WebSVN – Kolibri OS – Blame – /contrib/sdk/sources/ffmpeg/ffmpeg-2.8/libavcodec/x86/hevc_sao.asm

Rev	Author	Line No.	Line
6147	serge	1	;******************************************************************************
		2	;* SIMD optimized SAO functions for HEVC decoding
		3	;*
		4	;* Copyright (c) 2013 Pierre-Edouard LEPERE
		5	;* Copyright (c) 2014 James Almer
		6	;*
		7	;* This file is part of FFmpeg.
		8	;*
		9	;* FFmpeg is free software; you can redistribute it and/or
		10	;* modify it under the terms of the GNU Lesser General Public
		11	;* License as published by the Free Software Foundation; either
		12	;* version 2.1 of the License, or (at your option) any later version.
		13	;*
		14	;* FFmpeg is distributed in the hope that it will be useful,
		15	;* but WITHOUT ANY WARRANTY; without even the implied warranty of
		16	;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
		17	;* Lesser General Public License for more details.
		18	;*
		19	;* You should have received a copy of the GNU Lesser General Public
		20	;* License along with FFmpeg; if not, write to the Free Software
		21	;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
		22	;******************************************************************************
		23
		24	%include "libavutil/x86/x86util.asm"
		25
		26	SECTION_RODATA 32
		27
		28	pw_mask10: times 16 dw 0x03FF
		29	pw_mask12: times 16 dw 0x0FFF
		30	pw_m2: times 16 dw -2
		31	pb_edge_shuffle: times 2 db 1, 2, 0, 3, 4, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1
		32	pb_eo: db -1, 0, 1, 0, 0, -1, 0, 1, -1, -1, 1, 1, 1, -1, -1, 1
		33	cextern pw_m1
		34	cextern pw_1
		35	cextern pw_2
		36	cextern pb_1
		37	cextern pb_2
		38
		39	SECTION .text
		40
		41	%define MAX_PB_SIZE 64
		42	%define PADDING_SIZE 32 ; AV_INPUT_BUFFER_PADDING_SIZE
		43
		44	;******************************************************************************
		45	;SAO Band Filter
		46	;******************************************************************************
		47
		48	%macro HEVC_SAO_BAND_FILTER_INIT 1
		49	and leftq, 31
		50	movd xm0, leftd
		51	add leftq, 1
		52	and leftq, 31
		53	movd xm1, leftd
		54	add leftq, 1
		55	and leftq, 31
		56	movd xm2, leftd
		57	add leftq, 1
		58	and leftq, 31
		59	movd xm3, leftd
		60
		61	SPLATW m0, xm0
		62	SPLATW m1, xm1
		63	SPLATW m2, xm2
		64	SPLATW m3, xm3
		65	%if mmsize > 16
		66	SPLATW m4, [offsetq + 2]
		67	SPLATW m5, [offsetq + 4]
		68	SPLATW m6, [offsetq + 6]
		69	SPLATW m7, [offsetq + 8]
