1 /*****************************************************************************
2 * cpu.c: h264 encoder library
3 *****************************************************************************
4 * Copyright (C) 2003 Laurent Aimar
5 * $Id: cpu.c,v 1.1 2004/06/03 19:27:06 fenrir Exp $
7 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA.
22 *****************************************************************************/
24 #if defined(HAVE_PTHREAD) && defined(SYS_LINUX)
29 #include <kernel/OS.h>
31 #if defined(SYS_MACOSX) || defined(SYS_FREEBSD)
32 #include <sys/types.h>
33 #include <sys/sysctl.h>
41 } x264_cpu_names[] = {
42 {"MMX", X264_CPU_MMX},
43 {"MMX2", X264_CPU_MMX|X264_CPU_MMXEXT},
44 {"MMXEXT", X264_CPU_MMX|X264_CPU_MMXEXT},
45 {"SSE", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE},
46 {"SSE1", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE},
47 {"SSE2", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE|X264_CPU_SSE2},
48 {"SSE3", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE|X264_CPU_SSE2|X264_CPU_SSE3},
49 {"SSSE3", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE|X264_CPU_SSE2|X264_CPU_SSE3|X264_CPU_SSSE3},
50 {"SSE4", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE|X264_CPU_SSE2|X264_CPU_SSE3|X264_CPU_SSSE3|X264_CPU_SSE4},
51 {"3DNow", X264_CPU_3DNOW},
52 {"Altivec", X264_CPU_ALTIVEC},
53 {"Cache32", X264_CPU_CACHELINE_SPLIT|X264_CPU_CACHELINE_32},
54 {"Cache64", X264_CPU_CACHELINE_SPLIT|X264_CPU_CACHELINE_64},
59 extern int x264_cpu_cpuid_test( void );
60 extern uint32_t x264_cpu_cpuid( uint32_t op, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx );
62 uint32_t x264_cpu_detect( void )
65 uint32_t eax, ebx, ecx, edx;
66 uint32_t vendor[4] = {0};
71 if( !x264_cpu_cpuid_test() )
75 x264_cpu_cpuid( 0, &eax, vendor+0, vendor+2, vendor+1 );
79 x264_cpu_cpuid( 1, &eax, &ebx, &ecx, &edx );
85 cpu |= X264_CPU_MMXEXT|X264_CPU_SSE;
91 cpu |= X264_CPU_SSSE3;
95 x264_cpu_cpuid( 0x80000000, &eax, &ebx, &ecx, &edx );
96 max_extended_cap = eax;
98 if( !strcmp((char*)vendor, "AuthenticAMD") && max_extended_cap >= 0x80000001 )
100 x264_cpu_cpuid( 0x80000001, &eax, &ebx, &ecx, &edx );
102 cpu |= X264_CPU_3DNOW;
104 cpu |= X264_CPU_MMXEXT;
107 if( !strcmp((char*)vendor, "GenuineIntel") || !strcmp((char*)vendor, "CyrixInstead") )
108 cpu |= X264_CPU_CACHELINE_SPLIT;
109 /* cacheline size is specified in 3 places, any of which may be missing */
110 x264_cpu_cpuid( 1, &eax, &ebx, &ecx, &edx );
111 cache = (ebx&0xff00)>>5; // cflush size
112 if( !cache && max_extended_cap >= 0x80000006 )
114 x264_cpu_cpuid( 0x80000006, &eax, &ebx, &ecx, &edx );
115 cache = ecx&0xff; // cacheline size
119 // Cache and TLB Information
120 static const char cache32_ids[] = { 0x0a, 0x0c, 0x41, 0x42, 0x43, 0x44, 0x45, 0x82, 0x83, 0x84, 0x85, 0 };
121 static const char cache64_ids[] = { 0x22, 0x23, 0x25, 0x29, 0x2c, 0x46, 0x47, 0x49, 0x60, 0x66, 0x67, 0x68, 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7c, 0x7f, 0x86, 0x87, 0 };
125 x264_cpu_cpuid( 2, buf+0, buf+1, buf+2, buf+3 );
132 if( strchr( cache32_ids, buf[j]&0xff ) )
134 if( strchr( cache64_ids, buf[j]&0xff ) )
138 } while( ++i < max );
142 cpu |= X264_CPU_CACHELINE_32;
144 cpu |= X264_CPU_CACHELINE_64;
149 #elif defined( ARCH_PPC )
152 #include <sys/sysctl.h>
153 uint32_t x264_cpu_detect( void )
157 int selectors[2] = { CTL_HW, HW_VECTORUNIT };
159 size_t length = sizeof( has_altivec );
160 int error = sysctl( selectors, 2, &has_altivec, &length, NULL, 0 );
162 if( error == 0 && has_altivec != 0 )
164 cpu |= X264_CPU_ALTIVEC;
170 #elif defined( SYS_LINUX )
173 static sigjmp_buf jmpbuf;
174 static volatile sig_atomic_t canjump = 0;
176 static void sigill_handler( int sig )
180 signal( sig, SIG_DFL );
185 siglongjmp( jmpbuf, 1 );
188 uint32_t x264_cpu_detect( void )
190 static void (* oldsig)( int );
192 oldsig = signal( SIGILL, sigill_handler );
193 if( sigsetjmp( jmpbuf, 1 ) )
195 signal( SIGILL, oldsig );
200 asm volatile( "mtspr 256, %0\n\t"
206 signal( SIGILL, oldsig );
208 return X264_CPU_ALTIVEC;
214 uint32_t x264_cpu_detect( void )
222 void x264_emms( void )
228 int x264_cpu_num_processors( void )
230 #if !defined(HAVE_PTHREAD)
233 #elif defined(_WIN32)
234 return pthread_num_processors_np();
236 #elif defined(SYS_LINUX)
240 memset( &p_aff, 0, sizeof(p_aff) );
241 sched_getaffinity( 0, sizeof(p_aff), &p_aff );
242 for( np = 0, bit = 0; bit < sizeof(p_aff); bit++ )
243 np += (((uint8_t *)&p_aff)[bit / 8] >> (bit % 8)) & 1;
246 #elif defined(SYS_BEOS)
248 get_system_info( &info );
249 return info.cpu_count;
251 #elif defined(SYS_MACOSX) || defined(SYS_FREEBSD)
253 size_t length = sizeof( numberOfCPUs );
254 if( sysctlbyname("hw.ncpu", &numberOfCPUs, &length, NULL, 0) )