AltiVec support in postproc/ + altivec optimizations for yuv2yuvX patch by...

AltiVec support in postproc/ + altivec optimizations for yuv2yuvX patch by (Romain Dolbeau <dolbeau at irisa dot fr>)

Originally committed as revision 12018 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc

postproc/swscale.c

@@ -673,6 +673,12 @@ static inline void yuv2packedXinC(SwsContext *c, int16_t *lumFilter, int16_t **l
 #define COMPILE_C
 #endif
 
+#ifdef ARCH_POWERPC
+#ifdef HAVE_ALTIVEC
+#define COMPILE_ALTIVEC
+#endif //HAVE_ALTIVEC
+#endif //ARCH_POWERPC
+
 #ifdef ARCH_X86
 
 #if (defined (HAVE_MMX) && !defined (HAVE_3DNOW) && !defined (HAVE_MMX2)) || defined (RUNTIME_CPUDETECT)
@@ -696,10 +702,20 @@ static inline void yuv2packedXinC(SwsContext *c, int16_t *lumFilter, int16_t **l
 #undef HAVE_MMX
 #undef HAVE_MMX2
 #undef HAVE_3DNOW
+#undef HAVE_ALTIVEC
 #define RENAME(a) a ## _C
 #include "swscale_template.c"
 #endif
 
+#ifdef ARCH_POWERPC
+#ifdef COMPILE_ALTIVEC
+#undef RENAME
+#define HAVE_ALTIVEC
+#define RENAME(a) a ## _altivec
+#include "swscale_template.c"
+#endif
+#endif //ARCH_POWERPC
+
 #ifdef ARCH_X86
 
 //X86 versions
@@ -1309,6 +1325,12 @@ static SwsFunc getSwsFunc(int flags){
 		return swScale_C;
 
 #else
+#ifdef ARCH_POWERPC
+	if(flags & SWS_CPU_CAPS_ALTIVEC)
+	  return swScale_altivec;
+	else
+	  return swScale_C;
+#endif
 	return swScale_C;
 #endif
 #else //RUNTIME_CPUDETECT
@@ -1318,6 +1340,8 @@ static SwsFunc getSwsFunc(int flags){
 	return swScale_3DNow;
 #elif defined (HAVE_MMX)
 	return swScale_MMX;
+#elif defined (HAVE_ALTIVEC)
+	return swScale_altivec;
 #else
 	return swScale_C;
 #endif
@@ -1720,13 +1744,15 @@ SwsContext *sws_getContext(int srcW, int srcH, int origSrcFormat, int dstW, int
 #endif
 
 #ifndef RUNTIME_CPUDETECT //ensure that the flags match the compiled variant if cpudetect is off
-	flags &= ~(SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2|SWS_CPU_CAPS_3DNOW);
+	flags &= ~(SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2|SWS_CPU_CAPS_3DNOW|SWS_CPU_CAPS_ALTIVEC);
 #ifdef HAVE_MMX2
 	flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2;
 #elif defined (HAVE_3DNOW)
 	flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_3DNOW;
 #elif defined (HAVE_MMX)
 	flags |= SWS_CPU_CAPS_MMX;
+#elif defined (HAVE_ALTIVEC)
+	flags |= SWS_CPU_CAPS_ALTIVEC;
 #endif
 #endif
 	if(clip_table[512] != 255) globalInit();
@@ -2033,7 +2059,9 @@ SwsContext *sws_getContext(int srcW, int srcH, int origSrcFormat, int dstW, int
 			MSG_INFO("using 3DNOW\n");
 		else if(flags & SWS_CPU_CAPS_MMX)
 			MSG_INFO("using MMX\n");
-		else
+		else if(flags & SWS_CPU_CAPS_ALTIVEC)
+			MSG_INFO("using AltiVec\n");
+		else 
 			MSG_INFO("using C\n");
 	}
 

postproc/swscale.h

@@ -60,6 +60,7 @@ extern "C" {
 #define SWS_CPU_CAPS_MMX   0x80000000
 #define SWS_CPU_CAPS_MMX2  0x20000000
 #define SWS_CPU_CAPS_3DNOW 0x40000000
+#define SWS_CPU_CAPS_ALTIVEC 0x10000000
 
