generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
0b0bc8d098
vpx/vp8/common/rtcd_defs.sh

487 lines
23 KiB
Bash
Raw Normal View History

New RTCD implementation This is a proof of concept RTCD implementation to replace the current system of nested includes, prototypes, INVOKE macros, etc. Currently only the decoder specific functions are implemented in the new system. Additional functions will be added in subsequent commits. Overview: RTCD "functions" are implemented as either a global function pointer or a macro (when only one eligible specialization available). Functions which have RTCD specializations are listed using a simple DSL identifying the function's base name, its prototype, and the architecture extensions that specializations are available for. Advantages over the old system: - No INVOKE macros. A call to an RTCD function looks like an ordinary function call. - No need to pass vtables around. - If there is only one eligible function to call, the function is called directly, rather than indirecting through a function pointer. - Supports the notion of "required" extensions, so in combination with the above, on x86_64 if the best function available is sse2 or lower it will be called directly, since all x86_64 platforms implement sse2. - Elides all references to functions which will never be called, which could reduce binary size. For example if sse2 is required and there are both mmx and sse2 implementations of a certain function, the code will have no link time references to the mmx code. - Significantly easier to add a new function, just one file to edit. Disadvantages: - Requires global writable data (though this is not a new requirement) - 1 new generated source file. Change-Id: Iae6edab65315f79c168485c96872641c5aa09d55
2011-08-19 20:06:00 +02:00
common_forward_decls() {
cat <<EOF
struct blockd;
RTCD: add recon functions This commit continues the process of converting to the new RTCD system. Change-Id: I9bfcf9bef65c3d4ba0fb9a3e1532bad1463a10d6
2012-01-13 01:55:44 +01:00
struct macroblockd;
RTCD: add loopfilter functions This commit continues the process of converting to the new RTCD system. Change-Id: Ic8a4047d72ff3a54ec98977dd90e70c13213db71
2012-01-13 01:15:42 +01:00
struct loop_filter_info;
RTCD: add quantizer functions This commit continues the process of converting to the new RTCD system. Change-Id: Iba9df4c03a508e51c37201c621be43523fae87d9
2012-01-13 01:55:44 +01:00
/* Encoder forward decls */
struct block;
struct macroblock;
RTCD: add motion search functions This commit continues the process of converting to the new RTCD system. Change-Id: Ia5828b7ecc80db55b21916704aa3d54cbb98f625
2012-01-13 01:55:44 +01:00
struct variance_vtable;
union int_mv;
New RTCD implementation This is a proof of concept RTCD implementation to replace the current system of nested includes, prototypes, INVOKE macros, etc. Currently only the decoder specific functions are implemented in the new system. Additional functions will be added in subsequent commits. Overview: RTCD "functions" are implemented as either a global function pointer or a macro (when only one eligible specialization available). Functions which have RTCD specializations are listed using a simple DSL identifying the function's base name, its prototype, and the architecture extensions that specializations are available for. Advantages over the old system: - No INVOKE macros. A call to an RTCD function looks like an ordinary function call. - No need to pass vtables around. - If there is only one eligible function to call, the function is called directly, rather than indirecting through a function pointer. - Supports the notion of "required" extensions, so in combination with the above, on x86_64 if the best function available is sse2 or lower it will be called directly, since all x86_64 platforms implement sse2. - Elides all references to functions which will never be called, which could reduce binary size. For example if sse2 is required and there are both mmx and sse2 implementations of a certain function, the code will have no link time references to the mmx code. - Significantly easier to add a new function, just one file to edit. Disadvantages: - Requires global writable data (though this is not a new requirement) - 1 new generated source file. Change-Id: Iae6edab65315f79c168485c96872641c5aa09d55
2011-08-19 20:06:00 +02:00
EOF
}
forward_decls common_forward_decls
RTCD: add loopfilter functions This commit continues the process of converting to the new RTCD system. Change-Id: Ic8a4047d72ff3a54ec98977dd90e70c13213db71
2012-01-13 01:15:42 +01:00
#
# Dequant
#
New RTCD implementation This is a proof of concept RTCD implementation to replace the current system of nested includes, prototypes, INVOKE macros, etc. Currently only the decoder specific functions are implemented in the new system. Additional functions will be added in subsequent commits. Overview: RTCD "functions" are implemented as either a global function pointer or a macro (when only one eligible specialization available). Functions which have RTCD specializations are listed using a simple DSL identifying the function's base name, its prototype, and the architecture extensions that specializations are available for. Advantages over the old system: - No INVOKE macros. A call to an RTCD function looks like an ordinary function call. - No need to pass vtables around. - If there is only one eligible function to call, the function is called directly, rather than indirecting through a function pointer. - Supports the notion of "required" extensions, so in combination with the above, on x86_64 if the best function available is sse2 or lower it will be called directly, since all x86_64 platforms implement sse2. - Elides all references to functions which will never be called, which could reduce binary size. For example if sse2 is required and there are both mmx and sse2 implementations of a certain function, the code will have no link time references to the mmx code. - Significantly easier to add a new function, just one file to edit. Disadvantages: - Requires global writable data (though this is not a new requirement) - 1 new generated source file. Change-Id: Iae6edab65315f79c168485c96872641c5aa09d55
2011-08-19 20:06:00 +02:00
prototype void vp8_dequantize_b "struct blockd*, short *dqc"
specialize vp8_dequantize_b mmx media neon
vp8_dequantize_b_media=vp8_dequantize_b_v6
prototype void vp8_dequant_idct_add "short *input, short *dq, unsigned char *output, int stride"
specialize vp8_dequant_idct_add mmx media neon
vp8_dequant_idct_add_media=vp8_dequant_idct_add_v6
prototype void vp8_dequant_idct_add_y_block "short *q, short *dq, unsigned char *dst, int stride, char *eobs"
specialize vp8_dequant_idct_add_y_block mmx sse2 media neon
vp8_dequant_idct_add_y_block_media=vp8_dequant_idct_add_y_block_v6
prototype void vp8_dequant_idct_add_uv_block "short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs"
specialize vp8_dequant_idct_add_uv_block mmx sse2 media neon
vp8_dequant_idct_add_uv_block_media=vp8_dequant_idct_add_uv_block_v6
RTCD: add loopfilter functions This commit continues the process of converting to the new RTCD system. Change-Id: Ic8a4047d72ff3a54ec98977dd90e70c13213db71
