1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/gfx/ycbcr/yuv_convert_sse2.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,47 @@
1.4 +// Copyright (c) 2010 The Chromium Authors. All rights reserved.
1.5 +// Use of this source code is governed by a BSD-style license that can be
1.6 +// found in the LICENSE file.
1.7 +
1.8 +#include <emmintrin.h>
1.9 +#include "yuv_row.h"
1.10 +
1.11 +namespace mozilla {
1.12 +namespace gfx {
1.13 +
1.14 +// FilterRows combines two rows of the image using linear interpolation.
1.15 +// SSE2 version does 16 pixels at a time.
1.16 +void FilterRows_SSE2(uint8* ybuf, const uint8* y0_ptr, const uint8* y1_ptr,
1.17 + int source_width, int source_y_fraction) {
1.18 + __m128i zero = _mm_setzero_si128();
1.19 + __m128i y1_fraction = _mm_set1_epi16(source_y_fraction);
1.20 + __m128i y0_fraction = _mm_set1_epi16(256 - source_y_fraction);
1.21 +
1.22 + const __m128i* y0_ptr128 = reinterpret_cast<const __m128i*>(y0_ptr);
1.23 + const __m128i* y1_ptr128 = reinterpret_cast<const __m128i*>(y1_ptr);
1.24 + __m128i* dest128 = reinterpret_cast<__m128i*>(ybuf);
1.25 + __m128i* end128 = reinterpret_cast<__m128i*>(ybuf + source_width);
1.26 +
1.27 + do {
1.28 + __m128i y0 = _mm_loadu_si128(y0_ptr128);
1.29 + __m128i y1 = _mm_loadu_si128(y1_ptr128);
1.30 + __m128i y2 = _mm_unpackhi_epi8(y0, zero);
1.31 + __m128i y3 = _mm_unpackhi_epi8(y1, zero);
1.32 + y0 = _mm_unpacklo_epi8(y0, zero);
1.33 + y1 = _mm_unpacklo_epi8(y1, zero);
1.34 + y0 = _mm_mullo_epi16(y0, y0_fraction);
1.35 + y1 = _mm_mullo_epi16(y1, y1_fraction);
1.36 + y2 = _mm_mullo_epi16(y2, y0_fraction);
1.37 + y3 = _mm_mullo_epi16(y3, y1_fraction);
1.38 + y0 = _mm_add_epi16(y0, y1);
1.39 + y2 = _mm_add_epi16(y2, y3);
1.40 + y0 = _mm_srli_epi16(y0, 8);
1.41 + y2 = _mm_srli_epi16(y2, 8);
1.42 + y0 = _mm_packus_epi16(y0, y2);
1.43 + *dest128++ = y0;
1.44 + ++y0_ptr128;
1.45 + ++y1_ptr128;
1.46 + } while (dest128 < end128);
1.47 +}
1.48 +
1.49 +}
1.50 +}
