Commit 60760f71 authored by James Zern's avatar James Zern

fix vp9_satd_sse2

accumulate satd in 32-bits
+ add unit test

Change-Id: I6748183df3662ddb9d635f9641f9586f2fd38ad5
parent 3e0138ed
......@@ -356,6 +356,14 @@ INSTANTIATE_TEST_CASE_P(
make_tuple(16, &vp9_int_pro_col_sse2, &vp9_int_pro_col_c),
make_tuple(32, &vp9_int_pro_col_sse2, &vp9_int_pro_col_c),
make_tuple(64, &vp9_int_pro_col_sse2, &vp9_int_pro_col_c)));
INSTANTIATE_TEST_CASE_P(
SSE2, SatdTest,
::testing::Values(
make_tuple(16, &vp9_satd_sse2),
make_tuple(64, &vp9_satd_sse2),
make_tuple(256, &vp9_satd_sse2),
make_tuple(1024, &vp9_satd_sse2)));
#endif
#if HAVE_NEON
......
......@@ -210,7 +210,7 @@ add_proto qw/void vp9_hadamard_16x16/, "int16_t const *src_diff, int src_stride,
specialize qw/vp9_hadamard_16x16 sse2/;
add_proto qw/int vp9_satd/, "const int16_t *coeff, int length";
specialize qw/vp9_satd/;
specialize qw/vp9_satd sse2/;
add_proto qw/void vp9_int_pro_row/, "int16_t *hbuf, uint8_t const *ref, const int ref_stride, const int height";
specialize qw/vp9_int_pro_row sse2 neon/;
......
......@@ -283,35 +283,31 @@ void vp9_hadamard_16x16_sse2(int16_t const *src_diff, int src_stride,
}
}
// 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);
__m128i val = _mm_xor_si128(sum, sign);
sum = _mm_sub_epi16(val, sign);
coeff += 8;
for (i = 8; i < length; i += 8) {
__m128i src_line = _mm_load_si128((const __m128i *)coeff);
sign = _mm_srai_epi16(src_line, 15);
val = _mm_xor_si128(src_line, sign);
val = _mm_sub_epi16(val, sign);
sum = _mm_add_epi16(sum, val);
const __m128i zero = _mm_setzero_si128();
__m128i accum = zero;
for (i = 0; i < length; i += 8) {
const __m128i src_line = _mm_load_si128((const __m128i *)coeff);
const __m128i inv = _mm_sub_epi16(zero, src_line);
const __m128i abs = _mm_max_epi16(src_line, inv); // abs(src_line)
const __m128i abs_lo = _mm_unpacklo_epi16(abs, zero);
const __m128i abs_hi = _mm_unpackhi_epi16(abs, zero);
const __m128i sum = _mm_add_epi32(abs_lo, abs_hi);
accum = _mm_add_epi32(accum, sum);
coeff += 8;
}
val = _mm_srli_si128(sum, 8);
sum = _mm_add_epi16(sum, val);
val = _mm_srli_epi64(sum, 32);
sum = _mm_add_epi16(sum, val);
val = _mm_srli_epi32(sum, 16);
sum = _mm_add_epi16(sum, val);
{ // cascading summation of accum
__m128i hi = _mm_srli_si128(accum, 8);
accum = _mm_add_epi32(accum, hi);
hi = _mm_srli_epi64(accum, 32);
accum = _mm_add_epi32(accum, hi);
}
return _mm_extract_epi16(sum, 0);
return _mm_cvtsi128_si32(accum);
}
#endif
void vp9_int_pro_row_sse2(int16_t *hbuf, uint8_t const*ref,
const int ref_stride, const int height) {
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment