From 61d793ef13ac2ef8f9c2b41b71430f21fac80337 Mon Sep 17 00:00:00 2001
From: Mike Melanson <mike@multimedia.cx>
Date: Sun, 1 Feb 2004 05:31:16 +0000
Subject: sync to ffmpeg build 4699

CVS patchset: 6090
CVS date: 2004/02/01 05:31:16
---
 src/libffmpeg/libavcodec/ppc/dsputil_altivec.c | 40 +++++++++++---------------
 src/libffmpeg/libavcodec/ppc/dsputil_altivec.h | 16 +++++------
 src/libffmpeg/libavcodec/ppc/dsputil_ppc.c     | 24 +++++++++++-----
 src/libffmpeg/libavcodec/ppc/dsputil_ppc.h     |  1 +
 src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c   |  3 +-
 5 files changed, 43 insertions(+), 41 deletions(-)

(limited to 'src/libffmpeg/libavcodec/ppc')

diff --git a/src/libffmpeg/libavcodec/ppc/dsputil_altivec.c b/src/libffmpeg/libavcodec/ppc/dsputil_altivec.c
index 635480784..d1a294365 100644
--- a/src/libffmpeg/libavcodec/ppc/dsputil_altivec.c
+++ b/src/libffmpeg/libavcodec/ppc/dsputil_altivec.c
@@ -45,7 +45,7 @@ static void sigill_handler (int sig)
 }
 #endif /* CONFIG_DARWIN */
 
-int pix_abs16x16_x2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
+int sad16_x2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
@@ -57,7 +57,7 @@ int pix_abs16x16_x2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
 
     s = 0;
     sad = (vector unsigned int)vec_splat_u32(0);
-    for(i=0;i<16;i++) {
+    for(i=0;i<h;i++) {
         /*
            Read unaligned pixels into our vectors. The vectors are as follows:
            pix1v: pix1[0]-pix1[15]
@@ -92,7 +92,7 @@ int pix_abs16x16_x2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
     return s;
 }
 
-int pix_abs16x16_y2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
+int sad16_y2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
@@ -118,7 +118,7 @@ int pix_abs16x16_y2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
     tv = (vector unsigned char *) &pix2[0];
     pix2v = vec_perm(tv[0], tv[1], vec_lvsl(0, &pix2[0]));
     
-    for(i=0;i<16;i++) {
+    for(i=0;i<h;i++) {
         /*
            Read unaligned pixels into our vectors. The vectors are as follows:
            pix1v: pix1[0]-pix1[15]
@@ -152,7 +152,7 @@ int pix_abs16x16_y2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
     return s;    
 }
 
-int pix_abs16x16_xy2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
+int sad16_xy2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
@@ -194,7 +194,7 @@ int pix_abs16x16_xy2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
     t1 = vec_add(pix2hv, pix2ihv);
     t2 = vec_add(pix2lv, pix2ilv);
     
-    for(i=0;i<16;i++) {
+    for(i=0;i<h;i++) {
         /*
            Read unaligned pixels into our vectors. The vectors are as follows:
            pix1v: pix1[0]-pix1[15]
@@ -253,7 +253,7 @@ int pix_abs16x16_xy2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
     return s;
 }
 
-int pix_abs16x16_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
+int sad16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
@@ -266,7 +266,7 @@ int pix_abs16x16_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
     sad = (vector unsigned int)vec_splat_u32(0);
 
 
-    for(i=0;i<16;i++) {
+    for(i=0;i<h;i++) {
 	/* Read potentially unaligned pixels into t1 and t2 */
         perm1 = vec_lvsl(0, pix1);
         pix1v = (vector unsigned char *) pix1;
@@ -295,7 +295,7 @@ int pix_abs16x16_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
     return s;
 }
 
-int pix_abs8x8_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
+int sad8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
@@ -309,7 +309,7 @@ int pix_abs8x8_altivec(uint8_t *pix1, uint8_t *pix2, int line_size)
 
     permclear = (vector unsigned char)AVV(255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0);
 
