[Mplayerxp-cvslog] SF.net SVN: mplayerxp:[116] mplayerxp
Brought to you by:
olov
Menu
▾
▴
[Mplayerxp-cvslog] SF.net SVN: mplayerxp:[116] mplayerxp
|
From: <nic...@us...> - 2010-01-21 14:58:45
|
Revision: 116
http://mplayerxp.svn.sourceforge.net/mplayerxp/?rev=116&view=rev
Author: nickols_k
Date: 2010-01-21 14:58:37 +0000 (Thu, 21 Jan 2010)
Log Message:
-----------
minor accelerate audio-filters + bugfixes
Modified Paths:
--------------
TOOLS/Makefile
TOOLS/asmopt.c
TOOLS/asmopt_template.h
TOOLS/asmoptf.c
TOOLS/asmoptf_template.h
mplayerxp/configure
mplayerxp/dec_ahead.c
mplayerxp/dec_ahead.h
mplayerxp/libmpcodecs/ad_mp3.c
mplayerxp/libmpdemux/demuxer_r.c
mplayerxp/mplayer.c
mplayerxp/postproc/af.c
mplayerxp/postproc/af.h
mplayerxp/postproc/af_ao2.c
mplayerxp/postproc/af_center.c
mplayerxp/postproc/af_channels.c
mplayerxp/postproc/af_comp.c
mplayerxp/postproc/af_crystality.c
mplayerxp/postproc/af_delay.c
mplayerxp/postproc/af_dummy.c
mplayerxp/postproc/af_dyn.c
mplayerxp/postproc/af_echo3d.c
mplayerxp/postproc/af_equalizer.c
mplayerxp/postproc/af_export.c
mplayerxp/postproc/af_extrastereo.c
mplayerxp/postproc/af_ffenc.c
mplayerxp/postproc/af_format.c
mplayerxp/postproc/af_gate.c
mplayerxp/postproc/af_hrtf.c
mplayerxp/postproc/af_karaoke.c
mplayerxp/postproc/af_lp.c
mplayerxp/postproc/af_pan.c
mplayerxp/postproc/af_raw.c
mplayerxp/postproc/af_resample.c
mplayerxp/postproc/af_scaletempo.c
mplayerxp/postproc/af_sinesuppress.c
mplayerxp/postproc/af_sub.c
mplayerxp/postproc/af_surround.c
mplayerxp/postproc/af_volnorm.c
mplayerxp/postproc/af_volume.c
mplayerxp/postproc/dsp.c
mplayerxp/postproc/dsp.h
mplayerxp/postproc/dsp_accel.h
mplayerxp/pvector/pvector.h
mplayerxp/pvector/pvector_f32_x86.h
mplayerxp/pvector/pvector_inc.h
Modified: TOOLS/Makefile
===================================================================
--- TOOLS/Makefile 2010-01-20 18:48:33 UTC (rev 115)
+++ TOOLS/Makefile 2010-01-21 14:58:37 UTC (rev 116)
@@ -16,7 +16,7 @@
../mplayerxp/osdep/libosdep.a ../mplayerxp/mp_msg.o ../mplayerxp/nls/libnls.a ../mplayerxp/cpudetect.o
COMMON_LIBS = $(MP_LIBS) $(EXTRALIBS) -lm
-CFLAGS = $(OPTFLAGS) -I ../mplayerxp -I../mplayerxp/libmpdemux -I../mplayerxp/libvo $(EXTRA_INC)
+CFLAGS = $(OPTFLAGS) -I./ -I../mplayerxp -I../mplayerxp/libmpdemux -I../mplayerxp/libvo $(EXTRA_INC)
all: $(OBJS)
Modified: TOOLS/asmopt.c
===================================================================
--- TOOLS/asmopt.c 2010-01-20 18:48:33 UTC (rev 115)
+++ TOOLS/asmopt.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -20,49 +20,10 @@
#include "../mplayerxp/mp_config.h"
#include "../mplayerxp/cpudetect.h"
-#undef OPTIMIZE_AVX
-#undef OPTIMIZE_SSE4
-#undef OPTIMIZE_SSSE3
-#undef OPTIMIZE_SSE3
-#undef OPTIMIZE_SSE2
-#undef OPTIMIZE_SSE
-#undef OPTIMIZE_MMX2
-#undef OPTIMIZE_MMX
-#define RENAME(a) a ## _C
-#include "asmopt_template.h"
+#define PVECTOR_TESTING
+#define PVECTOR_ACCEL_H "asmopt_template.h"
+#include "../mplayerxp/pvector/pvector_inc.h"
-
-#ifdef __MMX__
-#define OPTIMIZE_MMX
-#undef RENAME
-#define RENAME(a) a ## _MMX
-#include "asmopt_template.h"
-#endif
-#ifdef __SSE__
-#define OPTIMIZE_MMX2
-#undef RENAME
-#define RENAME(a) a ## _MMX2
-#include "asmopt_template.h"
-#endif
-#ifdef __SSE2__
-#define OPTIMIZE_SSE2
-#undef RENAME
-#define RENAME(a) a ## _SSE2
-#include "asmopt_template.h"
-#endif
-#ifdef __SSE3__
-#define OPTIMIZE_SSE3
-#undef RENAME
-#define RENAME(a) a ## _SSE3
-#include "asmopt_template.h"
-#endif
-#ifdef __SSE4_1__
-#define OPTIMIZE_SSE4
-#undef RENAME
-#define RENAME(a) a ## _SSE4
-#include "asmopt_template.h"
-#endif
-
#define ARR_SIZE (1024*64*2)*10
unsigned verbose=1;
extern CpuCaps gCpuCaps;
@@ -136,7 +97,7 @@
gCpuCaps.has3DNow, gCpuCaps.has3DNowExt,
gCpuCaps.hasSSE, gCpuCaps.hasSSE2);
- test_simd("asmopt.gen" ,"GENERIC:",convert_C);
+ test_simd("asmopt.gen" ,"GENERIC:",convert_c);
// ordered per speed fasterst first
#ifdef __SSE3__
if(gCpuCaps.hasSSE3) test_simd("asmopt.sse3","SSE3 :",convert_SSE3);
@@ -144,8 +105,8 @@
#ifdef __SSE2__
if(gCpuCaps.hasSSE2) test_simd("asmopt.sse2","SSE2 :",convert_SSE2);
#endif
-#ifdef __MMX2__
- if(gCpuCaps.hasMMX2) test_simd("asmopt.mmx2","MMX2 :",convert_MMX2);
+#ifdef __SSE__
+ if(gCpuCaps.hasMMX2) test_simd("asmopt.mmx2","MMX2 :",convert_SSE);
#endif
#ifdef __MMX__
if(gCpuCaps.hasMMX) test_simd("asmopt.mmx", "MMX :",convert_MMX);
Modified: TOOLS/asmopt_template.h
===================================================================
--- TOOLS/asmopt_template.h 2010-01-20 18:48:33 UTC (rev 115)
+++ TOOLS/asmopt_template.h 2010-01-21 14:58:37 UTC (rev 116)
@@ -10,35 +10,39 @@
MMX2 : cpu clocks=5874563 = 2659us...[OK]
MMX : cpu clocks=6092012 = 2757us...[OK]
*/
-void RENAME(convert)(unsigned char *dstbase,unsigned char *src,unsigned char *srca,unsigned int asize)
+void PVECTOR_RENAME(convert)(unsigned char *dstbase,unsigned char *src,unsigned char *srca,unsigned int asize)
{
#ifdef HAVE_INT_PVECTOR
- __ivec vzero = _ivec_setzero();
+ __ivec izero = _ivec_setzero();
#endif
- unsigned x,w;
- x = 0;
- w = asize;
+ uint8_t *out_data = dstbase;
+ uint8_t *in_data = src;
+
+ unsigned i,len;
+ i = 0;
+ len = asize;
#ifdef HAVE_INT_PVECTOR
- if(w>=__IVEC_SIZE)
- for(;x<w;x+=__IVEC_SIZE){
- __ivec vmsk,vdest,vsrc,vsrca,vt[4];
- vdest = _ivec_loadu(&dstbase[x]);
- vsrc = _ivec_loada(&src[x]);
- vsrca = _ivec_loada(&srca[x]);
- vmsk = _ivec_not(_ivec_cmpeq_s8(vsrca,vzero));
- vt[0] = _ivec_u16_from_lou8(vdest);
- vt[1] = _ivec_u16_from_hiu8(vdest);
- vt[2] = _ivec_u16_from_lou8(vsrca);
- vt[3] = _ivec_u16_from_hiu8(vsrca);
- vt[0] = _ivec_srl_s16_imm(_ivec_mullo_s16(vt[0],vt[2]),8);
- vt[1] = _ivec_srl_s16_imm(_ivec_mullo_s16(vt[1],vt[3]),8);
- vt[0] = _ivec_add_s8(_ivec_u8_from_u16(vt[0],vt[1]),vsrc);
- _ivec_storeu(&dstbase[x],_ivec_blend_u8(vdest,vt[0],vmsk));
- }
+ for(;i<len;i++) {
+ ((uint16_t*)out_data)[i]=((uint16_t)((uint8_t*)in_data)[i])<<8;
+ if((((long)out_data)&(__IVEC_SIZE-1))==0) break;
+ }
+ if((len-i)>=__IVEC_SIZE)
+ for(;i<len;i+=__IVEC_SIZE){
+ __ivec ind,itmp[2];
+ ind = _ivec_loadu(&((uint8_t *)in_data)[i]);
+#if 0 /* slower but portable on non-x86 CPUs version */
+ itmp[0]= _ivec_sll_s16_imm(_ivec_u16_from_lou8(ind),8);
+ itmp[1]= _ivec_sll_s16_imm(_ivec_u16_from_hiu8(ind),8);
+#else
+ itmp[0]= _ivec_interleave_lo_u8(izero,ind);
+ itmp[1]= _ivec_interleave_hi_u8(izero,ind);
#endif
- for(;x<(unsigned)w;x++){
- if(srca[x]) dstbase[x]=((dstbase[x]*srca[x])>>8)+src[x];
- }
+ _ivec_storea(&((uint16_t*)out_data)[i],itmp[0]);
+ _ivec_storea(&((uint16_t*)out_data)[i+__IVEC_SIZE/2],itmp[1]);
+ }
+#endif
+ for(;i<len;i++)
+ ((uint16_t*)out_data)[i]=((uint16_t)((uint8_t*)in_data)[i])<<8;
#ifdef HAVE_INT_PVECTOR
_ivec_empty();
_ivec_sfence();
Modified: TOOLS/asmoptf.c
===================================================================
