#ifdef SYS_BEOS
#include <kernel/OS.h>
#endif
-#ifdef SYS_MACOSX
+#if defined(SYS_MACOSX) || defined(SYS_FREEBSD)
#include <sys/types.h>
#include <sys/sysctl.h>
#endif
-#include <string.h>
-
#include "common.h"
-#if defined(ARCH_X86) || defined(ARCH_X86_64)
+const struct {
+ const char name[8];
+ int flags;
+} x264_cpu_names[] = {
+ {"MMX", X264_CPU_MMX},
+ {"MMX2", X264_CPU_MMX|X264_CPU_MMXEXT},
+ {"MMXEXT", X264_CPU_MMX|X264_CPU_MMXEXT},
+ {"SSE", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE},
+ {"SSE1", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE},
+ {"SSE2", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE|X264_CPU_SSE2},
+ {"SSE3", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE|X264_CPU_SSE2|X264_CPU_SSE3},
+ {"SSSE3", X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_SSE|X264_CPU_SSE2|X264_CPU_SSE3|X264_CPU_SSSE3},
+ {"3DNow", X264_CPU_3DNOW},
+ {"Altivec", X264_CPU_ALTIVEC},
+ {"Cache32", X264_CPU_CACHELINE_SPLIT|X264_CPU_CACHELINE_32},
+ {"Cache64", X264_CPU_CACHELINE_SPLIT|X264_CPU_CACHELINE_64},
+ {"", 0},
+};
+
+#ifdef HAVE_MMX
extern int x264_cpu_cpuid_test( void );
extern uint32_t x264_cpu_cpuid( uint32_t op, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx );
-extern void x264_emms( void );
uint32_t x264_cpu_detect( void )
{
uint32_t cpu = 0;
-
uint32_t eax, ebx, ecx, edx;
- int b_amd;
-
+ uint32_t vendor[4] = {0};
+ int max_extended_cap;
+ int cache;
+#ifndef ARCH_X86_64
if( !x264_cpu_cpuid_test() )
- {
- /* No cpuid */
return 0;
- }
+#endif
- x264_cpu_cpuid( 0, &eax, &ebx, &ecx, &edx);
+ x264_cpu_cpuid( 0, &eax, vendor+0, vendor+2, vendor+1 );
if( eax == 0 )
- {
return 0;
- }
- b_amd = (ebx == 0x68747541) && (ecx == 0x444d4163) && (edx == 0x69746e65);
x264_cpu_cpuid( 1, &eax, &ebx, &ecx, &edx );
- if( (edx&0x00800000) == 0 )
- {
- /* No MMX */
+ if( edx&0x00800000 )
+ cpu |= X264_CPU_MMX;
+ else
return 0;
- }
- cpu = X264_CPU_MMX;
- if( (edx&0x02000000) )
- {
- /* SSE - identical to AMD MMX extensions */
+ if( edx&0x02000000 )
cpu |= X264_CPU_MMXEXT|X264_CPU_SSE;
- }
- if( (edx&0x04000000) )
- {
- /* Is it OK ? */
+ if( edx&0x04000000 )
cpu |= X264_CPU_SSE2;
- }
+ if( ecx&0x00000001 )
+ cpu |= X264_CPU_SSE3;
+ if( ecx&0x00000200 )
+ cpu |= X264_CPU_SSSE3;
x264_cpu_cpuid( 0x80000000, &eax, &ebx, &ecx, &edx );
- if( eax < 0x80000001 )
+ max_extended_cap = eax;
+
+ if( !strcmp((char*)vendor, "AuthenticAMD") && max_extended_cap >= 0x80000001 )
{
- /* no extended capabilities */
- return cpu;
+ x264_cpu_cpuid( 0x80000001, &eax, &ebx, &ecx, &edx );
+ if( edx&0x80000000 )
+ cpu |= X264_CPU_3DNOW;
+ if( edx&0x00400000 )
+ cpu |= X264_CPU_MMXEXT;
}
- x264_cpu_cpuid( 0x80000001, &eax, &ebx, &ecx, &edx );
- if( edx&0x80000000 )
+ if( !strcmp((char*)vendor, "GenuineIntel") || !strcmp((char*)vendor, "CyrixInstead") )
+ cpu |= X264_CPU_CACHELINE_SPLIT;
+ /* cacheline size is specified in 3 places, any of which may be missing */
