;****************************************************************************** ;* VP9 IDCT SIMD optimizations ;* ;* Copyright (C) 2013 Clément Bœsch <u pkh me> ;* Copyright (C) 2013 Ronald S. Bultje <rsbultje gmail com> ;* ;* This file is part of FFmpeg. ;* ;* FFmpeg is free software; you can redistribute it and/or ;* modify it under the terms of the GNU Lesser General Public ;* License as published by the Free Software Foundation; either ;* version 2.1 of the License, or (at your option) any later version. ;* ;* FFmpeg is distributed in the hope that it will be useful, ;* but WITHOUT ANY WARRANTY; without even the implied warranty of ;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU ;* Lesser General Public License for more details. ;* ;* You should have received a copy of the GNU Lesser General Public ;* License along with FFmpeg; if not, write to the Free Software ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA ;****************************************************************************** %include "libavutil/x86/x86util.asm" %include "vp9itxfm_template.asm" SECTION_RODATA 32 %macro VP9_IDCT_COEFFS 2-3 0 const pw_m%1_%2 times 8 dw -%1, %2 const pw_%2_%1 times 8 dw %2, %1 %if %3 == 1 const pw_m%2_m%1 times 8 dw -%2, -%1 %if %1 != %2 const pw_m%2_%1 times 8 dw -%2, %1 const pw_%1_%2 times 8 dw %1, %2 %endif %endif %if %1 < 11585 pw_m%1x2: times 16 dw -%1*2 %elif %1 > 11585 pw_%1x2: times 16 dw %1*2 %else const pw_%1x2 times 16 dw %1*2 %endif %if %2 != %1 pw_%2x2: times 16 dw %2*2 %endif %endmacro VP9_IDCT_COEFFS 16364, 804 VP9_IDCT_COEFFS 16305, 1606 VP9_IDCT_COEFFS 16069, 3196, 1 VP9_IDCT_COEFFS 15893, 3981 VP9_IDCT_COEFFS 15137, 6270, 1 VP9_IDCT_COEFFS 14811, 7005 VP9_IDCT_COEFFS 14449, 7723 VP9_IDCT_COEFFS 13160, 9760 VP9_IDCT_COEFFS 11585, 11585, 1 VP9_IDCT_COEFFS 11003, 12140 VP9_IDCT_COEFFS 10394, 12665 VP9_IDCT_COEFFS 9102, 13623, 1 VP9_IDCT_COEFFS 8423, 14053 VP9_IDCT_COEFFS 5520, 15426 VP9_IDCT_COEFFS 4756, 15679 VP9_IDCT_COEFFS 2404, 16207 const pw_5283_13377 times 4 dw 5283, 13377 const pw_9929_13377 times 4 dw 9929, 13377 const pw_15212_m13377 times 4 dw 15212, -13377 const pw_15212_9929 times 4 dw 15212, 9929 const pw_m5283_m15212 times 4 dw -5283, -15212 const pw_13377x2 times 8 dw 13377*2 const pw_m13377_13377 times 4 dw -13377, 13377 const pw_13377_0 times 4 dw 13377, 0 cextern pw_8 cextern pw_16 cextern pw_32 cextern pw_512 cextern pw_1024 cextern pw_2048 cextern pw_m1 cextern pd_8192 SECTION .text %macro VP9_UNPACK_MULSUB_2D_4X 6 ; dst1 [src1], dst2 [src2], dst3, dst4, mul1, mul2 punpckhwd m%4, m%2, m%1 punpcklwd m%2, m%1 pmaddwd m%3, m%4, [pw_m%5_%6] pmaddwd m%4, [pw_%6_%5] pmaddwd m%1, m%2, [pw_m%5_%6] pmaddwd m%2, [pw_%6_%5] %endmacro %macro VP9_RND_SH_SUMSUB_BA 6 ; dst1 [src1], dst2 [src2], src3, src4, tmp, round SUMSUB_BA d, %1, %2, %5 SUMSUB_BA d, %3, %4, %5 paddd m%1, %6 paddd m%2, %6 paddd m%3, %6 paddd m%4, %6 psrad m%1, 14 psrad m%2, 14 psrad m%3, 14 psrad m%4, 14 packssdw m%1, m%3 packssdw m%2, m%4 %endmacro %macro VP9_STORE_2X 5-6 dstq ; reg1, reg2, tmp1, tmp2, zero, dst %if mmsize == 32 pmovzxbw m%3, [%6] pmovzxbw m%4, [%6+strideq] %else movh m%3, [%6] movh m%4, [%6+strideq] punpcklbw m%3, m%5 punpcklbw m%4, m%5 %endif paddw m%3, m%1 paddw m%4, m%2 %if mmsize == 32 packuswb m%3, m%4 ; Intel... vpermq m%3, m%3, q3120 mova [%6], xm%3 vextracti128 [%6+strideq], m%3, 1 %elif mmsize == 16 packuswb m%3, m%4 movh [%6], m%3 movhps [%6+strideq], m%3 %else packuswb m%3, m%5 packuswb m%4, m%5 movh [%6], m%3 movh [%6+strideq], m%4 %endif %endmacro %macro ZERO_BLOCK 4 ; mem, stride, nnzcpl, zero_reg %assign %%y 0 %rep %3 %assign %%x 0 %rep %3*2/mmsize mova [%1+%%y+%%x], %4 %assign %%x (%%x+mmsize) %endrep %assign %%y (%%y+%2) %endrep %endmacro ;------------------------------------------------------------------------------------------- ; void vp9_iwht_iwht_4x4_add_<opt>(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob); ;------------------------------------------------------------------------------------------- INIT_MMX mmx cglobal vp9_iwht_iwht_4x4_add, 3, 3, 0, dst, stride, block, eob mova m0, [blockq+0*8] mova m1, [blockq+1*8] mova m2, [blockq+2*8] mova m3, [blockq+3*8] psraw m0, 2 psraw m1, 2 psraw m2, 2 psraw m3, 2 VP9_IWHT4_1D TRANSPOSE4x4W 0, 1, 2, 3, 4 VP9_IWHT4_1D pxor m4, m4 VP9_STORE_2X 0, 1, 5, 6, 4 lea dstq, [dstq+strideq*2] VP9_STORE_2X 2, 3, 5, 6, 4 ZERO_BLOCK blockq, 8, 4, m4 RET ;------------------------------------------------------------------------------------------- ; void vp9_idct_idct_4x4_add_<opt>(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob); ;------------------------------------------------------------------------------------------- ; 2x2 top left corner %macro VP9_IDCT4_2x2_1D 0 pmulhrsw m0, m5 ; m0=t1 mova m2, m0 ; m2=t0 mova m3, m1 pmulhrsw m1, m6 ; m1=t2 pmulhrsw m3, m7 ; m3=t3 VP9_IDCT4_1D_FINALIZE %endmacro %macro VP9_IDCT4_WRITEOUT 0 %if cpuflag(ssse3) mova m5, [pw_2048] pmulhrsw m0, m5 ; (x*2048 + (1<<14))>>15 <=> (x+8)>>4 pmulhrsw m1, m5 %else mova m5, [pw_8] paddw m0, m5 paddw m1, m5 psraw m0, 4 psraw m1, 4 %endif VP9_STORE_2X 0, 1, 6, 7, 4 lea dstq, [dstq+2*strideq] %if cpuflag(ssse3) pmulhrsw m2, m5 pmulhrsw m3, m5 %else paddw m2, m5 paddw m3, m5 psraw m2, 4 psraw m3, 4 %endif VP9_STORE_2X 2, 3, 6, 7, 4 %endmacro %macro IDCT_4x4_FN 1 INIT_MMX %1 cglobal vp9_idct_idct_4x4_add, 4, 4, 0, dst, stride, block, eob %if cpuflag(ssse3) cmp eobd, 4 ; 2x2 or smaller jg .idctfull cmp eobd, 1 ; faster path for when only DC is set jne .idct2x2 %else cmp eobd, 1 jg .idctfull %endif %if cpuflag(ssse3) movd m0, [blockq] mova m5, [pw_11585x2] pmulhrsw m0, m5 pmulhrsw m0, m5 %else DEFINE_ARGS dst, stride, block, coef movsx coefd, word [blockq] imul coefd, 11585 add coefd, 8192 sar coefd, 14 imul coefd, 11585 add coefd, (8 << 14) + 8192 sar coefd, 14 + 4 movd m0, coefd %endif pshufw m0, m0, 0 pxor m4, m4 movh [blockq], m4 %if cpuflag(ssse3) pmulhrsw m0, [pw_2048] ; (x*2048 + (1<<14))>>15 <=> (x+8)>>4 %endif VP9_STORE_2X 0, 0, 6, 7, 4 lea dstq, [dstq+2*strideq] VP9_STORE_2X 0, 0, 6, 7, 4 RET %if cpuflag(ssse3) ; faster path for when only top left 2x2 block is set .idct2x2: movd m0, [blockq+0] movd m1, [blockq+8] mova m5, [pw_11585x2] mova m6, [pw_6270x2] mova m7, [pw_15137x2] VP9_IDCT4_2x2_1D ; partial 2x4 transpose punpcklwd m0, m1 punpcklwd m2, m3 SBUTTERFLY dq, 0, 2, 1 SWAP 1, 2 VP9_IDCT4_2x2_1D pxor m4, m4 ; used for the block reset, and VP9_STORE_2X movh [blockq+ 0], m4 movh [blockq+ 8], m4 VP9_IDCT4_WRITEOUT RET %endif .idctfull: ; generic full 4x4 idct/idct mova m0, [blockq+ 0] mova m1, [blockq+ 8] mova m2, [blockq+16] mova m3, [blockq+24] %if cpuflag(ssse3) mova m6, [pw_11585x2] %endif mova m7, [pd_8192] ; rounding VP9_IDCT4_1D TRANSPOSE4x4W 0, 1, 2, 3, 4 VP9_IDCT4_1D pxor m4, m4 ; used for the block reset, and VP9_STORE_2X mova [blockq+ 0], m4 mova [blockq+ 8], m4 mova [blockq+16], m4 mova [blockq+24], m4 VP9_IDCT4_WRITEOUT RET %endmacro IDCT_4x4_FN mmxext IDCT_4x4_FN ssse3 ;------------------------------------------------------------------------------------------- ; void vp9_iadst_iadst_4x4_add_<opt>(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob); ;------------------------------------------------------------------------------------------- %macro IADST4_FN 5 INIT_MMX %5 cglobal vp9_%1_%3_4x4_add, 3, 3, 0, dst, stride, block, eob %if WIN64 && notcpuflag(ssse3) INIT_XMM cpuname WIN64_SPILL_XMM 8 INIT_MMX cpuname %endif movdqa xmm5, [pd_8192] mova m0, [blockq+ 0] mova m1, [blockq+ 8] mova m2, [blockq+16] mova m3, [blockq+24] %if cpuflag(ssse3) mova m6, [pw_11585x2] %endif %ifnidn %1%3, iadstiadst movdq2q m7, xmm5 %endif VP9_%2_1D TRANSPOSE4x4W 0, 1, 2, 3, 4 VP9_%4_1D pxor m4, m4 ; used for the block reset, and VP9_STORE_2X mova [blockq+ 0], m4 mova [blockq+ 8], m4 mova [blockq+16], m4 mova [blockq+24], m4 VP9_IDCT4_WRITEOUT RET %endmacro IADST4_FN idct, IDCT4, iadst, IADST4, sse2 IADST4_FN iadst, IADST4, idct, IDCT4, sse2 IADST4_FN iadst, IADST4, iadst, IADST4, sse2 IADST4_FN idct, IDCT4, iadst, IADST4, ssse3 IADST4_FN iadst, IADST4, idct, IDCT4, ssse3 IADST4_FN iadst, IADST4, iadst, IADST4, ssse3 %macro SCRATCH 3 %if ARCH_X86_64 SWAP %1, %2 %else mova [%3], m%1 %endif %endmacro %macro UNSCRATCH 3 %if ARCH_X86_64 SWAP %1, %2 %else mova m%1, [%3] %endif %endmacro ;------------------------------------------------------------------------------------------- ; void vp9_idct_idct_8x8_add_<opt>(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob); ;------------------------------------------------------------------------------------------- %macro VP9_IDCT8_1D_FINALIZE 0 SUMSUB_BA w, 3, 6, 5 ; m3=t0+t7, m6=t0-t7 SUMSUB_BA w, 1, 2, 5 ; m1=t1+t6, m2=t1-t6 SUMSUB_BA w, 7, 0, 5 ; m7=t2+t5, m0=t2-t5 UNSCRATCH 5, 8, blockq+ 0 SCRATCH 2, 8, blockq+ 0 SUMSUB_BA w, 5, 4, 2 ; m5=t3+t4, m4=t3-t4 SWAP 7, 6, 2 SWAP 3, 5, 0 %if ARCH_X86_64 SWAP 6, 8 %endif %endmacro ; x86-32 ; - in: m0/m4 is in mem ; - out: m6 is in mem ; x86-64: ; - everything is in registers (m0-7) %macro VP9_IDCT8_1D 0 %if ARCH_X86_64 SWAP 0, 8 SWAP 4, 9 %endif VP9_UNPACK_MULSUB_2W_4X 5, 3, 9102, 13623, D_8192_REG, 0, 4 ; m5=t5a, m3=t6a VP9_UNPACK_MULSUB_2W_4X 1, 7, 16069, 3196, D_8192_REG, 0, 4 ; m1=t4a, m7=t7a SUMSUB_BA w, 5, 1, 0 ; m5=t4a+t5a (t4), m1=t4a-t5a (t5a) SUMSUB_BA w, 3, 7, 0 ; m3=t7a+t6a (t7), m7=t7a-t6a (t6a) %if cpuflag(ssse3) SUMSUB_BA w, 1, 7, 0 ; m1=t6a+t5a (t6), m7=t6a-t5a (t5) pmulhrsw m1, W_11585x2_REG ; m1=t6 pmulhrsw m7, W_11585x2_REG ; m7=t5 %else VP9_UNPACK_MULSUB_2W_4X 7, 1, 11585, 11585, D_8192_REG, 0, 4 %endif VP9_UNPACK_MULSUB_2W_4X 2, 6, 15137, 6270, D_8192_REG, 0, 4 ; m2=t2a, m6=t3a UNSCRATCH 0, 8, blockq+ 0 ; IN(0) UNSCRATCH 4, 9, blockq+64 ; IN(4) SCRATCH 5, 8, blockq+ 0 %if cpuflag(ssse3) SUMSUB_BA w, 4, 0, 5 ; m4=IN(0)+IN(4) m0=IN(0)-IN(4) pmulhrsw m4, W_11585x2_REG ; m4=t0a pmulhrsw m0, W_11585x2_REG ; m0=t1a %else SCRATCH 7, 9, blockq+64 VP9_UNPACK_MULSUB_2W_4X 0, 4, 11585, 11585, D_8192_REG, 