--- TOOLS/asmoptf.c 2010-01-20 18:48:33 UTC (rev 115)
+++ TOOLS/asmoptf.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -14,6 +14,8 @@
#include <inttypes.h>
#include <unistd.h>
#include <fcntl.h>
+#define __USE_ISOC99
+#include <math.h>
#include <sys/mman.h>
#include <sys/time.h>
#include <limits.h>
@@ -21,54 +23,10 @@
#include "../mplayerxp/mp_config.h"
#include "../mplayerxp/cpudetect.h"
-#undef F32_OPTIMIZE_AVX
-#undef F32_OPTIMIZE_SSE4
-#undef F32_OPTIMIZE_SSSE3
-#undef F32_OPTIMIZE_SSE3
-#undef F32_OPTIMIZE_SSE2
-#undef F32_OPTIMIZE_SSE
-#undef F32_OPTIMIZE_3DNOW
-#define RENAME(a) a ## _C
-#include "asmoptf_template.h"
+#define PVECTOR_TESTING
+#define PVECTOR_ACCEL_H "asmoptf_template.h"
+#include "../mplayerxp/pvector/pvector_inc.h"
-
-#ifdef __3dNOW__
-#define F32_OPTIMIZE_3DNOW
-#undef RENAME
-#define RENAME(a) a ## _3DNOW
-#include "asmoptf_template.h"
-#endif
-#ifdef __SSE__
-#define F32_OPTIMIZE_SSE
-#undef RENAME
-#define RENAME(a) a ## _SSE
-#include "asmoptf_template.h"
-#endif
-#ifdef __SSE2__
-#define F32_OPTIMIZE_SSE2
-#undef RENAME
-#define RENAME(a) a ## _SSE2
-#include "asmoptf_template.h"
-#endif
-#ifdef __SSE3__
-#define F32_OPTIMIZE_SSE3
-#undef RENAME
-#define RENAME(a) a ## _SSE3
-#include "asmoptf_template.h"
-#endif
-#ifdef __SSE4_1__
-#define F32_OPTIMIZE_SSE4
-#undef RENAME
-#define RENAME(a) a ## _SSE4
-#include "asmoptf_template.h"
-#endif
-#ifdef __AVX__
-#define F32_OPTIMIZE_AVX
-#undef RENAME
-#define RENAME(a) a ## _AVX
-#include "asmoptf_template.h"
-#endif
-
#define ARR_SIZE (1024*64*2)*10
unsigned verbose=1;
extern CpuCaps gCpuCaps;
@@ -76,8 +34,8 @@
/* Fixme: put here any complexness of source array filling */
#define INIT_ARRAYS(x) \
{\
- for(i=0; i<x; i++) srca[i] = i; \
- for(i=0; i<x; i++) src[i] = i+64; \
+ for(i=0; i<x; i++) srca[i] = i*0.0001; \
+ for(i=0; i<x; i++) src[i] = (i+64)*0.0001; \
memset(dsta,0,sizeof(ARR_SIZE*sizeof(float))); \
}
@@ -143,7 +101,7 @@
gCpuCaps.has3DNow, gCpuCaps.has3DNowExt,
gCpuCaps.hasSSE, gCpuCaps.hasSSE2);
- test_simd("asmoptf.gen" ,"GENERIC:",convert_C);
+ test_simd("asmoptf.gen" ,"GENERIC:",convert_c);
// ordered per speed fasterst first
#ifdef __AVX__
if(gCpuCaps.hasAVX) test_simd("asmoptf.avx","AVX :",convert_AVX);
Modified: TOOLS/asmoptf_template.h
===================================================================
--- TOOLS/asmoptf_template.h 2010-01-20 18:48:33 UTC (rev 115)
+++ TOOLS/asmoptf_template.h 2010-01-21 14:58:37 UTC (rev 116)
@@ -10,34 +10,86 @@
MMX2 : cpu clocks=5874563 = 2659us...[OK]
MMX : cpu clocks=6092012 = 2757us...[OK]
*/
-void RENAME(convert)(uint8_t *dstbase,const uint8_t *src,const uint8_t *srca,unsigned int asize)
+#ifndef SATURATE
+#define SATURATE(x,_min,_max) {if((x)<(_min)) (x)=(_min); else if((x)>(_max)) (x)=(_max);}
+#endif
+
+static void __FASTCALL__ PVECTOR_RENAME(float2int32)(float* in, int32_t* out, unsigned len, int final)
{
+ register unsigned i;
+ float ftmp;
+#ifdef HAVE_F32_PVECTOR
+ __f32vec int_max;
+#endif
+ i=0;
+#ifdef HAVE_F32_PVECTOR
+ int_max = _f32vec_broadcast(INT_MAX-1);
+ for(;i<len;i++) {
+ ftmp=((float*)in)[i];
+ SATURATE(ftmp,-1.0,+1.0);
+ ((int32_t*)out)[i]=(int32_t)lrintf((INT_MAX-1)*ftmp);
+ if((((long)out)&(__F32VEC_SIZE-1))==0) break;
+ }
+ _ivec_empty();
+ if((len-i)>=__F32VEC_SIZE)
+ for(;i<len;i+=__F32VEC_SIZE/sizeof(float)) {
+ __f32vec tmp;
+ tmp = _f32vec_mul(int_max,_f32vec_loadu(&((float*)in)[i]));
+ if(final)
+ _f32vec_to_s32_stream(&((int32_t*)out)[i],tmp);
+ else
+ _f32vec_to_s32a(&((int32_t*)out)[i],tmp);
+ }
+ _ivec_sfence();
+ _ivec_empty();
+#endif
+ for(;i<len;i++) {
+ ftmp=((float*)in)[i];
+ SATURATE(ftmp,-1.0,+1.0);
+ ((int32_t*)out)[i]=(int32_t)lrintf((INT_MAX-1)*ftmp);
+ }
+}
+
+void PVECTOR_RENAME(convert)(uint8_t *dstbase,const uint8_t *src,const uint8_t *srca,unsigned int asize)
+{
unsigned i,len;
const uint8_t *in;
- uint8_t *out;
+ float ftmp;
+ int32_t __attribute__((aligned(__IVEC_SIZE))) i32_tmp[asize];
#ifdef HAVE_F32_PVECTOR
- __f32vec rev_imax = _f32vec_broadcast(1.0/INT_MAX);
+ __f32vec int_max = _f32vec_broadcast(SHRT_MAX);
#endif
- in = src;
+ uint8_t *out;
+ in = srca;
out = dstbase;
i=0;
len=asize;
#ifdef HAVE_F32_PVECTOR
for(;i<len;i++) {
- ((float*)out)[i]=(1.0/INT_MAX)*((float)((int32_t*)in)[i]);
+ ftmp=((float*)in)[i];
+ SATURATE(ftmp,-1.0,+1.0);
+ ((int16_t*)out)[i]=(int16_t)lrintf(SHRT_MAX*ftmp);
if((((long)out)&(__F32VEC_SIZE-1))==0) break;
}
- _f32vec_empty();
- if((len-i)>=__F32VEC_SIZE)
- for(;i<len;i+=__F32VEC_SIZE/sizeof(float)) {
- __f32vec tmp;
- tmp = _f32vec_mul(rev_imax,_f32vec_from_s32u(&((int32_t*)in)[i]));
- _f32vec_storea(&((float*)out)[i],tmp);
+ _ivec_empty();
+ if((len-i)>=__F32VEC_SIZE) {
+ PVECTOR_RENAME(float2int32)(in,i32_tmp,len-i,0);
+ for(;i<len;i+=__IVEC_SIZE*2/sizeof(int32_t)) {
+ __ivec tmp[2];
+ tmp[0] = _ivec_loada(&((int32_t*)i32_tmp)[i]));
+ tmp[1] = _ivec_loada(&((int32_t*)i32_tmp)[i+__F32VEC_SIZE]));
+ _f32vec_to_s32a(itmp[0],ftmp[0]);
+ _f32vec_to_s32a(itmp[1],ftmp[1]);
+ tmp=_ivec_s16_from_s32(_ivec_loada(itmp[1]),_ivec_loada(itmp[0]));
+ _ivec_storea(&((int16_t*)out)[i],tmp);
}
- _f32vec_sfence();
- _f32vec_empty();
+ }
+ _ivec_sfence();
+ _ivec_empty();
#endif
for(;i<len;i++) {
- ((float*)out)[i]=(1.0/INT_MAX)*((float)((int32_t*)in)[i]);
+ ftmp=((float*)in)[i];
+ SATURATE(ftmp,-1.0,+1.0);
+ ((int16_t*)out)[i]=(int16_t)lrintf(SHRT_MAX*ftmp);
}
}
Modified: mplayerxp/configure
===================================================================
--- mplayerxp/configure 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/configure 2010-01-21 14:58:37 UTC (rev 116)
@@ -360,7 +360,7 @@
_arch=$host_arch
x86_32 && _arch="x86_32"
x86_64 && _arch="x86_64"
-ffmpeg_args="--arch=$arch $ffmpeg_args"
+ffmpeg_args="--arch=$_arch $ffmpeg_args"
fi
test "$debug" != "0" && ffmpeg_args="$ffmpeg_args --enable-debug=$debug"
enabled $profile && ffmpeg_args="$ffmpeg_args --enable-profile"
Modified: mplayerxp/dec_ahead.c
===================================================================
--- mplayerxp/dec_ahead.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/dec_ahead.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -209,9 +209,7 @@
}
if(in_size<0)
{
- LOCK_VDECA();
- shva[dec_ahead_locked_frame].eof=1;
- UNLOCK_VDECA();
+ __MP_SYNCHRONIZE(vdeca_mutex,shva[dec_ahead_locked_frame].eof=1);
xp_eof=1;
if(has_xp_audio && enable_xp<XP_VAFull) {
while(!xp_audio_eof && !dec_ahead_in_lseek && !pthread_end_of_work) {
@@ -562,9 +560,7 @@
if(pthread_id && pthread_is_living && has_xp_video)
{
pthread_end_of_work=1;
- LOCK_VIDEO_DECODE();
- pthread_cond_signal(&video_decode_cond);
- UNLOCK_VIDEO_DECODE();
+ __MP_SYNCHRONIZE(video_decode_mutex,pthread_cond_signal(&video_decode_cond));
while(pthread_is_living && !force) usleep(0);
pthread_is_living=0;