2012-01-13 01:15:42 +01:00
#
# Loopfilter
#
prototype void vp8_loop_filter_mbv "unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi"
specialize vp8_loop_filter_mbv mmx sse2 media neon
vp8_loop_filter_mbv_media=vp8_loop_filter_mbv_armv6
prototype void vp8_loop_filter_bv "unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi"
specialize vp8_loop_filter_bv mmx sse2 media neon
vp8_loop_filter_bv_media=vp8_loop_filter_bv_armv6
prototype void vp8_loop_filter_mbh "unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi"
specialize vp8_loop_filter_mbh mmx sse2 media neon
vp8_loop_filter_mbh_media=vp8_loop_filter_mbh_armv6
prototype void vp8_loop_filter_bh "unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi"
specialize vp8_loop_filter_bh mmx sse2 media neon
vp8_loop_filter_bh_media=vp8_loop_filter_bh_armv6
prototype void vp8_loop_filter_simple_mbv "unsigned char *y, int ystride, const unsigned char *blimit"
specialize vp8_loop_filter_simple_mbv mmx sse2 media neon
vp8_loop_filter_simple_mbv_c=vp8_loop_filter_simple_vertical_edge_c
vp8_loop_filter_simple_mbv_mmx=vp8_loop_filter_simple_vertical_edge_mmx
vp8_loop_filter_simple_mbv_sse2=vp8_loop_filter_simple_vertical_edge_sse2
vp8_loop_filter_simple_mbv_media=vp8_loop_filter_simple_vertical_edge_armv6
vp8_loop_filter_simple_mbv_neon=vp8_loop_filter_mbvs_neon
prototype void vp8_loop_filter_simple_mbh "unsigned char *y, int ystride, const unsigned char *blimit"
specialize vp8_loop_filter_simple_mbh mmx sse2 media neon
vp8_loop_filter_simple_mbh_c=vp8_loop_filter_simple_horizontal_edge_c
vp8_loop_filter_simple_mbh_mmx=vp8_loop_filter_simple_horizontal_edge_mmx
vp8_loop_filter_simple_mbh_sse2=vp8_loop_filter_simple_horizontal_edge_sse2
vp8_loop_filter_simple_mbh_media=vp8_loop_filter_simple_horizontal_edge_armv6
vp8_loop_filter_simple_mbh_neon=vp8_loop_filter_mbhs_neon
prototype void vp8_loop_filter_simple_bv "unsigned char *y, int ystride, const unsigned char *blimit"
specialize vp8_loop_filter_simple_bv mmx sse2 media neon
vp8_loop_filter_simple_bv_c=vp8_loop_filter_bvs_c
vp8_loop_filter_simple_bv_mmx=vp8_loop_filter_bvs_mmx
vp8_loop_filter_simple_bv_sse2=vp8_loop_filter_bvs_sse2
vp8_loop_filter_simple_bv_media=vp8_loop_filter_bvs_armv6
vp8_loop_filter_simple_bv_neon=vp8_loop_filter_bvs_neon
prototype void vp8_loop_filter_simple_bh "unsigned char *y, int ystride, const unsigned char *blimit"
specialize vp8_loop_filter_simple_bh mmx sse2 media neon
vp8_loop_filter_simple_bh_c=vp8_loop_filter_bhs_c
vp8_loop_filter_simple_bh_mmx=vp8_loop_filter_bhs_mmx
vp8_loop_filter_simple_bh_sse2=vp8_loop_filter_bhs_sse2
vp8_loop_filter_simple_bh_media=vp8_loop_filter_bhs_armv6
vp8_loop_filter_simple_bh_neon=vp8_loop_filter_bhs_neon
RTCD: add remaining IDCT functions This commit continues the process of converting to the new RTCD system. Change-Id: I03c4dbf30dfd3558b0e256ff9d3ff4c012aadc80
2012-01-13 01:55:44 +01:00
#
# IDCT
#
#idct16
prototype void vp8_short_idct4x4llm "short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride"
specialize vp8_short_idct4x4llm mmx media neon
vp8_short_idct4x4llm_media=vp8_short_idct4x4llm_v6_dual
#iwalsh1
prototype void vp8_short_inv_walsh4x4_1 "short *input, short *output"
RTCD: add postproc functions This commit continues the process of converting to the new RTCD system. Change-Id: If54eb5cb5d1b0cac6c4c0633a9e99c93ca860ba2
2012-01-13 01:55:44 +01:00
# no asm yet
RTCD: add remaining IDCT functions This commit continues the process of converting to the new RTCD system. Change-Id: I03c4dbf30dfd3558b0e256ff9d3ff4c012aadc80
2012-01-13 01:55:44 +01:00
#iwalsh16