 #define SWS_MAX_REDUCE_CUTOFF 0.002
 

postproc/swscale_altivec_template.c

+/*
+  AltiVec-enhanced yuv2yuvX
+
+    Copyright (C) 2004 Romain Dolbeau <romain@dolbeau.org>
+    based on the equivalent C code in "postproc/swscale.c"
+
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+*/
+
+static const vector unsigned int altivec_vectorShiftInt19 = {19, 19, 19, 19};
+static inline void
+altivec_packIntArrayToCharArray(int *val, uint8_t* dest, int dstW) {
+  register int i;
+  if ((unsigned long)dest % 16) {
+    /* badly aligned store, we force store alignement */
+    /* and will handle load misalignement on val w/ vec_perm */
+    for (i = 0 ; (i < dstW) &&
+	   (((unsigned long)dest + i) % 16) ; i++) {
+      int t = val[i] >> 19;
+      dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
+    }
+    vector unsigned char perm1 = vec_lvsl(i << 2, val);
+    vector signed int v1 = vec_ld(i << 2, val);
+    for ( ; i < (dstW - 15); i+=16) {
+      int offset = i << 2;
+      vector signed int v2 = vec_ld(offset + 16, val);
+      vector signed int v3 = vec_ld(offset + 32, val);
+      vector signed int v4 = vec_ld(offset + 48, val);
+      vector signed int v5 = vec_ld(offset + 64, val);
+      vector signed int v12 = vec_perm(v1,v2,perm1);
+      vector signed int v23 = vec_perm(v2,v3,perm1);
+      vector signed int v34 = vec_perm(v3,v4,perm1);
+      vector signed int v45 = vec_perm(v4,v5,perm1);
+      
+      vector signed int vA = vec_sra(v12, altivec_vectorShiftInt19);
+      vector signed int vB = vec_sra(v23, altivec_vectorShiftInt19);
+      vector signed int vC = vec_sra(v34, altivec_vectorShiftInt19);
+      vector signed int vD = vec_sra(v45, altivec_vectorShiftInt19);
+      vector unsigned short vs1 = vec_packsu(vA, vB);
+      vector unsigned short vs2 = vec_packsu(vC, vD);
+      vector unsigned char vf = vec_packsu(vs1, vs2);
+      vec_st(vf, i, dest);
+      v1 = v5;
+    }
+  } else { // dest is properly aligned, great
+    for (i = 0; i < (dstW - 15); i+=16) {
+      int offset = i << 2;
+      vector signed int v1 = vec_ld(offset, val);
+      vector signed int v2 = vec_ld(offset + 16, val);
+      vector signed int v3 = vec_ld(offset + 32, val);
+      vector signed int v4 = vec_ld(offset + 48, val);
+      vector signed int v5 = vec_sra(v1, altivec_vectorShiftInt19);
+      vector signed int v6 = vec_sra(v2, altivec_vectorShiftInt19);
+      vector signed int v7 = vec_sra(v3, altivec_vectorShiftInt19);
+      vector signed int v8 = vec_sra(v4, altivec_vectorShiftInt19);
+      vector unsigned short vs1 = vec_packsu(v5, v6);
+      vector unsigned short vs2 = vec_packsu(v7, v8);
+      vector unsigned char vf = vec_packsu(vs1, vs2);
+      vec_st(vf, i, dest);
+    }
+  }
+  for ( ; i < dstW ; i++) {
+    int t = val[i] >> 19;
+    dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
+  }
+}
+
+static inline void
+yuv2yuvX_altivec_real(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
+		      int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
+		      uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW)
+{
+  const vector signed int vini = {(1 << 18), (1 << 18), (1 << 18), (1 << 18)};
+  register int i, j;
+  {
+    int __attribute__ ((aligned (16))) val[dstW];
+    
+    for (i = 0; i < (dstW -7); i+=4) {
+      vec_st(vini, i << 2, val);
+    }
+    for (; i < dstW; i++) {
+      val[i] = (1 << 18);
+    }
+    
+    for (j = 0; j < lumFilterSize; j++) {
+      vector signed short vLumFilter = vec_ld(j << 1, lumFilter);
+      vector unsigned char perm0 = vec_lvsl(j << 1, lumFilter);
+      vLumFilter = vec_perm(vLumFilter, vLumFilter, perm0);
+      vLumFilter = vec_splat(vLumFilter, 0); // lumFilter[j] is loaded 8 times in vLumFilter
+      
+      vector unsigned char perm = vec_lvsl(0, lumSrc[j]);
+      vector signed short l1 = vec_ld(0, lumSrc[j]);
+      
+      for (i = 0; i < (dstW - 7); i+=8) {
+	int offset = i << 2;
+	vector signed short l2 = vec_ld((i << 1) + 16, lumSrc[j]);
+	
+	vector signed int v1 = vec_ld(offset, val);