pthread_attr_destroy(&our_attr);
@@ -577,9 +573,7 @@
a_pthread_end_of_work=1;
xp_audio_eof=1;
if(a_pthread_is_living) {
- LOCK_AUDIO_DECODE();
- pthread_cond_signal(&audio_decode_cond);
- UNLOCK_AUDIO_DECODE();
+ __MP_SYNCHRONIZE(audio_decode_mutex,pthread_cond_signal(&audio_decode_cond));
}
while(a_pthread_is_living && !force) usleep(0);
a_pthread_is_living=0;
@@ -711,9 +705,7 @@
if (is_reset_vcache)
UNLOCK_VDECODING(); /* Release lock from vo_x11 */
if(pthread_is_living) {
- LOCK_VIDEO_DECODE();
- pthread_cond_signal(&video_decode_cond);
- UNLOCK_VIDEO_DECODE();
+ __MP_SYNCHRONIZE(video_decode_mutex,pthread_cond_signal(&video_decode_cond));
while(dec_ahead_in_lseek==PreSeek)
usleep(1);
}
@@ -737,9 +729,7 @@
xp_thread_decode_audio();
if(pthread_is_living) {
- LOCK_VIDEO_DECODE();
- pthread_cond_signal(&video_decode_cond);
- UNLOCK_VIDEO_DECODE();
+ __MP_SYNCHRONIZE(video_decode_mutex,pthread_cond_signal(&video_decode_cond));
while(dec_ahead_in_lseek==Seek)
usleep(1);
while(abs_dec_ahead_locked_frame == abs_dec_ahead_active_frame && !xp_eof)
@@ -748,17 +738,11 @@
dec_ahead_in_lseek = NoSeek;
- if(a_pthread_is_living) {
- LOCK_AUDIO_DECODE();
- pthread_cond_signal(&audio_decode_cond);
- UNLOCK_AUDIO_DECODE();
- }
+ if(a_pthread_is_living)
+ __MP_SYNCHRONIZE(audio_decode_mutex,pthread_cond_signal(&audio_decode_cond));
- if(pthread_audio_is_living) {
- LOCK_AUDIO_PLAY();
- pthread_cond_signal(&audio_play_cond);
- UNLOCK_AUDIO_PLAY();
- }
+ if(pthread_audio_is_living)
+ __MP_SYNCHRONIZE(audio_play_mutex,pthread_cond_signal(&audio_play_cond));
}
/* Audio stuff */
Modified: mplayerxp/dec_ahead.h
===================================================================
--- mplayerxp/dec_ahead.h 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/dec_ahead.h 2010-01-21 14:58:37 UTC (rev 116)
@@ -45,22 +45,30 @@
#endif
#define LOCK_VDEC_ACTIVE() { MSG_D(DA_PREFIX"LOCK_VDEC_ACTIVE\n"); pthread_mutex_lock(&vdec_active_mutex); }
#define UNLOCK_VDEC_ACTIVE() { MSG_D(DA_PREFIX"UNLOCK_VDEC_ACTIVE\n"); pthread_mutex_unlock(&vdec_active_mutex); }
+
#define LOCK_VDEC_LOCKED() { MSG_D(DA_PREFIX"LOCK_VDEC_LOCKED\n"); pthread_mutex_lock(&vdec_locked_mutex); }
#define UNLOCK_VDEC_LOCKED() { MSG_D(DA_PREFIX"UNLOCK_VDEC_LOCKED\n"); pthread_mutex_unlock(&vdec_locked_mutex); }
+
#define LOCK_VDECODING() { MSG_D(DA_PREFIX"LOCK_VDECODING\n"); pthread_mutex_lock(&vdecoding_mutex); }
#define UNLOCK_VDECODING() { MSG_D(DA_PREFIX"UNLOCK_VDECODING\n"); pthread_mutex_unlock(&vdecoding_mutex); }
+
#define LOCK_VREADING() { MSG_D(DA_PREFIX"LOCK_VREADING\n"); pthread_mutex_lock(&vreading_mutex); }
#define UNLOCK_VREADING() { MSG_D(DA_PREFIX"UNLOCK_VREADING\n"); pthread_mutex_unlock(&vreading_mutex); }
+
#define LOCK_VDECA() { MSG_D(DA_PREFIX"LOCK_VDECA\n"); pthread_mutex_lock(&vdeca_mutex); }
#define UNLOCK_VDECA() { MSG_D(DA_PREFIX"UNLOCK_VDECA\n"); pthread_mutex_unlock(&vdeca_mutex); }
+
#define LOCK_AUDIO_PLAY() { MSG_D(DA_PREFIX"LOCK_AUDIO_PLAY\n"); pthread_mutex_lock(&audio_play_mutex); }
#define UNLOCK_AUDIO_PLAY() { MSG_D(DA_PREFIX"UNLOCK_AUDIO_PLAY\n"); pthread_mutex_unlock(&audio_play_mutex); }
+
#define LOCK_AUDIO_DECODE() { MSG_D(DA_PREFIX"LOCK_AUDIO_DECODE\n"); pthread_mutex_lock(&audio_decode_mutex); }
#define UNLOCK_AUDIO_DECODE() { MSG_D(DA_PREFIX"UNLOCK_AUDIO_DECODE\n"); pthread_mutex_unlock(&audio_decode_mutex); }
+
#define LOCK_VIDEO_DECODE() { MSG_D(DA_PREFIX"LOCK_VIDEO_DECODE\n"); pthread_mutex_lock(&video_decode_mutex); }
#define UNLOCK_VIDEO_DECODE() { MSG_D(DA_PREFIX"UNLOCK_VIDEO_DECODE\n"); pthread_mutex_unlock(&video_decode_mutex); }
-#define __MP_ATOMIC(OP) { pthread_mutex_t loc_mutex; pthread_mutex_lock(&loc_mutex); OP; pthread_mutex_unlock(&loc_mutex); }
+#define __MP_ATOMIC(OP) { static pthread_mutex_t loc_mutex; pthread_mutex_lock(&loc_mutex); OP; pthread_mutex_unlock(&loc_mutex); }
+#define __MP_SYNCHRONIZE(mtx,OP) { pthread_mutex_lock(&mtx); OP; pthread_mutex_unlock(&mtx); }
typedef struct sh_video_attr
{
Modified: mplayerxp/libmpcodecs/ad_mp3.c
===================================================================
--- mplayerxp/libmpcodecs/ad_mp3.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/libmpcodecs/ad_mp3.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -331,7 +331,7 @@
,fi.flags&MPG123_ORIGINAL?"Yes":"No"
,fi.flags&MPG123_CRC?"Yes":"No"
,fi.flags&MPG123_PRIVATE?"Yes":"No"
- ,fi.emphasis,mpg123_current_decoder(sh->context));
+ ,fi.emphasis,mpg123_current_decoder(priv->mh));
}
return 1;
}
@@ -354,10 +354,10 @@
int decode_audio(sh_audio_t *sh,unsigned char *buf,int minlen,int maxlen,float *pts)
{
mp3_priv_t *priv=sh->context;
- unsigned char *indata=NULL,*outdata;
+ unsigned char *indata=NULL,*outdata=NULL;
int err=MPG123_OK,indata_size=0;
off_t offset,cpos;
- size_t done=0,total=0;
+ size_t done=0;
minlen=1;
/* decode one frame per call to be compatible with old logic:
***************************
@@ -377,7 +377,6 @@
cpos = mpg123_tell(priv->mh);
*pts = priv->pts+((float)(cpos-priv->pos)/sh->i_bps);
err=mpg123_decode_frame(priv->mh,&offset,&outdata,&done);
- total+=done;
if(!((err==MPG123_OK)||(err==MPG123_NEED_MORE))) {
MSG_ERR("mpg123_read = %s done = %u minlen = %u\n",mpg123_plain_strerror(err),done,minlen);
}
Modified: mplayerxp/libmpdemux/demuxer_r.c
===================================================================
--- mplayerxp/libmpdemux/demuxer_r.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/libmpdemux/demuxer_r.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -9,6 +9,7 @@
#include "demuxer_r.h"
#include "../osdep/timer.h"
#include "../mplayer.h"
+#include "../dec_ahead.h"
pthread_mutex_t demuxer_mutex=PTHREAD_MUTEX_INITIALIZER;
@@ -170,26 +171,20 @@
int demuxer_switch_audio_r(demuxer_t *d, int id)
{
int retval;
- LOCK_DEMUXER();
- retval=demuxer_switch_audio(d,id);
- UNLOCK_DEMUXER();
+ __MP_SYNCHRONIZE(demuxer_mutex,retval=demuxer_switch_audio(d,id));
return retval;
}
int demuxer_switch_video_r(demuxer_t *d, int id)
{
int retval;
- LOCK_DEMUXER();
- retval=demuxer_switch_video(d,id);
- UNLOCK_DEMUXER();
+ __MP_SYNCHRONIZE(demuxer_mutex,retval=demuxer_switch_video(d,id));
return retval;
}
int demuxer_switch_subtitle_r(demuxer_t *d, int id)
{
int retval;
- LOCK_DEMUXER();
- retval=demuxer_switch_subtitle(d,id);
- UNLOCK_DEMUXER();
+ __MP_SYNCHRONIZE(demuxer_mutex,retval=demuxer_switch_subtitle(d,id));
return retval;
}
Modified: mplayerxp/mplayer.c
===================================================================
--- mplayerxp/mplayer.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/mplayer.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -768,9 +768,7 @@
volatile int dec_ahead_active_frame;
for(;;)
{
- LOCK_VDEC_ACTIVE();
- vo_get_active_frame(&dec_ahead_active_frame);
- UNLOCK_VDEC_ACTIVE();