prototype void vp8_short_inv_walsh4x4 "short *input, short *output"
specialize vp8_short_inv_walsh4x4 mmx sse2 media neon
vp8_short_inv_walsh4x4_media=vp8_short_inv_walsh4x4_v6
#idct1_scalar_add
prototype void vp8_dc_only_idct_add "short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride"
specialize vp8_dc_only_idct_add mmx media neon
vp8_dc_only_idct_add_media=vp8_dc_only_idct_add_v6
RTCD: add recon functions This commit continues the process of converting to the new RTCD system. Change-Id: I9bfcf9bef65c3d4ba0fb9a3e1532bad1463a10d6
2012-01-13 01:55:44 +01:00
#
# RECON
#
prototype void vp8_copy_mem16x16 "unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch"
specialize vp8_copy_mem16x16 mmx sse2 media neon
vp8_copy_mem16x16_media=vp8_copy_mem16x16_v6
prototype void vp8_copy_mem8x8 "unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch"
specialize vp8_copy_mem8x8 mmx media neon
vp8_copy_mem8x8_media=vp8_copy_mem8x8_v6
prototype void vp8_copy_mem8x4 "unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch"
specialize vp8_copy_mem8x4 mmx media neon
vp8_copy_mem8x4_media=vp8_copy_mem8x4_v6
prototype void vp8_build_intra_predictors_mby "struct macroblockd *x"
specialize vp8_build_intra_predictors_mby sse2 ssse3 neon
prototype void vp8_build_intra_predictors_mby_s "struct macroblockd *x"
specialize vp8_build_intra_predictors_mby_s sse2 ssse3 neon
prototype void vp8_build_intra_predictors_mbuv "struct macroblockd *x"
specialize vp8_build_intra_predictors_mbuv sse2 ssse3
prototype void vp8_build_intra_predictors_mbuv_s "struct macroblockd *x"
specialize vp8_build_intra_predictors_mbuv_s sse2 ssse3
prototype void vp8_intra4x4_predict "unsigned char *src, int src_stride, int b_mode, unsigned char *dst, int dst_stride"
specialize vp8_intra4x4_predict media
vp8_intra4x4_predict_media=vp8_intra4x4_predict_armv6
RTCD: add postproc functions This commit continues the process of converting to the new RTCD system. Change-Id: If54eb5cb5d1b0cac6c4c0633a9e99c93ca860ba2
2012-01-13 01:55:44 +01:00
#
# Postproc
#
if [ "$CONFIG_POSTPROC" = "yes" ]; then
prototype void vp8_mbpost_proc_down "unsigned char *dst, int pitch, int rows, int cols,int flimit"
specialize vp8_mbpost_proc_down mmx sse2
vp8_mbpost_proc_down_sse2=vp8_mbpost_proc_down_xmm
prototype void vp8_mbpost_proc_across_ip "unsigned char *dst, int pitch, int rows, int cols,int flimit"
specialize vp8_mbpost_proc_across_ip sse2
vp8_mbpost_proc_across_ip_sse2=vp8_mbpost_proc_across_ip_xmm
prototype void vp8_post_proc_down_and_across "unsigned char *src, unsigned char *dst, int src_pitch, int dst_pitch, int rows, int cols, int flimit"
specialize vp8_post_proc_down_and_across mmx sse2
vp8_post_proc_down_and_across_sse2=vp8_post_proc_down_and_across_xmm
prototype void vp8_plane_add_noise "unsigned char *s, char *noise, char blackclamp[16], char whiteclamp[16], char bothclamp[16], unsigned int w, unsigned int h, int pitch"
specialize vp8_plane_add_noise mmx sse2
vp8_plane_add_noise_sse2=vp8_plane_add_noise_wmt
prototype void vp8_blend_mb_inner "unsigned char *y, unsigned char *u, unsigned char *v, int y1, int u1, int v1, int alpha, int stride"
# no asm yet
prototype void vp8_blend_mb_outer "unsigned char *y, unsigned char *u, unsigned char *v, int y1, int u1, int v1, int alpha, int stride"
# no asm yet
prototype void vp8_blend_b "unsigned char *y, unsigned char *u, unsigned char *v, int y1, int u1, int v1, int alpha, int stride"
# no asm yet
fi
RTCD: add subpixel functions This commit continues the process of converting to the new RTCD system. Change-Id: I6c519ab61e4f4e0ebcc796f2df061f945c48cefe
2012-01-13 01:55:44 +01:00