		70	%else
		71	movq m7, [offsetq + 2]
		72	SPLATW m4, m7, 0
		73	SPLATW m5, m7, 1
		74	SPLATW m6, m7, 2
		75	SPLATW m7, m7, 3
		76	%endif
		77
		78	%if ARCH_X86_64
		79	%if %1 > 8
		80	mova m13, [pw_mask %+ %1]
		81	%endif
		82	pxor m14, m14
		83
		84	%else ; ARCH_X86_32
		85	mova [rsp+mmsize*0], m0
		86	mova [rsp+mmsize*1], m1
		87	mova [rsp+mmsize*2], m2
		88	mova [rsp+mmsize*3], m3
		89	mova [rsp+mmsize*4], m4
		90	mova [rsp+mmsize*5], m5
		91	mova [rsp+mmsize*6], m6
		92	pxor m0, m0
		93	%if %1 > 8
		94	mova m1, [pw_mask %+ %1]
		95	%endif
		96	%assign MMSIZE mmsize
		97	%define m14 m0
		98	%define m13 m1
		99	%define m9 m2
		100	%define m8 m3
		101	%endif ; ARCH
		102	DEFINE_ARGS dst, src, dststride, srcstride, offset, height
		103	mov heightd, r7m
		104	%endmacro
		105
		106	%macro HEVC_SAO_BAND_FILTER_COMPUTE 3
		107	psraw %2, %3, %1-5
		108	%if ARCH_X86_64
		109	pcmpeqw m10, %2, m0
		110	pcmpeqw m11, %2, m1
		111	pcmpeqw m12, %2, m2
		112	pcmpeqw %2, m3
		113	pand m10, m4
		114	pand m11, m5
		115	pand m12, m6
		116	pand %2, m7
		117	por m10, m11
		118	por m12, %2
		119	por m10, m12
		120	paddw %3, m10
		121	%else ; ARCH_X86_32
		122	pcmpeqw m4, %2, [rsp+MMSIZE*0]
		123	pcmpeqw m5, %2, [rsp+MMSIZE*1]
		124	pcmpeqw m6, %2, [rsp+MMSIZE*2]
		125	pcmpeqw %2, [rsp+MMSIZE*3]
		126	pand m4, [rsp+MMSIZE*4]
		127	pand m5, [rsp+MMSIZE*5]
		128	pand m6, [rsp+MMSIZE*6]
		129	pand %2, m7
		130	por m4, m5
		131	por m6, %2
		132	por m4, m6
		133	paddw %3, m4
		134	%endif ; ARCH
		135	%endmacro
		136
		137	;void ff_hevc_sao_band_filter__8_(uint8_t _dst, uint8_t _src, ptrdiff_t _stride_dst, ptrdiff_t _stride_src,
		138	; int16_t *sao_offset_val, int sao_left_class, int width, int height);
		139	%macro HEVC_SAO_BAND_FILTER_8 2
		140	cglobal hevc_sao_band_filter_%1_8, 6, 6, 15, 7mmsizeARCH_X86_32, dst, src, dststride, srcstride, offset, left
		141	HEVC_SAO_BAND_FILTER_INIT 8
		142
		143	align 16
		144	.loop:
		145	%if %1 == 8
		146	movq m8, [srcq]
		147	punpcklbw m8, m14
		148	HEVC_SAO_BAND_FILTER_COMPUTE 8, m9, m8
		149	packuswb m8, m14
		150	movq [dstq], m8
		151	%endif ; %1 == 8
		152
		153	%assign i 0
		154	%rep %2
		155	mova m13, [srcq + i]
		156	punpcklbw m8, m13, m14
		157	HEVC_SAO_BAND_FILTER_COMPUTE 8, m9, m8
		158	punpckhbw m13, m14
		159	HEVC_SAO_BAND_FILTER_COMPUTE 8, m9, m13
		160	packuswb m8, m13
		161	mova [dstq + i], m8
		162	%assign i i+mmsize
		163	%endrep
		164
		165	%if %1 == 48
		166	INIT_XMM cpuname
		167