+ __MP_SYNCHRONIZE(vdec_active_mutex,vo_get_active_frame(&dec_ahead_active_frame));
LOCK_VDECA();
if(shva[dec_ahead_active_frame].eof) break;
usleep(0);
@@ -1524,9 +1522,7 @@
int delay_corrected=1;
int final_frame=0;
int num_frames_decoded = 0;
- LOCK_VDEC_ACTIVE();
- vo_get_active_frame(&dec_ahead_active_frame);
- UNLOCK_VDEC_ACTIVE();
+ __MP_SYNCHRONIZE(vdec_active_mutex,vo_get_active_frame(&dec_ahead_active_frame));
LOCK_VDECA();
final_frame = shva[dec_ahead_active_frame].eof;
sh_video->num_frames = shva[dec_ahead_active_frame].num_frames;
@@ -1722,9 +1718,7 @@
bench_dropped_frames ++;
}
}
- LOCK_VDEC_ACTIVE();
- abs_dec_ahead_active_frame++;
- UNLOCK_VDEC_ACTIVE();
+ __MP_SYNCHRONIZE(vdec_active_mutex,abs_dec_ahead_active_frame++);
pinfo[xp_id].current_module=NULL;
/*================ A-V TIMESTAMP CORRECTION: =========================*/
@@ -1763,12 +1757,8 @@
if(x> max_pts_correction) x= max_pts_correction;
if(default_max_pts_correction>=0)
max_pts_correction=default_max_pts_correction;
- else
- {
- LOCK_VDECA();
- max_pts_correction=shva[dec_ahead_active_frame].duration*0.10; // +-10% of time
- UNLOCK_VDECA();
- }
+ else // +-10% of time
+ __MP_SYNCHRONIZE(vdeca_mutex,max_pts_correction=shva[dec_ahead_active_frame].duration*0.10);
if(enable_xp>=XP_VAPlay)
pthread_mutex_lock(&audio_timer_mutex);
sh_audio->timer+=x;
@@ -2869,9 +2859,7 @@
volatile unsigned ada_blitted_frame;
do {
usleep(0);
- LOCK_VDEC_LOCKED();
- ada_blitted_frame = abs_dec_ahead_blitted_frame;
- UNLOCK_VDEC_LOCKED();
+ __MP_SYNCHRONIZE(vdec_locked_mutex,ada_blitted_frame = abs_dec_ahead_blitted_frame);
}while(ada_blitted_frame < xp_num_frames/2 && !xp_eof);
}
if(run_xp_players()!=0) exit_player("Can't run xp players!\n");
@@ -3018,9 +3006,7 @@
ao_resume(); // resume audio
if( enable_xp >= XP_VAPlay ) {
dec_ahead_in_pause=0;
- LOCK_AUDIO_PLAY();
- pthread_cond_signal(&audio_play_cond);
- UNLOCK_AUDIO_PLAY();
+ __MP_SYNCHRONIZE(audio_play_mutex,pthread_cond_signal(&audio_play_cond));
}
}
if (vo_inited && sh_video)
Modified: mplayerxp/postproc/af.c
===================================================================
--- mplayerxp/postproc/af.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -523,7 +523,7 @@
// Iterate through all filters
do{
MSG_DBG2("filtering %s\n",af->info->name);
- data=af->play(af,data);
+ data=af->play(af,data,af->next?0:1);
af=af->next;
}while(af && data);
return data;
Modified: mplayerxp/postproc/af.h
===================================================================
--- mplayerxp/postproc/af.h 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af.h 2010-01-21 14:58:37 UTC (rev 116)
@@ -50,7 +50,7 @@
const af_info_t* info;
int (* __FASTCALL__ control)(struct af_instance_s* af, int cmd, void* arg);
void (* __FASTCALL__ uninit)(struct af_instance_s* af);
- af_data_t* (* __FASTCALL__ play)(struct af_instance_s* af, af_data_t* data);
+ af_data_t* (* __FASTCALL__ play)(struct af_instance_s* af, af_data_t* data,int final);
void* setup; // setup data for this specific instance and filter
af_data_t* data; // configuration for outgoing data stream
struct af_instance_s* next;
Modified: mplayerxp/postproc/af_ao2.c
===================================================================
--- mplayerxp/postproc/af_ao2.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_ao2.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -148,7 +148,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
// Do something necessary to get rid of annoying warning during compile
if(!af)
Modified: mplayerxp/postproc/af_center.c
===================================================================
--- mplayerxp/postproc/af_center.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_center.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -70,7 +70,7 @@
}
// Filter data through filter
-static af_data_t* play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_center_t* s = af->setup; // Setup for this instance
Modified: mplayerxp/postproc/af_channels.c
===================================================================
--- mplayerxp/postproc/af_channels.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_channels.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -244,7 +244,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_data_t* l = af->data; // Local data
Modified: mplayerxp/postproc/af_comp.c
===================================================================
--- mplayerxp/postproc/af_comp.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_comp.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -106,7 +106,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_comp_t* s = (af_comp_t*)af->setup; // Setup for this instance
Modified: mplayerxp/postproc/af_crystality.c
===================================================================
--- mplayerxp/postproc/af_crystality.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_crystality.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -516,7 +516,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; /* Current working data */
echo3d(af->setup,(float*)c->audio, c->len);
Modified: mplayerxp/postproc/af_delay.c
===================================================================
--- mplayerxp/postproc/af_delay.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_delay.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -107,7 +107,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_delay_t* s = af->setup; // Setup for this instance
Modified: mplayerxp/postproc/af_dummy.c
===================================================================
--- mplayerxp/postproc/af_dummy.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_dummy.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -27,7 +27,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
// Do something necessary to get rid of annoying warning during compile
if(!af)
Modified: mplayerxp/postproc/af_dyn.c
===================================================================
--- mplayerxp/postproc/af_dyn.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_dyn.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -60,7 +60,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
register unsigned i = 0;
float *in = (float*)data->audio; // Audio data
Modified: mplayerxp/postproc/af_echo3d.c
===================================================================
--- mplayerxp/postproc/af_echo3d.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_echo3d.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -192,7 +192,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; /* Current working data */
Modified: mplayerxp/postproc/af_equalizer.c
===================================================================