5, 7 UNSCRATCH 7, 9, blockq+64 %endif SUMSUB_BA w, 6, 4, 5 ; m6=t0a+t3a (t0), m4=t0a-t3a (t3) SUMSUB_BA w, 2, 0, 5 ; m2=t1a+t2a (t1), m0=t1a-t2a (t2) VP9_IDCT8_1D_FINALIZE %endmacro %macro VP9_IDCT8_4x4_1D 0 pmulhrsw m0, W_11585x2_REG ; m0=t1a/t0a pmulhrsw m6, m2, [pw_15137x2] ; m6=t3a pmulhrsw m2, [pw_6270x2] ; m2=t2a pmulhrsw m7, m1, [pw_16069x2] ; m7=t7a pmulhrsw m1, [pw_3196x2] ; m1=t4a pmulhrsw m5, m3, [pw_m9102x2] ; m5=t5a pmulhrsw m3, [pw_13623x2] ; m3=t6a SUMSUB_BA w, 5, 1, 4 ; m1=t4a+t5a (t4), m5=t4a-t5a (t5a) SUMSUB_BA w, 3, 7, 4 ; m3=t7a+t6a (t7), m7=t7a-t6a (t6a) SUMSUB_BA w, 1, 7, 4 ; m1=t6a+t5a (t6), m7=t6a-t5a (t5) pmulhrsw m1, W_11585x2_REG ; m1=t6 pmulhrsw m7, W_11585x2_REG ; m7=t5 psubw m4, m0, m6 ; m4=t0a-t3a (t3) paddw m6, m0 ; m6=t0a+t3a (t0) SCRATCH 5, 8, blockq+ 0 SUMSUB_BA w, 2, 0, 5 ; m2=t1a+t2a (t1), m0=t1a-t2a (t2) VP9_IDCT8_1D_FINALIZE %endmacro %macro VP9_IDCT8_2x2_1D 1 pmulhrsw m0, W_11585x2_REG ; m0=t0 pmulhrsw m3, m1, W_16069x2_REG ; m3=t7 pmulhrsw m1, W_3196x2_REG ; m1=t4 psubw m7, m3, m1 ; t5 = t7a - t4a paddw m5, m3, m1 ; t6 = t7a + t4a pmulhrsw m7, W_11585x2_REG ; m7=t5 pmulhrsw m5, W_11585x2_REG ; m5=t6 SWAP 5, 1 ; merged VP9_IDCT8_1D_FINALIZE to make register-sharing w/ avx easier psubw m6, m0, m3 ; m6=t0-t7 paddw m3, m0 ; m3=t0+t7 psubw m2, m0, m1 ; m2=t1-t6 paddw m1, m0 ; m1=t1+t6 %if %1 == 1 punpcklwd m3, m1 %define SCRATCH_REG 1 %elif ARCH_X86_32 mova [blockq+ 0], m2 %define SCRATCH_REG 2 %else %define SCRATCH_REG 8 %endif psubw m4, m0, m5 ; m4=t3-t4 paddw m5, m0 ; m5=t3+t4 SUMSUB_BA w, 7, 0, SCRATCH_REG ; m7=t2+t5, m0=t2-t5 SWAP 7, 6, 2 SWAP 3, 5, 0 %undef SCRATCH_REG %endmacro %macro VP9_IDCT8_WRITEx2 6-8 5 ; line1, line2, tmp1, tmp2, zero, pw_1024/pw_16, shift %if cpuflag(ssse3) pmulhrsw m%1, %6 ; (x*1024 + (1<<14))>>15 <=> (x+16)>>5 pmulhrsw m%2, %6 %else paddw m%1, %6 paddw m%2, %6 psraw m%1, %7 psraw m%2, %7 %endif %if %0 <= 7 VP9_STORE_2X %1, %2, %3, %4, %5 %else VP9_STORE_2X %1, %2, %3, %4, %5, %8 %endif %endmacro ; x86-32: ; - m6 is in mem ; x86-64: ; - m8 holds m6 (SWAP) ; m6 holds zero %macro VP9_IDCT8_WRITEOUT 0 %if ARCH_X86_64 %if cpuflag(ssse3) mova m9, [pw_1024] %else mova m9, [pw_16] %endif %define ROUND_REG m9 %else %if cpuflag(ssse3) %define ROUND_REG [pw_1024] %else %define ROUND_REG [pw_16] %endif %endif SCRATCH 5, 10, blockq+16 SCRATCH 7, 11, blockq+32 VP9_IDCT8_WRITEx2 0, 1, 5, 7, 6, ROUND_REG lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 2, 3, 5, 7, 6, ROUND_REG lea dstq, [dstq+2*strideq] UNSCRATCH 5, 10, blockq+16 UNSCRATCH 7, 11, blockq+32 VP9_IDCT8_WRITEx2 4, 5, 0, 1, 6, ROUND_REG lea dstq, [dstq+2*strideq] UNSCRATCH 5, 8, blockq+ 0 VP9_IDCT8_WRITEx2 5, 7, 0, 1, 6, ROUND_REG %undef ROUND_REG %endmacro %macro VP9_IDCT_IDCT_8x8_ADD_XMM 2 INIT_XMM %1 cglobal vp9_idct_idct_8x8_add, 4, 4, %2, dst, stride, block, eob %if cpuflag(ssse3) %if ARCH_X86_64 mova m12, [pw_11585x2] ; often used %define W_11585x2_REG m12 %else %define W_11585x2_REG [pw_11585x2] %endif cmp eobd, 12 ; top left half or less jg .idctfull cmp eobd, 3 ; top left corner or less jg .idcthalf cmp eobd, 1 ; faster path for when only DC is set jne .idcttopleftcorner %else cmp eobd, 1 jg .idctfull %endif %if cpuflag(ssse3) movd m0, [blockq] pmulhrsw m0, W_11585x2_REG pmulhrsw m0, W_11585x2_REG %else DEFINE_ARGS dst, stride, block, coef movsx coefd, word [blockq] imul coefd, 11585 add coefd, 8192 sar coefd, 14 imul coefd, 11585 add coefd, (16 << 14) + 8192 sar coefd, 14 + 5 movd m0, coefd %endif SPLATW m0, m0, 0 pxor m4, m4 movd [blockq], m4 %if cpuflag(ssse3) pmulhrsw m0, [pw_1024] ; (x*1024 + (1<<14))>>15 <=> (x+16)>>5 %endif %rep 3 VP9_STORE_2X 0, 0, 6, 7, 4 lea dstq, [dstq+2*strideq] %endrep VP9_STORE_2X 0, 0, 6, 7, 4 RET %if cpuflag(ssse3) ; faster path for when only left corner is set (3 input: DC, right to DC, below ; to DC). Note: also working with a 2x2 block .idcttopleftcorner: movd m0, [blockq+0] movd m1, [blockq+16] %if ARCH_X86_64 mova m10, [pw_3196x2] mova m11, [pw_16069x2] %define W_3196x2_REG m10 %define W_16069x2_REG m11 %else %define W_3196x2_REG [pw_3196x2] %define W_16069x2_REG [pw_16069x2] %endif VP9_IDCT8_2x2_1D 1 ; partial 2x8 transpose ; punpcklwd m0, m1 already done inside idct punpcklwd m2, m3 punpcklwd m4, m5 punpcklwd m6, m7 punpckldq m0, m2 punpckldq m4, m6 SBUTTERFLY qdq, 0, 4, 1 SWAP 1, 4 VP9_IDCT8_2x2_1D 2 %if ARCH_X86_64 SWAP 6, 8 %endif pxor m6, m6 ; used for the block reset, and VP9_STORE_2X VP9_IDCT8_WRITEOUT %if ARCH_X86_64 movd [blockq+ 0], m6 movd [blockq+16], m6 %else mova [blockq+ 0], m6 mova [blockq+16], m6 mova [blockq+32], m6 %endif RET .idcthalf: movh m0, [blockq + 0] movh m1, [blockq +16] movh m2, [blockq +32] movh m3, [blockq +48] VP9_IDCT8_4x4_1D ; partial 4x8 transpose %if ARCH_X86_32 mova m6, [blockq+ 0] %endif punpcklwd m0, m1 punpcklwd m2, m3 punpcklwd m4, m5 punpcklwd m6, m7 SBUTTERFLY dq, 0, 2, 1 SBUTTERFLY dq, 4, 6, 5 SBUTTERFLY qdq, 0, 4, 1 SBUTTERFLY qdq, 2, 6, 5 SWAP 1, 4 SWAP 3, 6 VP9_IDCT8_4x4_1D %if ARCH_X86_64 SWAP 6, 8 %endif pxor m6, m6 VP9_IDCT8_WRITEOUT %if ARCH_X86_64 movh [blockq+ 0], m6 movh [blockq+16], m6 movh [blockq+32], m6 %else mova [blockq+ 0], m6 mova [blockq+16], m6 mova [blockq+32], m6 %endif movh [blockq+48], m6 RET %endif .idctfull: ; generic full 8x8 idct/idct %if ARCH_X86_64 mova m0, [blockq+ 0] ; IN(0) %endif mova m1, [blockq+ 16] ; IN(1) mova m2, [blockq+ 32] ; IN(2) mova m3, [blockq+ 48] ; IN(3) %if ARCH_X86_64 mova m4, [blockq+ 64] ; IN(4) %endif mova m5, [blockq+ 80] ; IN(5) mova m6, [blockq+ 96] ; IN(6) mova m7, [blockq+112] ; IN(7) %if ARCH_X86_64 mova m11, [pd_8192] ; rounding %define D_8192_REG m11 %else %define D_8192_REG [pd_8192] %endif VP9_IDCT8_1D %if ARCH_X86_64 TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, 8 %else TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, [blockq+0], [blockq+64], 1 mova [blockq+0], m0 %endif VP9_IDCT8_1D %if ARCH_X86_64 SWAP 6, 8 %endif pxor m6, m6 ; used for the block reset, and VP9_STORE_2X VP9_IDCT8_WRITEOUT ZERO_BLOCK blockq, 16, 8, m6 RET %undef W_11585x2_REG %endmacro VP9_IDCT_IDCT_8x8_ADD_XMM sse2, 12 VP9_IDCT_IDCT_8x8_ADD_XMM ssse3, 13 VP9_IDCT_IDCT_8x8_ADD_XMM avx, 13 ;--------------------------------------------------------------------------------------------- ; void vp9_iadst_iadst_8x8_add_<opt>(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob); ;--------------------------------------------------------------------------------------------- ; x86-32: ; - in: m0/3/4/7 are in mem [blockq+N*16] ; - out: m6 is in mem [blockq+0] ; x86-64: ; - everything is in registers %macro VP9_IADST8_1D 0 ; input/output=m0/1/2/3/4/5/6/7 %if ARCH_X86_64 SWAP 0, 8 SWAP 3, 9 SWAP 4, 10 SWAP 7, 11 %endif VP9_UNPACK_MULSUB_2D_4X 5, 2, 0, 3, 14449, 7723 ; m5/2=t3[d], m2/4=t2[d] VP9_UNPACK_MULSUB_2D_4X 1, 6, 4, 7, 4756, 15679 ; m1/4=t7[d], m6/7=t6[d] SCRATCH 4, 12, blockq+1*16 VP9_RND_SH_SUMSUB_BA 6, 2, 7, 3, 4, D_8192_REG ; m6=t2[w], m2=t6[w] UNSCRATCH 4, 12, blockq+1*16 VP9_RND_SH_SUMSUB_BA 1, 5, 4, 0, 3, D_8192_REG ; m1=t3[w], m5=t7[w] UNSCRATCH 0, 8, blockq+16*0 UNSCRATCH 3, 9, blockq+16*3 UNSCRATCH 4, 10, blockq+16*4 UNSCRATCH 7, 11, blockq+16*7 SCRATCH 1, 8, blockq+16*1 SCRATCH 2, 9, blockq+16*2 SCRATCH 5, 10, blockq+16*5 SCRATCH 6, 11, blockq+16*6 VP9_UNPACK_MULSUB_2D_4X 7, 0, 1, 2, 16305, 1606 ; m7/1=t1[d], m0/2=t0[d] VP9_UNPACK_MULSUB_2D_4X 3, 4, 5, 6, 10394, 12665 ; m3/5=t5[d], m4/6=t4[d] SCRATCH 1, 12, blockq+ 0*16 VP9_RND_SH_SUMSUB_BA 4, 0, 6, 2, 1, D_8192_REG ; m4=t0[w], m0=t4[w] UNSCRATCH 1, 12, blockq+ 0*16 VP9_RND_SH_SUMSUB_BA 3, 7, 5, 1, 2, D_8192_REG ; m3=t1[w], m7=t5[w] UNSCRATCH 2, 9, blockq+16*2 UNSCRATCH 5, 10, blockq+16*5 SCRATCH 3, 9, blockq+16*3 SCRATCH 4, 10, blockq+16*4 ; m4=t0, m3=t1, m6=t2, m1=t3, m0=t4, m7=t5, m2=t6, m5=t7 VP9_UNPACK_MULSUB_2D_4X 0, 7, 1, 3, 15137, 6270 ; m0/1=t5[d], m7/3=t4[d] VP9_UNPACK_MULSUB_2D_4X 5, 2, 4, 6, 6270, 15137 ; m5/4=t6[d], m2/6=t7[d] SCRATCH 1, 12, blockq+ 0*16 VP9_RND_SH_SUMSUB_BA 5, 7, 4, 3, 1, D_8192_REG UNSCRATCH 1, 12, blockq+ 0*16 PSIGNW m5, W_M1_REG ; m5=out1[w], m7=t6[w] VP9_RND_SH_SUMSUB_BA 2, 0, 6, 1, 3, D_8192_REG ; m2=out6[w], m0=t7[w] UNSCRATCH 1, 8, blockq+16*1 UNSCRATCH 3, 9, blockq+16*3 UNSCRATCH 4, 10, blockq+16*4 UNSCRATCH 6, 11, blockq+16*6 SCRATCH 2, 8, blockq+16*0 SUMSUB_BA w, 6, 4, 2 ; m6=out0[w], m4=t2[w] SUMSUB_BA w, 1, 3, 2 PSIGNW m1, W_M1_REG ; m1=out7[w], m3=t3[w] ; m6=out0, m5=out1, m4=t2, m3=t3, m7=t6, m0=t7, m2=out6, m1=out7 ; unfortunately, the code below overflows in some cases %if 0; cpuflag(ssse3) SUMSUB_BA w, 3, 4, 2 SUMSUB_BA w, 0, 7, 2 pmulhrsw m3, W_11585x2_REG pmulhrsw m7, W_11585x2_REG pmulhrsw m4, W_11585x2_REG ; out4 pmulhrsw m0, W_11585x2_REG ; out2 %else SCRATCH 5, 9, blockq+16*1 VP9_UNPACK_MULSUB_2W_4X 4, 3, 11585, 11585, D_8192_REG, 2, 5 VP9_UNPACK_MULSUB_2W_4X 7, 0, 11585, 11585, D_8192_REG, 2, 5 UNSCRATCH 5, 9, blockq+16*1 %endif PSIGNW m3, W_M1_REG ; out3 PSIGNW m7, W_M1_REG ; out5 ; m6=out0, m5=out1, m0=out2, m3=out3, m4=out4, m7=out5, m2=out6, m1=out7 %if ARCH_X86_64 SWAP 2, 8 %endif SWAP 0, 6, 2 SWAP 7, 1, 5 %endmacro %macro IADST8_FN 6 INIT_XMM %5 cglobal vp9_%1_%3_8x8_add, 3, 3, %6, dst, stride, block, eob %ifidn %1, idct %define first_is_idct 1 %else %define first_is_idct 0 %endif %ifidn %3, idct %define second_is_idct 1 %else %define second_is_idct 0 %endif %if ARCH_X86_64 mova m0, [blockq+ 0] ; IN(0) %endif mova m1, [blockq+ 16] ; IN(1) mova m2, [blockq+ 32] ; IN(2) %if ARCH_X86_64 || first_is_idct mova m3, [blockq+ 48] ; IN(3) %endif %if ARCH_X86_64 mova m4, [blockq+ 64] ; IN(4) %endif mova m5, [blockq+ 80] ; IN(5) mova m6, [blockq+ 96] ; IN(6) %if ARCH_X86_64 || first_is_idct mova m7, [blockq+112] ; IN(7) %endif %if ARCH_X86_64 %if cpuflag(ssse3) mova m15, [pw_11585x2] ; often used %endif mova m13, [pd_8192] ; rounding mova m14, [pw_m1] %define W_11585x2_REG m15 %define D_8192_REG m13 %define W_M1_REG m14 %else %define W_11585x2_REG [pw_11585x2] %define D_8192_REG [pd_8192] %define