		168	mova m13, [srcq + i]
		169	punpcklbw m8, m13, m14
		170	HEVC_SAO_BAND_FILTER_COMPUTE 8, m9, m8
		171	punpckhbw m13, m14
		172	HEVC_SAO_BAND_FILTER_COMPUTE 8, m9, m13
		173	packuswb m8, m13
		174	mova [dstq + i], m8
		175	%if cpuflag(avx2)
		176	INIT_YMM cpuname
		177	%endif
		178	%endif ; %1 == 48
		179
		180	add dstq, dststrideq ; dst += dststride
		181	add srcq, srcstrideq ; src += srcstride
		182	dec heightd ; cmp height
		183	jnz .loop ; height loop
		184	REP_RET
		185	%endmacro
		186
		187	;void ff_hevc_sao_band_filter___(uint8_t _dst, uint8_t _src, ptrdiff_t _stride_dst, ptrdiff_t _stride_src,
		188	; int16_t *sao_offset_val, int sao_left_class, int width, int height);
		189	%macro HEVC_SAO_BAND_FILTER_16 3
		190	cglobal hevc_sao_band_filter_%2_%1, 6, 6, 15, 7mmsizeARCH_X86_32, dst, src, dststride, srcstride, offset, left
		191	HEVC_SAO_BAND_FILTER_INIT %1
		192
		193	align 16
		194	.loop:
		195	%if %2 == 8
		196	movu m8, [srcq]
		197	HEVC_SAO_BAND_FILTER_COMPUTE %1, m9, m8
		198	CLIPW m8, m14, m13
		199	movu [dstq], m8
		200	%endif
		201
		202	%assign i 0
		203	%rep %3
		204	mova m8, [srcq + i]
		205	HEVC_SAO_BAND_FILTER_COMPUTE %1, m9, m8
		206	CLIPW m8, m14, m13
		207	mova [dstq + i], m8
		208
		209	mova m9, [srcq + i + mmsize]
		210	HEVC_SAO_BAND_FILTER_COMPUTE %1, m8, m9
		211	CLIPW m9, m14, m13
		212	mova [dstq + i + mmsize], m9
		213	%assign i i+mmsize*2
		214	%endrep
		215
		216	%if %2 == 48
		217	INIT_XMM cpuname
		218	mova m8, [srcq + i]
		219	HEVC_SAO_BAND_FILTER_COMPUTE %1, m9, m8
		220	CLIPW m8, m14, m13
		221	mova [dstq + i], m8
		222
		223	mova m9, [srcq + i + mmsize]
		224	HEVC_SAO_BAND_FILTER_COMPUTE %1, m8, m9
		225	CLIPW m9, m14, m13
		226	mova [dstq + i + mmsize], m9
		227	%if cpuflag(avx2)
		228	INIT_YMM cpuname
		229	%endif
		230	%endif ; %1 == 48
		231
		232	add dstq, dststrideq
		233	add srcq, srcstrideq
		234	dec heightd
		235	jg .loop
		236	REP_RET
		237	%endmacro
		238
		239	%macro HEVC_SAO_BAND_FILTER_FUNCS 0
		240	HEVC_SAO_BAND_FILTER_8 8, 0
		241	HEVC_SAO_BAND_FILTER_8 16, 1
		242	HEVC_SAO_BAND_FILTER_8 32, 2
		243	HEVC_SAO_BAND_FILTER_8 48, 2
		244	HEVC_SAO_BAND_FILTER_8 64, 4
		245
		246	HEVC_SAO_BAND_FILTER_16 10, 8, 0
		247	HEVC_SAO_BAND_FILTER_16 10, 16, 1
		248	HEVC_SAO_BAND_FILTER_16 10, 32, 2
		249	HEVC_SAO_BAND_FILTER_16 10, 48, 2
		250	HEVC_SAO_BAND_FILTER_16 10, 64, 4
		251
		252	HEVC_SAO_BAND_FILTER_16 12, 8, 0
		253	HEVC_SAO_BAND_FILTER_16 12, 16, 1
		254	HEVC_SAO_BAND_FILTER_16 12, 32, 2