#
# Subpixel
#
prototype void vp8_sixtap_predict16x16 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
specialize vp8_sixtap_predict16x16 mmx sse2 ssse3 media neon
vp8_sixtap_predict16x16_media=vp8_sixtap_predict16x16_armv6
prototype void vp8_sixtap_predict8x8 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
specialize vp8_sixtap_predict8x8 mmx sse2 ssse3 media neon
vp8_sixtap_predict8x8_media=vp8_sixtap_predict8x8_armv6
prototype void vp8_sixtap_predict8x4 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
specialize vp8_sixtap_predict8x4 mmx sse2 ssse3 media neon
vp8_sixtap_predict8x4_media=vp8_sixtap_predict8x4_armv6
prototype void vp8_sixtap_predict4x4 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
specialize vp8_sixtap_predict4x4 mmx ssse3 media neon
vp8_sixtap_predict4x4_media=vp8_sixtap_predict4x4_armv6
prototype void vp8_bilinear_predict16x16 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
specialize vp8_bilinear_predict16x16 mmx sse2 ssse3 media neon
vp8_bilinear_predict16x16_media=vp8_bilinear_predict16x16_armv6
prototype void vp8_bilinear_predict8x8 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
specialize vp8_bilinear_predict8x8 mmx sse2 ssse3 media neon
vp8_bilinear_predict8x8_media=vp8_bilinear_predict8x8_armv6
prototype void vp8_bilinear_predict8x4 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
specialize vp8_bilinear_predict8x4 mmx media neon
vp8_bilinear_predict8x4_media=vp8_bilinear_predict8x4_armv6
prototype void vp8_bilinear_predict4x4 "unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch"
specialize vp8_bilinear_predict4x4 mmx media neon
vp8_bilinear_predict4x4_media=vp8_bilinear_predict4x4_armv6
RTCD: add variance functions This commit continues the process of converting to the new RTCD system. Change-Id: Ie5c1aa480637e98dc3918fb562ff45c37a66c538
2012-01-13 01:55:44 +01:00
#
# Encoder functions below this point.
#
if [ "$CONFIG_VP8_ENCODER" = "yes" ]; then
#
# Whole-pixel Variance
#
prototype unsigned int vp8_variance4x4 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_variance4x4 mmx sse2
vp8_variance4x4_sse2=vp8_variance4x4_wmt
prototype unsigned int vp8_variance8x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_variance8x8 mmx sse2 media neon
vp8_variance8x8_sse2=vp8_variance8x8_wmt
vp8_variance8x8_media=vp8_variance8x8_armv6
prototype unsigned int vp8_variance8x16 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_variance8x16 mmx sse2 neon
vp8_variance8x16_sse2=vp8_variance8x16_wmt
prototype unsigned int vp8_variance16x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_variance16x8 mmx sse2 neon
vp8_variance16x8_sse2=vp8_variance16x8_wmt
prototype unsigned int vp8_variance16x16 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_variance16x16 mmx sse2 media neon
vp8_variance16x16_sse2=vp8_variance16x16_wmt
vp8_variance16x16_media=vp8_variance16x16_armv6
#
# Sub-pixel Variance
#
prototype unsigned int vp8_sub_pixel_variance4x4 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
specialize vp8_sub_pixel_variance4x4 mmx sse2
vp8_sub_pixel_variance4x4_sse2=vp8_sub_pixel_variance4x4_wmt
prototype unsigned int vp8_sub_pixel_variance8x8 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
specialize vp8_sub_pixel_variance8x8 mmx sse2 media neon
vp8_sub_pixel_variance8x8_sse2=vp8_sub_pixel_variance8x8_wmt
vp8_sub_pixel_variance8x8_media=vp8_sub_pixel_variance8x8_armv6
prototype unsigned int vp8_sub_pixel_variance8x16 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
specialize vp8_sub_pixel_variance8x16 mmx sse2
vp8_sub_pixel_variance8x16_sse2=vp8_sub_pixel_variance8x16_wmt
