libavcodec/ppc/mpegvideoencdsp.c : fix pix_norm1_altivec() and pix_sum_altivec() for POWER LE

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

1 files changed, 60 insertions, 0 deletions

diff --git a/libavcodec/ppc/mpegvideoencdsp.c b/libavcodec/ppc/mpegvideoencdsp.c
index 00ae2a6f30..e91ba5d25f 100644
--- a/libavcodec/ppc/mpegvideoencdsp.c
+++ b/libavcodec/ppc/mpegvideoencdsp.c
@@ -31,6 +31,34 @@
 
 #if HAVE_ALTIVEC
 
+#if HAVE_VSX
+static int pix_norm1_altivec(uint8_t *pix, int line_size)
+{
+    int i, s = 0;
+    const vector unsigned int zero =
+        (const vector unsigned int) vec_splat_u32(0);
+    vector unsigned int sv = (vector unsigned int) vec_splat_u32(0);
+    vector signed int sum;
+
+    for (i = 0; i < 16; i++) {
+        /* Read the potentially unaligned pixels. */
+        //vector unsigned char pixl = vec_ld(0,  pix);
+        //vector unsigned char pixr = vec_ld(15, pix);
+        //vector unsigned char pixv = vec_perm(pixl, pixr, perm);
+        vector unsigned char pixv = vec_vsx_ld(0,  pix);
+
+        /* Square the values, and add them to our sum. */
+        sv = vec_msum(pixv, pixv, sv);
+
+        pix += line_size;
+    }
+    /* Sum up the four partial sums, and put the result into s. */
+    sum = vec_sums((vector signed int) sv, (vector signed int) zero);
+    sum = vec_splat(sum, 3);
+    vec_vsx_st(sum, 0, &s);
+    return s;
+}
+#else
 static int pix_norm1_altivec(uint8_t *pix, int line_size)
 {
     int i, s = 0;
@@ -58,7 +86,37 @@ static int pix_norm1_altivec(uint8_t *pix, int line_size)
 
     return s;
 }
+#endif /* HAVE_VSX */
+
+#if HAVE_VSX
+static int pix_sum_altivec(uint8_t *pix, int line_size)
+{
+    int i, s;
+    const vector unsigned int zero =
+        (const vector unsigned int) vec_splat_u32(0);
+    vector unsigned int sad = (vector unsigned int) vec_splat_u32(0);
+    vector signed int sumdiffs;
+
+    for (i = 0; i < 16; i++) {
+        /* Read the potentially unaligned 16 pixels into t1. */
+        //vector unsigned char pixl = vec_ld(0,  pix);
+        //vector unsigned char pixr = vec_ld(15, pix);
+        //vector unsigned char t1   = vec_perm(pixl, pixr, perm);
+        vector unsigned char t1   = vec_vsx_ld(0,  pix);
+
+        /* Add each 4 pixel group together and put 4 results into sad. */
+        sad = vec_sum4s(t1, sad);
+
+        pix += line_size;
+    }
 
+    /* Sum up the four partial sums, and put the result into s. */
+    sumdiffs = vec_sums((vector signed int) sad, (vector signed int) zero);
+    sumdiffs = vec_splat(sumdiffs, 3);
+    vec_vsx_st(sumdiffs, 0, &s);
+    return s;
+}
+#else
 static int pix_sum_altivec(uint8_t *pix, int line_size)
 {
     int i, s;
@@ -88,6 +146,8 @@ static int pix_sum_altivec(uint8_t *pix, int line_size)
     return s;
 }
 
+#endif /* HAVE_VSX */
+
 #endif /* HAVE_ALTIVEC */
 
 av_cold void ff_mpegvideoencdsp_init_ppc(MpegvideoEncDSPContext *c,