRIX))) { // confirmed_safe_unsafe_usage | |
| 184 | + WelsCPUId (1, &uiFeatureA, &uiFeatureB, &uiFeatureC, &uiFeatureD); | |
| 185 | + | |
| 186 | + CacheLineSize = (uiFeatureB & 0xff00) >> | |
| 187 | + 5; // ((clflush_line_size >> 8) << 3), CLFLUSH_line_size * 8 = CacheLineSize_in_byte | |
| 188 | + | |
| 189 | + if (CacheLineSize == 128) { | |
| 190 | + uiCPU |= WELS_CPU_CACHELINE_128; | |
| 191 | + } else if (CacheLineSize == 64) { | |
| 192 | + uiCPU |= WELS_CPU_CACHELINE_64; | |
| 193 | + } else if (CacheLineSize == 32) { | |
| 194 | + uiCPU |= WELS_CPU_CACHELINE_32; | |
| 195 | + } else if (CacheLineSize == 16) { | |
| 196 | + uiCPU |= WELS_CPU_CACHELINE_16; | |
| 197 | + } | |
| 198 | + } | |
| 199 | + | |
| 200 | + return uiCPU; | |
| 201 | +} | |
| 202 | + | |
| 203 | + | |
| 204 | +void WelsCPURestore (const uint32_t kuiCPU) { | |
| 205 | + if (kuiCPU & (WELS_CPU_MMX | WELS_CPU_MMXEXT | WELS_CPU_3DNOW | WELS_CPU_3DNOWEXT)) { | |
| 206 | + WelsEmms(); | |
| 207 | + } | |
| 208 | +} | |
| 209 | + | |
| 210 | +#elif defined(HAVE_NEON) //For supporting both android platform and iOS platform | |
| 211 | +#if defined(ANDROID_NDK) | |
| 212 | +uint32_t WelsCPUFeatureDetect (int32_t* pNumberOfLogicProcessors) | |
| 213 | +{ | |
| 214 | + uint32_t uiCPU = 0; | |
| 215 | + AndroidCpuFamily cpuFamily = ANDROID_CPU_FAMILY_UNKNOWN; | |
| 216 | + uint64_t uiFeatures = 0; | |
| 217 | + cpuFamily = android_getCpuFamily(); | |
| 218 | + if (cpuFamily == ANDROID_CPU_FAMILY_ARM) { | |
| 219 | + uiFeatures = android_getCpuFeatures(); | |
| 220 | + if (uiFeatures & ANDROID_CPU_ARM_FEATURE_ARMv7){ | |
| 221 | + uiCPU |= WELS_CPU_ARMv7; | |
| 222 | + } | |
| 223 | + if (uiFeatures & ANDROID_CPU_ARM_FEATURE_VFPv3){ | |
| 224 | + uiCPU |= WELS_CPU_VFPv3; | |
| 225 | + } | |
| 226 | + if (uiFeatures & ANDROID_CPU_ARM_FEATURE_NEON){ | |
| 227 | + uiCPU |= WELS_CPU_NEON; | |
| 228 | + } | |
| 229 | + } | |
| 230 | + | |
| 231 | + if( pNumberOfLogicProcessors != NULL ){ | |
| 232 | + *pNumberOfLogicProcessors = android_getCpuCount(); | |
| 233 | + } | |
| 234 | + | |
| 235 | + return uiCPU; | |
| 236 | +} | |
| 237 | + | |
| 238 | +#elif defined(__APPLE__) | |
| 239 | +uint32_t WelsCPUFeatureDetect (int32_t* pNumberOfLogicProcessors) | |
| 240 | +{ | |
| 241 | + uint32_t uiCPU = 0; | |
| 242 | + | |
| 243 | +#if defined(__ARM_NEON__) | |
| 244 | + uiCPU |= WELS_CPU_ARMv7; | |
| 245 | + uiCPU |= WELS_CPU_VFPv3; | |
| 246 | + uiCPU |= WELS_CPU_NEON; | |
| 247 | +#endif | |
| 248 | + return uiCPU; | |
| 249 | +} | |
| 250 | +#elif defined(__linux__) | |
| 251 | + | |
| 252 | +/* Generic arm/linux cpu feature detection */ | |
| 253 | +uint32_t WelsCPUFeatureDetect (int32_t* pNumberOfLogicProcessors) { | |
| 254 | + FILE *f = fopen("/proc/cpuinfo", "r"); | |
| 255 | + | |
| 256 | + if (!f) | |
| 257 | + return 0; | |
| 258 | + | |
| 259 | + char buf[200]; | |
| 260 | + int flags = 0; | |
| 261 | + while (fgets(buf, sizeof(buf), f)) { | |
| 262 | + if (!strncmp(buf, "Features", strlen("Features"))) { | |
| 263 | + if (strstr(buf, " neon ")) | |
| 264 | + flags |= WELS_CPU_NEON; | |
| 265 | + if (strstr(buf, " vfpv3 ")) | |
| 266 | + flags |= WELS_CPU_VFPv3; | |
| 267 | + break; | |
| 268 | + } | |
| 269 | + } | |
| 270 | + fclose(f); | |
| 271 | + return flags; | |
| 272 | +} | |
| 273 | + | |
| 274 | +#else /* HAVE_NEON enabled but no runtime detection */ | |
| 275 | + | |
| 276 | +/* No runtime feature detection available, but built with HAVE_NEON - assume | |
| 277 | + * that NEON and all associated features are available. */ | |
| 278 | + | |
| 279 | +uint32_t WelsCPUFeatureDetect (int32_t* pNumberOfLogicProcessors) { | |
| 280 | + return WELS_CPU_ARMv7 | | |
| 281 | + WELS_CPU_VFPv3 | | |
| 282 | + WELS_CPU_NEON; | |
| 283 | +} | |
| 284 | +#endif | |
| 285 | +#else /* Neither X86_ASM nor HAVE_NEON */ | |
| 286 | + | |
| 287 | +uint32_t WelsCPUFeatureDetect (int32_t* pNumberOfLogicProcessors) { | |
| 288 | + return 0; | |
| 289 | +} | |
| 290 | + | |
| 291 | +#endif | |
| 292 | + | |
| 293 | + | |