		255	HEVC_SAO_BAND_FILTER_16 12, 48, 2
		256	HEVC_SAO_BAND_FILTER_16 12, 64, 4
		257	%endmacro
		258
		259	INIT_XMM sse2
		260	HEVC_SAO_BAND_FILTER_FUNCS
		261	INIT_XMM avx
		262	HEVC_SAO_BAND_FILTER_FUNCS
		263
		264	%if HAVE_AVX2_EXTERNAL
		265	INIT_XMM avx2
		266	HEVC_SAO_BAND_FILTER_8 8, 0
		267	HEVC_SAO_BAND_FILTER_8 16, 1
		268	INIT_YMM avx2
		269	HEVC_SAO_BAND_FILTER_8 32, 1
		270	HEVC_SAO_BAND_FILTER_8 48, 1
		271	HEVC_SAO_BAND_FILTER_8 64, 2
		272
		273	INIT_XMM avx2
		274	HEVC_SAO_BAND_FILTER_16 10, 8, 0
		275	HEVC_SAO_BAND_FILTER_16 10, 16, 1
		276	INIT_YMM avx2
		277	HEVC_SAO_BAND_FILTER_16 10, 32, 1
		278	HEVC_SAO_BAND_FILTER_16 10, 48, 1
		279	HEVC_SAO_BAND_FILTER_16 10, 64, 2
		280
		281	INIT_XMM avx2
		282	HEVC_SAO_BAND_FILTER_16 12, 8, 0
		283	HEVC_SAO_BAND_FILTER_16 12, 16, 1
		284	INIT_YMM avx2
		285	HEVC_SAO_BAND_FILTER_16 12, 32, 1
		286	HEVC_SAO_BAND_FILTER_16 12, 48, 1
		287	HEVC_SAO_BAND_FILTER_16 12, 64, 2
		288	%endif
		289
		290	;******************************************************************************
		291	;SAO Edge Filter
		292	;******************************************************************************
		293
		294	%define EDGE_SRCSTRIDE 2 * MAX_PB_SIZE + PADDING_SIZE
		295
		296	%macro HEVC_SAO_EDGE_FILTER_INIT 1
		297	%if WIN64
		298	movsxd eoq, dword eom
		299	%elif ARCH_X86_64
		300	movsxd eoq, eod
		301	%else
		302	mov eoq, r4m
		303	%endif
		304	lea tmp2q, [pb_eo]
		305	movsx a_strideq, byte [tmp2q+eoq*4+1]
		306	movsx b_strideq, byte [tmp2q+eoq*4+3]
		307	imul a_strideq, EDGE_SRCSTRIDE>>%1
		308	imul b_strideq, EDGE_SRCSTRIDE>>%1
		309	movsx tmpq, byte [tmp2q+eoq*4]
		310	add a_strideq, tmpq
		311	movsx tmpq, byte [tmp2q+eoq*4+2]
		312	add b_strideq, tmpq
		313	%endmacro
		314
		315	%macro HEVC_SAO_EDGE_FILTER_COMPUTE_8 1
		316	pminub m4, m1, m2
		317	pminub m5, m1, m3
		318	pcmpeqb m2, m4
		319	pcmpeqb m3, m5
		320	pcmpeqb m4, m1
		321	pcmpeqb m5, m1
		322	psubb m4, m2
		323	psubb m5, m3
		324	paddb m4, m6
		325	paddb m4, m5
		326
		327	pshufb m2, m0, m4
		328	%if %1 > 8
		329	punpckhbw m5, m7, m1
		330	punpckhbw m4, m2, m7
		331	punpcklbw m3, m7, m1
		332	punpcklbw m2, m7
		333	pmaddubsw m5, m4
		334	pmaddubsw m3, m2
		335	packuswb m3, m5
		336	%else
		337	punpcklbw m3, m7, m1
		338	punpcklbw m2, m7
		339	pmaddubsw m3, m2
		340	packuswb m3, m3
		341	%endif
		342	%endmacro
		343
		344	;void ff_hevc_sao_edge_filter__8_(uint8_t _dst, uint8_t _src, ptrdiff_t stride_dst, int16_t *sao_offset_val,