prototype unsigned int vp8_sub_pixel_variance16x8 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
specialize vp8_sub_pixel_variance16x8 mmx sse2 ssse3
vp8_sub_pixel_variance16x8_sse2=vp8_sub_pixel_variance16x8_wmt
prototype unsigned int vp8_sub_pixel_variance16x16 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
specialize vp8_sub_pixel_variance16x16 mmx sse2 ssse3 media neon
vp8_sub_pixel_variance16x16_sse2=vp8_sub_pixel_variance16x16_wmt
vp8_sub_pixel_variance16x16_media=vp8_sub_pixel_variance16x16_armv6
prototype unsigned int vp8_variance_halfpixvar16x16_h "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_variance_halfpixvar16x16_h mmx sse2 media neon
vp8_variance_halfpixvar16x16_h_sse2=vp8_variance_halfpixvar16x16_h_wmt
vp8_variance_halfpixvar16x16_h_media=vp8_variance_halfpixvar16x16_h_armv6
prototype unsigned int vp8_variance_halfpixvar16x16_v "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_variance_halfpixvar16x16_v mmx sse2 media neon
vp8_variance_halfpixvar16x16_v_sse2=vp8_variance_halfpixvar16x16_v_wmt
vp8_variance_halfpixvar16x16_v_media=vp8_variance_halfpixvar16x16_v_armv6
prototype unsigned int vp8_variance_halfpixvar16x16_hv "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_variance_halfpixvar16x16_hv mmx sse2 media neon
vp8_variance_halfpixvar16x16_hv_sse2=vp8_variance_halfpixvar16x16_hv_wmt
vp8_variance_halfpixvar16x16_hv_media=vp8_variance_halfpixvar16x16_hv_armv6
#
# Sum of squares (vector)
#
prototype unsigned int vp8_get_mb_ss "const short *"
specialize vp8_get_mb_ss mmx sse2
#
# SSE (Sum Squared Error)
#
prototype unsigned int vp8_sub_pixel_mse16x16 "const unsigned char *src_ptr, int source_stride, int xoffset, int yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse"
specialize vp8_sub_pixel_mse16x16 mmx sse2
vp8_sub_pixel_mse16x16_sse2=vp8_sub_pixel_mse16x16_wmt
prototype unsigned int vp8_mse16x16 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sse"
specialize vp8_mse16x16 mmx sse2 media neon
vp8_mse16x16_sse2=vp8_mse16x16_wmt
vp8_mse16x16_media=vp8_mse16x16_armv6
prototype unsigned int vp8_get4x4sse_cs "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride"
specialize vp8_get4x4sse_cs mmx neon
#
# Single block SAD
#
prototype unsigned int vp8_sad4x4 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, int max_sad"
specialize vp8_sad4x4 mmx sse2 neon
vp8_sad4x4_sse2=vp8_sad4x4_wmt
prototype unsigned int vp8_sad8x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, int max_sad"
specialize vp8_sad8x8 mmx sse2 neon
vp8_sad8x8_sse2=vp8_sad8x8_wmt
prototype unsigned int vp8_sad8x16 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, int max_sad"
specialize vp8_sad8x16 mmx sse2 neon
vp8_sad8x16_sse2=vp8_sad8x16_wmt
prototype unsigned int vp8_sad16x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, int max_sad"
specialize vp8_sad16x8 mmx sse2 neon
vp8_sad16x8_sse2=vp8_sad16x8_wmt
prototype unsigned int vp8_sad16x16 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, int max_sad"
specialize vp8_sad16x16 mmx sse2 sse3 media neon
vp8_sad16x16_sse2=vp8_sad16x16_wmt
vp8_sad16x16_media=vp8_sad16x16_armv6
#
# Multi-block SAD, comparing a reference to N blocks 1 pixel apart horizontally
#
prototype void vp8_sad4x4x3 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
specialize vp8_sad4x4x3 sse3
prototype void vp8_sad8x8x3 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
specialize vp8_sad8x8x3 sse3
prototype void vp8_sad8x16x3 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
specialize vp8_sad8x16x3 sse3
prototype void vp8_sad16x8x3 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