-    for(i=0;i<8;i++) {
+    for(i=0;i<h;i++) {
 	/* Read potentially unaligned pixels into t1 and t2
 	   Since we're reading 16 pixels, and actually only want 8,
 	   mask out the last 8 pixels. The 0s don't change the sum. */
@@ -374,9 +374,9 @@ int pix_norm1_altivec(uint8_t *pix, int line_size)
 /**
  * Sum of Squared Errors for a 8x8 block.
  * AltiVec-enhanced.
- * It's the pix_abs8x8_altivec code above w/ squaring added.
+ * It's the sad8_altivec code above w/ squaring added.
  */
-int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size)
+int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
@@ -391,7 +391,7 @@ int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size)
     permclear = (vector unsigned char)AVV(255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0);
 
     
-    for(i=0;i<8;i++) {
+    for(i=0;i<h;i++) {
 	/* Read potentially unaligned pixels into t1 and t2
 	   Since we're reading 16 pixels, and actually only want 8,
 	   mask out the last 8 pixels. The 0s don't change the sum. */
@@ -430,9 +430,9 @@ int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size)
 /**
  * Sum of Squared Errors for a 16x16 block.
  * AltiVec-enhanced.
- * It's the pix_abs16x16_altivec code above w/ squaring added.
+ * It's the sad16_altivec code above w/ squaring added.
  */
-int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size)
+int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
@@ -444,7 +444,7 @@ int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size)
     
     sum = (vector unsigned int)vec_splat_u32(0);
     
-    for(i=0;i<16;i++) {
+    for(i=0;i<h;i++) {
 	/* Read potentially unaligned pixels into t1 and t2 */
         perm1 = vec_lvsl(0, pix1);
         pix1v = (vector unsigned char *) pix1;
@@ -609,14 +609,6 @@ void diff_pixels_altivec(DCTELEM *restrict block, const uint8_t *s1,
     }
 }
 
-int sad16x16_altivec(void *s, uint8_t *a, uint8_t *b, int stride) {
-  return pix_abs16x16_altivec(a,b,stride);
-}
-
-int sad8x8_altivec(void *s, uint8_t *a, uint8_t *b, int stride) {
-  return pix_abs8x8_altivec(a,b,stride);
-}
-
 void add_bytes_altivec(uint8_t *dst, uint8_t *src, int w) {
 #ifdef ALTIVEC_USE_REFERENCE_C_CODE
     int i;
diff --git a/src/libffmpeg/libavcodec/ppc/dsputil_altivec.h b/src/libffmpeg/libavcodec/ppc/dsputil_altivec.h
index f04496d62..93448a1ad 100644
--- a/src/libffmpeg/libavcodec/ppc/dsputil_altivec.h
+++ b/src/libffmpeg/libavcodec/ppc/dsputil_altivec.h
@@ -24,16 +24,14 @@
 
 #ifdef HAVE_ALTIVEC
 
-extern int pix_abs16x16_x2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size);
-extern int pix_abs16x16_y2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size);
-extern int pix_abs16x16_xy2_altivec(uint8_t *pix1, uint8_t *pix2, int line_size);
-extern int pix_abs16x16_altivec(uint8_t *pix1, uint8_t *pix2, int line_size);
-extern int pix_abs8x8_altivec(uint8_t *pix1, uint8_t *pix2, int line_size);
-extern int sad16x16_altivec(void *s, uint8_t *a, uint8_t *b, int stride);
-extern int sad8x8_altivec(void *s, uint8_t *a, uint8_t *b, int stride);
+extern int sad16_x2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
+extern int sad16_y2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
+extern int sad16_xy2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
+extern int sad16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
+extern int sad8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
 extern int pix_norm1_altivec(uint8_t *pix, int line_size);
-extern int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size);
-extern int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size);
+extern int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
+extern int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
 extern int pix_sum_altivec(uint8_t * pix, int line_size);
 extern void diff_pixels_altivec(DCTELEM* block, const uint8_t* s1, const uint8_t* s2, int stride);
 extern void get_pixels_altivec(DCTELEM* block, const uint8_t * pixels, int line_size);
diff --git a/src/libffmpeg/libavcodec/ppc/dsputil_ppc.c b/src/libffmpeg/libavcodec/ppc/dsputil_ppc.c
index 7af2aa002..9882e401f 100644
--- a/src/libffmpeg/libavcodec/ppc/dsputil_ppc.c
+++ b/src/libffmpeg/libavcodec/ppc/dsputil_ppc.c
@@ -25,6 +25,7 @@
 #include "dsputil_altivec.h"
 #endif
 