--- mplayerxp/postproc/af_equalizer.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_equalizer.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -168,7 +168,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_equalizer_t* s = (af_equalizer_t*)af->setup; // Setup
Modified: mplayerxp/postproc/af_export.c
===================================================================
--- mplayerxp/postproc/af_export.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_export.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -194,7 +194,7 @@
af audio filter instance
data audio data
*/
-static af_data_t* __FASTCALL__ play( struct af_instance_s* af, af_data_t* data )
+static af_data_t* __FASTCALL__ play( struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_export_t* s = af->setup; // Setup for this instance
@@ -225,10 +225,10 @@
// Export buffer to mmaped area
if(flag){
// update buffer in mapped area
- memcpy(s->mmap_area + SIZE_HEADER, s->buf[0], sz * c->bps * nch);
+ stream_copy(s->mmap_area + SIZE_HEADER, s->buf[0], sz * c->bps * nch);
s->count++; // increment counter (to sync)
- memcpy(s->mmap_area + SIZE_HEADER - sizeof(s->count),
- &(s->count), sizeof(s->count));
+ stream_copy(s->mmap_area + SIZE_HEADER - sizeof(s->count),
+ &(s->count), sizeof(s->count));
}
// We don't modify data, just export it
Modified: mplayerxp/postproc/af_extrastereo.c
===================================================================
--- mplayerxp/postproc/af_extrastereo.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_extrastereo.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -75,7 +75,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_extrastereo_t *s = af->setup;
register int i = 0;
Modified: mplayerxp/postproc/af_ffenc.c
===================================================================
--- mplayerxp/postproc/af_ffenc.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_ffenc.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -154,7 +154,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
unsigned tlen,ilen,olen,delta;
af_ffenc_t *s=af->setup;
Modified: mplayerxp/postproc/af_format.c
===================================================================
--- mplayerxp/postproc/af_format.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_format.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -38,11 +38,11 @@
}af_format_t;
// Switch endianess
-static void endian(void* in, void* out, int len, int bps);
+static void endian(void* in, void* out, int len, int bps,int final);
// From singed to unsigned
-static void si2us(void* in, void* out, int len, int bps);
+static void si2us(void* in, void* out, int len, int bps,int final);
// From unsinged to signed
-static void us2si(void* in, void* out, int len, int bps);
+static void us2si(void* in, void* out, int len, int bps,int final);
static const struct fmt_alias_s
{
@@ -378,7 +378,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* l = af->data; // Local data
af_data_t* c = data; // Current working data
@@ -389,7 +389,7 @@
// Change to cpu native endian format
if((c->format&AF_FORMAT_END_MASK)!=AF_FORMAT_NE)
- endian(c->audio,c->audio,len,c->bps);
+ endian(c->audio,c->audio,len,c->bps,final);
// Conversion table
switch(c->format & ~AF_FORMAT_END_MASK){
@@ -398,14 +398,14 @@
if(AF_FORMAT_A_LAW == (l->format&AF_FORMAT_SPECIAL_MASK))
to_ulaw(l->audio, l->audio, len, 1, AF_FORMAT_SI);
if((l->format&AF_FORMAT_SIGN_MASK) == AF_FORMAT_US)
- si2us(l->audio,l->audio,len,l->bps);
+ si2us(l->audio,l->audio,len,l->bps,final);
break;
case(AF_FORMAT_A_LAW):
from_alaw(c->audio, l->audio, len, l->bps, l->format&AF_FORMAT_POINT_MASK);
if(AF_FORMAT_A_LAW == (l->format&AF_FORMAT_SPECIAL_MASK))
to_alaw(l->audio, l->audio, len, 1, AF_FORMAT_SI);
if((l->format&AF_FORMAT_SIGN_MASK) == AF_FORMAT_US)
- si2us(l->audio,l->audio,len,l->bps);
+ si2us(l->audio,l->audio,len,l->bps,final);
break;
case(AF_FORMAT_F):
switch(l->format&AF_FORMAT_SPECIAL_MASK){
@@ -416,9 +416,12 @@
to_alaw(c->audio, l->audio, len, c->bps, c->format&AF_FORMAT_POINT_MASK);
break;
default:
- float2int(c->audio, l->audio, len, l->bps);
- if((l->format&AF_FORMAT_SIGN_MASK) == AF_FORMAT_US)
- si2us(l->audio,l->audio,len,l->bps);
+ if((l->format&AF_FORMAT_SIGN_MASK) == AF_FORMAT_US) {
+ float2int(c->audio, l->audio, len, l->bps,0);
+ si2us(l->audio,l->audio,len,l->bps,final);
+ }
+ else
+ float2int(c->audio, l->audio, len, l->bps,final);
break;
}
break;
@@ -428,9 +431,9 @@
// Change signed/unsigned
if((c->format&AF_FORMAT_SIGN_MASK) != (l->format&AF_FORMAT_SIGN_MASK)){
if((c->format&AF_FORMAT_SIGN_MASK) == AF_FORMAT_US)
- us2si(c->audio,c->audio,len,c->bps);
+ us2si(c->audio,c->audio,len,c->bps,final);
else
- si2us(c->audio,c->audio,len,c->bps);
+ si2us(c->audio,c->audio,len,c->bps,final);
}
// Convert to special formats
switch(l->format&(AF_FORMAT_SPECIAL_MASK|AF_FORMAT_POINT_MASK)){
@@ -441,12 +444,12 @@
to_alaw(c->audio, l->audio, len, c->bps, c->format&AF_FORMAT_POINT_MASK);
break;
case(AF_FORMAT_F):
- int2float(c->audio, l->audio, len, c->bps);
+ int2float(c->audio, l->audio, len, c->bps,final);
break;
default:
// Change the number of bits
if(c->bps != l->bps)
- change_bps(c->audio,l->audio,len,c->bps,l->bps);
+ change_bps(c->audio,l->audio,len,c->bps,l->bps,final);
else
l->audio=c->audio;
break;
@@ -456,7 +459,7 @@
// Switch from cpu native endian to the correct endianess
if((l->format&AF_FORMAT_END_MASK)!=AF_FORMAT_NE)
- endian(l->audio,l->audio,len,l->bps);
+ endian(l->audio,l->audio,len,l->bps,final);
// Set output data
c->audio = l->audio;
@@ -515,7 +518,7 @@
}
// Function implementations used by play
-static void endian(void* in, void* out, int len, int bps)
+static void endian(void* in, void* out, int len, int bps,int final)
{
register int i;
switch(bps){
@@ -545,7 +548,7 @@
}
}
-static void si2us(void* in, void* out, int len, int bps)
+static void si2us(void* in, void* out, int len, int bps,int final)
{
register int i;
switch(bps){
@@ -568,7 +571,7 @@
}
}
-static void us2si(void* in, void* out, int len, int bps)
+static void us2si(void* in, void* out, int len, int bps,int final)
{
register int i;
switch(bps){
Modified: mplayerxp/postproc/af_gate.c
===================================================================
--- mplayerxp/postproc/af_gate.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_gate.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -102,7 +102,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_gate_t* s = (af_gate_t*)af->setup; // Setup for this instance
Modified: mplayerxp/postproc/af_hrtf.c
===================================================================
--- mplayerxp/postproc/af_hrtf.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_hrtf.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -383,7 +383,7 @@
2. A bass compensation is introduced to ensure that 0-200 Hz are not
damped (without any real 3D acoustical image, however).
