yading@10: /* yading@10: * MMX optimized DSP utils yading@10: * Copyright (c) 2007 Aurelien Jacobs yading@10: * yading@10: * This file is part of FFmpeg. yading@10: * yading@10: * FFmpeg is free software; you can redistribute it and/or yading@10: * modify it under the terms of the GNU Lesser General Public yading@10: * License as published by the Free Software Foundation; either yading@10: * version 2.1 of the License, or (at your option) any later version. yading@10: * yading@10: * FFmpeg is distributed in the hope that it will be useful, yading@10: * but WITHOUT ANY WARRANTY; without even the implied warranty of yading@10: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU yading@10: * Lesser General Public License for more details. yading@10: * yading@10: * You should have received a copy of the GNU Lesser General Public yading@10: * License along with FFmpeg; if not, write to the Free Software yading@10: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA yading@10: */ yading@10: yading@10: #ifndef AVCODEC_X86_DSPUTIL_MMX_H yading@10: #define AVCODEC_X86_DSPUTIL_MMX_H yading@10: yading@10: #include yading@10: #include yading@10: yading@10: #include "libavcodec/dsputil.h" yading@10: #include "libavutil/x86/asm.h" yading@10: yading@10: extern const uint64_t ff_bone; yading@10: extern const uint64_t ff_wtwo; yading@10: yading@10: extern const xmm_reg ff_pw_3; yading@10: extern const xmm_reg ff_pw_4; yading@10: extern const xmm_reg ff_pw_5; yading@10: extern const xmm_reg ff_pw_8; yading@10: extern const uint64_t ff_pw_15; yading@10: extern const xmm_reg ff_pw_16; yading@10: extern const xmm_reg ff_pw_18; yading@10: extern const uint64_t ff_pw_20; yading@10: extern const xmm_reg ff_pw_32; yading@10: extern const uint64_t ff_pw_42; yading@10: extern const uint64_t ff_pw_53; yading@10: extern const xmm_reg ff_pw_64; yading@10: extern const uint64_t ff_pw_96; yading@10: extern const uint64_t ff_pw_128; yading@10: extern const uint64_t ff_pw_255; yading@10: yading@10: extern const xmm_reg ff_pb_1; yading@10: extern const xmm_reg ff_pb_3; yading@10: extern const uint64_t ff_pb_3F; yading@10: extern const xmm_reg ff_pb_F8; yading@10: extern const uint64_t ff_pb_FC; yading@10: yading@10: extern const double ff_pd_1[2]; yading@10: extern const double ff_pd_2[2]; yading@10: yading@10: #define SBUTTERFLY(a,b,t,n,m)\ yading@10: "mov" #m " " #a ", " #t " \n\t" /* abcd */\ yading@10: "punpckl" #n " " #b ", " #a " \n\t" /* aebf */\ yading@10: "punpckh" #n " " #b ", " #t " \n\t" /* cgdh */\ yading@10: yading@10: #define TRANSPOSE4(a,b,c,d,t)\ yading@10: SBUTTERFLY(a,b,t,wd,q) /* a=aebf t=cgdh */\ yading@10: SBUTTERFLY(c,d,b,wd,q) /* c=imjn b=kolp */\ yading@10: SBUTTERFLY(a,c,d,dq,q) /* a=aeim d=bfjn */\ yading@10: SBUTTERFLY(t,b,c,dq,q) /* t=cgko c=dhlp */ yading@10: yading@10: #define MOVQ_WONE(regd) \ yading@10: __asm__ volatile ( \ yading@10: "pcmpeqd %%" #regd ", %%" #regd " \n\t" \ yading@10: "psrlw $15, %%" #regd ::) yading@10: yading@10: void ff_dsputilenc_init_mmx(DSPContext* c, AVCodecContext *avctx); yading@10: void ff_dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx); yading@10: yading@10: void ff_add_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size); yading@10: void ff_put_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size); yading@10: void ff_put_signed_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size); yading@10: yading@10: void ff_avg_pixels8_mmxext(uint8_t *block, const uint8_t *pixels, yading@10: ptrdiff_t line_size, int h); yading@10: void ff_put_pixels8_mmxext(uint8_t *block, const uint8_t *pixels, yading@10: ptrdiff_t line_size, int h); yading@10: void ff_avg_pixels16_sse2(uint8_t *block, const uint8_t *pixels, yading@10: ptrdiff_t line_size, int h); yading@10: void ff_put_pixels16_sse2(uint8_t *block, const uint8_t *pixels, yading@10: ptrdiff_t line_size, int h); yading@10: yading@10: void ff_put_cavs_qpel8_mc00_mmxext(uint8_t *dst, uint8_t *src, ptrdiff_t stride); yading@10: void ff_avg_cavs_qpel8_mc00_mmxext(uint8_t *dst, uint8_t *src, ptrdiff_t stride); yading@10: void ff_put_cavs_qpel16_mc00_mmxext(uint8_t *dst, uint8_t *src, ptrdiff_t stride); yading@10: void ff_avg_cavs_qpel16_mc00_mmxext(uint8_t *dst, uint8_t *src, ptrdiff_t stride); yading@10: yading@10: void ff_put_vc1_mspel_mc00_mmx(uint8_t *dst, const uint8_t *src, ptrdiff_t stride, int rnd); yading@10: yading@10: void ff_put_rv40_qpel8_mc33_mmx(uint8_t *block, uint8_t *pixels, ptrdiff_t stride); yading@10: void ff_put_rv40_qpel16_mc33_mmx(uint8_t *block, uint8_t *pixels, ptrdiff_t stride); yading@10: void ff_avg_rv40_qpel8_mc33_mmx(uint8_t *block, uint8_t *pixels, ptrdiff_t stride); yading@10: void ff_avg_rv40_qpel16_mc33_mmx(uint8_t *block, uint8_t *pixels, ptrdiff_t stride); yading@10: yading@10: void ff_mmx_idct(int16_t *block); yading@10: void ff_mmxext_idct(int16_t *block); yading@10: yading@10: yading@10: void ff_deinterlace_line_mmx(uint8_t *dst, yading@10: const uint8_t *lum_m4, const uint8_t *lum_m3, yading@10: const uint8_t *lum_m2, const uint8_t *lum_m1, yading@10: const uint8_t *lum, yading@10: int size); yading@10: yading@10: void ff_deinterlace_line_inplace_mmx(const uint8_t *lum_m4, yading@10: const uint8_t *lum_m3, yading@10: const uint8_t *lum_m2, yading@10: const uint8_t *lum_m1, yading@10: const uint8_t *lum, int size); yading@10: yading@10: #endif /* AVCODEC_X86_DSPUTIL_MMX_H */