W_M1_REG [pw_m1] %endif ; note different calling conventions for idct8 vs. iadst8 on x86-32 VP9_%2_1D %if ARCH_X86_64 TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, 8 %else TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, [blockq+0], [blockq+64], 1 mova [blockq+ 0], m0 %if second_is_idct == 0 mova [blockq+ 48], m3 mova [blockq+112], m7 %endif %endif VP9_%4_1D %if ARCH_X86_64 SWAP 6, 8 %endif pxor m6, m6 ; used for the block reset, and VP9_STORE_2X VP9_IDCT8_WRITEOUT ZERO_BLOCK blockq, 16, 8, m6 RET %undef W_11585x2_REG %undef first_is_idct %undef second_is_idct %endmacro IADST8_FN idct, IDCT8, iadst, IADST8, sse2, 15 IADST8_FN iadst, IADST8, idct, IDCT8, sse2, 15 IADST8_FN iadst, IADST8, iadst, IADST8, sse2, 15 IADST8_FN idct, IDCT8, iadst, IADST8, ssse3, 16 IADST8_FN idct, IDCT8, iadst, IADST8, avx, 16 IADST8_FN iadst, IADST8, idct, IDCT8, ssse3, 16 IADST8_FN iadst, IADST8, idct, IDCT8, avx, 16 IADST8_FN iadst, IADST8, iadst, IADST8, ssse3, 16 IADST8_FN iadst, IADST8, iadst, IADST8, avx, 16 ;--------------------------------------------------------------------------------------------- ; void vp9_idct_idct_16x16_add_<opt>(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob); ;--------------------------------------------------------------------------------------------- ; x86-64: ; at the end of this macro, m7 is stored in [%4+15*%5] ; everything else (t0-6 and t8-15) is stored in m0-6 and m8-15 ; the following sumsubs have not been done yet: ; SUMSUB_BA w, 6, 9, 15 ; t6, t9 ; SUMSUB_BA w, 7, 8, 15 ; t7, t8 ; or (x86-32) t0-t5 are in m0-m5, t10-t15 are in x11/9/7/5/3/1, ; and the following simsubs have not been done yet: ; SUMSUB_BA w, x13, x14, 7 ; t6, t9 ; SUMSUB_BA w, x15, x12, 7 ; t7, t8 %macro VP9_IDCT16_1D_START 6 ; src, nnzc, stride, scratch, scratch_stride, is_iadst %if %2 <= 4 mova m3, [%1+ 1*%3] ; IN(1) mova m0, [%1+ 3*%3] ; IN(3) pmulhrsw m4, m3, [pw_16305x2] ; t14-15 pmulhrsw m3, [pw_1606x2] ; t8-9 pmulhrsw m7, m0, [pw_m4756x2] ; t10-11 pmulhrsw m0, [pw_15679x2] ; t12-13 ; m8=t0, m9=t1, m10=t2, m11=t3, m12=t4, m14=t5, m13=t6, m15=t7 ; m3=t8, m5=t9, m1=t10, m7=t11, m0=t12, m6=t13, m2=t14, m4=t15 VP9_UNPACK_MULSUB_2W_4X 2, 5, 4, 3, 15137, 6270, [pd_8192], 1, 6 ; t9, t14 SCRATCH 4, 10, %4+ 1*%5 SCRATCH 5, 11, %4+ 7*%5 VP9_UNPACK_MULSUB_2W_4X 6, 1, 0, 7, 6270, m15137, [pd_8192], 4, 5 ; t10, t13 UNSCRATCH 5, 11, %4+ 7*%5 ; m15=t0, m14=t1, m13=t2, m12=t3, m11=t4, m10=t5, m9=t6, m8=t7 ; m7=t8, m6=t9, m2=t10, m3=t11, m4=t12, m5=t13, m1=t14, m0=t15 %else mova m5, [%1+ 1*%3] ; IN(1) mova m4, [%1+ 7*%3] ; IN(7) %if %2 <= 8 pmulhrsw m2, m5, [pw_16305x2] ; t15 pmulhrsw m5, [pw_1606x2] ; t8 pmulhrsw m3, m4, [pw_m10394x2] ; t9 pmulhrsw m4, [pw_12665x2] ; t14 %else mova m3, [%1+ 9*%3] ; IN(9) mova m2, [%1+15*%3] ; IN(15) ; m10=in0, m5=in1, m14=in2, m6=in3, m9=in4, m7=in5, m15=in6, m4=in7 ; m11=in8, m3=in9, m12=in10 m0=in11, m8=in12, m1=in13, m13=in14, m2=in15 VP9_UNPACK_MULSUB_2W_4X 5, 2, 16305, 1606, [pd_8192], 0, 1 ; t8, t15 VP9_UNPACK_MULSUB_2W_4X 3, 4, 10394, 12665, [pd_8192], 0, 1 ; t9, t14 %endif SUMSUB_BA w, 3, 5, 0 ; t8, t9 SUMSUB_BA w, 4, 2, 0 ; t15, t14 VP9_UNPACK_MULSUB_2W_4X 2, 5, 15137, 6270, [pd_8192], 0, 1 ; t9, t14 SCRATCH 4, 10, %4+ 1*%5 SCRATCH 5, 11, %4+ 7*%5 mova m6, [%1+ 3*%3] ; IN(3) mova m7, [%1+ 5*%3] ; IN(5) %if %2 <= 8 pmulhrsw m0, m7, [pw_14449x2] ; t13 pmulhrsw m7, [pw_7723x2] ; t10 pmulhrsw m1, m6, [pw_m4756x2] ; t11 pmulhrsw m6, [pw_15679x2] ; t12 %else mova m0, [%1+11*%3] ; IN(11) mova m1, [%1+13*%3] ; IN(13) VP9_UNPACK_MULSUB_2W_4X 7, 0, 14449, 7723, [pd_8192], 4, 5 ; t10, t13 VP9_UNPACK_MULSUB_2W_4X 1, 6, 4756, 15679, [pd_8192], 4, 5 ; t11, t12 %endif ; m11=t0, m10=t1, m9=t2, m8=t3, m14=t4, m12=t5, m15=t6, m13=t7 ; m5=t8, m3=t9, m7=t10, m1=t11, m6=t12, m0=t13, m4=t14, m2=t15 SUMSUB_BA w, 7, 1, 4 ; t11, t10 SUMSUB_BA w, 0, 6, 4 ; t12, t13 ; m8=t0, m9=t1, m10=t2, m11=t3, m12=t4, m14=t5, m13=t6, m15=t7 ; m3=t8, m5=t9, m1=t10, m7=t11, m0=t12, m6=t13, m2=t14, m4=t15 VP9_UNPACK_MULSUB_2W_4X 6, 1, 6270, m15137, [pd_8192], 4, 5 ; t10, t13 UNSCRATCH 5, 11, %4+ 7*%5 %endif ; m8=t0, m9=t1, m10=t2, m11=t3, m12=t4, m13=t5, m14=t6, m15=t7 ; m3=t8, m2=t9, m6=t10, m7=t11, m0=t12, m1=t13, m5=t14, m4=t15 SUMSUB_BA w, 7, 3, 4 ; t8, t11 ; backup first register mova [%4+15*%5], m7 SUMSUB_BA w, 6, 2, 7 ; t9, t10 UNSCRATCH 4, 10, %4+ 1*%5 SUMSUB_BA w, 0, 4, 7 ; t15, t12 SUMSUB_BA w, 1, 5, 7 ; t14. t13 ; m15=t0, m14=t1, m13=t2, m12=t3, m11=t4, m10=t5, m9=t6, m8=t7 ; m7=t8, m6=t9, m2=t10, m3=t11, m4=t12, m5=t13, m1=t14, m0=t15 %if cpuflag(ssse3) && %6 == 0 SUMSUB_BA w, 2, 5, 7 SUMSUB_BA w, 3, 4, 7 pmulhrsw m5, [pw_11585x2] ; t10 pmulhrsw m4, [pw_11585x2] ; t11 pmulhrsw m3, [pw_11585x2] ; t12 pmulhrsw m2, [pw_11585x2] ; t13 %else SCRATCH 6, 10, %4+ 1*%5 VP9_UNPACK_MULSUB_2W_4X 5, 2, 11585, 11585, [pd_8192], 6, 7 ; t10, t13 VP9_UNPACK_MULSUB_2W_4X 4, 3, 11585, 11585, [pd_8192], 6, 7 ; t11, t12 UNSCRATCH 6, 10, %4+ 1*%5 %endif ; m15=t0, m14=t1, m13=t2, m12=t3, m11=t4, m10=t5, m9=t6, m8=t7 ; m7=t8, m6=t9, m5=t10, m4=t11, m3=t12, m2=t13, m1=t14, m0=t15 SCRATCH 0, 8, %4+ 1*%5 SCRATCH 1, 9, %4+ 3*%5 SCRATCH 2, 10, %4+ 5*%5 SCRATCH 3, 11, %4+ 7*%5 SCRATCH 4, 12, %4+ 9*%5 SCRATCH 5, 13, %4+11*%5 SCRATCH 6, 14, %4+13*%5 ; even (tx8x8) %if %2 <= 4 mova m3, [%1+ 0*%3] ; IN(0) mova m4, [%1+ 2*%3] ; IN(2) pmulhrsw m3, [pw_11585x2] ; t0-t3 pmulhrsw m7, m4, [pw_16069x2] ; t6-7 pmulhrsw m4, [pw_3196x2] ; t4-5 %if 0 ; overflows :( paddw m6, m7, m4 psubw m5, m7, m4 pmulhrsw m5, [pw_11585x2] ; t5 pmulhrsw m6, [pw_11585x2] ; t6 %else VP9_UNPACK_MULSUB_2W_4X 5, 6, 7, 4, 11585, 11585, [pd_8192], 0, 1 ; t5, t6 %endif psubw m0, m3, m7 paddw m7, m3 psubw m1, m3, m6 paddw m6, m3 psubw m2, m3, m5 paddw m5, m3 %if ARCH_X86_32 SWAP 0, 7 %endif SCRATCH 7, 15, %4+12*%5 %else mova m6, [%1+ 2*%3] ; IN(2) mova m1, [%1+ 4*%3] ; IN(4) mova m7, [%1+ 6*%3] ; IN(6) %if %2 <= 8 pmulhrsw m0, m1, [pw_15137x2] ; t3 pmulhrsw m1, [pw_6270x2] ; t2 pmulhrsw m5, m6, [pw_16069x2] ; t7 pmulhrsw m6, [pw_3196x2] ; t4 pmulhrsw m4, m7, [pw_m9102x2] ; t5 pmulhrsw m7, [pw_13623x2] ; t6 %else mova m4, [%1+10*%3] ; IN(10) mova m0, [%1+12*%3] ; IN(12) mova m5, [%1+14*%3] ; IN(14) VP9_UNPACK_MULSUB_2W_4X 1, 0, 15137, 6270, [pd_8192], 2, 3 ; t2, t3 VP9_UNPACK_MULSUB_2W_4X 6, 5, 16069, 3196, [pd_8192], 2, 3 ; t4, t7 VP9_UNPACK_MULSUB_2W_4X 4, 7, 9102, 13623, [pd_8192], 2, 3 ; t5, t6 %endif SUMSUB_BA w, 4, 6, 2 ; t4, t5 SUMSUB_BA w, 7, 5, 2 ; t7, t6 %if cpuflag(ssse3) && %6 == 0 SUMSUB_BA w, 6, 5, 2 pmulhrsw m5, [pw_11585x2] ; t5 pmulhrsw m6, [pw_11585x2] ; t6 %else VP9_UNPACK_MULSUB_2W_4X 5, 6, 11585, 11585, [pd_8192], 2, 3 ; t5, t6 %endif SCRATCH 5, 15, %4+10*%5 mova m2, [%1+ 0*%3] ; IN(0) %if %2 <= 8 pmulhrsw m2, [pw_11585x2] ; t0 and t1 psubw m3, m2, m0 paddw m0, m2 SUMSUB_BA w, 7, 0, 5 ; t0, t7 %else mova m3, [%1+ 8*%3] ; IN(8) ; from 3 stages back %if cpuflag(ssse3) && %6 == 0 SUMSUB_BA w, 3, 2, 5 pmulhrsw m3, [pw_11585x2] ; t0 pmulhrsw m2, [pw_11585x2] ; t1 %else mova [%1+ 0*%3], m0 VP9_UNPACK_MULSUB_2W_4X 2, 3, 11585, 11585, [pd_8192], 5, 0 ; t0, t1 mova m0, [%1+ 0*%3] %endif ; from 2 stages back SUMSUB_BA w, 0, 3, 5 ; t0, t3 SUMSUB_BA w, 7, 0, 5 ; t0, t7 %endif UNSCRATCH 5, 15, %4+10*%5 %if ARCH_X86_32 SWAP 0, 7 %endif SCRATCH 7, 15, %4+12*%5 SUMSUB_BA w, 1, 2, 7 ; t1, t2 ; from 1 stage back SUMSUB_BA w, 6, 1, 7 ; t1, t6 SUMSUB_BA w, 5, 2, 7 ; t2, t5 %endif SUMSUB_BA w, 4, 3, 7 ; t3, t4 %if ARCH_X86_64 SWAP 0, 8 SWAP 1, 9 SWAP 2, 10 SWAP 3, 11 SWAP 4, 12 SWAP 5, 13 SWAP 6, 14 SUMSUB_BA w, 0, 15, 7 ; t0, t15 SUMSUB_BA w, 1, 14, 7 ; t1, t14 SUMSUB_BA w, 2, 13, 7 ; t2, t13 SUMSUB_BA w, 3, 12, 7 ; t3, t12 SUMSUB_BA w, 4, 11, 7 ; t4, t11 SUMSUB_BA w, 5, 10, 7 ; t5, t10 %else SWAP 1, 6 SWAP 2, 5 SWAP 3, 4 mova [%4+14*%5], m6 %macro %%SUMSUB_BA_STORE 5 ; reg, from_mem, to_mem, scratch, scratch_stride mova m6, [%4+%2*%5] SUMSUB_BA w, 6, %1, 7 SWAP %1, 6 mova [%4+%3*%5], m6 %endmacro %%SUMSUB_BA_STORE 0, 1, 1, %4, %5 ; t0, t15 %%SUMSUB_BA_STORE 1, 3, 3, %4, %5 ; t1, t14 %%SUMSUB_BA_STORE 2, 5, 5, %4, %5 ; t2, t13 %%SUMSUB_BA_STORE 3, 7, 7, %4, %5 ; t3, t12 %%SUMSUB_BA_STORE 4, 9, 9, %4, %5 ; t4, t11 %%SUMSUB_BA_STORE 5, 11, 11, %4, %5 ; t5, t10 %endif %endmacro %macro VP9_IDCT16_1D 2-4 16, 1 ; src, pass, nnzc, is_iadst %if %2 == 1 VP9_IDCT16_1D_START %1, %3, 32, tmpq, 16, %4 %if ARCH_X86_64 ; backup a different register mova m7, [tmpq+15*16] mova [tmpq+ 1*16], m15 SUMSUB_BA w, 6, 9, 15 ; t6, t9 SUMSUB_BA w, 7, 8, 15 ; t7, t8 TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, 15 mova [tmpq+ 0], m0 mova [tmpq+ 32], m1 mova [tmpq+ 64], m2 mova [tmpq+ 96], m3 mova [tmpq+128], m4 mova [tmpq+160], m5 mova [tmpq+192], m6 mova [tmpq+224], m7 mova m15, [tmpq+ 1*16] TRANSPOSE8x8W 8, 9, 10, 11, 12, 13, 14, 15, 0 mova [tmpq+ 16], m8 mova [tmpq+ 48], m9 mova [tmpq+ 80], m10 mova [tmpq+112], m11 mova [tmpq+144], m12 mova [tmpq+176], m13 mova [tmpq+208], m14 mova [tmpq+240], m15 %else mova m6, [tmpq+13*16] mova m7, [tmpq+14*16] SUMSUB_BA w, 6, 7 ; t6, t9 mova [tmpq+14*16], m6 mova [tmpq+13*16], m7 mova m7, [tmpq+15*16] mova m6, [tmpq+12*16] SUMSUB_BA w, 7, 6 ; t7, t8 mova [tmpq+15*16], m6 TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, [tmpq+14*16], [tmpq+ 8*16], 1 mova [tmpq+ 0*16], m0 mova [tmpq+ 2*16], m1 mova [tmpq+ 4*16], m2 mova [tmpq+ 6*16], m3 mova [tmpq+10*16], m5 mova [tmpq+12*16], m6 mova [tmpq+14*16], m7 mova m0, [tmpq+15*16] mova m1, [tmpq+13*16] mova m2, [tmpq+11*16] mova m3, [tmpq+ 9*16] mova m4, [tmpq+ 7*16] mova m5, [tmpq+ 5*16] mova m7, [tmpq+ 1*16] TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, [tmpq+ 3*16], [tmpq+ 9*16], 1 mova [tmpq+ 1*16], m0 mova [tmpq+ 3*16], m1 mova [tmpq+ 5*16], m2 mova [tmpq+ 7*16], m3 mova [tmpq+11*16], m5 mova [tmpq+13*16], m6 mova [tmpq+15*16], m7 %endif %else ; %2 == 2 VP9_IDCT16_1D_START %1, %3, 32, %1, 32, %4 %if cpuflag(ssse3) %define ROUND_REG [pw_512] %else %define ROUND_REG [pw_32] %endif pxor m7, m7 %if ARCH_X86_64 ; backup more registers