*/
-static af_data_t* __FASTCALL__ play(struct af_instance_s *af, af_data_t *data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s *af, af_data_t *data,int final)
{
af_hrtf_t *s = af->setup;
real_t *in = data->audio; // Input audio data
Modified: mplayerxp/postproc/af_karaoke.c
===================================================================
--- mplayerxp/postproc/af_karaoke.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_karaoke.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -37,7 +37,7 @@
}
// Filter data through filter
-static af_data_t* play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
float* a = c->audio; // Audio data
Modified: mplayerxp/postproc/af_lp.c
===================================================================
--- mplayerxp/postproc/af_lp.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_lp.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -58,7 +58,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
// Do something necessary to get rid of annoying warning during compile
if(!af)
Modified: mplayerxp/postproc/af_pan.c
===================================================================
--- mplayerxp/postproc/af_pan.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_pan.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -123,7 +123,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_data_t* l = af->data; // Local data
Modified: mplayerxp/postproc/af_raw.c
===================================================================
--- mplayerxp/postproc/af_raw.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_raw.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -142,7 +142,7 @@
af audio filter instance
data audio data
*/
-static af_data_t* __FASTCALL__ play( struct af_instance_s* af, af_data_t* data )
+static af_data_t* __FASTCALL__ play( struct af_instance_s* af, af_data_t* data,int final)
{
af_raw_t* s = af->setup; // Setup for this instance
if(s->fd) fwrite(data->audio,data->len,1,s->fd);
Modified: mplayerxp/postproc/af_resample.c
===================================================================
--- mplayerxp/postproc/af_resample.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_resample.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -348,7 +348,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
int len = 0; // Length of output data
af_data_t* c = data; // Current working data
Modified: mplayerxp/postproc/af_scaletempo.c
===================================================================
--- mplayerxp/postproc/af_scaletempo.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_scaletempo.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -165,7 +165,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_scaletempo_t* s = af->setup;
int offset_in;
@@ -202,6 +202,11 @@
bytes_off = s->best_overlap_offset(s);
s->output_overlap(s, pout, bytes_off);
}
+ if(final)
+ stream_copy(pout + s->bytes_overlap,
+ s->buf_queue + bytes_off + s->bytes_overlap,
+ s->bytes_standing);
+ else
memcpy(pout + s->bytes_overlap,
s->buf_queue + bytes_off + s->bytes_overlap,
s->bytes_standing);
Modified: mplayerxp/postproc/af_sinesuppress.c
===================================================================
--- mplayerxp/postproc/af_sinesuppress.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_sinesuppress.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -31,9 +31,9 @@
double pos;
}af_sinesuppress_t;
-static af_data_t* play_s16(struct af_instance_s* af, af_data_t* data);
+static af_data_t* play_s16(struct af_instance_s* af, af_data_t* data,int final);
#if 0
-static af_data_t* play_float(struct af_instance_s* af, af_data_t* data);
+static af_data_t* play_float(struct af_instance_s* af, af_data_t* data,int final);
#endif
// Initialization and runtime control
@@ -90,7 +90,7 @@
}
// Filter data through filter
-static af_data_t* play_s16(struct af_instance_s* af, af_data_t* data)
+static af_data_t* play_s16(struct af_instance_s* af, af_data_t* data,int final)
{
af_sinesuppress_t *s = af->setup;
register int i = 0;
@@ -121,7 +121,7 @@
}
#if 0
-static af_data_t* play_float(struct af_instance_s* af, af_data_t* data)
+static af_data_t* play_float(struct af_instance_s* af, af_data_t* data,int final)
{
af_sinesuppress_t *s = af->setup;
register int i = 0;
Modified: mplayerxp/postproc/af_sub.c
===================================================================
--- mplayerxp/postproc/af_sub.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_sub.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -136,7 +136,7 @@
#endif
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_sub_t* s = af->setup; // Setup for this instance
Modified: mplayerxp/postproc/af_surround.c
===================================================================
--- mplayerxp/postproc/af_surround.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_surround.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -168,7 +168,7 @@
//static int amp_L = 0, amp_R = 0, amp_C = 0, amp_S = 0;
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data){
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final){
af_surround_t* s = (af_surround_t*)af->setup;
float* m = steering_matrix[0];
float* in = data->audio; // Input audio data
Modified: mplayerxp/postproc/af_volnorm.c
===================================================================
--- mplayerxp/postproc/af_volnorm.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_volnorm.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -116,7 +116,7 @@
free(af->setup);
}
-static void __FASTCALL__ method1_int16(af_volnorm_t *s, af_data_t *c)
+static void __FASTCALL__ method1_int16(af_volnorm_t *s, af_data_t *c,int final)
{
register int i = 0;
int16_t *data = (int16_t*)c->audio; // Audio data
@@ -158,7 +158,7 @@
s->lastavg = (1.0 - SMOOTH_LASTAVG) * s->lastavg + SMOOTH_LASTAVG * newavg;
}
-static void __FASTCALL__ method1_float(af_volnorm_t *s, af_data_t *c)
+static void __FASTCALL__ method1_float(af_volnorm_t *s, af_data_t *c,int final)
{
register int i = 0;
float *data = (float*)c->audio; // Audio data
@@ -195,7 +195,7 @@
s->lastavg = (1.0 - SMOOTH_LASTAVG) * s->lastavg + SMOOTH_LASTAVG * newavg;
}
-static void __FASTCALL__ method2_int16(af_volnorm_t *s, af_data_t *c)
+static void __FASTCALL__ method2_int16(af_volnorm_t *s, af_data_t *c,int final)
{
register int i = 0;
int16_t *data = (int16_t*)c->audio; // Audio data
@@ -245,7 +245,7 @@
s->idx = (s->idx + 1) % NSAMPLES;
}
-static void __FASTCALL__ method2_float(af_volnorm_t *s, af_data_t *c)
+static void __FASTCALL__ method2_float(af_volnorm_t *s, af_data_t *c,int final)
{
register int i = 0;
float *data = (float*)c->audio; // Audio data
@@ -292,23 +292,23 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_volnorm_t *s = af->setup;
