diff options
author | Mans Rullgard <mans@mansr.com> | 2013-01-23 20:24:53 -0800 |
---|---|---|
committer | Luca Barbato <lu_zero@gentoo.org> | 2013-01-24 10:44:43 +0100 |
commit | e9d817351b28e62868528476971afe4cde1fa280 (patch) | |
tree | f17f5c43d1ce5856a275dd4b40a970d15b7a9eac /libavcodec/sh4 | |
parent | 7a95afe433b2a692f490b98948c082e62ffc1d27 (diff) | |
download | ffmpeg-e9d817351b28e62868528476971afe4cde1fa280.tar.gz |
dsputil: Separate h264 qpel
The sh4 optimizations are removed, because the code is
100% identical to the C code, so it is unlikely to
provide any real practical benefit.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Diffstat (limited to 'libavcodec/sh4')
-rw-r--r-- | libavcodec/sh4/dsputil_align.c | 9 | ||||
-rw-r--r-- | libavcodec/sh4/qpel.c | 401 |
2 files changed, 0 insertions, 410 deletions
diff --git a/libavcodec/sh4/dsputil_align.c b/libavcodec/sh4/dsputil_align.c index 9058e4f4ba..f723f76743 100644 --- a/libavcodec/sh4/dsputil_align.c +++ b/libavcodec/sh4/dsputil_align.c @@ -396,15 +396,6 @@ void ff_dsputil_init_align(DSPContext* c, AVCodecContext *avctx) dspfunc(avg_qpel, 1, 8); /* dspfunc(avg_no_rnd_qpel, 1, 8); */ - if (!high_bit_depth) { - dspfunc(put_h264_qpel, 0, 16); - dspfunc(put_h264_qpel, 1, 8); - dspfunc(put_h264_qpel, 2, 4); - dspfunc(avg_h264_qpel, 0, 16); - dspfunc(avg_h264_qpel, 1, 8); - dspfunc(avg_h264_qpel, 2, 4); - } - #undef dspfunc if (!high_bit_depth) { c->put_h264_chroma_pixels_tab[0]= put_h264_chroma_mc8_sh4; diff --git a/libavcodec/sh4/qpel.c b/libavcodec/sh4/qpel.c index 20540f7b93..055d184580 100644 --- a/libavcodec/sh4/qpel.c +++ b/libavcodec/sh4/qpel.c @@ -842,407 +842,6 @@ QPEL_MC(0, avg_ , _ , op_avg) #undef op_put #undef op_put_no_rnd -#define H264_LOWPASS(OPNAME, OP, OP2) \ -static inline void OPNAME ## h264_qpel_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,int w,int h){\ - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\ - do {\ - int srcB,srcA,src0,src1,src2,src3,src4,src5,src6;\ - uint8_t *s = src-2;\ - srcB = *s++;\ - srcA = *s++;\ - src0 = *s++;\ - src1 = *s++;\ - src2 = *s++;\ - src3 = *s++;\ - OP(dst[0], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\ - src4 = *s++;\ - OP(dst[1], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\ - src5 = *s++;\ - OP(dst[2], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\ - src6 = *s++;\ - OP(dst[3], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\ - if (w>4) { /* it optimized */ \ - int src7,src8,src9,src10; \ - src7 = *s++;\ - OP(dst[4], (src4+src5)*20 - (src3+src6)*5 + (src2+src7));\ - src8 = *s++;\ - OP(dst[5], (src5+src6)*20 - (src4+src7)*5 + (src3+src8));\ - src9 = *s++;\ - OP(dst[6], (src6+src7)*20 - (src5+src8)*5 + (src4+src9));\ - src10 = *s++;\ - OP(dst[7], (src7+src8)*20 - (src6+src9)*5 + (src5+src10));\ - if (w>8) { \ - int src11,src12,src13,src14,src15,src16,src17,src18; \ - src11 = *s++;\ - OP(dst[8] , (src8 +src9 )*20 - (src7 +src10)*5 + (src6 +src11));\ - src12 = *s++;\ - OP(dst[9] , (src9 +src10)*20 - (src8 +src11)*5 + (src7 +src12));\ - src13 = *s++;\ - OP(dst[10], (src10+src11)*20 - (src9 +src12)*5 + (src8 +src13));\ - src14 = *s++;\ - OP(dst[11], (src11+src12)*20 - (src10+src13)*5 + (src9 +src14));\ - src15 = *s++;\ - OP(dst[12], (src12+src13)*20 - (src11+src14)*5 + (src10+src15));\ - src16 = *s++;\ - OP(dst[13], (src13+src14)*20 - (src12+src15)*5 + (src11+src16));\ - src17 = *s++;\ - OP(dst[14], (src14+src15)*20 - (src13+src16)*5 + (src12+src17));\ - src18 = *s++;\ - OP(dst[15], (src15+src16)*20 - (src14+src17)*5 + (src13+src18));\ - } \ - } \ - dst+=dstStride;\ - src+=srcStride;\ - }while(--h);\ -}\ -\ -static inline void OPNAME ## h264_qpel_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,int w,int h){\ - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\ - do{\ - int srcB,srcA,src0,src1,src2,src3,src4,src5,src6;\ - uint8_t *s = src-2*srcStride,*d=dst;\ - srcB = *s; s+=srcStride;\ - srcA = *s; s+=srcStride;\ - src0 = *s; s+=srcStride;\ - src1 = *s; s+=srcStride;\ - src2 = *s; s+=srcStride;\ - src3 = *s; s+=srcStride;\ - OP(*d, (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));d+=dstStride;\ - src4 = *s; s+=srcStride;\ - OP(*d, (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));d+=dstStride;\ - src5 = *s; s+=srcStride;\ - OP(*d, (src2+src3)*20 - (src1+src4)*5 + (src0+src5));d+=dstStride;\ - src6 = *s; s+=srcStride;\ - OP(*d, (src3+src4)*20 - (src2+src5)*5 + (src1+src6));d+=dstStride;\ - if (h>4) { \ - int src7,src8,src9,src10; \ - src7 = *s; s+=srcStride;\ - OP(*d, (src4+src5)*20 - (src3+src6)*5 + (src2+src7));d+=dstStride;\ - src8 = *s; s+=srcStride;\ - OP(*d, (src5+src6)*20 - (src4+src7)*5 + (src3+src8));d+=dstStride;\ - src9 = *s; s+=srcStride;\ - OP(*d, (src6+src7)*20 - (src5+src8)*5 + (src4+src9));d+=dstStride;\ - src10 = *s; s+=srcStride;\ - OP(*d, (src7+src8)*20 - (src6+src9)*5 + (src5+src10));d+=dstStride;\ - if (h>8) { \ - int src11,src12,src13,src14,src15,src16,src17,src18; \ - src11 = *s; s+=srcStride;\ - OP(*d , (src8 +src9 )*20 - (src7 +src10)*5 + (src6 +src11));d+=dstStride;\ - src12 = *s; s+=srcStride;\ - OP(*d , (src9 +src10)*20 - (src8 +src11)*5 + (src7 +src12));d+=dstStride;\ - src13 = *s; s+=srcStride;\ - OP(*d, (src10+src11)*20 - (src9 +src12)*5 + (src8 +src13));d+=dstStride;\ - src14 = *s; s+=srcStride;\ - OP(*d, (src11+src12)*20 - (src10+src13)*5 + (src9 +src14));d+=dstStride;\ - src15 = *s; s+=srcStride;\ - OP(*d, (src12+src13)*20 - (src11+src14)*5 + (src10+src15));d+=dstStride;\ - src16 = *s; s+=srcStride;\ - OP(*d, (src13+src14)*20 - (src12+src15)*5 + (src11+src16));d+=dstStride;\ - src17 = *s; s+=srcStride;\ - OP(*d, (src14+src15)*20 - (src13+src16)*5 + (src12+src17));d+=dstStride;\ - src18 = *s; s+=srcStride;\ - OP(*d, (src15+src16)*20 - (src14+src17)*5 + (src13+src18));d+=dstStride;\ - } \ - } \ - dst++;\ - src++;\ - }while(--w);\ -}\ -\ -static inline void OPNAME ## h264_qpel_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride,int w,int h){\ - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\ - int i;\ - src -= 2*srcStride;\ - i= h+5; \ - do {\ - int srcB,srcA,src0,src1,src2,src3,src4,src5,src6;\ - uint8_t *s = src-2;\ - srcB = *s++;\ - srcA = *s++;\ - src0 = *s++;\ - src1 = *s++;\ - src2 = *s++;\ - src3 = *s++;\ - tmp[0] = ((src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\ - src4 = *s++;\ - tmp[1] = ((src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\ - src5 = *s++;\ - tmp[2] = ((src2+src3)*20 - (src1+src4)*5 + (src0+src5));\ - src6 = *s++;\ - tmp[3] = ((src3+src4)*20 - (src2+src5)*5 + (src1+src6));\ - if (w>4) { /* it optimized */ \ - int src7,src8,src9,src10; \ - src7 = *s++;\ - tmp[4] = ((src4+src5)*20 - (src3+src6)*5 + (src2+src7));\ - src8 = *s++;\ - tmp[5] = ((src5+src6)*20 - (src4+src7)*5 + (src3+src8));\ - src9 = *s++;\ - tmp[6] = ((src6+src7)*20 - (src5+src8)*5 + (src4+src9));\ - src10 = *s++;\ - tmp[7] = ((src7+src8)*20 - (src6+src9)*5 + (src5+src10));\ - if (w>8) { \ - int src11,src12,src13,src14,src15,src16,src17,src18; \ - src11 = *s++;\ - tmp[8] = ((src8 +src9 )*20 - (src7 +src10)*5 + (src6 +src11));\ - src12 = *s++;\ - tmp[9] = ((src9 +src10)*20 - (src8 +src11)*5 + (src7 +src12));\ - src13 = *s++;\ - tmp[10] = ((src10+src11)*20 - (src9 +src12)*5 + (src8 +src13));\ - src14 = *s++;\ - tmp[11] = ((src11+src12)*20 - (src10+src13)*5 + (src9 +src14));\ - src15 = *s++;\ - tmp[12] = ((src12+src13)*20 - (src11+src14)*5 + (src10+src15));\ - src16 = *s++;\ - tmp[13] = ((src13+src14)*20 - (src12+src15)*5 + (src11+src16));\ - src17 = *s++;\ - tmp[14] = ((src14+src15)*20 - (src13+src16)*5 + (src12+src17));\ - src18 = *s++;\ - tmp[15] = ((src15+src16)*20 - (src14+src17)*5 + (src13+src18));\ - } \ - } \ - tmp+=tmpStride;\ - src+=srcStride;\ - }while(--i);\ - tmp -= tmpStride*(h+5-2);\ - i = w; \ - do {\ - int tmpB,tmpA,tmp0,tmp1,tmp2,tmp3,tmp4,tmp5,tmp6;\ - int16_t *s = tmp-2*tmpStride; \ - uint8_t *d=dst;\ - tmpB = *s; s+=tmpStride;\ - tmpA = *s; s+=tmpStride;\ - tmp0 = *s; s+=tmpStride;\ - tmp1 = *s; s+=tmpStride;\ - tmp2 = *s; s+=tmpStride;\ - tmp3 = *s; s+=tmpStride;\ - OP2(*d, (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));d+=dstStride;\ - tmp4 = *s; s+=tmpStride;\ - OP2(*d, (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));d+=dstStride;\ - tmp5 = *s; s+=tmpStride;\ - OP2(*d, (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));d+=dstStride;\ - tmp6 = *s; s+=tmpStride;\ - OP2(*d, (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));d+=dstStride;\ - if (h>4) { \ - int tmp7,tmp8,tmp9,tmp10; \ - tmp7 = *s; s+=tmpStride;\ - OP2(*d, (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));d+=dstStride;\ - tmp8 = *s; s+=tmpStride;\ - OP2(*d, (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));d+=dstStride;\ - tmp9 = *s; s+=tmpStride;\ - OP2(*d, (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));d+=dstStride;\ - tmp10 = *s; s+=tmpStride;\ - OP2(*d, (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));d+=dstStride;\ - if (h>8) { \ - int tmp11,tmp12,tmp13,tmp14,tmp15,tmp16,tmp17,tmp18; \ - tmp11 = *s; s+=tmpStride;\ - OP2(*d , (tmp8 +tmp9 )*20 - (tmp7 +tmp10)*5 + (tmp6 +tmp11));d+=dstStride;\ - tmp12 = *s; s+=tmpStride;\ - OP2(*d , (tmp9 +tmp10)*20 - (tmp8 +tmp11)*5 + (tmp7 +tmp12));d+=dstStride;\ - tmp13 = *s; s+=tmpStride;\ - OP2(*d, (tmp10+tmp11)*20 - (tmp9 +tmp12)*5 + (tmp8 +tmp13));d+=dstStride;\ - tmp14 = *s; s+=tmpStride;\ - OP2(*d, (tmp11+tmp12)*20 - (tmp10+tmp13)*5 + (tmp9 +tmp14));d+=dstStride;\ - tmp15 = *s; s+=tmpStride;\ - OP2(*d, (tmp12+tmp13)*20 - (tmp11+tmp14)*5 + (tmp10+tmp15));d+=dstStride;\ - tmp16 = *s; s+=tmpStride;\ - OP2(*d, (tmp13+tmp14)*20 - (tmp12+tmp15)*5 + (tmp11+tmp16));d+=dstStride;\ - tmp17 = *s; s+=tmpStride;\ - OP2(*d, (tmp14+tmp15)*20 - (tmp13+tmp16)*5 + (tmp12+tmp17));d+=dstStride;\ - tmp18 = *s; s+=tmpStride;\ - OP2(*d, (tmp15+tmp16)*20 - (tmp14+tmp17)*5 + (tmp13+tmp18));d+=dstStride;\ - } \ - } \ - dst++;\ - tmp++;\ - }while(--i);\ -}\ -\ -static void OPNAME ## h264_qpel4_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ - OPNAME ## h264_qpel_h_lowpass(dst,src,dstStride,srcStride,4,4); \ -}\ -static void OPNAME ## h264_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ - OPNAME ## h264_qpel_h_lowpass(dst,src,dstStride,srcStride,8,8); \ -}\ -static void OPNAME ## h264_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ - OPNAME ## h264_qpel_h_lowpass(dst,src,dstStride,srcStride,16,16); \ -}\ -\ -static void OPNAME ## h264_qpel4_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ - OPNAME ## h264_qpel_v_lowpass(dst,src,dstStride,srcStride,4,4); \ -}\ -static void OPNAME ## h264_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ - OPNAME ## h264_qpel_v_lowpass(dst,src,dstStride,srcStride,8,8); \ -}\ -static void OPNAME ## h264_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ - OPNAME ## h264_qpel_v_lowpass(dst,src,dstStride,srcStride,16,16); \ -}\ -static void OPNAME ## h264_qpel4_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\ - OPNAME ## h264_qpel_hv_lowpass(dst,tmp,src,dstStride,tmpStride,srcStride,4,4); \ -}\ -static void OPNAME ## h264_qpel8_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\ - OPNAME ## h264_qpel_hv_lowpass(dst,tmp,src,dstStride,tmpStride,srcStride,8,8); \ -}\ -static void OPNAME ## h264_qpel16_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\ - OPNAME ## h264_qpel_hv_lowpass(dst,tmp,src,dstStride,tmpStride,srcStride,16,16); \ -}\ - -#define H264_MC(OPNAME, SIZE) \ -static void OPNAME ## h264_qpel ## SIZE ## _mc00_sh4 (uint8_t *dst, uint8_t *src, int stride){\ - OPNAME ## pixels ## SIZE ## _c(dst, src, stride, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc10_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t half[SIZE*SIZE];\ - put_h264_qpel ## SIZE ## _h_lowpass(half, src, SIZE, stride);\ - OPNAME ## pixels ## SIZE ## _l2_aligned2(dst, src, half, stride, stride, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc20_sh4(uint8_t *dst, uint8_t *src, int stride){\ - OPNAME ## h264_qpel ## SIZE ## _h_lowpass(dst, src, stride, stride);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc30_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t half[SIZE*SIZE];\ - put_h264_qpel ## SIZE ## _h_lowpass(half, src, SIZE, stride);\ - OPNAME ## pixels ## SIZE ## _l2_aligned2(dst, src+1, half, stride, stride, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc01_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - uint8_t half[SIZE*SIZE];\ - copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\ - put_h264_qpel ## SIZE ## _v_lowpass(half, full_mid, SIZE, SIZE);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, full_mid, half, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc02_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\ - OPNAME ## h264_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc03_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - uint8_t half[SIZE*SIZE];\ - copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\ - put_h264_qpel ## SIZE ## _v_lowpass(half, full_mid, SIZE, SIZE);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, full_mid+SIZE, half, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc11_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - uint8_t halfH[SIZE*SIZE];\ - uint8_t halfV[SIZE*SIZE];\ - put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\ - copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\ - put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc31_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - uint8_t halfH[SIZE*SIZE];\ - uint8_t halfV[SIZE*SIZE];\ - put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\ - copy_block ## SIZE (full, src - stride*2 + 1, SIZE, stride, SIZE + 5);\ - put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc13_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - uint8_t halfH[SIZE*SIZE];\ - uint8_t halfV[SIZE*SIZE];\ - put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\ - copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\ - put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc33_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - uint8_t halfH[SIZE*SIZE];\ - uint8_t halfV[SIZE*SIZE];\ - put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\ - copy_block ## SIZE (full, src - stride*2 + 1, SIZE, stride, SIZE + 5);\ - put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc22_sh4(uint8_t *dst, uint8_t *src, int stride){\ - int16_t tmp[SIZE*(SIZE+5)];\ - OPNAME ## h264_qpel ## SIZE ## _hv_lowpass(dst, tmp, src, stride, SIZE, stride);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc21_sh4(uint8_t *dst, uint8_t *src, int stride){\ - int16_t tmp[SIZE*(SIZE+5)];\ - uint8_t halfH[SIZE*SIZE];\ - uint8_t halfHV[SIZE*SIZE];\ - put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\ - put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfHV, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc23_sh4(uint8_t *dst, uint8_t *src, int stride){\ - int16_t tmp[SIZE*(SIZE+5)];\ - uint8_t halfH[SIZE*SIZE];\ - uint8_t halfHV[SIZE*SIZE];\ - put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\ - put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfHV, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc12_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - int16_t tmp[SIZE*(SIZE+5)];\ - uint8_t halfV[SIZE*SIZE];\ - uint8_t halfHV[SIZE*SIZE];\ - copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\ - put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\ - put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfV, halfHV, stride, SIZE, SIZE, SIZE);\ -}\ -\ -static void OPNAME ## h264_qpel ## SIZE ## _mc32_sh4(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[SIZE*(SIZE+5)];\ - uint8_t * const full_mid= full + SIZE*2;\ - int16_t tmp[SIZE*(SIZE+5)];\ - uint8_t halfV[SIZE*SIZE];\ - uint8_t halfHV[SIZE*SIZE];\ - copy_block ## SIZE (full, src - stride*2 + 1, SIZE, stride, SIZE + 5);\ - put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\ - put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\ - OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfV, halfHV, stride, SIZE, SIZE, SIZE);\ -}\ - -#define op_avg(a, b) a = (((a)+cm[((b) + 16)>>5]+1)>>1) -//#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7) -#define op_put(a, b) a = cm[((b) + 16)>>5] -#define op2_avg(a, b) a = (((a)+cm[((b) + 512)>>10]+1)>>1) -#define op2_put(a, b) a = cm[((b) + 512)>>10] - -H264_LOWPASS(put_ , op_put, op2_put) -H264_LOWPASS(avg_ , op_avg, op2_avg) -H264_MC(put_, 4) -H264_MC(put_, 8) -H264_MC(put_, 16) -H264_MC(avg_, 4) -H264_MC(avg_, 8) -H264_MC(avg_, 16) - -#undef op_avg -#undef op_put -#undef op2_avg -#undef op2_put - static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){ uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; |