mova [%1+ 2*32], m8 mova [%1+ 3*32], m9 VP9_IDCT8_WRITEx2 0, 1, 8, 9, 7, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 2, 3, 8, 9, 7, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 4, 5, 8, 9, 7, ROUND_REG, 6 lea dstq, [dstq+strideq*2] ; restore from cache SWAP 0, 7 ; move zero from m7 to m0 mova m7, [%1+15*32] mova m8, [%1+ 2*32] mova m9, [%1+ 3*32] SUMSUB_BA w, 6, 9, 3 ; t6, t9 SUMSUB_BA w, 7, 8, 3 ; t7, t8 VP9_IDCT8_WRITEx2 6, 7, 3, 4, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 8, 9, 3, 4, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 10, 11, 1, 2, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 12, 13, 1, 2, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 14, 15, 1, 2, 0, ROUND_REG, 6 %else mova [tmpq+ 0*32], m5 VP9_IDCT8_WRITEx2 0, 1, 5, 6, 7, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 2, 3, 5, 6, 7, ROUND_REG, 6 lea dstq, [dstq+strideq*2] SWAP 0, 7 ; move zero from m7 to m0 mova m5, [tmpq+ 0*32] VP9_IDCT8_WRITEx2 4, 5, 1, 2, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m4, [tmpq+13*32] mova m7, [tmpq+14*32] mova m5, [tmpq+15*32] mova m6, [tmpq+12*32] SUMSUB_BADC w, 4, 7, 5, 6, 1 VP9_IDCT8_WRITEx2 4, 5, 1, 2, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 6, 7, 1, 2, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m4, [tmpq+11*32] mova m5, [tmpq+ 9*32] mova m6, [tmpq+ 7*32] mova m7, [tmpq+ 5*32] VP9_IDCT8_WRITEx2 4, 5, 1, 2, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 6, 7, 1, 2, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m4, [tmpq+ 3*32] mova m5, [tmpq+ 1*32] VP9_IDCT8_WRITEx2 4, 5, 1, 2, 0, ROUND_REG, 6 lea dstq, [dstq+strideq*2] %endif %undef ROUND_REG %endif ; %2 == 1/2 %endmacro %macro VP9_STORE_2XFULL 6-7 strideq; dc, tmp1, tmp2, tmp3, tmp4, zero, stride mova m%3, [dstq] mova m%5, [dstq+%7] punpcklbw m%2, m%3, m%6 punpckhbw m%3, m%6 punpcklbw m%4, m%5, m%6 punpckhbw m%5, m%6 paddw m%2, m%1 paddw m%3, m%1 paddw m%4, m%1 paddw m%5, m%1 packuswb m%2, m%3 packuswb m%4, m%5 mova [dstq], m%2 mova [dstq+%7], m%4 %endmacro %macro VP9_IDCT_IDCT_16x16_ADD_XMM 1 INIT_XMM %1 cglobal vp9_idct_idct_16x16_add, 4, 6, 16, 512, dst, stride, block, eob %if cpuflag(ssse3) ; 2x2=eob=3, 4x4=eob=10 cmp eobd, 38 jg .idctfull cmp eobd, 1 ; faster path for when only DC is set jne .idct8x8 %else cmp eobd, 1 ; faster path for when only DC is set jg .idctfull %endif ; dc-only %if cpuflag(ssse3) movd m0, [blockq] mova m1, [pw_11585x2] pmulhrsw m0, m1 pmulhrsw m0, m1 %else DEFINE_ARGS dst, stride, block, coef movsx coefd, word [blockq] imul coefd, 11585 add coefd, 8192 sar coefd, 14 imul coefd, 11585 add coefd, (32 << 14) + 8192 sar coefd, 14 + 6 movd m0, coefd %endif SPLATW m0, m0, q0000 %if cpuflag(ssse3) pmulhrsw m0, [pw_512] %endif pxor m5, m5 movd [blockq], m5 %rep 7 VP9_STORE_2XFULL 0, 1, 2, 3, 4, 5 lea dstq, [dstq+2*strideq] %endrep VP9_STORE_2XFULL 0, 1, 2, 3, 4, 5 RET DEFINE_ARGS dst, stride, block, cnt, dst_bak, tmp %if cpuflag(ssse3) .idct8x8: mov tmpq, rsp VP9_IDCT16_1D blockq, 1, 8, 0 mov cntd, 2 mov dst_bakq, dstq .loop2_8x8: VP9_IDCT16_1D tmpq, 2, 8, 0 lea dstq, [dst_bakq+8] add tmpq, 16 dec cntd jg .loop2_8x8 ; at the end of the loop, m0 should still be zero ; use that to zero out block coefficients ZERO_BLOCK blockq, 32, 8, m0 RET %endif .idctfull: mov cntd, 2 mov tmpq, rsp .loop1_full: VP9_IDCT16_1D blockq, 1, 16, 0 add blockq, 16 add tmpq, 256 dec cntd jg .loop1_full sub blockq, 32 mov cntd, 2 mov tmpq, rsp mov dst_bakq, dstq .loop2_full: VP9_IDCT16_1D tmpq, 2, 16, 0 lea dstq, [dst_bakq+8] add tmpq, 16 dec cntd jg .loop2_full ; at the end of the loop, m0 should still be zero ; use that to zero out block coefficients ZERO_BLOCK blockq, 32, 16, m0 RET %endmacro VP9_IDCT_IDCT_16x16_ADD_XMM sse2 VP9_IDCT_IDCT_16x16_ADD_XMM ssse3 VP9_IDCT_IDCT_16x16_ADD_XMM avx %macro VP9_IDCT16_YMM_1D 0 VP9_UNPACK_MULSUB_2W_4X 1, 15, 16305, 1606, [pd_8192], 0, 4 ; t8, t15 VP9_UNPACK_MULSUB_2W_4X 9, 7, 10394, 12665, [pd_8192], 0, 4 ; t9, t14 SUMSUB_BA w, 9, 1, 0 ; t8, t9 SUMSUB_BA w, 7, 15, 0 ; t15, t14 VP9_UNPACK_MULSUB_2W_4X 15, 1, 15137, 6270, [pd_8192], 0, 4 ; t9, t14 VP9_UNPACK_MULSUB_2W_4X 5, 11, 14449, 7723, [pd_8192], 0, 4 ; t10, t13 VP9_UNPACK_MULSUB_2W_4X 13, 3, 4756, 15679, [pd_8192], 0, 4 ; t11, t12 SUMSUB_BA w, 5, 13, 0 ; t11, t10 SUMSUB_BA w, 11, 3, 0 ; t12, t13 VP9_UNPACK_MULSUB_2W_4X 3, 13, 6270, m15137, [pd_8192], 0, 4 ; t10, t13 SUMSUB_BA w, 5, 9, 0 ; t8, t11 SUMSUB_BA w, 3, 15, 0 ; t9, t10 SUMSUB_BA w, 11, 7, 0 ; t15, t12 SUMSUB_BA w, 13, 1, 0 ; t14, t13 SUMSUB_BA w, 15, 1, 0 SUMSUB_BA w, 9, 7, 0 pmulhrsw m1, [pw_11585x2] ; t10 pmulhrsw m7, [pw_11585x2] ; t11 pmulhrsw m9, [pw_11585x2] ; t12 pmulhrsw m15, [pw_11585x2] ; t13 ; even (tx8x8) mova m4, [blockq+128] mova [blockq+128], m5 VP9_UNPACK_MULSUB_2W_4X 4, 12, 15137, 6270, [pd_8192], 0, 5 ; t2, t3 VP9_UNPACK_MULSUB_2W_4X 2, 14, 16069, 3196, [pd_8192], 0, 5 ; t4, t7 VP9_UNPACK_MULSUB_2W_4X 10, 6, 9102, 13623, [pd_8192], 0, 5 ; t5, t6 mova m0, [blockq+ 0] SUMSUB_BA w, 8, 0, 5 pmulhrsw m8, [pw_11585x2] ; t0 pmulhrsw m0, [pw_11585x2] ; t1 SUMSUB_BA w, 10, 2, 5 ; t4, t5 SUMSUB_BA w, 6, 14, 5 ; t7, t6 SUMSUB_BA w, 12, 8, 5 ; t0, t3 SUMSUB_BA w, 4, 0, 5 ; t1, t2 SUMSUB_BA w, 2, 14, 5 pmulhrsw m14, [pw_11585x2] ; t5 pmulhrsw m2, [pw_11585x2] ; t6 SUMSUB_BA w, 6, 12, 5 ; t0, t7 SUMSUB_BA w, 2, 4, 5 ; t1, t6 SUMSUB_BA w, 14, 0, 5 ; t2, t5 SUMSUB_BA w, 10, 8, 5 ; t3, t4 ; final stage SUMSUB_BA w, 11, 6, 5 ; out0, out15 SUMSUB_BA w, 13, 2, 5 ; out1, out14 SUMSUB_BA w, 15, 14, 5 ; out2, out13 SUMSUB_BA w, 9, 10, 5 ; out3, out12 SUMSUB_BA w, 7, 8, 5 ; out4, out11 SUMSUB_BA w, 1, 0, 5 ; out5, out10 SUMSUB_BA w, 3, 4, 5 ; out6, out9 mova m5, [blockq+128] mova [blockq+192], m3 SUMSUB_BA w, 5, 12, 3 ; out7, out8 SWAP 0, 11, 8, 12, 10 SWAP 1, 13, 14, 2, 15, 6, 3, 9, 4, 7, 5 %endmacro ; this is almost identical to VP9_STORE_2X, but it does two rows ; for slightly improved interleaving, and it omits vpermq since the ; input is DC so all values are identical %macro VP9_STORE_YMM_DC_4X 6 ; reg, tmp1, tmp2, tmp3, tmp4, zero mova xm%2, [dstq] mova xm%4, [dstq+strideq*2] vinserti128 m%2, m%2, [dstq+strideq], 1 vinserti128 m%4, m%4, [dstq+stride3q], 1 punpckhbw m%3, m%2, m%6 punpcklbw m%2, m%6 punpckhbw m%5, m%4, m%6 punpcklbw m%4, m%6 paddw m%3, m%1 paddw m%2, m%1 paddw m%5, m%1 paddw m%4, m%1 packuswb m%2, m%3 packuswb m%4, m%5 mova [dstq], xm%2 mova [dstq+strideq*2], xm%4 vextracti128 [dstq+strideq], m%2, 1 vextracti128 [dstq+stride3q], m%4, 1 %endmacro %if ARCH_X86_64 && HAVE_AVX2_EXTERNAL INIT_YMM avx2 cglobal vp9_idct_idct_16x16_add, 4, 4, 16, dst, stride, block, eob cmp eobd, 1 ; faster path for when only DC is set jg .idctfull ; dc-only mova m1, [pw_11585x2] vpbroadcastw m0, [blockq] pmulhrsw m0, m1 pmulhrsw m0, m1 pxor m5, m5 pmulhrsw m0, [pw_512] movd [blockq], xm5 DEFINE_ARGS dst, stride, stride3, cnt mov cntd, 4 lea stride3q, [strideq*3] .loop_dc: VP9_STORE_YMM_DC_4X 0, 1, 2, 3, 4, 5 lea dstq, [dstq+4*strideq] dec cntd jg .loop_dc RET DEFINE_ARGS dst, stride, block, eob .idctfull: mova m1, [blockq+ 32] mova m2, [blockq+ 64] mova m3, [blockq+ 96] mova m5, [blockq+160] mova m6, [blockq+192] mova m7, [blockq+224] mova m8, [blockq+256] mova m9, [blockq+288] mova m10, [blockq+320] mova m11, [blockq+352] mova m12, [blockq+384] mova m13, [blockq+416] mova m14, [blockq+448] mova m15, [blockq+480] VP9_IDCT16_YMM_1D TRANSPOSE16x16W 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, \ [blockq+192], [blockq+128], 1 mova [blockq+ 0], m0 VP9_IDCT16_YMM_1D mova [blockq+224], m7 ; store VP9_IDCT8_WRITEx2 0, 1, 6, 7, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 2, 3, 6, 7, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 4, 5, 6, 7, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] mova m6, [blockq+192] mova m7, [blockq+224] VP9_IDCT8_WRITEx2 6, 7, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 8, 9, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 10, 11, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 12, 13, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 14, 15, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] ; at the end of the loop, m0 should still be zero ; use that to zero out block coefficients pxor m0, m0 ZERO_BLOCK blockq, 32, 16, m0 RET %endif ;--------------------------------------------------------------------------------------------- ; void vp9_iadst_iadst_16x16_add_<opt>(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob); ;--------------------------------------------------------------------------------------------- %macro VP9_IADST16_1D 2 ; src, pass %assign %%str 16*%2 mova m0, [%1+ 0*32] ; in0 mova m1, [%1+15*32] ; in15 mova m2, [%1+ 7*32] ; in7 mova m3, [%1+ 8*32] ; in8 VP9_UNPACK_MULSUB_2D_4X 1, 0, 4, 5, 16364, 804 ; m1/4=t1[d], m0/5=t0[d] VP9_UNPACK_MULSUB_2D_4X 2, 3, 7, 6, 11003, 12140 ; m2/7=t9[d], m3/6=t8[d] SCRATCH 4, 8, tmpq+ 0*%%str VP9_RND_SH_SUMSUB_BA 3, 0, 6, 5, 4, [pd_8192] ; m3=t0[w], m0=t8[w] UNSCRATCH 4, 8, tmpq+ 0*%%str VP9_RND_SH_SUMSUB_BA 2, 1, 7, 4, 5, [pd_8192] ; m2=t1[w], m1=t9[w] SCRATCH 0, 10, tmpq+ 0*%%str SCRATCH 1, 11, tmpq+15*%%str mova [tmpq+ 7*%%str], m2 mova [tmpq+ 8*%%str], m3 mova m1, [%1+ 2*32] ; in2 mova m0, [%1+13*32] ; in13 mova m3, [%1+ 5*32] ; in5 mova m2, [%1+10*32] ; in10 VP9_UNPACK_MULSUB_2D_4X 0, 1, 6, 7, 15893, 3981 ; m0/6=t3[d], m1/7=t2[d] VP9_UNPACK_MULSUB_2D_4X 3, 2, 4, 5, 8423, 14053 ; m3/4=t11[d], m2/5=t10[d] SCRATCH 4, 12, tmpq+ 2*%%str VP9_RND_SH_SUMSUB_BA 2, 1, 5, 7, 4, [pd_8192] ; m2=t2[w], m1=t10[w] UNSCRATCH 4, 12, tmpq+ 2*%%str VP9_RND_SH_SUMSUB_BA 3, 0, 4, 6, 5, [pd_8192] ; m3=t3[w], m0=t11[w] SCRATCH 0, 12, tmpq+ 2*%%str SCRATCH 1, 13, tmpq+13*%%str mova [tmpq+ 5*%%str], m2 mova [tmpq+10*%%str], m3 mova m2, [%1+ 4*32] ; in4 mova m3, [%1+11*32] ; in11 mova m0, [%1+ 3*32] ; in3 mova m1, [%1+12*32] ; in12 VP9_UNPACK_MULSUB_2D_4X 3, 2, 7, 6, 14811, 7005 ; m3/7=t5[d], m2/6=t4[d] VP9_UNPACK_MULSUB_2D_4X 0, 1, 4, 5, 5520, 15426 ; m0/4=t13[d], m1/5=t12[d] SCRATCH 4, 9, tmpq+ 4*%%str VP9_RND_SH_SUMSUB_BA 1, 2, 5, 6, 4, [pd_8192] ; m1=t4[w], m2=t12[w] UNSCRATCH 4, 9, tmpq+ 4*%%str VP9_RND_SH_SUMSUB_BA 0, 3, 4, 7, 6, [pd_8192] ; m0=t5[w], m3=t13[w] SCRATCH 0, 8, tmpq+ 4*%%str mova [tmpq+11*%%str], m1 ; t4:m1->r11 UNSCRATCH 0, 10, tmpq+ 0*%%str UNSCRATCH 1, 11, tmpq+15*%%str ; round 2 interleaved part 1 VP9_UNPACK_MULSUB_2D_4X 0, 1, 6, 7, 16069, 3196 ; m1/7=t8[d], m0/6=t9[d] VP9_UNPACK_MULSUB_2D_4X 3, 2, 5, 4, 3196, 16069 ; m3/5=t12[d], m2/4=t13[d] SCRATCH 4, 9, tmpq+ 3*%%str VP9_RND_SH_SUMSUB_BA 3, 1, 5, 7, 4, [pd_8192] ; m3=t8[w], m1=t12[w] UNSCRATCH 4, 9, tmpq+ 3*%%str VP9_RND_SH_SUMSUB_BA 2, 0, 4, 6, 5, [pd_8192] ; m2=t9[w], m0=t13[w] SCRATCH 0, 10, tmpq+ 0*%%str SCRATCH 1, 11, tmpq+15*%%str SCRATCH 2, 14, tmpq+ 3*%%str SCRATCH 3, 15, tmpq+12*%%str mova m2, [%1+ 6*32] ; in6 mova m3, [%1+ 9*32] ; in9 mova m0, [%1+ 1*32] ; in1 mova m1, [%1+14*32] ; in14 VP9_UNPACK_MULSUB_2D_4X 3, 2, 7, 6, 13160, 9760 ; m3/7=t7[d], m2/6=t6[d] VP9_UNPACK_MULSUB_2D_4X 0, 1, 4, 5, 2404, 16207 ; m0/4=t15[d], m1/5=t14[d] SCRATCH 4, 9, tmpq+ 6*%%str VP9_RND_SH_SUMSUB_BA 1, 2, 5, 6, 4, [pd_8192] ; m1=t6[w], m2=t14[w] UNSCRATCH 4, 9, tmpq+ 6*%%str VP9_RND_SH_SUMSUB_BA 0, 3, 4, 7, 6, [pd_8192] ; m0=t7[w], m3=t15[w] ; r8=t0, r7=t1, r5=t2, r10=t3, r11=t4, m8|r4=t5, m1=t6, m0=t7 ; m10|r0=t8, m11|r15=t9, m13|r13=t10, m12|r2=t11, m14|r3=t12, m15|r12=t13, m2=t14, m3=t15 UNSCRATCH 4, 12, tmpq+ 2*%%str UNSCRATCH 5, 13, tmpq+13*%%str SCRATCH 0, 12, tmpq+ 1*%%str SCRATCH 1, 13, tmpq+14*%%str ; remainder of round 2 (rest of t8-15) VP9_UNPACK_MULSUB_2D_4X 5, 4, 6, 7, 9102, 13623 ; m5/6=t11[d], m4/7=t10[d] VP9_UNPACK_MULSUB_2D_4X 3, 2, 1, 0, 13623, 9102 ; m3/1=t14[d], m2/0=t15[d] SCRATCH 0, 9, tmpq+ 6*%%str VP9_RND_SH_SUMSUB_BA 3, 4, 1, 7, 0, [pd_8192] ; m3=t10[w], m4=t14[w] UNSCRATCH 0, 9, tmpq+ 6*%%str VP9_RND_SH_SUMSUB_BA 2, 5, 0, 6, 1, [pd_8192] ; m2=t11[w], m5=t15[w] ; m15|r12=t8, m14|r3=t9, m3=t10, m2=t11, m11|r15=t12, m10|r0=t13, m4=t14, m5=t15 UNSCRATCH 6, 14, tmpq+ 3*%%str UNSCRATCH 7, 15, tmpq+12*%%str SUMSUB_BA w, 3, 7, 1 PSIGNW m3, [pw_m1] ; m3=out1[w], m7=t10[w] SUMSUB_BA w, 2, 6, 1 ; m2=out14[w], m6=t11[w] ; unfortunately, the code below overflows in some cases, e.g. ; http://downloads.webmproject.org/test_data/libvpx/vp90-2-14-resize-fp-tiles-16-8.webm %if 0; cpuflag(ssse3) SUMSUB_BA w, 7, 6, 1 pmulhrsw m7, [pw_11585x2] ; m7=out6[w] pmulhrsw m6, [pw_11585x2] ; m6=out9[w] %else VP9_UNPACK_MULSUB_2W_4X 6, 7, 11585, 11585, [pd_8192], 1, 0 %endif mova [tmpq+ 3*%%str], m6 mova [tmpq+ 6*%%str], m7 UNSCRATCH 6, 10, tmpq+ 0*%%str UNSCRATCH 7, 11, tmpq+15*%%str mova [tmpq+13*%%str], m2 SCRATCH 3, 11, tmpq+ 9*%%str VP9_UNPACK_MULSUB_2D_4X 7, 6, 2, 3, 15137, 6270 ; m6/3=t13[d], m7/2=t12[d] VP9_UNPACK_MULSUB_2D_4X 5, 4, 1, 0, 6270, 15137 ; m5/1=t14[d], m4/0=t15[d] SCRATCH 0, 9, tmpq+ 2*%%str VP9_RND_SH_SUMSUB_BA 5, 6, 1, 3, 0, [pd_8192] ; m5=out2[w], m6=t14[w] UNSCRATCH 0, 9, tmpq+ 2*%%str VP9_RND_SH_SUMSUB_BA 4, 7, 0, 2, 1, [pd_8192] PSIGNW m4, [pw_m1] ; m4=out13[w], m7=t15[w] ; unfortunately, the code below overflows in some cases %if 0; cpuflag(ssse3) SUMSUB_BA w, 7, 6, 1 pmulhrsw m7, [pw_m11585x2] ; m7=out5[w] pmulhrsw m6, [pw_11585x2] ; m6=out10[w] %else PSIGNW m7, [pw_m1] VP9_UNPACK_MULSUB_2W_4X 7, 6, 11585, 11585, [pd_8192], 1, 0 %endif ; m11|r13=out1, m5=out2, m7=out5, r15=out6, r3=out9, m6=out10, m4=out13, r2=out14 mova m2, [tmpq+ 8*%%str] mova m3, [tmpq+ 7*%%str] mova m1, [tmpq+11*%%str] mova [tmpq+ 7*%%str], m6 mova [tmpq+11*%%str], m4 mova m4, [tmpq+ 5*%%str] SCRATCH 5, 14, tmpq+ 5*%%str SCRATCH 7, 15, tmpq+ 8*%%str UNSCRATCH 6, 8, tmpq+ 4*%%str UNSCRATCH 5, 12, tmpq+ 1*%%str UNSCRATCH 7, 13, tmpq+14*%%str ; m2=t0, m3=t1, m9=t2, m0=t3, m1=t4, m8=t5, m13=t6, m12=t7 ; m11|r13=out1, m5=out2, m7=out5, r15=out6, r3=out9, r10=out10, r11=out13, r2=out14 SUMSUB_BA w, 1, 2, 0 ; m1=t0[w], m2=t4[w] mova m0, [tmpq+10*%%str] SCRATCH 1, 12, tmpq+ 1*%%str SUMSUB_BA w, 6, 3, 1 ; m8=t1[w], m3=t5[w] SCRATCH 6, 13, tmpq+ 4*%%str SUMSUB_BA w, 7, 4, 1 ; m13=t2[w], m9=t6[w] SCRATCH 7, 8, tmpq+10*%%str SUMSUB_BA w, 5, 0, 1 ; m12=t3[w], m0=t7[w] SCRATCH 5, 9, tmpq+14*%%str VP9_UNPACK_MULSUB_2D_4X 2, 3, 7, 5, 15137, 6270 ; m2/6=t5[d], m3/10=t4[d] VP9_UNPACK_MULSUB_2D_4X 0, 4, 1, 6, 6270, 15137 ; m0/14=t6[d], m9/15=t7[d] SCRATCH 6, 10, tmpq+ 0*%%str VP9_RND_SH_SUMSUB_BA 0, 3, 1, 5, 6, [pd_8192] UNSCRATCH 6, 10, tmpq+ 0*%%str PSIGNW m0, [pw_m1] ; m0=out3[w], m3=t6[w] VP9_RND_SH_SUMSUB_BA 4, 2, 6, 7, 5, [pd_8192] ; m9=out12[w], m2=t7[w] UNSCRATCH 1, 8, tmpq+10*%%str UNSCRATCH 5, 9, tmpq+14*%%str UNSCRATCH 6, 12, tmpq+ 1*%%str UNSCRATCH 7, 13, tmpq+ 4*%%str SCRATCH 4, 9, tmpq+14*%%str SUMSUB_BA w, 1, 6, 4 ; m13=out0[w], m1=t2[w] SUMSUB_BA w, 5, 7, 4 PSIGNW m5, [pw_m1] ; m12=out15[w], m8=t3[w] ; unfortunately, the code below overflows in some cases, e.g. ; http://downloads.webmproject.org/test_data/libvpx/vp90-2-14-resize-fp-tiles-16-8-4-2-1.webm %if 0 ; cpuflag(ssse3) SUMSUB_BA w, 7, 6, 4 pmulhrsw m7, [pw_m11585x2] ; m8=out7[w] pmulhrsw m6, [pw_11585x2] ; m1=out8[w] SWAP 6, 7 SUMSUB_BA w, 3, 2, 4 pmulhrsw m3, [pw_11585x2] ; m3=out4[w] pmulhrsw m2, [pw_11585x2] ; m2=out11[w] %else SCRATCH 5, 8, tmpq+10*%%str VP9_UNPACK_MULSUB_2W_4X 6, 7, 11585, m11585, [pd_8192], 5, 4 VP9_UNPACK_MULSUB_2W_4X 2, 3, 11585, 11585, [pd_8192], 5, 4 UNSCRATCH 5, 8, tmpq+10*%%str %endif ; m13=out0, m0=out3, m3=out4, m8=out7, m1=out8, m2=out11, m9=out12, m12=out15 ; m11|r13=out1, m5=out2, m7=out5, r15=out6, r3=out9, r10=out10, r11=out13, r2=out14 %if %2 == 1 %if ARCH_X86_64 mova m13, [tmpq+ 6*%%str] TRANSPOSE8x8W 1, 11, 14, 0, 3, 15, 13, 6, 10 mova [tmpq+ 0*16], m1 mova [tmpq+ 2*16], m11 mova [tmpq+ 4*16], m14 mova [tmpq+ 6*16], m0 mova m1, [tmpq+ 3*%%str] mova m11, [tmpq+ 7*%%str] mova m14, [tmpq+11*%%str] mova m0, [tmpq+13*%%str] mova [tmpq+ 8*16], m3 mova [tmpq+10*16], m15 mova [tmpq+12*16], m13 mova [tmpq+14*16], m6 TRANSPOSE8x8W 7, 1, 11, 2, 9, 14, 0, 5, 10 mova [tmpq+ 1*16], m7 mova [tmpq+ 3*16], m1 mova [tmpq+ 5*16], m11 mova [tmpq+ 7*16], m2 mova [tmpq+ 9*16], m9 mova [tmpq+11*16], m14 mova [tmpq+13*16], m0 mova [tmpq+15*16], m5 %else mova [tmpq+12*%%str], m2 mova [tmpq+ 1*%%str], m5 mova [tmpq+15*%%str], m7 mova m2, [tmpq+ 9*%%str] mova m5, [tmpq+ 5*%%str] mova m7, [tmpq+ 8*%%str] TRANSPOSE8x8W 1, 2, 5, 0, 3, 7, 4, 6, [tmpq+ 6*%%str], [tmpq+ 8*%%str], 1 mova [tmpq+ 0*16], m1 mova [tmpq+ 2*16], m2 mova [tmpq+ 4*16], m5 mova [tmpq+ 6*16], m0 mova [tmpq+10*16], m7 mova m3, [tmpq+12*%%str] mova [tmpq+12*16], m4 mova m4, [tmpq+14*%%str] mova [tmpq+14*16], m6 mova m0, [tmpq+15*%%str] mova m1, [tmpq+ 3*%%str] mova m2, [tmpq+ 7*%%str] mova m5, [tmpq+11*%%str] mova m7, [tmpq+ 1*%%str] TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, [tmpq+13*%%str], [tmpq+ 9*%%str], 1 mova [tmpq+ 1*16], m0 mova [tmpq+ 3*16], m1 mova [tmpq+ 5*16], m2 mova [tmpq+ 7*16], m3 mova [tmpq+11*16], m5 mova [tmpq+13*16], m6 mova [tmpq+15*16], m7 %endif %else pxor m4, m4 %if cpuflag(ssse3) %define ROUND_REG [pw_512] %else %define ROUND_REG [pw_32] %endif %if ARCH_X86_64 mova m12, [tmpq+ 6*%%str] VP9_IDCT8_WRITEx2 1, 11, 10, 8, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 14, 0, 10, 8, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 3, 15, 10, 8, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 12, 6, 10, 8, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m1, [tmpq+ 3*%%str] mova m11, [tmpq+ 7*%%str] mova m14, [tmpq+11*%%str] mova m0, [tmpq+13*%%str] VP9_IDCT8_WRITEx2 7, 1, 10, 8, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 11, 2, 10, 8, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 9, 14, 10, 8, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] VP9_IDCT8_WRITEx2 0, 5, 10, 8, 4, ROUND_REG, 6 %else mova [tmpq+ 0*%%str], m2 mova [tmpq+ 1*%%str], m5 mova [tmpq+ 2*%%str], m7 mova m2, [tmpq+ 9*%%str] VP9_IDCT8_WRITEx2 1, 2, 5, 7, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m5, [tmpq+ 5*%%str] VP9_IDCT8_WRITEx2 5, 0, 1, 2, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m5, [tmpq+ 8*%%str] VP9_IDCT8_WRITEx2 3, 5, 1, 2, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m5, [tmpq+ 6*%%str] VP9_IDCT8_WRITEx2 5, 6, 1, 2, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m0, [tmpq+ 2*%%str] mova m3, [tmpq+ 3*%%str] VP9_IDCT8_WRITEx2 0, 3, 1, 2, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m0, [tmpq+ 7*%%str] mova m3, [tmpq+ 0*%%str] VP9_IDCT8_WRITEx2 0, 3, 1, 2, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m0, [tmpq+14*%%str] mova m3, [tmpq+11*%%str] VP9_IDCT8_WRITEx2 0, 3, 1, 2, 4, ROUND_REG, 6 lea dstq, [dstq+strideq*2] mova m0, [tmpq+13*%%str] mova m3, [tmpq+ 1*%%str] VP9_IDCT8_WRITEx2 0, 3, 1, 2, 4, ROUND_REG, 6 %endif SWAP 0, 4 ; zero %undef ROUND_REG %endif %endmacro %macro IADST16_FN 5 INIT_XMM %5 cglobal vp9_%1_%3_16x16_add, 3, 6, 16, 512, dst, stride, block, cnt, dst_bak, tmp mov cntd, 2 mov tmpq, rsp .loop1_full: VP9_%2_1D blockq, 1 add blockq, 16 add tmpq, 256 dec cntd jg .loop1_full sub blockq, 32 mov cntd, 2 mov tmpq, rsp mov dst_bakq, dstq .loop2_full: VP9_%4_1D tmpq, 2 lea dstq, [dst_bakq+8] add tmpq, 16 dec cntd jg .loop2_full ; at the end of the loop, m0 should still be zero ; use that to zero out block coefficients ZERO_BLOCK blockq, 32, 16, m0 RET %endmacro IADST16_FN idct, IDCT16, iadst, IADST16, sse2 IADST16_FN iadst, IADST16, idct, IDCT16, sse2 IADST16_FN iadst, IADST16, iadst, IADST16, sse2 IADST16_FN idct, IDCT16, iadst, IADST16, ssse3 IADST16_FN iadst, IADST16, idct, IDCT16, ssse3 IADST16_FN iadst, IADST16, iadst, IADST16, ssse3 IADST16_FN idct, IDCT16, iadst, IADST16, avx IADST16_FN iadst, IADST16, idct, IDCT16, avx IADST16_FN iadst, IADST16, iadst, IADST16, avx ; in: data in m[0-15] except m0/m4, which are in [blockq+0] and [blockq+128] ; out: m[0-15] except m6, which is in [blockq+192] ; uses blockq as scratch space %macro VP9_IADST16_YMM_1D 0 mova [blockq+ 32], m3 mova [blockq+ 64], m7 mova [blockq+ 96], m8 ; first