if(af->data->format == (AF_FORMAT_SI | AF_FORMAT_NE))
{
if (s->method)
- method2_int16(s, data);
+ method2_int16(s, data,final);
else
- method1_int16(s, data);
+ method1_int16(s, data,final);
}
else if(af->data->format == (AF_FORMAT_F | AF_FORMAT_NE))
{
if (s->method)
- method2_float(s, data);
+ method2_float(s, data,final);
else
- method1_float(s, data);
+ method1_float(s, data,final);
}
return data;
}
Modified: mplayerxp/postproc/af_volume.c
===================================================================
--- mplayerxp/postproc/af_volume.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/af_volume.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -135,7 +135,7 @@
}
// Filter data through filter
-static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data)
+static af_data_t* __FASTCALL__ play(struct af_instance_s* af, af_data_t* data,int final)
{
af_data_t* c = data; // Current working data
af_volume_t* s = (af_volume_t*)af->setup; // Setup for this instance
Modified: mplayerxp/postproc/dsp.c
===================================================================
--- mplayerxp/postproc/dsp.c 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/dsp.c 2010-01-21 14:58:37 UTC (rev 116)
@@ -664,7 +664,7 @@
bp->prev = bp->pprev = 0.0;
}
-static void __FASTCALL__ init_change_bps(const void* in, void* out, unsigned len, unsigned inbps, unsigned outbps)
+static void __FASTCALL__ init_change_bps(const void* in, void* out, unsigned len, unsigned inbps, unsigned outbps,int final)
{
#ifdef __SSE4_1__
if(gCpuCaps.hasSSE41) change_bps = change_bps_SSE4;
@@ -689,11 +689,11 @@
#endif
#endif /* __x86_64__ */
change_bps = change_bps_c;
- (*change_bps)(in,out,len,inbps,outbps);
+ (*change_bps)(in,out,len,inbps,outbps,final);
}
-void (* __FASTCALL__ change_bps)(const void* in, void* out, unsigned len, unsigned inbps, unsigned outbps)=init_change_bps;
+void (* __FASTCALL__ change_bps)(const void* in, void* out, unsigned len, unsigned inbps, unsigned outbps,int final)=init_change_bps;
-static void __FASTCALL__ init_float2int(void* in, void* out, int len, int bps)
+static void __FASTCALL__ init_float2int(void* in, void* out, int len, int bps,int final)
{
#ifdef __AVX__
if(gCpuCaps.hasAVX) float2int = float2int_AVX;
@@ -726,11 +726,11 @@
#endif
#endif /*__x86_64__*/
float2int = float2int_c;
- (*float2int)(in,out,len,bps);
+ (*float2int)(in,out,len,bps,final);
}
-void (* __FASTCALL__ float2int)(void* in, void* out, int len, int bps)=init_float2int;
+void (* __FASTCALL__ float2int)(void* in, void* out, int len, int bps,int final)=init_float2int;
-static void __FASTCALL__ init_int2float(void* in, void* out, int len, int bps)
+static void __FASTCALL__ init_int2float(void* in, void* out, int len, int bps,int final)
{
#ifdef __AVX__
if(gCpuCaps.hasAVX) int2float = int2float_AVX;
@@ -763,9 +763,9 @@
#endif
#endif /*__x86_64__*/
int2float = int2float_c;
- (*int2float)(in,out,len,bps);
+ (*int2float)(in,out,len,bps,final);
}
-void (* __FASTCALL__ int2float)(void* in, void* out, int len, int bps)=init_int2float;
+void (* __FASTCALL__ int2float)(void* in, void* out, int len, int bps,int final)=init_int2float;
static int32_t __FASTCALL__ FIR_i16_init(int16_t *x,int16_t *w)
Modified: mplayerxp/postproc/dsp.h
===================================================================
--- mplayerxp/postproc/dsp.h 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/dsp.h 2010-01-21 14:58:37 UTC (rev 116)
@@ -143,9 +143,9 @@
}
/* some mmx_optimized stuff */
-extern void (* __FASTCALL__ change_bps)(const void* in, void* out, unsigned len, unsigned inbps, unsigned outbps);
-extern void (* __FASTCALL__ float2int)(void* in, void* out, int len, int bps);
-extern void (* __FASTCALL__ int2float)(void* in, void* out, int len, int bps);
+extern void (* __FASTCALL__ change_bps)(const void* in, void* out, unsigned len, unsigned inbps, unsigned outbps,int final);
+extern void (* __FASTCALL__ float2int)(void* in, void* out, int len, int bps,int final);
+extern void (* __FASTCALL__ int2float)(void* in, void* out, int len, int bps,int final);
extern int32_t (* __FASTCALL__ FIR_i16)(int16_t *x,int16_t *w);
extern float (* __FASTCALL__ FIR_f32)(float *x,float *w);
Modified: mplayerxp/postproc/dsp_accel.h
===================================================================
--- mplayerxp/postproc/dsp_accel.h 2010-01-20 18:48:33 UTC (rev 115)
+++ mplayerxp/postproc/dsp_accel.h 2010-01-21 14:58:37 UTC (rev 116)
@@ -1,66 +1,163 @@
/* DSP acceleration routines */
#include "pvector/pvector.h"
-#ifdef HAVE_INT_PVECTOR
-static __inline __m64 __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_load)(const void *__P)
-{
- return *(const __m64 *)__P;
-}
-#undef _m_load
-#define _m_load PVECTOR_RENAME(_m_load)
-static __inline __m64 __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_load_half)(const void *__P)
+static void __FASTCALL__ PVECTOR_RENAME(int8_to_int16)(const int8_t* in_data, int16_t* out_data, unsigned len, int final)
{
- return _mm_cvtsi32_si64 (*(const int *)__P);
+#ifdef HAVE_INT_PVECTOR
+ __ivec izero = _ivec_setzero();
+#endif
+ unsigned i;
+ i = 0;
+#ifdef HAVE_INT_PVECTOR
+ for(;i<len;i++) {
+ ((uint16_t*)out_data)[i]=((uint16_t)((uint8_t*)in_data)[i])<<8;
+ if((((long)out_data)&(__IVEC_SIZE-1))==0) break;
+ }
+ if((len-i)>=__IVEC_SIZE)
+ for(;i<len;i+=__IVEC_SIZE){
+ __ivec ind,itmp[2];
+ ind = _ivec_loadu(&((uint8_t *)in_data)[i]);
+#if 0 /* slower but portable on non-x86 CPUs version */
+ itmp[0]= _ivec_sll_s16_imm(_ivec_u16_from_lou8(ind),8);
+ itmp[1]= _ivec_sll_s16_imm(_ivec_u16_from_hiu8(ind),8);
+#else
+ itmp[0]= _ivec_interleave_lo_u8(izero,ind);
+ itmp[1]= _ivec_interleave_hi_u8(izero,ind);
+#endif
+ if(final) {
+ _ivec_stream(&((uint16_t*)out_data)[i],itmp[0]);
+ _ivec_stream(&((uint16_t*)out_data)[i+__IVEC_SIZE/2],itmp[1]);
+ } else {
+ _ivec_storea(&((uint16_t*)out_data)[i],itmp[0]);
+ _ivec_storea(&((uint16_t*)out_data)[i+__IVEC_SIZE/2],itmp[1]);
+ }
+ }
+#endif
+ for(;i<len;i++)
+ ((uint16_t*)out_data)[i]=((uint16_t)((uint8_t*)in_data)[i])<<8;
+#ifdef HAVE_INT_PVECTOR
+ _ivec_empty();
+ _ivec_sfence();
+#endif
}
-#undef _m_load_half
-#define _m_load_half PVECTOR_RENAME(_m_load_half)
-static __inline void __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_store)(void *__P, __m64 src)
+static void __FASTCALL__ PVECTOR_RENAME(int16_to_int8)(const int16_t* in_data, int8_t* out_data, unsigned len, int final)
{
- *(__m64 *)__P = src;
+ unsigned i;
+ i = 0;
+#ifdef HAVE_INT_PVECTOR
+ for(;i<len;i++) {
+ ((uint8_t*)out_data)[i]=(uint8_t)((((uint16_t*)in_data)[i])>>8);
+ if((((long)out_data)&(__IVEC_SIZE-1))==0) break;
+ }
+ if((len-i)>=__IVEC_SIZE)
+ for(;i<len;i+=__IVEC_SIZE){
+ __ivec outd,itmp[2];
+ itmp[0] = _ivec_sra_s16_imm(_ivec_loadu(&((uint16_t*)in_data)[i]),8);
