--- xvid.c 2005/09/13 12:12:15 1.64 +++ xvid.c 2006/11/12 01:40:36 1.72 @@ -19,7 +19,7 @@ * along with this program ; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA * - * $Id: xvid.c,v 1.64 2005/09/13 12:12:15 suxen_drol Exp $ + * $Id: xvid.c,v 1.72 2006/11/12 01:40:36 chl Exp $ * ****************************************************************************/ @@ -40,6 +40,7 @@ #include "utils/mbfunctions.h" #include "quant/quant.h" #include "motion/motion.h" +#include "motion/gmc.h" #include "motion/sad.h" #include "utils/emms.h" #include "utils/timer.h" @@ -126,7 +127,7 @@ /* detect cpu flags */ static unsigned int -detect_cpu_flags() +detect_cpu_flags(void) { /* enable native assembly optimizations by default */ unsigned int cpu_flags = XVID_CPU_ASM; @@ -183,7 +184,7 @@ idct = idct_int32; /* Only needed on PPC Altivec archs */ - sadInit = 0; + sadInit = NULL; /* Restore FPU context : emms_c is a nop functions */ emms = emms_c; @@ -256,6 +257,7 @@ yv12_to_yv12 = yv12_to_yv12_c; rgb555_to_yv12 = rgb555_to_yv12_c; rgb565_to_yv12 = rgb565_to_yv12_c; + rgb_to_yv12 = rgb_to_yv12_c; bgr_to_yv12 = bgr_to_yv12_c; bgra_to_yv12 = bgra_to_yv12_c; abgr_to_yv12 = abgr_to_yv12_c; @@ -277,6 +279,7 @@ /* All colorspace transformation functions YV12->User format */ yv12_to_rgb555 = yv12_to_rgb555_c; yv12_to_rgb565 = yv12_to_rgb565_c; + yv12_to_rgb = yv12_to_rgb_c; yv12_to_bgr = yv12_to_bgr_c; yv12_to_bgra = yv12_to_bgra_c; yv12_to_abgr = yv12_to_abgr_c; @@ -306,6 +309,8 @@ sse8_16bit = sse8_16bit_c; sse8_8bit = sse8_8bit_c; + init_GMC(cpu_flags); + #if defined(ARCH_IS_IA32) if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || @@ -376,7 +381,9 @@ /* image input xxx_to_yv12 related functions */ yv12_to_yv12 = yv12_to_yv12_mmx; bgr_to_yv12 = bgr_to_yv12_mmx; + rgb_to_yv12 = rgb_to_yv12_mmx; bgra_to_yv12 = bgra_to_yv12_mmx; + rgba_to_yv12 = rgba_to_yv12_mmx; yuyv_to_yv12 = yuyv_to_yv12_mmx; uyvy_to_yv12 = uyvy_to_yv12_mmx; @@ -436,7 +443,6 @@ interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm; /* Quantization */ - quant_mpeg_intra = quant_mpeg_intra_xmm; quant_mpeg_inter = quant_mpeg_inter_xmm; dequant_h263_intra = dequant_h263_intra_xmm; @@ -636,7 +642,7 @@ * features there really are. */ if (cpu_flags & XVID_CPU_ASM) { /* SIMD state flusher */ - emms = emms_3dn; + emms = emms_mmx; /* DCT operators */ fdct = fdct_skal_x86_64; @@ -673,7 +679,7 @@ quant_h263_inter = quant_h263_inter_x86_64; dequant_h263_intra = dequant_h263_intra_x86_64; dequant_h263_inter = dequant_h263_inter_x86_64; - quant_mpeg_intra = quant_mpeg_intra_x86_64; + /*quant_mpeg_intra = quant_mpeg_intra_x86_64; fix me! */ quant_mpeg_inter = quant_mpeg_inter_x86_64; dequant_mpeg_intra = dequant_mpeg_intra_x86_64; dequant_mpeg_inter = dequant_mpeg_inter_x86_64; @@ -712,13 +718,21 @@ return XVID_ERR_VERSION; info->actual_version = XVID_VERSION; - info->build = "xvid-1.1.0-beta2"; + info->build = "xvid-1.2.0-dev"; info->cpu_flags = detect_cpu_flags(); + info->num_threads = 0; -#if defined(_SMP) && defined(WIN32) - info->num_threads = pthread_num_processors_np();; -#else - info->num_threads = 0; +#if defined(WIN32) + { + DWORD dwProcessAffinityMask, dwSystemAffinityMask; + if (GetProcessAffinityMask(GetCurrentProcess(), &dwProcessAffinityMask, &dwSystemAffinityMask)) { + int i; + for(i=0; i<32; i++) { + if ((dwProcessAffinityMask & (1<num_threads++; + } + } + } #endif return 0;