half of round 1 VP9_UNPACK_MULSUB_2D_4X 9, 6, 0, 3, 13160, 9760 ; m9/x=t7[d], m6/x=t6[d] VP9_UNPACK_MULSUB_2D_4X 1, 14, 4, 7, 2404, 16207 ; m1/x=t15[d], m14/x=t14[d] VP9_RND_SH_SUMSUB_BA 14, 6, 7, 3, 8, [pd_8192] ; m14=t6[w], m6=t14[w] VP9_RND_SH_SUMSUB_BA 1, 9, 4, 0, 8, [pd_8192] ; m1=t7[w], m9=t15[w] VP9_UNPACK_MULSUB_2D_4X 13, 2, 4, 7, 15893, 3981 ; m13/x=t3[d], m2/x=t2[d] VP9_UNPACK_MULSUB_2D_4X 5, 10, 0, 3, 8423, 14053 ; m5/x=t11[d], m10/x=t10[d] VP9_RND_SH_SUMSUB_BA 10, 2, 3, 7, 8, [pd_8192] ; m10=t2[w], m2=t10[w] VP9_RND_SH_SUMSUB_BA 5, 13, 0, 4, 8, [pd_8192] ; m5=t3[w], m13=t11[w] ; half of round 2 t8-15 VP9_UNPACK_MULSUB_2D_4X 2, 13, 4, 7, 9102, 13623 ; m2/x=t11[d], m13/x=t10[d] VP9_UNPACK_MULSUB_2D_4X 9, 6, 3, 0, 13623, 9102 ; m9/x=t14[d], m6/x=t15[d] VP9_RND_SH_SUMSUB_BA 9, 13, 3, 7, 8, [pd_8192] ; m9=t10[w], m13=t14[w] VP9_RND_SH_SUMSUB_BA 6, 2, 0, 4, 8, [pd_8192] ; m6=t11[w], m2=t15[w] SUMSUB_BA w, 14, 10, 8 ; m14=t2, m10=t6 SUMSUB_BA w, 1, 5, 8 ; m1=t3, m5=t7 mova m0, [blockq+ 0] mova m4, [blockq+128] mova m3, [blockq+ 32] mova m7, [blockq+ 64] mova m8, [blockq+ 96] mova [blockq+ 0], m1 mova [blockq+128], m14 mova [blockq+ 32], m6 mova [blockq+ 64], m9 mova [blockq+ 96], m10 ; second half of round 1 VP9_UNPACK_MULSUB_2D_4X 15, 0, 1, 9, 16364, 804 ; m15/x=t1[d], m0/x=t0[d] VP9_UNPACK_MULSUB_2D_4X 7, 8, 10, 6, 11003, 12140 ; m7/x=t9[d], m8/x=t8[d] VP9_RND_SH_SUMSUB_BA 8, 0, 6, 9, 14, [pd_8192] ; m8=t0[w], m0=t8[w] VP9_RND_SH_SUMSUB_BA 7, 15, 10, 1, 14, [pd_8192] ; m7=t1[w], m15=t9[w] VP9_UNPACK_MULSUB_2D_4X 11, 4, 10, 6, 14811, 7005 ; m11/x=t5[d], m4/x=t4[d] VP9_UNPACK_MULSUB_2D_4X 3, 12, 1, 9, 5520, 15426 ; m3/x=t13[d], m12/x=t12[d] VP9_RND_SH_SUMSUB_BA 12, 4, 9, 6, 14, [pd_8192] ; m12=t4[w], m4=t12[w] VP9_RND_SH_SUMSUB_BA 3, 11, 1, 10, 14, [pd_8192] ; m3=t5[w], m11=t13[w] ; second half of round 2 t8-15 VP9_UNPACK_MULSUB_2D_4X 0, 15, 6, 10, 16069, 3196 ; m15/x=t8[d], m0/x=t9[d] VP9_UNPACK_MULSUB_2D_4X 11, 4, 9, 1, 3196, 16069 ; m11/x=t12[d], m4/x=t13[d] VP9_RND_SH_SUMSUB_BA 11, 15, 9, 10, 14, [pd_8192] ; m11=t8[w], m15=t12[w] VP9_RND_SH_SUMSUB_BA 4, 0, 1, 6, 14, [pd_8192] ; m4=t9[w], m0=t13[w] SUMSUB_BA w, 12, 8, 14 ; m12=t0, m8=t4 SUMSUB_BA w, 3, 7, 14 ; m3=t1, m7=t5 mova m10, [blockq+ 96] mova [blockq+ 96], m12 ; round 3 VP9_UNPACK_MULSUB_2D_4X 15, 0, 9, 12, 15137, 6270 ; m15/x=t13[d], m0/x=t12[d] VP9_UNPACK_MULSUB_2D_4X 2, 13, 1, 6, 6270, 15137 ; m2/x=t14[d], m13/x=t15[d] VP9_RND_SH_SUMSUB_BA 2, 0, 1, 12, 14, [pd_8192] ; m2=out2[w], m0=t14a[w] VP9_RND_SH_SUMSUB_BA 13, 15, 6, 9, 14, [pd_8192] PSIGNW m13, [pw_m1] ; m13=out13[w], m15=t15a[w] VP9_UNPACK_MULSUB_2D_4X 8, 7, 12, 9, 15137, 6270 ; m8/x=t5[d], m7/x=t4[d] VP9_UNPACK_MULSUB_2D_4X 5, 10, 1, 6, 6270, 15137 ; m5/x=t6[d], m10/x=t7[d] VP9_RND_SH_SUMSUB_BA 5, 7, 1, 9, 14, [pd_8192] PSIGNW m5, [pw_m1] ; m5=out3[w], m7=t6[w] VP9_RND_SH_SUMSUB_BA 10, 8, 6, 12, 14, [pd_8192] ; m10=out12[w], m8=t7[w] mova m1, [blockq+ 0] mova m14, [blockq+128] mova m6, [blockq+ 32] mova m9, [blockq+ 64] mova m12, [blockq+ 96] mova [blockq+ 0], m10 mova [blockq+128], m5 SUMSUB_BA w, 14, 12, 5 ; m14=out0, m12=t2a SUMSUB_BA w, 1, 3, 5 PSIGNW m1, [pw_m1] ; m1=out15, m3=t3a SUMSUB_BA w, 9, 11, 5 PSIGNW m9, [pw_m1] ; m9=out1, m11=t10 SUMSUB_BA w, 6, 4, 5 ; m6=out14, m4=t11 VP9_UNPACK_MULSUB_2W_4X 4, 11, 11585, 11585, [pd_8192], 5, 10 ; m4=out9, m11=out6 mova m5, [blockq+128] mova [blockq+192], m11 PSIGNW m15, [pw_m1] VP9_UNPACK_MULSUB_2W_4X 15, 0, 11585, 11585, [pd_8192], 10, 11 ; m15=out5, m0=out10 PSIGNW m3, [pw_m1] VP9_UNPACK_MULSUB_2W_4X 3, 12, 11585, 11585, [pd_8192], 10, 11 ; m3=out7,m12=out8 VP9_UNPACK_MULSUB_2W_4X 8, 7, 11585, 11585, [pd_8192], 10, 11 ; m8=out11,m7=out4 mova m10, [blockq+ 0] SWAP 0, 14, 6, 11, 8, 12, 10 SWAP 1, 9, 15, 4, 7, 3, 5 SWAP 5, 9, 15 %endmacro %if ARCH_X86_64 && HAVE_AVX2_EXTERNAL %macro IADST16_YMM_FN 4 INIT_YMM avx2 cglobal vp9_%1_%3_16x16_add, 4, 4, 16, dst, stride, block, eob mova m1, [blockq+ 32] mova m2, [blockq+ 64] mova m3, [blockq+ 96] mova m5, [blockq+160] mova m6, [blockq+192] mova m7, [blockq+224] mova m8, [blockq+256] mova m9, [blockq+288] mova m10, [blockq+320] mova m11, [blockq+352] mova m12, [blockq+384] mova m13, [blockq+416] mova m14, [blockq+448] mova m15, [blockq+480] VP9_%2_YMM_1D TRANSPOSE16x16W 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, \ [blockq+192], [blockq+128], 1 mova [blockq+ 0], m0 VP9_%4_YMM_1D mova [blockq+224], m7 ; store VP9_IDCT8_WRITEx2 0, 1, 6, 7, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 2, 3, 6, 7, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 4, 5, 6, 7, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] mova m6, [blockq+192] mova m7, [blockq+224] VP9_IDCT8_WRITEx2 6, 7, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 8, 9, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 10, 11, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 12, 13, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] VP9_IDCT8_WRITEx2 14, 15, 1, 2, unused, [pw_512], 6 lea dstq, [dstq+2*strideq] ; at the end of the loop, m0 should still be zero ; use that to zero out block coefficients pxor m0, m0 ZERO_BLOCK blockq, 32, 16, m0 RET %endmacro IADST16_YMM_FN idct, IDCT16, iadst, IADST16 IADST16_YMM_FN iadst, IADST16, idct, IDCT16 IADST16_YMM_FN iadst, IADST16, iadst, IADST16 %endif ;--------------------------------------------------------------------------------------------- ; void vp9_idct_idct_32x32_add_<opt>(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob); ;--------------------------------------------------------------------------------------------- %macro VP9_IDCT32_1D 2-3 32 ; src, pass, nnzc %if %2 == 1 %assign %%str mmsize %else %assign %%str 64 %endif ; first do t0-15, this can be done identical to idct16x16 VP9_IDCT16_1D_START %1, %3/2, 64*2, tmpq, 2*%%str, 1 ; store everything on stack to make space available for t16-31 ; we store interleaved with the output of the second half (t16-31) ; so we don't need to allocate extra stack space mova [tmpq+ 0*%%str], m0 ; t0 mova [tmpq+ 4*%%str], m1 ; t1 mova [tmpq+ 8*%%str], m2 ; t2 mova [tmpq+12*%%str], m3 ; t3 mova [tmpq+16*%%str], m4 ; t4 mova [tmpq+20*%%str], m5 ; t5 %if ARCH_X86_64 mova [tmpq+22*%%str], m10 ; t10 mova [tmpq+18*%%str], m11 ; t11 mova [tmpq+14*%%str], m12 ; t12 mova [tmpq+10*%%str], m13 ; t13 mova [tmpq+ 6*%%str], m14 ; t14 mova [tmpq+ 2*%%str], m15 ; t15 %endif mova m0, [tmpq+ 30*%%str] UNSCRATCH 1, 6, tmpq+26*%%str UNSCRATCH 2, 8, tmpq+24*%%str UNSCRATCH 3, 9, tmpq+28*%%str SUMSUB_BA w, 1, 3, 4 ; t6, t9 SUMSUB_BA w, 0, 2, 4 ; t7, t8 mova [tmpq+24*%%str], m1 ; t6 mova [tmpq+28*%%str], m0 ; t7 mova [tmpq+30*%%str], m2 ; t8 mova [tmpq+26*%%str], m3 ; t9 ; then, secondly, do t16-31 %if %3 <= 8 mova m4, [%1+ 1*64] mova m7, [%1+ 7*64] pmulhrsw m1, m4, [pw_16364x2] ;t31 pmulhrsw m4, [pw_804x2] ;t16 VP9_UNPACK_MULSUB_2W_4X 5, 0, 1, 4, 16069, 3196, [pd_8192], 6, 2 ; t17, t30 pmulhrsw m3, m7, [pw_m5520x2] ;t19 pmulhrsw m7, [pw_15426x2] ;t28 SCRATCH 4, 13, tmpq+ 1*%%str SCRATCH 5, 12, tmpq+15*%%str VP9_UNPACK_MULSUB_2W_4X 2, 6, 7, 3, 3196, m16069, [pd_8192], 4, 5 ; t18, t29 %else mova m0, [%1+ 1*64] mova m1, [%1+15*64] %if %3 <= 16 pmulhrsw m5, m0, [pw_16364x2] pmulhrsw m0, [pw_804x2] pmulhrsw m4, m1, [pw_m11003x2] pmulhrsw m1, [pw_12140x2] %else mova m4, [%1+17*64] mova m5, [%1+31*64] VP9_UNPACK_MULSUB_2W_4X 0, 5, 16364, 804, [pd_8192], 2, 3 ; t16, t31 VP9_UNPACK_MULSUB_2W_4X 4, 1, 11003, 12140, [pd_8192], 2, 3 ; t17, t30 %endif SUMSUB_BA w, 4, 0, 2 SUMSUB_BA w, 1, 5, 2 VP9_UNPACK_MULSUB_2W_4X 5, 0, 16069, 3196, [pd_8192], 2, 3 ; t17, t30 SCRATCH 4, 13, tmpq+ 1*%%str SCRATCH 5, 12, tmpq+15*%%str mova m2, [%1+ 7*64] mova m3, [%1+ 9*64] %if %3 <= 16 pmulhrsw m7, m3, [pw_14811x2] pmulhrsw m3, [pw_7005x2] pmulhrsw m6, m2, [pw_m5520x2] pmulhrsw m2, [pw_15426x2] %else mova m7, [%1+23*64] mova m6, [%1+25*64] VP9_UNPACK_MULSUB_2W_4X 3, 7, 14811, 7005, [pd_8192], 4, 5 ; t18, t29 VP9_UNPACK_MULSUB_2W_4X 6, 2, 5520, 15426, [pd_8192], 4, 5 ; t19, t28 %endif SUMSUB_BA w, 3, 6, 4 SUMSUB_BA w, 7, 2, 4 VP9_UNPACK_MULSUB_2W_4X 2, 6, 3196, m16069, [pd_8192], 4, 5 ; t18, t29 %endif UNSCRATCH 5, 12, tmpq+15*%%str SUMSUB_BA w, 6, 0, 4 mova [tmpq+25*%%str], m6 ; t19 UNSCRATCH 4, 13, tmpq+ 1*%%str SUMSUB_BA w, 7, 1, 6 SUMSUB_BA w, 3, 4, 6 mova [tmpq+23*%%str], m3 ; t16 SUMSUB_BA w, 2, 5, 6 VP9_UNPACK_MULSUB_2W_4X 0, 5, 15137, 6270, [pd_8192], 6, 3 ; t18, t29 VP9_UNPACK_MULSUB_2W_4X 1, 4, 15137, 6270, [pd_8192], 6, 3 ; t19, t28 SCRATCH 0, 10, tmpq+ 1*%%str SCRATCH 1, 11, tmpq+ 7*%%str SCRATCH 2, 9, tmpq+ 9*%%str SCRATCH 4, 14, tmpq+15*%%str SCRATCH 5, 15, tmpq+17*%%str SCRATCH 7, 13, tmpq+31*%%str %if %3 <= 8 mova m0, [%1+ 5*64] mova m3, [%1+ 3*64] pmulhrsw m5, m0, [pw_15893x2] ;t27 pmulhrsw m0, [pw_3981x2] ;t20 VP9_UNPACK_MULSUB_2W_4X 1, 4, 5, 0, 9102, 13623, [pd_8192], 7, 2 ; t21, t26 pmulhrsw m6, m3, [pw_m2404x2] ;t23 pmulhrsw m3, [pw_16207x2] ;t24 SCRATCH 5, 8, tmpq+ 5*%%str SCRATCH 4, 12, tmpq+11*%%str VP9_UNPACK_MULSUB_2W_4X 7, 2, 3, 6, 13623, m9102, [pd_8192], 4, 5 ; t22, t25 %else mova m4, [%1+ 5*64] mova m5, [%1+11*64] %if %3 <= 16 pmulhrsw m1, m4, [pw_15893x2] pmulhrsw m4, [pw_3981x2] pmulhrsw m0, m5, [pw_m8423x2] pmulhrsw m5, [pw_14053x2] %else mova m0, [%1+21*64] mova m1, [%1+27*64] VP9_UNPACK_MULSUB_2W_4X 4, 1, 15893, 3981, [pd_8192], 2, 3 ; t20, t27 VP9_UNPACK_MULSUB_2W_4X 0, 5, 8423, 14053, [pd_8192], 2, 3 ; t21, t26 %endif SUMSUB_BA w, 0, 4, 2 SUMSUB_BA w, 5, 1, 2 VP9_UNPACK_MULSUB_2W_4X 1, 4, 9102, 13623, [pd_8192], 2, 3 ; t21, t26 SCRATCH 5, 8, tmpq+ 5*%%str SCRATCH 4, 12, tmpq+11*%%str mova m7, [%1+ 3*64] mova m6, [%1+13*64] %if %3 <= 16 pmulhrsw m3, m6, [pw_13160x2] pmulhrsw m6, [pw_9760x2] pmulhrsw m2, m7, [pw_m2404x2] pmulhrsw m7, [pw_16207x2] %else mova m2, [%1+29*64] mova m3, [%1+19*64] VP9_UNPACK_MULSUB_2W_4X 6, 3, 13160, 9760, [pd_8192], 4, 5 ; t22, t25 VP9_UNPACK_MULSUB_2W_4X 2, 7, 2404, 16207, [pd_8192], 4, 5 ; t23, t24 %endif SUMSUB_BA w, 6, 2, 4 SUMSUB_BA w, 3, 7, 