+ itmp[1] = _ivec_sra_s16_imm(_ivec_loadu(&((uint16_t*)in_data)[i+__IVEC_SIZE/2]),8);
+ outd = _ivec_s8_from_s16(itmp[0],itmp[1]);
+ if(final)
+ _ivec_stream(&((uint8_t*)out_data)[i],outd);
+ else
+ _ivec_storea(&((uint8_t*)out_data)[i],outd);
+ }
+#endif
+ for(;i<len;i++)
+ ((uint8_t*)out_data)[i]=(uint8_t)((((uint16_t*)in_data)[i])>>8);
+#ifdef HAVE_INT_PVECTOR
+ _ivec_empty();
+ _ivec_sfence();
+#endif
}
-#undef _m_store
-#define _m_store PVECTOR_RENAME(_m_store)
-static __inline void __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_store_half)(void *__P, __m64 src)
+static void __FASTCALL__ PVECTOR_RENAME(int16_to_int32)(const int16_t* in_data, int32_t* out_data, unsigned len, int final)
{
- *(int *)__P = _mm_cvtsi64_si32(src);
-}
-#undef _m_store_half
-#define _m_store_half PVECTOR_RENAME(_m_store_half)
-
-static __inline void __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_movntq)(void *__P, __m64 src)
-{
-#ifdef HAVE_MMX2
- _mm_stream_pi(__P,src);
+#ifdef HAVE_INT_PVECTOR
+ __ivec izero = _ivec_setzero();
+ unsigned len_mm,j;
+#endif
+ unsigned i;
+ i=0;
+#ifdef HAVE_INT_PVECTOR
+ j=0;
+ len_mm=len&(~(__IVEC_SIZE-1));
+ for(;i<len;i++,j+=2){
+ ((uint32_t*)out_data)[i]=((uint32_t)((uint16_t*)in_data)[i])<<16;
+ if((((long)out_data)&(__IVEC_SIZE-1))==0) break;
+ }
+ if((len_mm-i)>=__IVEC_SIZE)
+ for(;i<len_mm;i+=__IVEC_SIZE/2,j+=__IVEC_SIZE)
+ {
+ __ivec ind,tmp[2];
+ ind = _ivec_loadu(&((uint8_t *)in_data)[j]);
+#if 0 /* slower but portable on non-x86 CPUs version */
+ tmp[0]= _ivec_sll_s32_imm(_ivec_u32_from_lou16(ind),16);
+ tmp[1]= _ivec_sll_s32_imm(_ivec_u32_from_hiu16(ind),16);
#else
- _m_store(__P,src);
+ tmp[0]= _ivec_interleave_lo_u16(izero,ind);
+ tmp[1]= _ivec_interleave_hi_u16(izero,ind);
#endif
-}
-#undef _m_movntq
-#define _m_movntq PVECTOR_RENAME(_m_movntq)
+ if(final) {
+ _ivec_stream(&((uint8_t *)out_data)[j*2],tmp[0]);
+ _ivec_stream(&((uint8_t *)out_data)[j*2+__IVEC_SIZE],tmp[1]);
+ } else {
+ _ivec_storea(&((uint8_t *)out_data)[j*2],tmp[0]);
+ _ivec_storea(&((uint8_t *)out_data)[j*2+__IVEC_SIZE],tmp[1]);
+ }
+ }
#endif
+ for(;i<len;i++)
+ ((uint32_t*)out_data)[i]=((uint32_t)((uint16_t*)in_data)[i])<<16;
+#ifdef HAVE_INT_PVECTOR
+ _ivec_sfence();
+ _ivec_empty();
+#endif
+}
-static void __FASTCALL__ PVECTOR_RENAME(change_bps)(const void* in_data, void* out_data, unsigned len, unsigned inbps, unsigned outbps)
+static void __FASTCALL__ PVECTOR_RENAME(int32_to_int16)(const int32_t* in_data, int16_t* out_data, unsigned len, int final)
{
#ifdef HAVE_INT_PVECTOR
- __ivec izero = _ivec_setzero();
- unsigned len_mm,j;
+ unsigned len_mm,j;
#endif
+ unsigned i;
+ i=0;
+#ifdef HAVE_INT_PVECTOR
+ j=0;
+ len_mm=len&(~(__IVEC_SIZE-1));
+ for(;i<len;i++,j+=2){
+ ((uint16_t*)out_data)[i]=(uint16_t)((((uint32_t*)in_data)[i])>>16);
+ if((((long)out_data)&(__IVEC_SIZE-1))==0) break;
+ }
+ if((len-i)>=__IVEC_SIZE)
+ for(;i<len_mm;i+=__IVEC_SIZE/2,j+=__IVEC_SIZE)
+ {
+ __ivec ind[2],tmp;
+ ind[0]= _ivec_sra_s32_imm(_ivec_loadu(&((uint8_t *)in_data)[j*2]),16);
+ ind[1]= _ivec_sra_s32_imm(_ivec_loadu(&((uint8_t *)in_data)[j*2+__IVEC_SIZE]),16);
+ tmp = _ivec_s16_from_s32(ind[0],ind[1]);
+ if(final)
+ _ivec_stream(&((uint8_t *)out_data)[j],tmp);
+ else
+ _ivec_storea(&((uint8_t *)out_data)[j],tmp);
+ }
+#endif
+ for(;i<len;i++)
+ ((uint16_t*)out_data)[i]=(uint16_t)((((uint32_t*)in_data)[i])>>16);
+#ifdef HAVE_INT_PVECTOR
+ _ivec_sfence();
+ _ivec_empty();
+#endif
+}
+
+static void __FASTCALL__ PVECTOR_RENAME(change_bps)(const void* in_data, void* out_data, unsigned len, unsigned inbps, unsigned outbps,int final)
+{
unsigned i;
// Change the number of bits
switch(inbps){
case 1:
switch(outbps){
case 2:
- i=0;
- for(;i<len;i++)
- ((uint16_t*)out_data)[i]=((uint16_t)((uint8_t*)in_data)[i])<<8;
+ PVECTOR_RENAME(int8_to_int16)(in_data,out_data,len,final);
break;
case 3:
for(i=0;i<len;i++)
@@ -78,9 +175,7 @@
case 2:
switch(outbps){
case 1:
- i=0;
- for(;i<len;i++)
- ((uint8_t*)out_data)[i]=(uint8_t)((((uint16_t*)in_data)[i])>>8);
+ PVECTOR_RENAME(int16_to_int8)(in_data,out_data,len,final);
break;
case 3:
for(i=0;i<len;i++)
@@ -89,32 +184,7 @@
((uint8_t*)out_data)[3*i+2]=(((uint8_t*)in_data)[2*i+1]);
break;
case 4:
- i=0;
-#ifdef HAVE_INT_PVECTOR
- j=0;
- len_mm=len&(~(__IVEC_SIZE-1));
- for(;i<len;i++,j+=2){
- ((uint32_t*)out_data)[i]=((uint32_t)((uint16_t*)in_data)[i])<<16;
- if((((long)out_data)&(__IVEC_SIZE-1))==0) break;
- }
- if((len_mm-i)>=__IVEC_SIZE)
- for(;i<len_mm;i+=__IVEC_SIZE/2,j+=__IVEC_SIZE)
- {
- __ivec ind,tmp[2];
- ind = _ivec_loadu(&((uint8_t *)in_data)[j]);
-#if 0 /* slower but portable on non-x86 CPUs version */
- tmp[0]= _ivec_sll_s32_imm(_ivec_u32_from_lou16(ind),16);
- tmp[1]= _ivec_sll_s32_imm(_ivec_u32_from_hiu16(ind),16);
-#else
- tmp[0]= _ivec_interleave_lo_u16(izero,ind);
- tmp[1]= _ivec_interleave_hi_u16(izero,ind);
-#endif
- _ivec_storea(&((uint8_t *)out_data)[j*2],tmp[0]);
- _ivec_storea(&((uint8_t *)out_data)[j*2+__IVEC_SIZE],tmp[1]);
- }
-#endif
- for(;i<len;i++)
- ((uint32_t*)out_data)[i]=((uint32_t)((uint16_t*)in_data)[i])<<16;
+ PVECTOR_RENAME(int16_to_int32)(in_data,out_data,len,final);
break;
}
break;
@@ -150,26 +220,7 @@
((uint8_t*)out_data)[i]=(uint8_t)((((uint32_t*)in_data)[i])>>24);
break;
case 2:
- i=0;
-#ifdef HAVE_INT_PVECTOR
- j=0;
- len_mm=len&(~(__IVEC_SIZE-1));
- for(;i<len;i++,j+=2){
- ((uint16_t*)out_data)[i]=(uint16_t)((((uint32_t*)in_data)[i])>>16);
- if((((long)out_data)&(__IVEC_SIZE-1))==0) break;
- }
- if((len-i)>=__IVEC_SIZE)
- for(;i<len_mm;i+=__IVEC_SIZE/2,j+=__IVEC_SIZE)
- {
- __ivec ind[2],tmp;
- ind[0]= _ivec_sra_s32_imm(_ivec_loadu(&((uint8_t *)in_data)[j*2]),16);
- ind[1]= _ivec_sra_s32_imm(_ivec_loadu(&((uint8_t *)in_data)[j*2+__IVEC_SIZE]),16);
- tmp = _ivec_s16_from_s32(ind[0],ind[1]);
- _ivec_storea(&((uint8_t *)out_data)[j],tmp);
- }
-#endif
- for(;i<len;i++)
- ((uint16_t*)out_data)[i]=(uint16_t)((((uint32_t*)in_data)[i])>>16);
+ PVECTOR_RENAME(int32_to_int16)(in_data,out_data,len,final);
break;
case 3:
for(i=0;i<len;i++)
@@ -180,12 +231,53 @@
}
break;
}
+}
+
#ifdef HAVE_INT_PVECTOR
- _ivec_sfence();
- _ivec_empty();
-#endif
+static __inline __m64 __attribute__((__gnu_inline__, __always_inline__))
+PVECTOR_RENAME(_m_load)(const void *__P)
+{
+ return *(const __m64 *)__P;
}
+#undef _m_load
+#define _m_load PVECTOR_RENAME(_m_load)
+static __inline __m64 __attribute__((__gnu_inline__, __always_inline__))
+PVECTOR_RENAME(_m_load_half)(const void *__P)
+{
+ return _mm_cvtsi32_si64 (*(const int *)__P);
+}
+#undef _m_load_half
+#define _m_load_half PVECTOR_RENAME(_m_load_half)
+
+static __inline void __attribute__((__gnu_inline__, __always_inline__))
+PVECTOR_RENAME(_m_store)(void *__P, __m64...
[truncated message content] |