		345	; int eo, int width, int height);
		346	%macro HEVC_SAO_EDGE_FILTER_8 2-3
		347	%if ARCH_X86_64
		348	cglobal hevc_sao_edge_filter_%1_8, 4, 9, 8, dst, src, dststride, offset, eo, a_stride, b_stride, height, tmp
		349	%define tmp2q heightq
		350	HEVC_SAO_EDGE_FILTER_INIT 0
		351	mov heightd, r6m
		352
		353	%else ; ARCH_X86_32
		354	cglobal hevc_sao_edge_filter_%1_8, 1, 6, 8, dst, src, dststride, a_stride, b_stride, height
		355	%define eoq srcq
		356	%define tmpq heightq
		357	%define tmp2q dststrideq
		358	%define offsetq heightq
		359	HEVC_SAO_EDGE_FILTER_INIT 0
		360	mov srcq, srcm
		361	mov offsetq, r3m
		362	mov dststrideq, dststridem
		363	%endif ; ARCH
		364
		365	%if mmsize > 16
		366	vbroadcasti128 m0, [offsetq]
		367	%else
		368	movu m0, [offsetq]
		369	%endif
		370	mova m1, [pb_edge_shuffle]
		371	packsswb m0, m0
		372	mova m7, [pb_1]
		373	pshufb m0, m1
		374	mova m6, [pb_2]
		375	%if ARCH_X86_32
		376	mov heightd, r6m
		377	%endif
		378
		379	align 16
		380	.loop:
		381
		382	%if %1 == 8
		383	movq m1, [srcq]
		384	movq m2, [srcq + a_strideq]
		385	movq m3, [srcq + b_strideq]
		386	HEVC_SAO_EDGE_FILTER_COMPUTE_8 %1
		387	movq [dstq], m3
		388	%endif
		389
		390	%assign i 0
		391	%rep %2
		392	mova m1, [srcq + i]
		393	movu m2, [srcq + a_strideq + i]
		394	movu m3, [srcq + b_strideq + i]
		395	HEVC_SAO_EDGE_FILTER_COMPUTE_8 %1
		396	mov%3 [dstq + i], m3
		397	%assign i i+mmsize
		398	%endrep
		399
		400	%if %1 == 48
		401	INIT_XMM cpuname
		402
		403	mova m1, [srcq + i]
		404	movu m2, [srcq + a_strideq + i]
		405	movu m3, [srcq + b_strideq + i]
		406	HEVC_SAO_EDGE_FILTER_COMPUTE_8 %1
		407	mova [dstq + i], m3
		408	%if cpuflag(avx2)
		409	INIT_YMM cpuname
		410	%endif
		411	%endif
		412
		413	add dstq, dststrideq
		414	add srcq, EDGE_SRCSTRIDE
		415	dec heightd
		416	jg .loop
		417	RET
		418	%endmacro
		419
		420	%macro PMINUW 4
		421	%if cpuflag(sse4)
		422	pminuw %1, %2, %3
		423	%else
		424	psubusw %4, %2, %3
		425	psubw %1, %2, %4
		426	%endif
		427	%endmacro
		428
		429	%macro HEVC_SAO_EDGE_FILTER_COMPUTE_10 0
		430	PMINUW m4, m1, m2, m6
		431	PMINUW m5, m1, m3, m7
		432	pcmpeqw m2, m4
		433	pcmpeqw m3, m5
		434	pcmpeqw m4, m1
		435	pcmpeqw m5, m1
		436	psubw m4, m2
		437	psubw m5, m3
		438
		439	paddw m4, m5
		440	pcmpeqw m2, m4, [pw_m2]
		441	%if ARCH_X86_64
		442	pcmpeqw m3, m4, m13
		443	pcmpeqw m5, m4, m0
		444	pcmpeqw m6, m4, m14
		445	pcmpeqw m7, m4, m15
		446	pand m2, m8
		447	pand m3, m9