specialize vp8_sad16x8x3 sse3 ssse3
prototype void vp8_sad16x16x3 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int *sad_array"
specialize vp8_sad16x16x3 sse3 ssse3
# Note the only difference in the following prototypes is that they return into
# an array of short
prototype void vp8_sad4x4x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
specialize vp8_sad4x4x8 sse4_1
vp8_sad4x4x8_sse4_1=vp8_sad4x4x8_sse4
prototype void vp8_sad8x8x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
specialize vp8_sad8x8x8 sse4_1
vp8_sad8x8x8_sse4_1=vp8_sad8x8x8_sse4
prototype void vp8_sad8x16x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
specialize vp8_sad8x16x8 sse4_1
vp8_sad8x16x8_sse4_1=vp8_sad8x16x8_sse4
prototype void vp8_sad16x8x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
specialize vp8_sad16x8x8 sse4_1
vp8_sad16x8x8_sse4_1=vp8_sad16x8x8_sse4
prototype void vp8_sad16x16x8 "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, unsigned short *sad_array"
specialize vp8_sad16x16x8 sse4_1
vp8_sad16x16x8_sse4_1=vp8_sad16x16x8_sse4
#
# Multi-block SAD, comparing a reference to N independent blocks
#
prototype void vp8_sad4x4x4d "const unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr[4], int ref_stride, unsigned int *sad_array"
specialize vp8_sad4x4x4d sse3
prototype void vp8_sad8x8x4d "const unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr[4], int ref_stride, unsigned int *sad_array"
specialize vp8_sad8x8x4d sse3
prototype void vp8_sad8x16x4d "const unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr[4], int ref_stride, unsigned int *sad_array"
specialize vp8_sad8x16x4d sse3
prototype void vp8_sad16x8x4d "const unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr[4], int ref_stride, unsigned int *sad_array"
specialize vp8_sad16x8x4d sse3
prototype void vp8_sad16x16x4d "const unsigned char *src_ptr, int source_stride, unsigned char *ref_ptr[4], int ref_stride, unsigned int *sad_array"
specialize vp8_sad16x16x4d sse3
#
# Block copy
#
case $arch in
x86*)
prototype void vp8_copy32xn "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride, int n"
specialize vp8_copy32xn sse2 sse3
;;
esac
#
# Structured Similarity (SSIM)
#
if [ "$CONFIG_INTERNAL_STATS" = "yes" ]; then
[ $arch = "x86_64" ] && sse2_on_x86_64=sse2
prototype void vp8_ssim_parms_8x8 "unsigned char *s, int sp, unsigned char *r, int rp, unsigned long *sum_s, unsigned long *sum_r, unsigned long *sum_sq_s, unsigned long *sum_sq_r, unsigned long *sum_sxr"
specialize vp8_ssim_parms_8x8 $sse2_on_x86_64
prototype void vp8_ssim_parms_16x16 "unsigned char *s, int sp, unsigned char *r, int rp, unsigned long *sum_s, unsigned long *sum_r, unsigned long *sum_sq_s, unsigned long *sum_sq_r, unsigned long *sum_sxr"
specialize vp8_ssim_parms_16x16 $sse2_on_x86_64
fi
RTCD: add FDCT functions This commit continues the process of converting to the new RTCD system. Change-Id: I3f9c07db65eb206f6363d21bdb80e871570da767
2012-01-13 01:55:44 +01:00
#
# Forward DCT
#
prototype void vp8_short_fdct4x4 "short *input, short *output, int pitch"
specialize vp8_short_fdct4x4 mmx sse2 media neon
vp8_short_fdct4x4_media=vp8_short_fdct4x4_armv6
prototype void vp8_short_fdct8x4 "short *input, short *output, int pitch"
specialize vp8_short_fdct8x4 mmx sse2 media neon
vp8_short_fdct8x4_media=vp8_short_fdct8x4_armv6
prototype void vp8_short_walsh4x4 "short *input, short *output, int pitch"
specialize vp8_short_walsh4x4 mmx sse2 media neon
vp8_short_walsh4x4_media=vp8_short_walsh4x4_armv6
