diff options
| author | Francois-Olivier Devaux <fodevaux@users.noreply.github.com> | 2010-04-08 17:22:58 +0000 |
|---|---|---|
| committer | Francois-Olivier Devaux <fodevaux@users.noreply.github.com> | 2010-04-08 17:22:58 +0000 |
| commit | 627f484bce7b572dfcbd13f90f5f6fb083d1008c (patch) | |
| tree | 44e55f6d330d2b0c522f57cae8afcde8d9d33b4c /libopenjpeg/mct.c | |
| parent | d03779ee2a035509737ec452a2f2c75ef4fef032 (diff) | |
Significant optimizations of MCT, DWT, MQ and T1 modules by Peter Wimmer (thanks Peter)
Diffstat (limited to 'libopenjpeg/mct.c')
| -rw-r--r-- | libopenjpeg/mct.c | 42 |
1 files changed, 42 insertions, 0 deletions
diff --git a/libopenjpeg/mct.c b/libopenjpeg/mct.c index ca21744f..870993b0 100644 --- a/libopenjpeg/mct.c +++ b/libopenjpeg/mct.c @@ -29,6 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ +#ifdef __SSE__ +#include <xmmintrin.h> +#endif + #include "opj_includes.h" /* <summary> */ @@ -127,6 +131,44 @@ void mct_decode_real( int n) { int i; +#ifdef __SSE__ + __m128 vrv, vgu, vgv, vbu; + vrv = _mm_set1_ps(1.402f); + vgu = _mm_set1_ps(0.34413f); + vgv = _mm_set1_ps(0.71414f); + vbu = _mm_set1_ps(1.772f); + for (i = 0; i < (n >> 3); ++i) { + __m128 vy, vu, vv; + __m128 vr, vg, vb; + + vy = _mm_load_ps(c0); + vu = _mm_load_ps(c1); + vv = _mm_load_ps(c2); + vr = _mm_add_ps(vy, _mm_mul_ps(vv, vrv)); + vg = _mm_sub_ps(_mm_sub_ps(vy, _mm_mul_ps(vu, vgu)), _mm_mul_ps(vv, vgv)); + vb = _mm_add_ps(vy, _mm_mul_ps(vu, vbu)); + _mm_store_ps(c0, vr); + _mm_store_ps(c1, vg); + _mm_store_ps(c2, vb); + c0 += 4; + c1 += 4; + c2 += 4; + + vy = _mm_load_ps(c0); + vu = _mm_load_ps(c1); + vv = _mm_load_ps(c2); + vr = _mm_add_ps(vy, _mm_mul_ps(vv, vrv)); + vg = _mm_sub_ps(_mm_sub_ps(vy, _mm_mul_ps(vu, vgu)), _mm_mul_ps(vv, vgv)); + vb = _mm_add_ps(vy, _mm_mul_ps(vu, vbu)); + _mm_store_ps(c0, vr); + _mm_store_ps(c1, vg); + _mm_store_ps(c2, vb); + c0 += 4; + c1 += 4; + c2 += 4; + } + n &= 7; +#endif for(i = 0; i < n; ++i) { float y = c0[i]; float u = c1[i]; |