4 VP9_UNPACK_MULSUB_2W_4X 7, 2, 13623, m9102, [pd_8192], 4, 5 ; t22, t25 %endif ; m4=t16, m5=t17, m9=t18, m8=t19, m0=t20, m1=t21, m13=t22, m12=t23, ; m3=t24, m2=t25, m14=t26, m15=t27, m7=t28, m6=t29, m10=t30, m11=t31 UNSCRATCH 4, 12, tmpq+11*%%str SUMSUB_BA w, 0, 6, 5 SUMSUB_BA w, 4, 2, 5 UNSCRATCH 5, 8, tmpq+ 5*%%str SCRATCH 4, 8, tmpq+11*%%str SUMSUB_BA w, 1, 7, 4 SUMSUB_BA w, 5, 3, 4 SCRATCH 5, 12, tmpq+ 5*%%str VP9_UNPACK_MULSUB_2W_4X 3, 6, 6270, m15137, [pd_8192], 4, 5 ; t20, t27 VP9_UNPACK_MULSUB_2W_4X 2, 7, 6270, m15137, [pd_8192], 4, 5 ; t21, t26 ; m8[s]=t16, m9=t17, m5=t18, m4[s]=t19, m12=t20, m13=t21, m1=t22, m0=t23, ; m15=t24, m14=t25, m2=t26, m3=t27, m11=t28, m10=t29, m6=t30, m7=t31 UNSCRATCH 5, 9, tmpq+ 9*%%str mova m4, [tmpq+23*%%str] ; t16 %if ARCH_X86_64 SUMSUB_BA w, 1, 5, 9 SUMSUB_BA w, 0, 4, 9 %else SUMSUB_BADC w, 1, 5, 0, 4 %endif mova [tmpq+29*%%str], m1 ; t17 mova [tmpq+21*%%str], m0 ; t16 UNSCRATCH 0, 10, tmpq+ 1*%%str UNSCRATCH 1, 11, tmpq+ 7*%%str %if ARCH_X86_64 SUMSUB_BA w, 2, 0, 9 SUMSUB_BA w, 3, 1, 9 %else SUMSUB_BADC w, 2, 0, 3, 1 %endif mova [tmpq+ 9*%%str], m2 ; t18 mova [tmpq+13*%%str], m3 ; t19 SCRATCH 0, 10, tmpq+23*%%str SCRATCH 1, 11, tmpq+27*%%str UNSCRATCH 2, 14, tmpq+15*%%str UNSCRATCH 3, 15, tmpq+17*%%str SUMSUB_BA w, 6, 2, 0 SUMSUB_BA w, 7, 3, 0 SCRATCH 6, 14, tmpq+ 3*%%str SCRATCH 7, 15, tmpq+ 7*%%str UNSCRATCH 0, 8, tmpq+11*%%str mova m1, [tmpq+25*%%str] ; t19 UNSCRATCH 6, 12, tmpq+ 5*%%str UNSCRATCH 7, 13, tmpq+31*%%str %if ARCH_X86_64 SUMSUB_BA w, 0, 1, 9 SUMSUB_BA w, 6, 7, 9 %else SUMSUB_BADC w, 0, 1, 6, 7 %endif ; m0=t16, m1=t17, m2=t18, m3=t19, m11=t20, m10=t21, m9=t22, m8=t23, ; m7=t24, m6=t25, m5=t26, m4=t27, m12=t28, m13=t29, m14=t30, m15=t31 %if 0; cpuflag(ssse3) %if ARCH_X86_64 SUMSUB_BA w, 4, 7, 8 SUMSUB_BA w, 5, 1, 8 %else SUMSUB_BADC w, 4, 7, 5, 1 %endif pmulhrsw m7, [pw_11585x2] pmulhrsw m4, [pw_11585x2] pmulhrsw m1, [pw_11585x2] pmulhrsw m5, [pw_11585x2] mova [tmpq+ 5*%%str], m7 ; t23 SCRATCH 1, 13, tmpq+25*%%str UNSCRATCH 7, 10, tmpq+23*%%str UNSCRATCH 1, 11, tmpq+27*%%str %if ARCH_X86_64 SUMSUB_BA w, 7, 3, 10 SUMSUB_BA w, 1, 2, 10 %else SUMSUB_BADC w, 7, 3, 1, 2 %endif pmulhrsw m3, [pw_11585x2] pmulhrsw m7, [pw_11585x2] pmulhrsw m2, [pw_11585x2] pmulhrsw m1, [pw_11585x2] %else SCRATCH 0, 8, tmpq+15*%%str SCRATCH 6, 9, tmpq+17*%%str VP9_UNPACK_MULSUB_2W_4X 7, 4, 11585, 11585, [pd_8192], 0, 6 mova [tmpq+ 5*%%str], m7 ; t23 UNSCRATCH 7, 10, tmpq+23*%%str VP9_UNPACK_MULSUB_2W_4X 1, 5, 11585, 11585, [pd_8192], 0, 6 SCRATCH 1, 13, tmpq+25*%%str UNSCRATCH 1, 11, tmpq+27*%%str VP9_UNPACK_MULSUB_2W_4X 3, 7, 11585, 11585, [pd_8192], 0, 6 VP9_UNPACK_MULSUB_2W_4X 2, 1, 11585, 11585, [pd_8192], 0, 6 UNSCRATCH 0, 8, tmpq+15*%%str UNSCRATCH 6, 9, tmpq+17*%%str %endif ; m0=t16, m1=t17, m2=t18, m3=t19, m4=t20, m5=t21, m6=t22, m7=t23, ; m8=t24, m9=t25, m10=t26, m11=t27, m12=t28, m13=t29, m14=t30, m15=t31 ; then do final pass to sumsub+store the two halves %if %2 == 1 mova [tmpq+17*%%str], m2 ; t20 mova [tmpq+ 1*%%str], m3 ; t21 %if ARCH_X86_64 mova [tmpq+25*%%str], m13 ; t22 mova m8, [tmpq+ 0*%%str] ; t0 mova m9, [tmpq+ 4*%%str] ; t1 mova m12, [tmpq+ 8*%%str] ; t2 mova m11, [tmpq+12*%%str] ; t3 mova m2, [tmpq+16*%%str] ; t4 mova m3, [tmpq+20*%%str] ; t5 mova m13, [tmpq+24*%%str] ; t6 SUMSUB_BA w, 6, 8, 10 mova [tmpq+ 3*%%str], m8 ; t15 SUMSUB_BA w, 0, 9, 8 SUMSUB_BA w, 15, 12, 8 SUMSUB_BA w, 14, 11, 8 SUMSUB_BA w, 1, 2, 8 SUMSUB_BA w, 7, 3, 8 SUMSUB_BA w, 5, 13, 8 mova m10, [tmpq+28*%%str] ; t7 SUMSUB_BA w, 4, 10, 8 %if cpuflag(avx2) ; the "shitty" about this idct is that the final pass does the outermost ; interleave sumsubs (t0/31, t1/30, etc) but the tN for the 16x16 need ; to be sequential, which means I need to load/store half of the sumsub ; intermediates back to/from memory to get a 16x16 transpose going... ; This would be easier if we had more (e.g. 32) YMM regs here. mova [tmpq+ 7*%%str], m9 mova [tmpq+11*%%str], m12 mova [tmpq+15*%%str], m11 mova [tmpq+19*%%str], m2 mova [tmpq+23*%%str], m3 mova [tmpq+27*%%str], m13 mova [tmpq+31*%%str], m10 mova [tmpq+12*%%str], m5 mova m13, [tmpq+30*%%str] ; t8 mova m12, [tmpq+26*%%str] ; t9 mova m11, [tmpq+22*%%str] ; t10 mova m10, [tmpq+18*%%str] ; t11 mova m9, [tmpq+17*%%str] ; t20 mova m8, [tmpq+ 1*%%str] ; t21 mova m3, [tmpq+25*%%str] ; t22 mova m2, [tmpq+ 5*%%str] ; t23 SUMSUB_BA w, 9, 10, 5 SUMSUB_BA w, 8, 11, 5 SUMSUB_BA w, 3, 12, 5 SUMSUB_BA w, 2, 13, 5 mova [tmpq+ 1*%%str], m10 mova [tmpq+ 5*%%str], m11 mova [tmpq+17*%%str], m12 mova [tmpq+25*%%str], m13 mova m13, [tmpq+14*%%str] ; t12 mova m12, [tmpq+10*%%str] ; t13 mova m11, [tmpq+ 9*%%str] ; t18 mova m10, [tmpq+13*%%str] ; t19 SUMSUB_BA w, 11, 12, 5 SUMSUB_BA w, 10, 13, 5 mova [tmpq+ 9*%%str], m13 mova [tmpq+13*%%str], m12 mova [tmpq+10*%%str], m10 mova [tmpq+14*%%str], m11 mova m13, [tmpq+ 6*%%str] ; t14 mova m12, [tmpq+ 2*%%str] ; t15 mova m11, [tmpq+21*%%str] ; t16 mova m10, [tmpq+29*%%str] ; t17 SUMSUB_BA w, 11, 12, 5 SUMSUB_BA w, 10, 13, 5 mova [tmpq+21*%%str], m12 mova [tmpq+29*%%str], m13 mova m12, [tmpq+10*%%str] mova m13, [tmpq+14*%%str] TRANSPOSE16x16W 6, 0, 15, 14, 1, 7, 5, 4, \ 2, 3, 8, 9, 12, 13, 10, 11, \ [tmpq+12*%%str], [tmpq+ 8*%%str], 1 mova [tmpq+ 0*%%str], m6 mova [tmpq+ 2*%%str], m0 mova [tmpq+ 4*%%str], m15 mova [tmpq+ 6*%%str], m14 mova [tmpq+10*%%str], m7 mova [tmpq+12*%%str], m5 mova [tmpq+14*%%str], m4 mova [tmpq+16*%%str], m2 mova [tmpq+18*%%str], m3 mova [tmpq+20*%%str], m8 mova [tmpq+22*%%str], m9 mova [tmpq+24*%%str], m12 mova [tmpq+26*%%str], m13 mova [tmpq+28*%%str], m10 mova [tmpq+30*%%str], m11 mova m0, [tmpq+21*%%str] mova m1, [tmpq+29*%%str] mova m2, [tmpq+13*%%str] mova m3, [tmpq+ 9*%%str] mova m4, [tmpq+ 1*%%str] mova m5, [tmpq+ 5*%%str] mova m7, [tmpq+25*%%str] mova m8, [tmpq+31*%%str] mova m9, [tmpq+27*%%str] mova m10, [tmpq+23*%%str] mova m11, [tmpq+19*%%str] mova m12, [tmpq+15*%%str] mova m13, [tmpq+11*%%str] mova m14, [tmpq+ 7*%%str] mova m15, [tmpq+ 3*%%str] TRANSPOSE16x16W 0, 1, 2, 3, 4, 5, 6, 7, \ 8, 9, 10, 11, 12, 13, 14, 15, \ [tmpq+17*%%str], [tmpq+ 9*%%str], 1 mova [tmpq+ 1*%%str], m0 mova [tmpq+ 3*%%str], m1 mova [tmpq+ 5*%%str], m2 mova [tmpq+ 7*%%str], m3 mova [tmpq+11*%%str], m5 mova [tmpq+13*%%str], m6 mova [tmpq+15*%%str], m7 mova [tmpq+17*%%str], m8 mova [tmpq+19*%%str], m9 mova [tmpq+21*%%str], m10 mova [tmpq+23*%%str], m11 mova [tmpq+25*%%str], m12 mova [tmpq+27*%%str], m13 mova [tmpq+29*%%str], m14 mova [tmpq+31*%%str], m15 %else ; !avx2 TRANSPOSE8x8W 6, 0, 15, 14, 1, 7, 5, 4, 8 mova [tmpq+ 0*%%str], m6 mova [tmpq+ 4*%%str], m0 mova [tmpq+ 8*%%str], m15 mova [tmpq+12*%%str], m14 mova [tmpq+16*%%str], m1 mova [tmpq+20*%%str], m7 mova [tmpq+24*%%str], m5 mova [tmpq+28*%%str], m4 mova m8, [tmpq+ 3*%%str] ; t15 TRANSPOSE8x8W 10, 13, 3, 2, 11, 12, 9, 8, 0 mova [tmpq+ 3*%%str], m10 mova [tmpq+ 7*%%str], m13 mova [tmpq+11*%%str], m3 mova [tmpq+15*%%str], m2 mova [tmpq+19*%%str], m11 mova [tmpq+23*%%str], m12 mova [tmpq+27*%%str], m9 mova [tmpq+31*%%str], m8 mova m15, [tmpq+30*%%str] ; t8 mova m14, [tmpq+26*%%str] ; t9 mova m13, [tmpq+22*%%str] ; t10 mova m12, [tmpq+18*%%str] ; t11 mova m11, [tmpq+14*%%str] ; t12 mova m10, [tmpq+10*%%str] ; t13 mova m9, [tmpq+ 6*%%str] ; t14 mova m8, [tmpq+ 2*%%str] ; t15 mova m7, [tmpq+21*%%str] ; t16 mova m6, [tmpq+29*%%str] ; t17 mova m5, [tmpq+ 9*%%str] ; t18 mova m4, [tmpq+13*%%str] ; t19 mova m3, [tmpq+17*%%str] ; t20 mova m2, [tmpq+ 1*%%str] ; t21 mova m1, [tmpq+25*%%str] ; t22 SUMSUB_BA w, 7, 8, 0 mova [tmpq+ 2*%%str], m8 mova m0, [tmpq+ 5*%%str] ; t23 SUMSUB_BA w, 6, 9, 8 SUMSUB_BA w, 5, 10, 8 SUMSUB_BA w, 4, 11, 8 SUMSUB_BA w, 3, 12, 8 SUMSUB_BA w, 2, 13, 8 SUMSUB_BA w, 1, 14, 8 SUMSUB_BA w, 0, 15, 8 TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, 8 mova [tmpq+ 1*%%str], m0 mova [tmpq+ 5*%%str], m1 mova [tmpq+ 9*%%str], m2 mova [tmpq+13*%%str], m3 mova [tmpq+17*%%str], m4 mova [tmpq+21*%%str], m5 mova [tmpq+25*%%str], m6 mova [tmpq+29*%%str], m7 mova m8, [tmpq+ 2*%%str] TRANSPOSE8x8W 8, 9, 10, 11, 12, 13, 14, 15, 0 mova [tmpq+ 2*%%str], m8 mova [tmpq+ 6*%%str], m9 mova [tmpq+10*%%str], m10 mova [tmpq+14*%%str], m11 mova [tmpq+18*%%str], m12 mova [tmpq+22*%%str], m13 mova [tmpq+26*%%str], m14 mova [tmpq+30*%%str], m15 %endif ; avx2 %else mova m2, [tmpq+24*%%str] ; t6 mova m3, [tmpq+28*%%str] ; t7 SUMSUB_BADC w, 5, 2, 4, 3 mova [tmpq+24*%%str], m5 mova [tmpq+23*%%str], m2 mova [tmpq+28*%%str], m4 mova [tmpq+19*%%str], m3 mova m2, [tmpq+16*%%str] ; t4 mova m3, [tmpq+20*%%str] ; t5 SUMSUB_BA w, 1, 2, 5 SUMSUB_BA w, 7, 3, 5 mova [tmpq+15*%%str], m2 mova [tmpq+11*%%str], m3 mova m2, [tmpq+ 0*%%str] ; t0 mova m3, [tmpq+ 4*%%str] ; t1 SUMSUB_BA w, 6, 2, 5 SUMSUB_BA w, 0, 3, 5 mova [tmpq+31*%%str], m2 mova [tmpq+27*%%str], m3 mova m2, [tmpq+ 8*%%str] ; t2 mova m3, [tmpq+12*%%str] ; t3 mova m5, [tmpq+ 7*%%str] mova m4, [tmpq+ 3*%%str] SUMSUB_BADC w, 5, 2, 4, 3 mova [tmpq+ 7*%%str], m2 mova [tmpq+ 3*%%str], m3 mova m3, [tmpq+28*%%str] TRANSPOSE8x8W 6, 0, 5, 4, 1, 7, 2, 3, [tmpq+24*%%str], [tmpq+16*%%str], 1 mova [tmpq+ 0*%%str], m6 mova [tmpq+ 4*%%str], m0 mova [tmpq+ 8*%%str], m5 mova [tmpq+12*%%str], m4 mova [tmpq+20*%%str], m7 mova [tmpq+24*%%str], m2 mova [tmpq+28*%%str], m3 mova m6, [tmpq+19*%%str] mova m0, [tmpq+23*%%str] mova m5, [tmpq+11*%%str] mova m4, [tmpq+15*%%str] mova m1, [tmpq+ 3*%%str] mova m7, [tmpq+ 7*%%str] mova m3, [tmpq+31*%%str] TRANSPOSE8x8W 6, 0, 5, 4, 1, 7, 2, 3, [tmpq+27*%%str], [tmpq+19*%%str], 1 mova [tmpq+ 3*%%str], m6 mova [tmpq+ 7*%%str], m0 mova [tmpq+11*%%str], m5 mova [tmpq+15*%%str], m4 mova [tmpq+23*%%str], m7 mova [tmpq+27*%%str], m2 mova [tmpq+31*%%str], m3 mova m1, [tmpq+ 6*%%str] ; t14 mova m0, [tmpq+ 2*%%str] ; t15 mova m7, [tmpq+21*%%str] ; t16 mova m6, [tmpq+29*%%str] ; t17 SUMSUB_BA w, 7, 0, 2 SUMSUB_BA w, 6, 1, 2 mova [tmpq+29*%%str], m7 mova [tmpq+ 2*%%str], m0 mova [tmpq+21*%%str], m6 mova [tmpq+ 6*%%str], m1 mova m1, [tmpq+14*%%str] ; t12 mova m0, [tmpq+10*%%str] ; t13 mova