Index: postproc/swscale.c
===================================================================
RCS file: /cvsroot/mplayer/main/postproc/swscale.c,v
retrieving revision 1.138
diff -u -r1.138 swscale.c
--- postproc/swscale.c	14 Mar 2004 18:04:59 -0000	1.138
+++ postproc/swscale.c	25 Mar 2004 14:22:37 -0000
@@ -1947,7 +1947,10 @@
 
 	/* precalculate horizontal scaler filter coefficients */
 	{
-		const int filterAlign= (flags & SWS_CPU_CAPS_MMX) ? 4 : 1;
+		const int filterAlign=
+		  (flags & SWS_CPU_CAPS_MMX) ? 4 :
+		  (flags & SWS_CPU_CAPS_ALTIVEC) ? 4 :
+		  1;
 
 		initFilter(&c->hLumFilter, &c->hLumFilterPos, &c->hLumFilterSize, c->lumXInc,
 				 srcW      ,       dstW, filterAlign, 1<<14,
Index: postproc/swscale_altivec_template.c
===================================================================
RCS file: /cvsroot/mplayer/main/postproc/swscale_altivec_template.c,v
retrieving revision 1.1
diff -u -r1.1 swscale_altivec_template.c
--- postproc/swscale_altivec_template.c	11 Mar 2004 17:58:11 -0000	1.1
+++ postproc/swscale_altivec_template.c	25 Mar 2004 14:22:37 -0000
@@ -20,7 +20,19 @@
     Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */
 