RTCD: add quantizer functions This commit continues the process of converting to the new RTCD system. Change-Id: Iba9df4c03a508e51c37201c621be43523fae87d9
2012-01-13 01:55:44 +01:00
#
# Quantizer
#
prototype void vp8_regular_quantize_b "struct block *, struct blockd *"
specialize vp8_regular_quantize_b sse2 sse4_1
vp8_regular_quantize_b_sse4_1=vp8_regular_quantize_b_sse4
prototype void vp8_fast_quantize_b "struct block *, struct blockd *"
specialize vp8_fast_quantize_b sse2 ssse3 media neon
vp8_fast_quantize_b_media=vp8_fast_quantize_b_armv6
prototype void vp8_regular_quantize_b_pair "struct block *b1, struct block *b2, struct blockd *d1, struct blockd *d2"
# no asm yet
prototype void vp8_fast_quantize_b_pair "struct block *b1, struct block *b2, struct blockd *d1, struct blockd *d2"
specialize vp8_fast_quantize_b_pair neon
prototype void vp8_quantize_mb "struct macroblock *"
specialize vp8_quantize_mb neon
prototype void vp8_quantize_mby "struct macroblock *"
specialize vp8_quantize_mby neon
prototype void vp8_quantize_mbuv "struct macroblock *"
specialize vp8_quantize_mbuv neon
RTCD: add block subtraction functions This commit continues the process of converting to the new RTCD system. Change-Id: Id8a287fdd4bd050ea4452e1582ad85520f3081be
2012-01-13 01:55:44 +01:00
#
# Block subtraction
#
prototype int vp8_block_error "short *coeff, short *dqcoeff"
specialize vp8_block_error mmx sse2
vp8_block_error_sse2=vp8_block_error_xmm
prototype int vp8_mbblock_error "struct macroblock *mb, int dc"
specialize vp8_mbblock_error mmx sse2
vp8_mbblock_error_sse2=vp8_mbblock_error_xmm
prototype int vp8_mbuverror "struct macroblock *mb"
specialize vp8_mbuverror mmx sse2
vp8_mbuverror_sse2=vp8_mbuverror_xmm
prototype void vp8_subtract_b "struct block *be, struct blockd *bd, int pitch"
specialize vp8_subtract_b mmx sse2 media neon
vp8_subtract_b_media=vp8_subtract_b_armv6
prototype void vp8_subtract_mby "short *diff, unsigned char *src, int src_stride, unsigned char *pred, int pred_stride"
specialize vp8_subtract_mby mmx sse2 media neon
vp8_subtract_mby_media=vp8_subtract_mby_armv6
prototype void vp8_subtract_mbuv "short *diff, unsigned char *usrc, unsigned char *vsrc, int src_stride, unsigned char *upred, unsigned char *vpred, int pred_stride"
specialize vp8_subtract_mbuv mmx sse2 media neon
vp8_subtract_mbuv_media=vp8_subtract_mbuv_armv6
RTCD: add motion search functions This commit continues the process of converting to the new RTCD system. Change-Id: Ia5828b7ecc80db55b21916704aa3d54cbb98f625
2012-01-13 01:55:44 +01:00
#
# Motion search
#
prototype int vp8_full_search_sad "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, int sad_per_bit, int distance, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"
specialize vp8_full_search_sad sse3 sse4_1
vp8_full_search_sad_sse3=vp8_full_search_sadx3
vp8_full_search_sad_sse4_1=vp8_full_search_sadx8
prototype int vp8_refining_search_sad "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, int sad_per_bit, int distance, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"
specialize vp8_refining_search_sad sse3
vp8_refining_search_sad_sse3=vp8_refining_search_sadx4
prototype int vp8_diamond_search_sad "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, union int_mv *best_mv, int search_param, int sad_per_bit, int *num00, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"
vp8_diamond_search_sad_sse3=vp8_diamond_search_sadx4
RTCD: add variance functions This commit continues the process of converting to the new RTCD system. Change-Id: Ie5c1aa480637e98dc3918fb562ff45c37a66c538
2012-01-13 01:55:44 +01:00
# End of encoder only functions
fi
Reference in New Issue Copy Permalink