m5, [tmpq+ 9*%%str] ; t18 mova m4, [tmpq+13*%%str] ; t19 SUMSUB_BA w, 5, 0, 2 SUMSUB_BA w, 4, 1, 2 mova [tmpq+10*%%str], m0 mova [tmpq+14*%%str], m1 mova m1, [tmpq+22*%%str] ; t10 mova m0, [tmpq+18*%%str] ; t11 mova m3, [tmpq+17*%%str] ; t20 mova m2, [tmpq+ 1*%%str] ; t21 SUMSUB_BA w, 3, 0, 6 SUMSUB_BA w, 2, 1, 6 mova [tmpq+18*%%str], m0 mova [tmpq+22*%%str], m1 mova m7, [tmpq+30*%%str] ; t8 mova m6, [tmpq+26*%%str] ; t9 mova m1, [tmpq+25*%%str] ; t22 mova m0, [tmpq+ 5*%%str] ; t23 SUMSUB_BADC w, 1, 6, 0, 7 mova [tmpq+26*%%str], m6 mova [tmpq+30*%%str], m7 mova m7, [tmpq+29*%%str] TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, [tmpq+21*%%str], [tmpq+17*%%str], 1 mova [tmpq+ 1*%%str], m0 mova [tmpq+ 5*%%str], m1 mova [tmpq+ 9*%%str], m2 mova [tmpq+13*%%str], m3 mova [tmpq+21*%%str], m5 mova [tmpq+25*%%str], m6 mova [tmpq+29*%%str], m7 mova m0, [tmpq+ 2*%%str] mova m1, [tmpq+ 6*%%str] mova m2, [tmpq+10*%%str] mova m3, [tmpq+14*%%str] mova m4, [tmpq+18*%%str] mova m5, [tmpq+22*%%str] mova m7, [tmpq+30*%%str] TRANSPOSE8x8W 0, 1, 2, 3, 4, 5, 6, 7, [tmpq+26*%%str], [tmpq+18*%%str], 1 mova [tmpq+ 2*%%str], m0 mova [tmpq+ 6*%%str], m1 mova [tmpq+10*%%str], m2 mova [tmpq+14*%%str], m3 mova [tmpq+22*%%str], m5 mova [tmpq+26*%%str], m6 mova [tmpq+30*%%str], m7 %endif %else ; t0-7 is in [tmpq+{0,4,8,12,16,20,24,28}*%%str] ; t8-15 is in [tmpq+{2,6,10,14,18,22,26,30}*%%str] ; t16-19 and t23 is in [tmpq+{1,5,9,13,29}*%%str] ; t20-22 is in m4-6 ; t24-31 is in m8-15 %if cpuflag(ssse3) %define ROUND_REG [pw_512] %else %define ROUND_REG [pw_32] %endif %macro %%STORE_2X2 7-8 1 ; src[1-4], tmp[1-2], zero, inc_dst_ptrs SUMSUB_BA w, %4, %1, %5 SUMSUB_BA w, %3, %2, %5 VP9_IDCT8_WRITEx2 %4, %3, %5, %6, %7, ROUND_REG, 6 %if %8 == 1 add dstq, stride2q %endif VP9_IDCT8_WRITEx2 %2, %1, %5, %6, %7, ROUND_REG, 6, dst_endq %if %8 == 1 sub dst_endq, stride2q %endif %endmacro %if ARCH_X86_64 pxor m10, m10 ; store t0-1 and t30-31 mova m8, [tmpq+ 0*%%str] mova m9, [tmpq+ 4*%%str] %%STORE_2X2 8, 9, 0, 6, 12, 11, 10 ; store t2-3 and t28-29 mova m8, [tmpq+ 8*%%str] mova m9, [tmpq+12*%%str] %%STORE_2X2 8, 9, 14, 15, 12, 11, 10 ; store t4-5 and t26-27 mova m8, [tmpq+16*%%str] mova m9, [tmpq+20*%%str] %%STORE_2X2 8, 9, 7, 1, 12, 11, 10 ; store t6-7 and t24-25 mova m8, [tmpq+24*%%str] mova m9, [tmpq+28*%%str] %%STORE_2X2 8, 9, 4, 5, 12, 11, 10 ; store t8-9 and t22-23 mova m8, [tmpq+30*%%str] mova m9, [tmpq+26*%%str] mova m0, [tmpq+ 5*%%str] %%STORE_2X2 8, 9, 13, 0, 12, 11, 10 ; store t10-11 and t20-21 mova m8, [tmpq+22*%%str] mova m9, [tmpq+18*%%str] %%STORE_2X2 8, 9, 2, 3, 12, 11, 10 ; store t12-13 and t18-19 mova m8, [tmpq+14*%%str] mova m9, [tmpq+10*%%str] mova m5, [tmpq+13*%%str] mova m4, [tmpq+ 9*%%str] %%STORE_2X2 8, 9, 4, 5, 12, 11, 10 ; store t14-17 mova m8, [tmpq+ 6*%%str] mova m9, [tmpq+ 2*%%str] mova m5, [tmpq+29*%%str] mova m4, [tmpq+21*%%str] %%STORE_2X2 8, 9, 4, 5, 12, 11, 10, 0 SWAP 1, 10 ; zero %else mova [tmpq+ 1*%%str], m1 mova [tmpq+11*%%str], m2 mova [tmpq+15*%%str], m3 mova [tmpq+17*%%str], m4 mova [tmpq+19*%%str], m5 pxor m1, m1 ; store t0-1 and t30-31 mova m2, [tmpq+ 0*%%str] mova m3, [tmpq+ 4*%%str] %%STORE_2X2 2, 3, 0, 6, 4, 5, 1 ; store t2-3 and t28-29 mova m2, [tmpq+ 8*%%str] mova m3, [tmpq+12*%%str] mova m0, [tmpq+ 3*%%str] mova m6, [tmpq+ 7*%%str] %%STORE_2X2 2, 3, 0, 6, 4, 5, 1 ; store t4-5 and t26-27 mova m2, [tmpq+16*%%str] mova m3, [tmpq+20*%%str] mova m0, [tmpq+ 1*%%str] %%STORE_2X2 2, 3, 7, 0, 4, 5, 1 ; store t6-7 and t24-25 mova m2, [tmpq+24*%%str] mova m3, [tmpq+28*%%str] mova m0, [tmpq+17*%%str] mova m6, [tmpq+19*%%str] %%STORE_2X2 2, 3, 0, 6, 4, 5, 1 ; store t8-9 and t22-23 mova m2, [tmpq+30*%%str] mova m3, [tmpq+26*%%str] mova m0, [tmpq+25*%%str] mova m6, [tmpq+ 5*%%str] %%STORE_2X2 2, 3, 0, 6, 4, 5, 1 ; store t10-11 and t20-21 mova m2, [tmpq+22*%%str] mova m3, [tmpq+18*%%str] mova m0, [tmpq+11*%%str] mova m6, [tmpq+15*%%str] %%STORE_2X2 2, 3, 0, 6, 4, 5, 1 ; store t12-13 and t18-19 mova m2, [tmpq+14*%%str] mova m3, [tmpq+10*%%str] mova m6, [tmpq+13*%%str] mova m0, [tmpq+ 9*%%str] %%STORE_2X2 2, 3, 0, 6, 4, 5, 1 ; store t14-17 mova m2, [tmpq+ 6*%%str] mova m3, [tmpq+ 2*%%str] mova m6, [tmpq+29*%%str] mova m0, [tmpq+21*%%str] %%STORE_2X2 2, 3, 0, 6, 4, 5, 1, 0 %endif %undef ROUND_REG %endif %endmacro %macro VP9_IDCT_IDCT_32x32_ADD_XMM 1 INIT_XMM %1 cglobal vp9_idct_idct_32x32_add, 0, 6 + ARCH_X86_64 * 3, 16, 2048, dst, stride, block, eob movifnidn eobd, dword eobm %if cpuflag(ssse3) cmp eobd, 135 jg .idctfull cmp eobd, 34 jg .idct16x16 cmp eobd, 1 jg .idct8x8 %else cmp eobd, 1 jg .idctfull %endif ; dc-only case movifnidn blockq, blockmp movifnidn dstq, dstmp movifnidn strideq, stridemp %if cpuflag(ssse3) movd m0, [blockq] mova m1, [pw_11585x2] pmulhrsw m0, m1 pmulhrsw m0, m1 %else DEFINE_ARGS dst, stride, block, coef movsx coefd, word [blockq] imul coefd, 11585 add coefd, 8192 sar coefd, 14 imul coefd, 11585 add coefd, (32 << 14) + 8192 sar coefd, 14 + 6 movd m0, coefd %endif SPLATW m0, m0, q0000 %if cpuflag(ssse3) pmulhrsw m0, [pw_512] %endif pxor m5, m5 movd [blockq], m5 %rep 31 VP9_STORE_2XFULL 0, 1, 2, 3, 4, 5, mmsize add dstq, strideq %endrep VP9_STORE_2XFULL 0, 1, 2, 3, 4, 5, mmsize RET %if ARCH_X86_64 DEFINE_ARGS dst_bak, stride, block, cnt, dst, stride30, dst_end, stride2, tmp %else %define dst_bakq r0mp %endif %if cpuflag(ssse3) .idct8x8: %if ARCH_X86_32 DEFINE_ARGS block, u1, u2, u3, u4, tmp mov blockq, r2mp %endif mov tmpq, rsp VP9_IDCT32_1D blockq, 1, 8 %if ARCH_X86_32 DEFINE_ARGS dst, stride, stride30, dst_end, stride2, tmp mov strideq, r1mp %define cntd dword r3m %endif mov stride30q, strideq ; stride lea stride2q, [strideq*2] ; stride*2 shl stride30q, 5 ; stride*32 mov cntd, 4 sub stride30q, stride2q ; stride*30 .loop2_8x8: mov dstq, dst_bakq lea dst_endq, [dstq+stride30q] VP9_IDCT32_1D tmpq, 2, 8 add dst_bakq, 8 add tmpq, 16 dec cntd jg .loop2_8x8 ; at the end of the loop, m7 should still be zero ; use that to zero out block coefficients %if ARCH_X86_32 DEFINE_ARGS block mov blockq, r2mp %endif ZERO_BLOCK blockq, 64, 8, m1 RET .idct16x16: %if ARCH_X86_32 DEFINE_ARGS block, tmp, cnt mov blockq, r2mp %endif mov cntd, 2 mov tmpq, rsp .loop1_16x16: VP9_IDCT32_1D blockq, 1, 16 add blockq, 16 add tmpq, 512 dec cntd jg .loop1_16x16 %if ARCH_X86_64 sub blockq, 32 %else DEFINE_ARGS dst, stride, stride30, dst_end, stride2, tmp mov strideq, r1mp %define cntd dword r3m %endif mov stride30q, strideq ; stride lea stride2q, [strideq*2] ; stride*2 shl stride30q, 5 ; stride*32 mov cntd, 4 mov tmpq, rsp sub stride30q, stride2q ; stride*30 .loop2_16x16: mov dstq, dst_bakq lea dst_endq, [dstq+stride30q] VP9_IDCT32_1D tmpq, 2, 16 add dst_bakq, 8 add tmpq, 16 dec cntd jg .loop2_16x16 ; at the end of the loop, m7 should still be zero ; use that to zero out block coefficients %if ARCH_X86_32 DEFINE_ARGS block mov blockq, r2mp %endif ZERO_BLOCK blockq, 64, 16, m1 RET %endif .idctfull: %if ARCH_X86_32 DEFINE_ARGS block, tmp, cnt mov blockq, r2mp %endif mov cntd, 4 mov tmpq, rsp .loop1_full: VP9_IDCT32_1D blockq, 1 add blockq, 16 add tmpq, 512 dec cntd jg .loop1_full %if ARCH_X86_64 sub blockq, 64 %else DEFINE_ARGS dst, stride, stride30, dst_end, stride2, tmp mov strideq, r1mp %define cntd dword r3m %endif mov stride30q, strideq ; stride lea stride2q, [strideq*2] ; stride*2 shl stride30q, 5 ; stride*32 mov cntd, 4 mov tmpq, rsp sub stride30q, stride2q ; stride*30 .loop2_full: mov dstq, dst_bakq lea dst_endq, [dstq+stride30q] VP9_IDCT32_1D tmpq, 2 add dst_bakq, 8 add tmpq, 16 dec cntd jg .loop2_full ; at the end of the loop, m7 should still be zero ; use that to zero out block coefficients %if ARCH_X86_32 DEFINE_ARGS block mov blockq, r2mp %endif ZERO_BLOCK blockq, 64, 32, m1 RET %endmacro VP9_IDCT_IDCT_32x32_ADD_XMM sse2 VP9_IDCT_IDCT_32x32_ADD_XMM ssse3 VP9_IDCT_IDCT_32x32_ADD_XMM avx ; this is almost identical to VP9_STORE_2X, but it does two rows ; for slightly improved interleaving, and it omits vpermq since the ; input is DC so all values are identical %macro VP9_STORE_YMM_DC_2X2 6 ; reg, tmp1, tmp2, tmp3, tmp4, zero mova m%2, [dstq] mova m%4, [dstq+strideq] punpckhbw m%3, m%2, m%6 punpcklbw m%2, m%6 punpckhbw m%5, m%4, m%6 punpcklbw m%4, m%6 paddw m%3, m%1 paddw m%2, m%1 paddw m%5, m%1 paddw m%4, m%1 packuswb m%2, m%3 packuswb m%4, m%5 mova [dstq+strideq*0], m%2 mova [dstq+strideq*1], m%4 %endmacro %if ARCH_X86_64 && HAVE_AVX2_EXTERNAL INIT_YMM avx2 cglobal vp9_idct_idct_32x32_add, 4, 9, 16, 2048, dst, stride, block, eob cmp eobd, 135 jg .idctfull cmp eobd, 1 jg .idct16x16 ; dc-only case mova m1, [pw_11585x2] vpbroadcastw m0, [blockq] pmulhrsw m0, m1 pmulhrsw m0, m1 pxor m5, m5 pmulhrsw m0, [pw_512] movd [blockq], xm5 DEFINE_ARGS dst, stride, cnt mov cntd, 16 .loop_dc: VP9_STORE_YMM_DC_2X2 0, 1, 2, 3, 4, 5 lea dstq, [dstq+2*strideq] dec cntd jg .loop_dc RET DEFINE_ARGS dst_bak, stride, block, cnt, dst, stride30, dst_end, stride2, tmp .idct16x16: mov tmpq, rsp VP9_IDCT32_1D blockq, 1, 16 mov stride30q, strideq ; stride lea stride2q, [strideq*2] ; stride*2 shl stride30q, 5 ; stride*32 mov cntd, 2 sub stride30q, stride2q ; stride*30 .loop2_16x16: mov dstq, dst_bakq lea dst_endq, [dstq+stride30q] VP9_IDCT32_1D tmpq, 2, 16 add dst_bakq, 16 add tmpq, 32 dec cntd jg .loop2_16x16 ; at the end of the loop, m1 should still be zero ; use that to zero out block coefficients ZERO_BLOCK blockq, 64, 16, m1 RET .idctfull: mov cntd, 2 mov tmpq, rsp .loop1_full: VP9_IDCT32_1D blockq, 1 add blockq, 32 add tmpq, 1024 dec cntd jg .loop1_full sub blockq, 64 mov stride30q, strideq ; stride lea stride2q, [strideq*2] ; stride*2 shl stride30q, 5 ; stride*32 mov cntd, 2 mov tmpq, rsp sub stride30q, stride2q ; stride*30 .loop2_full: mov dstq, dst_bakq lea dst_endq, [dstq+stride30q] VP9_IDCT32_1D tmpq, 2 add dst_bakq, 16 add tmpq, 32 dec cntd jg .loop2_full ; at the end of the loop, m1 should still be zero ; use that to zero out block coefficients ZERO_BLOCK blockq, 64, 32, m1 RET %endif