		448	pand m5, m10
		449	pand m6, m11
		450	pand m7, m12
		451	%else
		452	pcmpeqw m3, m4, [pw_m1]
		453	pcmpeqw m5, m4, m0
		454	pcmpeqw m6, m4, [pw_1]
		455	pcmpeqw m7, m4, [pw_2]
		456	pand m2, [rsp+MMSIZE*0]
		457	pand m3, [rsp+MMSIZE*1]
		458	pand m5, [rsp+MMSIZE*2]
		459	pand m6, [rsp+MMSIZE*3]
		460	pand m7, [rsp+MMSIZE*4]
		461	%endif
		462	paddw m2, m3
		463	paddw m5, m6
		464	paddw m2, m7
		465	paddw m2, m1
		466	paddw m2, m5
		467	%endmacro
		468
		469	;void ff_hevc_sao_edge_filter___(uint8_t _dst, uint8_t _src, ptrdiff_t stride_dst, int16_t *sao_offset_val,
		470	; int eo, int width, int height);
		471	%macro HEVC_SAO_EDGE_FILTER_16 3
		472	%if ARCH_X86_64
		473	cglobal hevc_sao_edge_filter_%2_%1, 4, 9, 16, dst, src, dststride, offset, eo, a_stride, b_stride, height, tmp
		474	%define tmp2q heightq
		475	HEVC_SAO_EDGE_FILTER_INIT 1
		476	mov heightd, r6m
		477	add a_strideq, a_strideq
		478	add b_strideq, b_strideq
		479
		480	%else ; ARCH_X86_32
		481	cglobal hevc_sao_edge_filter_%2_%1, 1, 6, 8, 5*mmsize, dst, src, dststride, a_stride, b_stride, height
		482	%assign MMSIZE mmsize
		483	%define eoq srcq
		484	%define tmpq heightq
		485	%define tmp2q dststrideq
		486	%define offsetq heightq
		487	%define m8 m1
		488	%define m9 m2
		489	%define m10 m3
		490	%define m11 m4
		491	%define m12 m5
		492	HEVC_SAO_EDGE_FILTER_INIT 1
		493	mov srcq, srcm
		494	mov offsetq, r3m
		495	mov dststrideq, dststridem
		496	add a_strideq, a_strideq
		497	add b_strideq, b_strideq
		498
		499	%endif ; ARCH
		500
		501	%if cpuflag(avx2)
		502	SPLATW m8, [offsetq+2]
		503	SPLATW m9, [offsetq+4]
		504	SPLATW m10, [offsetq+0]
		505	SPLATW m11, [offsetq+6]
		506	SPLATW m12, [offsetq+8]
		507	%else
		508	movq m10, [offsetq+0]
		509	movd m12, [offsetq+6]
		510	SPLATW m8, xm10, 1
		511	SPLATW m9, xm10, 2
		512	SPLATW m10, xm10, 0
		513	SPLATW m11, xm12, 0
		514	SPLATW m12, xm12, 1
		515	%endif
		516	pxor m0, m0
		517	%if ARCH_X86_64
		518	mova m13, [pw_m1]
		519	mova m14, [pw_1]
		520	mova m15, [pw_2]
		521	%else
		522	mov heightd, r6m
		523	mova [rsp+mmsize*0], m8
		524	mova [rsp+mmsize*1], m9
		525	mova [rsp+mmsize*2], m10
		526	mova [rsp+mmsize*3], m11
		527	mova [rsp+mmsize*4], m12
		528	%endif
		529
		530	align 16
		531	.loop:
		532
		533	%if %2 == 8
		534	mova m1, [srcq]
		535	movu m2, [srcq+a_strideq]
		536	movu m3, [srcq+b_strideq]
		537
		538	HEVC_SAO_EDGE_FILTER_COMPUTE_10
		539	CLIPW m2, m0, [pw_mask %+ %1]
		540	movu [dstq], m2