-static const vector unsigned int altivec_vectorShiftInt19 = {19, 19, 19, 19};
+#ifdef CONFIG_DARWIN
+static const vector signed int vzero =
+  (vector signed int)(0, 0, 0, 0);
+static const vector unsigned int altivec_vectorShiftInt19 =
+  (vector unsigned int)(19, 19, 19, 19);
+#else
+static const vector signed int vzero =
+  (vector signed int){0,0,0,0};
+static const vector unsigned int altivec_vectorShiftInt19 =
+  (vector unsigned int){19, 19, 19, 19};
+
+#endif
+
 static inline void
 altivec_packIntArrayToCharArray(int *val, uint8_t* dest, int dstW) {
   register int i;
@@ -201,3 +213,166 @@
     altivec_packIntArrayToCharArray(v,vDest,chrDstW);
   }
 }
+
+static inline void hScale_altivec_real(int16_t *dst, int dstW, uint8_t *src, int srcW, int xInc, int16_t *filter, int16_t *filterPos, int filterSize) {
+  register int i;
+  int __attribute__ ((aligned (16))) tempo[4];
+
+  switch (filterSize) {
+  default:
+    //fprintf(stderr, "filterSize: %d\n", filterSize);
+    for(i=0; i<dstW; i++) {
+      register int j;
+      register int srcPos = filterPos[i];
+      register int val = 0;
+      for(j=0; j<filterSize; j++) {
+	val += ((int)src[srcPos + j])*filter[filterSize*i + j];
+      }
+      dst[i] = MIN(MAX(0, val>>7), (1<<15)-1);
+    }
+    break;
+ 
+  case 4:
+    {
+      for(i=0; i<dstW; i++) {
+	register int j;
+	register int srcPos = filterPos[i];
+
+	vector unsigned char src_v0 = vec_ld(srcPos, src);
+	vector unsigned char src_v1;
+	if ((((int)src + srcPos)% 16) > 12) {
+	  src_v1 = vec_ld(srcPos + 16, src);
+	}
+	vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
+
+	vector signed short src_v = // vec_unpackh sign-extends...
+	  (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+	// now put our elements in the even slots
+	src_v = vec_mergeh(src_v, (vector signed short)vzero);
+
+	vector signed short filter_v = vec_ld(i << 3, filter);
+// the 3 above is 2 (filterSize == 4) + 1 (sizeof(short) == 2)
+
+// the neat trick : we only care for half the elements,
+// high or low depending on (i<<3)%16 (it's 0 or 8 here),
+// and we're going to use vec_mule anyway, so we chose
+// carefully how to "unpack" the elements into the even slots
+	if ((i << 3) % 16)
+	  filter_v = vec_mergel(filter_v,(vector signed short)vzero);
+	else
+	  filter_v = vec_mergeh(filter_v,(vector signed short)vzero);
+
+	vector signed int val_vEven = vec_mule(src_v, filter_v);
+	vector signed int val_s = vec_sums(val_vEven, vzero);
+	vec_st(val_s, 0, tempo);
+	dst[i] = MIN(MAX(0, tempo[3]>>7), (1<<15)-1);
+      }
+    }
+    break;
+
+  case 8:
+    {
+      for(i=0; i<dstW; i++) {
+	register int j;
+	register int srcPos = filterPos[i];
+
+	vector unsigned char src_v0 = vec_ld(srcPos, src);
+	vector unsigned char src_v1;
+	if ((((int)src + srcPos)% 16) > 8) {
+	  src_v1 = vec_ld(srcPos + 16, src);
+	}
+	vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
+
+	vector signed short src_v = // vec_unpackh sign-extends...
+	  (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+	vector signed short filter_v = vec_ld(i << 4, filter);
+// the 4 above is 3 (filterSize == 8) + 1 (sizeof(short) == 2)
+	vector signed int val_vEven = vec_mule(src_v, filter_v);
+	vector signed int val_vOdd = vec_mulo(src_v, filter_v);
+	vector signed int val_v = vec_add(val_vEven, val_vOdd);
+	vector signed int val_s = vec_sums(val_v, vzero);
+	vec_st(val_s, 0, tempo);
+	dst[i] = MIN(MAX(0, tempo[3]>>7), (1<<15)-1);
+      }
+    }
+    break;
+    
+  case 12:
+    {
+      for(i=0; i<dstW; i++) {
+	register int j;
+	register int srcPos = filterPos[i];
+
+	vector unsigned char src_v0 = vec_ld(srcPos, src);
+	vector unsigned char src_v1 = vec_ld(srcPos + 16, src);
+	vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
+
+	vector signed short src_vA = // vec_unpackh sign-extends...
+	  (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+	vector signed short src_vB = // vec_unpackh sign-extends...
+	  (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));
+
+	vector signed short filter_v0 = vec_ld(i * 24, filter);
+       	vector signed short filter_v1 = vec_ld((i * 24) + 16, filter);
+	if ((i * 24)%16) { // this is 8
+	  filter_v0 = vec_sld(filter_v0, filter_v1, 8);
+	  filter_v1 = vec_sld(filter_v1, filter_v1, 8);
+	}
+
+	vector signed int val_vEven0 = vec_mule(src_vA, filter_v0);
+	vector signed int val_vOdd0 = vec_mulo(src_vA, filter_v0);
+	vector signed int val_vEven1 = vec_mule(src_vB, filter_v1);
+	vector signed int val_vOdd1 = vec_mulo(src_vB, filter_v1);
+
+	// 12 is like 16, except we must drop the last 4 products,
+	// half of which are in "val_vEven1", the others in "val_vOdd1"
+	vector signed int val_vTruncated1 = vec_mergeh(val_vOdd1, val_vEven1);
+	vector signed int val_vA = vec_add(val_vEven0, val_vOdd0);
+	vector signed int val_v = vec_add(val_vTruncated1, val_vA);
+
+	vector signed int val_s = vec_sums(val_v, vzero);
+
+	vec_st(val_s, 0, tempo);
+
+	dst[i] = MIN(MAX(0, tempo[3]>>7), (1<<15)-1);
+      }
+    }
+    break;
+    
+  case 16:
+    {
+      for(i=0; i<dstW; i++) {
+	register int j;
+	register int srcPos = filterPos[i];
+
+	vector unsigned char src_v0 = vec_ld(srcPos, src);
+	vector unsigned char src_v1 = vec_ld(srcPos + 16, src);
+	vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
+
+	vector signed short src_vA = // vec_unpackh sign-extends...
+	  (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+	vector signed short src_vB = // vec_unpackh sign-extends...
+	  (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));
+
+	vector signed short filter_v0 = vec_ld(i << 5, filter);
+       	vector signed short filter_v1 = vec_ld((i << 5) + 16, filter);
+// the 5 above are 4 (filterSize == 16) + 1 (sizeof(short) == 2)
+
+	vector signed int val_vEven0 = vec_mule(src_vA, filter_v0);
+	vector signed int val_vOdd0 = vec_mulo(src_vA, filter_v0);
+	vector signed int val_vEven1 = vec_mule(src_vB, filter_v1);
+	vector signed int val_vOdd1 = vec_mulo(src_vB, filter_v1);
+
+	vector signed int val_vA = vec_add(val_vEven0, val_vOdd0);
+	vector signed int val_vB = vec_add(val_vEven1, val_vOdd1);
+	vector signed int val_v = vec_add(val_vA, val_vB);
+
+	vector signed int val_s = vec_sums(val_v, vzero);
+
+	vec_st(val_s, 0, tempo);
+	dst[i] = MIN(MAX(0, tempo[3]>>7), (1<<15)-1);
+      }
+    }
+    break;
+  }
+}
Index: postproc/swscale_template.c
===================================================================
RCS file: /cvsroot/mplayer/main/postproc/swscale_template.c,v
retrieving revision 1.105
diff -u -r1.105 swscale_template.c
--- postproc/swscale_template.c	11 Mar 2004 17:58:11 -0000	1.105
+++ postproc/swscale_template.c	25 Mar 2004 14:22:37 -0000
@@ -2154,6 +2154,9 @@
 		);
 	}
 #else
+#ifdef HAVE_ALTIVEC
+	hScale_altivec_real(dst, dstW, src, srcW, xInc, filter, filterPos, filterSize);
+#else
 	int i;
 	for(i=0; i<dstW; i++)
 	{
@@ -2171,6 +2174,7 @@
 //		dst[i] = val>>7;
 	}
 #endif
+#endif
 }
       // *** horizontal scale Y line to temp buffer
 static inline void RENAME(hyscale)(uint16_t *dst, int dstWidth, uint8_t *src, int srcW, int xInc,