vp8: change build_intra4x4_predictors() to use vpx_dsp.

I've added a few new functions (d45e, d63e, he, ve) to cover the
filtered h/v 4x4 predictors that are vp8-specific, the "correct"
d45 with the correctly filtered bottom-right pixel (as opposed to
the unfiltered version in vp9), and the "broken" d63 with weirdly
filtered bottom-right pixels (which is correctly filtered in vp9).

There may be a minor performance impact on all systems because we
have to do an extra copy of the Above pixel array to incorporate
the topleft pixel in the same array (thus fitting the vpx_dsp API).
In addition, armv6 will have a more serious performance impact b/c
I removed the armv6/vp8-specific assembly. I'm not sure anyone
cares...

Change-Id: I7f9e5ebee11d8e21aca2cd517a69eefc181b2e86

2 files changed, 89 insertions, 0 deletions

diff --git a/vpx_dsp/intrapred.c b/vpx_dsp/intrapred.c
index 9ba0f6444..aafd8bfb7 100644
--- a/vpx_dsp/intrapred.c
+++ b/vpx_dsp/intrapred.c
@@ -247,6 +247,38 @@ static INLINE void dc_predictor(uint8_t *dst, ptrdiff_t stride, int bs,
   }
 }
 
+void vpx_he_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
+                            const uint8_t *above, const uint8_t *left) {
+  const int H = above[-1];
+  const int I = left[0];
+  const int J = left[1];
+  const int K = left[2];
+  const int L = left[3];
+
+  memset(dst + stride * 0, AVG3(H, I, J), 4);
+  memset(dst + stride * 1, AVG3(I, J, K), 4);
+  memset(dst + stride * 2, AVG3(J, K, L), 4);
+  memset(dst + stride * 3, AVG3(K, L, L), 4);
+}
+
+void vpx_ve_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
+                            const uint8_t *above, const uint8_t *left) {
+  const int H = above[-1];
+  const int I = above[0];
+  const int J = above[1];
+  const int K = above[2];
+  const int L = above[3];
+  const int M = above[4];
+
+  dst[0] = AVG3(H, I, J);
+  dst[1] = AVG3(I, J, K);
+  dst[2] = AVG3(J, K, L);
+  dst[3] = AVG3(K, L, M);
+  memcpy(dst + stride * 1, dst, 4);
+  memcpy(dst + stride * 2, dst, 4);
+  memcpy(dst + stride * 3, dst, 4);
+}
+
 void vpx_d207_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
                               const uint8_t *above, const uint8_t *left) {
   const int I = left[0];
@@ -287,6 +319,30 @@ void vpx_d63_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
               DST(3, 3) = AVG3(E, F, G);  // differs from vp8
 }
 
+void vpx_d63e_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
+                              const uint8_t *above, const uint8_t *left) {
+  const int A = above[0];
+  const int B = above[1];
+  const int C = above[2];
+  const int D = above[3];
+  const int E = above[4];
+  const int F = above[5];
+  const int G = above[6];
+  const int H = above[7];
+  (void)left;
+  DST(0, 0) =             AVG2(A, B);
+  DST(1, 0) = DST(0, 2) = AVG2(B, C);
+  DST(2, 0) = DST(1, 2) = AVG2(C, D);
+  DST(3, 0) = DST(2, 2) = AVG2(D, E);
+              DST(3, 2) = AVG3(E, F, G);
+
+  DST(0, 1) =             AVG3(A, B, C);
+  DST(1, 1) = DST(0, 3) = AVG3(B, C, D);
+  DST(2, 1) = DST(1, 3) = AVG3(C, D, E);
+  DST(3, 1) = DST(2, 3) = AVG3(D, E, F);
+              DST(3, 3) = AVG3(F, G, H);
+}
+
 void vpx_d45_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
                              const uint8_t *above, const uint8_t *left) {
   const int A = above[0];
@@ -308,6 +364,27 @@ void vpx_d45_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
                                       DST(3, 3) = H;  // differs from vp8
 }
 
+void vpx_d45e_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
+                              const uint8_t *above, const uint8_t *left) {
+  const int A = above[0];
+  const int B = above[1];
+  const int C = above[2];
+  const int D = above[3];
+  const int E = above[4];
+  const int F = above[5];
+  const int G = above[6];
+  const int H = above[7];
+  (void)stride;
+  (void)left;
+  DST(0, 0)                                     = AVG3(A, B, C);
+  DST(1, 0) = DST(0, 1)                         = AVG3(B, C, D);
+  DST(2, 0) = DST(1, 1) = DST(0, 2)             = AVG3(C, D, E);
+  DST(3, 0) = DST(2, 1) = DST(1, 2) = DST(0, 3) = AVG3(D, E, F);
+              DST(3, 1) = DST(2, 2) = DST(1, 3) = AVG3(E, F, G);
+                          DST(3, 2) = DST(2, 3) = AVG3(F, G, H);
+                                      DST(3, 3) = AVG3(G, H, H);
+}
+
 void vpx_d117_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride,
                               const uint8_t *above, const uint8_t *left) {
   const int I = left[0];
diff --git a/vpx_dsp/vpx_dsp_rtcd_defs.pl b/vpx_dsp/vpx_dsp_rtcd_defs.pl
index be7d5c2c6..352e1ff56 100644
--- a/vpx_dsp/vpx_dsp_rtcd_defs.pl
+++ b/vpx_dsp/vpx_dsp_rtcd_defs.pl
@@ -60,12 +60,21 @@ specialize qw/vpx_d207_predictor_4x4/, "$ssse3_x86inc";
 add_proto qw/void vpx_d45_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d45_predictor_4x4 neon/, "$ssse3_x86inc";
 
+add_proto qw/void vpx_d45e_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
+specialize qw/vpx_d45e_predictor_4x4/;
+
 add_proto qw/void vpx_d63_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d63_predictor_4x4/, "$ssse3_x86inc";
 
+add_proto qw/void vpx_d63e_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
+specialize qw/vpx_d63e_predictor_4x4/;
+
 add_proto qw/void vpx_h_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_h_predictor_4x4 neon dspr2 msa/, "$ssse3_x86inc";
 
+add_proto qw/void vpx_he_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
+specialize qw/vpx_he_predictor_4x4/;
+
 add_proto qw/void vpx_d117_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d117_predictor_4x4/;
 
@@ -78,6 +87,9 @@ specialize qw/vpx_d153_predictor_4x4/, "$ssse3_x86inc";
 add_proto qw/void vpx_v_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_v_predictor_4x4 neon msa/, "$sse_x86inc";
 
+add_proto qw/void vpx_ve_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
+specialize qw/vpx_ve_predictor_4x4/;
+
 add_proto qw/void vpx_tm_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_tm_predictor_4x4 neon dspr2 msa/, "$sse_x86inc";