vp9_satd: return an int

the final sum may use up to 26 bits

+ add a unit test
+ disable the sse2 as the result will rollover; this will be fixed in a
future commit

Change-Id: I2a49811dfaa06abfd9fa1e1e65ed7cd68e4c97ce

3 files changed, 7 insertions, 4 deletions

diff --git a/vp9/encoder/vp9_avg.c b/vp9/encoder/vp9_avg.c
index a9a4c3050..7baa09ae5 100644
--- a/vp9/encoder/vp9_avg.c
+++ b/vp9/encoder/vp9_avg.c
@@ -117,14 +117,14 @@ void vp9_hadamard_16x16_c(int16_t const *src_diff, int src_stride,
 
 // coeff: 16 bits, dynamic range [-32640, 32640].
 // length: value range {16, 64, 256, 1024}.
-int16_t vp9_satd_c(const int16_t *coeff, int length) {
+int vp9_satd_c(const int16_t *coeff, int length) {
   int i;
   int satd = 0;
   for (i = 0; i < length; ++i)
     satd += abs(coeff[i]);
 
   // satd: 26 bits, dynamic range [-32640 * 1024, 32640 * 1024]
-  return (int16_t)satd;
+  return satd;
 }
 
 // Integer projection onto row vectors.
diff --git a/vp9/encoder/vp9_pickmode.c b/vp9/encoder/vp9_pickmode.c
index 9db044f6b..266c15f4e 100644
--- a/vp9/encoder/vp9_pickmode.c
+++ b/vp9/encoder/vp9_pickmode.c
@@ -673,7 +673,7 @@ static void block_yrd(VP9_COMP *cpi, MACROBLOCK *x, int *rate, int64_t *dist,
         if (*eob == 1)
           *rate += (int)abs(qcoeff[0]);
         else if (*eob > 1)
-          *rate += (int)vp9_satd((const int16_t *)qcoeff, step << 4);
+          *rate += vp9_satd((const int16_t *)qcoeff, step << 4);
 
         *dist += vp9_block_error_fp(coeff, dqcoeff, step << 4) >> shift;
       }
diff --git a/vp9/encoder/x86/vp9_avg_intrin_sse2.c b/vp9/encoder/x86/vp9_avg_intrin_sse2.c
index 4531d794a..489bdfe55 100644
--- a/vp9/encoder/x86/vp9_avg_intrin_sse2.c
+++ b/vp9/encoder/x86/vp9_avg_intrin_sse2.c
@@ -283,7 +283,9 @@ void vp9_hadamard_16x16_sse2(int16_t const *src_diff, int src_stride,
   }
 }
 
-int16_t vp9_satd_sse2(const int16_t *coeff, int length) {
+// TODO(jingning): the sum needs to be accumulated in 32-bits to avoid rollover.
+#if 0
+int vp9_satd_sse2(const int16_t *coeff, int length) {
   int i;
   __m128i sum = _mm_load_si128((const __m128i *)coeff);
   __m128i sign = _mm_srai_epi16(sum, 15);
@@ -309,6 +311,7 @@ int16_t vp9_satd_sse2(const int16_t *coeff, int length) {
 
   return _mm_extract_epi16(sum, 0);
 }
+#endif
 
 void vp9_int_pro_row_sse2(int16_t *hbuf, uint8_t const*ref,
                           const int ref_stride, const int height) {