+	vector signed int v2 = vec_ld(offset + 16, val);
+	
+	vector signed short ls = vec_perm(l1, l2, perm); // lumSrc[j][i] ... lumSrc[j][i+7]
+	
+	vector signed int i1 = vec_mule(vLumFilter, ls);
+	vector signed int i2 = vec_mulo(vLumFilter, ls);
+	
+	vector signed int vf1 = vec_mergeh(i1, i2);
+	vector signed int vf2 = vec_mergel(i1, i2); // lumSrc[j][i] * lumFilter[j] ... lumSrc[j][i+7] * lumFilter[j]
+	
+	vector signed int vo1 = vec_add(v1, vf1);
+	vector signed int vo2 = vec_add(v2, vf2);
+	
+	vec_st(vo1, offset, val);
+	vec_st(vo2, offset + 16, val);
+	
+	l1 = l2;
+      }
+      for ( ; i < dstW; i++) {
+	val[i] += lumSrc[j][i] * lumFilter[j];
+      }
+    }
+    altivec_packIntArrayToCharArray(val,dest,dstW);
+  }
+  if (uDest != 0) {
+    int  __attribute__ ((aligned (16))) u[chrDstW];
+    int  __attribute__ ((aligned (16))) v[chrDstW];
+
+    for (i = 0; i < (chrDstW -7); i+=4) {
+      vec_st(vini, i << 2, u);
+      vec_st(vini, i << 2, v);
+    }
+    for (; i < chrDstW; i++) {
+      u[i] = (1 << 18);
+      v[i] = (1 << 18);
+    }
+    
+    for (j = 0; j < chrFilterSize; j++) {
+      vector signed short vChrFilter = vec_ld(j << 1, chrFilter);
+      vector unsigned char perm0 = vec_lvsl(j << 1, chrFilter);
+      vChrFilter = vec_perm(vChrFilter, vChrFilter, perm0);
+      vChrFilter = vec_splat(vChrFilter, 0); // chrFilter[j] is loaded 8 times in vChrFilter
+      
+      vector unsigned char perm = vec_lvsl(0, chrSrc[j]);
+      vector signed short l1 = vec_ld(0, chrSrc[j]);
+      vector signed short l1_V = vec_ld(2048 << 1, chrSrc[j]);
+      
+      for (i = 0; i < (chrDstW - 7); i+=8) {
+	int offset = i << 2;
+	vector signed short l2 = vec_ld((i << 1) + 16, chrSrc[j]);
+	vector signed short l2_V = vec_ld(((i + 2048) << 1) + 16, chrSrc[j]);
+	
+	vector signed int v1 = vec_ld(offset, u);
+	vector signed int v2 = vec_ld(offset + 16, u);
+	vector signed int v1_V = vec_ld(offset, v);
+	vector signed int v2_V = vec_ld(offset + 16, v);
+	
+	vector signed short ls = vec_perm(l1, l2, perm); // chrSrc[j][i] ... chrSrc[j][i+7]
+	vector signed short ls_V = vec_perm(l1_V, l2_V, perm); // chrSrc[j][i+2048] ... chrSrc[j][i+2055]
+	
+	vector signed int i1 = vec_mule(vChrFilter, ls);
+	vector signed int i2 = vec_mulo(vChrFilter, ls);
+	vector signed int i1_V = vec_mule(vChrFilter, ls_V);
+	vector signed int i2_V = vec_mulo(vChrFilter, ls_V);
+	
+	vector signed int vf1 = vec_mergeh(i1, i2);
+	vector signed int vf2 = vec_mergel(i1, i2); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
+	vector signed int vf1_V = vec_mergeh(i1_V, i2_V);
+	vector signed int vf2_V = vec_mergel(i1_V, i2_V); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
+	
+	vector signed int vo1 = vec_add(v1, vf1);
+	vector signed int vo2 = vec_add(v2, vf2);
+	vector signed int vo1_V = vec_add(v1_V, vf1_V);
+	vector signed int vo2_V = vec_add(v2_V, vf2_V);
+	
+	vec_st(vo1, offset, u);
+	vec_st(vo2, offset + 16, u);
+	vec_st(vo1_V, offset, v);
+	vec_st(vo2_V, offset + 16, v);
+	
+	l1 = l2;
+	l1_V = l2_V;
+      }
+      for ( ; i < chrDstW; i++) {
+	u[i] += chrSrc[j][i] * chrFilter[j];
+	v[i] += chrSrc[j][i + 2048] * chrFilter[j];
+      } 
+    }
+    altivec_packIntArrayToCharArray(u,uDest,chrDstW);
+    altivec_packIntArrayToCharArray(v,vDest,chrDstW);
+  }
+}

postproc/swscale_template.c

@@ -59,6 +59,10 @@
 #define MOVNTQ(a,b) "movq " #a ", " #b " \n\t"
 #endif
 
+#ifdef HAVE_ALTIVEC
+#include "swscale_altivec_template.c"
+#endif
+
 #define YSCALEYUV2YV12X(x, offset) \
 			"xorl %%eax, %%eax		\n\t"\
 			"movq "VROUNDER_OFFSET"(%0), %%mm3\n\t"\
@@ -766,9 +770,15 @@ static inline void RENAME(yuv2yuvX)(SwsContext *c, int16_t *lumFilter, int16_t *
 			: "%eax", "%edx", "%esi"
 		);
 #else
+#ifdef HAVE_ALTIVEC
+yuv2yuvX_altivec_real(lumFilter, lumSrc, lumFilterSize,
+		      chrFilter, chrSrc, chrFilterSize,
+		      dest, uDest, vDest, dstW, chrDstW);
+#else //HAVE_ALTIVEC
 yuv2yuvXinC(lumFilter, lumSrc, lumFilterSize,
 	    chrFilter, chrSrc, chrFilterSize,
 	    dest, uDest, vDest, dstW, chrDstW);
+#endif //!HAVE_ALTIVEC
 #endif
 }