		541	%endif
		542
		543	%assign i 0
		544	%rep %3
		545	mova m1, [srcq + i]
		546	movu m2, [srcq+a_strideq + i]
		547	movu m3, [srcq+b_strideq + i]
		548	HEVC_SAO_EDGE_FILTER_COMPUTE_10
		549	CLIPW m2, m0, [pw_mask %+ %1]
		550	mova [dstq + i], m2
		551
		552	mova m1, [srcq + i + mmsize]
		553	movu m2, [srcq+a_strideq + i + mmsize]
		554	movu m3, [srcq+b_strideq + i + mmsize]
		555	HEVC_SAO_EDGE_FILTER_COMPUTE_10
		556	CLIPW m2, m0, [pw_mask %+ %1]
		557	mova [dstq + i + mmsize], m2
		558	%assign i i+mmsize*2
		559	%endrep
		560
		561	%if %2 == 48
		562	INIT_XMM cpuname
		563	mova m1, [srcq + i]
		564	movu m2, [srcq+a_strideq + i]
		565	movu m3, [srcq+b_strideq + i]
		566	HEVC_SAO_EDGE_FILTER_COMPUTE_10
		567	CLIPW m2, m0, [pw_mask %+ %1]
		568	mova [dstq + i], m2
		569
		570	mova m1, [srcq + i + mmsize]
		571	movu m2, [srcq+a_strideq + i + mmsize]
		572	movu m3, [srcq+b_strideq + i + mmsize]
		573	HEVC_SAO_EDGE_FILTER_COMPUTE_10
		574	CLIPW m2, m0, [pw_mask %+ %1]
		575	mova [dstq + i + mmsize], m2
		576	%if cpuflag(avx2)
		577	INIT_YMM cpuname
		578	%endif
		579	%endif
		580
		581	add dstq, dststrideq
		582	add srcq, EDGE_SRCSTRIDE
		583	dec heightd
		584	jg .loop
		585	RET
		586	%endmacro
		587
		588	INIT_XMM ssse3
		589	HEVC_SAO_EDGE_FILTER_8 8, 0
		590	HEVC_SAO_EDGE_FILTER_8 16, 1, a
		591	HEVC_SAO_EDGE_FILTER_8 32, 2, a
		592	HEVC_SAO_EDGE_FILTER_8 48, 2, a
		593	HEVC_SAO_EDGE_FILTER_8 64, 4, a
		594
		595	%if HAVE_AVX2_EXTERNAL
		596	INIT_YMM avx2
		597	HEVC_SAO_EDGE_FILTER_8 32, 1, a
		598	HEVC_SAO_EDGE_FILTER_8 48, 1, u
		599	HEVC_SAO_EDGE_FILTER_8 64, 2, a
		600	%endif
		601
		602	INIT_XMM sse2
		603	HEVC_SAO_EDGE_FILTER_16 10, 8, 0
		604	HEVC_SAO_EDGE_FILTER_16 10, 16, 1
		605	HEVC_SAO_EDGE_FILTER_16 10, 32, 2
		606	HEVC_SAO_EDGE_FILTER_16 10, 48, 2
		607	HEVC_SAO_EDGE_FILTER_16 10, 64, 4
		608
		609	HEVC_SAO_EDGE_FILTER_16 12, 8, 0
		610	HEVC_SAO_EDGE_FILTER_16 12, 16, 1
		611	HEVC_SAO_EDGE_FILTER_16 12, 32, 2
		612	HEVC_SAO_EDGE_FILTER_16 12, 48, 2
		613	HEVC_SAO_EDGE_FILTER_16 12, 64, 4
		614
		615	%if HAVE_AVX2_EXTERNAL
		616	INIT_YMM avx2
		617	HEVC_SAO_EDGE_FILTER_16 10, 32, 1
		618	HEVC_SAO_EDGE_FILTER_16 10, 48, 1
		619	HEVC_SAO_EDGE_FILTER_16 10, 64, 2
		620
		621	HEVC_SAO_EDGE_FILTER_16 12, 32, 1
		622	HEVC_SAO_EDGE_FILTER_16 12, 48, 1
		623	HEVC_SAO_EDGE_FILTER_16 12, 64, 2
		624	%endif

Subversion Repositories Kolibri OS

(root)/contrib/sdk/sources/ffmpeg/ffmpeg-2.8/libavcodec/x86/hevc_sao.asm – Rev 6147