+ x264_cpu_cpuid( 1, &eax, &ebx, &ecx, &edx );
+ cache = (ebx&0xff00)>>5; // cflush size
+ if( !cache && max_extended_cap >= 0x80000006 )
{
- cpu |= X264_CPU_3DNOW;
+ x264_cpu_cpuid( 0x80000006, &eax, &ebx, &ecx, &edx );
+ cache = ecx&0xff; // cacheline size
}
- if( b_amd && (edx&0x00400000) )
+ if( !cache )
{
- /* AMD MMX extensions */
- cpu |= X264_CPU_MMXEXT;
+ // Cache and TLB Information
+ static const char cache32_ids[] = { 0x0a, 0x0c, 0x41, 0x42, 0x43, 0x44, 0x45, 0x82, 0x83, 0x84, 0x85, 0 };
+ 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 };
+ uint32_t buf[4];
+ int max, i=0, j;
+ do {
+ x264_cpu_cpuid( 2, buf+0, buf+1, buf+2, buf+3 );
+ max = buf[0]&0xff;
+ buf[0] &= ~0xff;
+ for(j=0; j<4; j++)
+ if( !(buf[j]>>31) )
+ while( buf[j] )
+ {
+ if( strchr( cache32_ids, buf[j]&0xff ) )
+ cache = 32;
+ if( strchr( cache64_ids, buf[j]&0xff ) )
+ cache = 64;
+ buf[j] >>= 8;
+ }
+ } while( ++i < max );
}
- return cpu;
-}
+ if( cache == 32 )
+ cpu |= X264_CPU_CACHELINE_32;
+ if( cache == 64 )
+ cpu |= X264_CPU_CACHELINE_64;
-void x264_cpu_restore( uint32_t cpu )
-{
- if( cpu&(X264_CPU_MMX|X264_CPU_MMXEXT|X264_CPU_3DNOW|X264_CPU_3DNOWEXT) )
- {
- x264_emms();
- }
+ return cpu;
}
#elif defined( ARCH_PPC )
}
#elif defined( SYS_LINUX )
-uint32_t x264_cpu_detect( void )
+#include <signal.h>
+#include <setjmp.h>
+static sigjmp_buf jmpbuf;
+static volatile sig_atomic_t canjump = 0;
+
+static void sigill_handler( int sig )
{
- /* FIXME (Linux PPC) */
- return X264_CPU_ALTIVEC;
+ if( !canjump )
+ {
+ signal( sig, SIG_DFL );
+ raise( sig );
+ }
+
+ canjump = 0;
+ siglongjmp( jmpbuf, 1 );
}
-#endif
-void x264_cpu_restore( uint32_t cpu )
+uint32_t x264_cpu_detect( void )
{
+ static void (* oldsig)( int );
+
+ oldsig = signal( SIGILL, sigill_handler );
+ if( sigsetjmp( jmpbuf, 1 ) )
+ {
+ signal( SIGILL, oldsig );
+ return 0;
+ }
+
+ canjump = 1;
+ asm volatile( "mtspr 256, %0\n\t"
+ "vand 0, 0, 0\n\t"
+ :
+ : "r"(-1) );
+ canjump = 0;
+
+ signal( SIGILL, oldsig );
+
+ return X264_CPU_ALTIVEC;
}
+#endif
#else
return 0;
}
-void x264_cpu_restore( uint32_t cpu )
+#endif
+
+#ifndef HAVE_MMX
+void x264_emms( void )
{
}
-
#endif
#if !defined(HAVE_PTHREAD)
return 1;
-#elif defined(SYS_LINUX) || defined(WIN32)
+#elif defined(_WIN32)
+ return pthread_num_processors_np();
+
+#elif defined(SYS_LINUX)
unsigned int bit;
int np;
-#if defined(WIN32)
- uint32_t p_aff, s_aff;
- GetProcessAffinityMask( GetCurrentProcess(), &p_aff, &s_aff );
-#else
cpu_set_t p_aff;
memset( &p_aff, 0, sizeof(p_aff) );
sched_getaffinity( 0, sizeof(p_aff), &p_aff );
-#endif
for( np = 0, bit = 0; bit < sizeof(p_aff); bit++ )
np += (((uint8_t *)&p_aff)[bit / 8] >> (bit % 8)) & 1;
return np;
get_system_info( &info );
return info.cpu_count;
-#elif defined(SYS_MACOSX)
+#elif defined(SYS_MACOSX) || defined(SYS_FREEBSD)
int numberOfCPUs;
size_t length = sizeof( numberOfCPUs );
if( sysctlbyname("hw.ncpu", &numberOfCPUs, &length, NULL, 0) )
projects / x264 / blobdiff