+extern void fdct_altivec(int16_t *block);
 extern void idct_put_altivec(uint8_t *dest, int line_size, int16_t *block);
 extern void idct_add_altivec(uint8_t *dest, int line_size, int16_t *block);
 
@@ -48,6 +49,7 @@ static unsigned char* perfname[] = {
   "fft_calc_altivec",
   "gmc1_altivec",
   "dct_unquantize_h263_altivec",
+  "fdct_altivec",
   "idct_add_altivec",
   "idct_put_altivec",
   "put_pixels16_altivec",
@@ -238,13 +240,13 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx)
         mm_flags |= MM_ALTIVEC;
         
         // Altivec specific optimisations
-        c->pix_abs16x16_x2 = pix_abs16x16_x2_altivec;
-        c->pix_abs16x16_y2 = pix_abs16x16_y2_altivec;
-        c->pix_abs16x16_xy2 = pix_abs16x16_xy2_altivec;
-        c->pix_abs16x16 = pix_abs16x16_altivec;
-        c->pix_abs8x8 = pix_abs8x8_altivec;
-        c->sad[0]= sad16x16_altivec;
-        c->sad[1]= sad8x8_altivec;
+        c->pix_abs[0][1] = sad16_x2_altivec;
+        c->pix_abs[0][2] = sad16_y2_altivec;
+        c->pix_abs[0][3] = sad16_xy2_altivec;
+        c->pix_abs[0][0] = sad16_altivec;
+        c->pix_abs[1][0] = sad8_altivec;
+        c->sad[0]= sad16_altivec;
+        c->sad[1]= sad8_altivec;
         c->pix_norm1 = pix_norm1_altivec;
         c->sse[1]= sse8_altivec;
         c->sse[0]= sse16_altivec;
@@ -270,6 +272,14 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx)
         
 	c->gmc1 = gmc1_altivec;
 
+#ifdef CONFIG_ENCODERS
+	if (avctx->dct_algo == FF_DCT_AUTO ||
+	    avctx->dct_algo == FF_DCT_ALTIVEC)
+	{
+	    c->fdct = fdct_altivec;
+	}
+#endif //CONFIG_ENCODERS
+
         if ((avctx->idct_algo == FF_IDCT_AUTO) ||
                 (avctx->idct_algo == FF_IDCT_ALTIVEC))
         {
diff --git a/src/libffmpeg/libavcodec/ppc/dsputil_ppc.h b/src/libffmpeg/libavcodec/ppc/dsputil_ppc.h
index 4cb299dd9..d672edfcb 100644
--- a/src/libffmpeg/libavcodec/ppc/dsputil_ppc.h
+++ b/src/libffmpeg/libavcodec/ppc/dsputil_ppc.h
@@ -40,6 +40,7 @@ enum powerpc_perf_index {
   altivec_fft_num = 0,
   altivec_gmc1_num,
   altivec_dct_unquantize_h263_num,
+  altivec_fdct,
   altivec_idct_add_num,
   altivec_idct_put_num,
   altivec_put_pixels16_num,
diff --git a/src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c b/src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c
index 18e86dce9..a83a0aa5c 100644
--- a/src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c
+++ b/src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c
@@ -72,7 +72,8 @@ void MPV_common_init_ppc(MpegEncContext *s)
                 (s->avctx->dct_algo == FF_DCT_ALTIVEC))
         {
             s->dct_quantize = dct_quantize_altivec;
-            s->dct_unquantize_h263 = dct_unquantize_h263_altivec;
+            s->dct_unquantize_h263_intra = dct_unquantize_h263_altivec;
+            s->dct_unquantize_h263_inter = dct_unquantize_h263_altivec;
         }
     } else
 #endif
-- 
cgit v1.2.3