WebSVN – Kolibri OS – Blame – /contrib/sdk/sources/ffmpeg/libavcodec/x86/h264_chromamc.asm

Rev	Author	Line No.	Line
4349	Serge	1	;******************************************************************************
		2	;* MMX/SSSE3-optimized functions for H264 chroma MC
		3	;* Copyright (c) 2005 Zoltan Hidvegi ,
		4	;* 2005-2008 Loren Merritt
		5	;*
		6	;* This file is part of FFmpeg.
		7	;*
		8	;* FFmpeg is free software; you can redistribute it and/or
		9	;* modify it under the terms of the GNU Lesser General Public
		10	;* License as published by the Free Software Foundation; either
		11	;* version 2.1 of the License, or (at your option) any later version.
		12	;*
		13	;* FFmpeg is distributed in the hope that it will be useful,
		14	;* but WITHOUT ANY WARRANTY; without even the implied warranty of
		15	;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
		16	;* Lesser General Public License for more details.
		17	;*
		18	;* You should have received a copy of the GNU Lesser General Public
		19	;* License along with FFmpeg; if not, write to the Free Software
		20	;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
		21	;******************************************************************************
		22
		23	%include "libavutil/x86/x86util.asm"
		24
		25	SECTION_RODATA
		26
		27	rnd_rv40_2d_tbl: times 4 dw 0
		28	times 4 dw 16
		29	times 4 dw 32
		30	times 4 dw 16
		31	times 4 dw 32
		32	times 4 dw 28
		33	times 4 dw 32
		34	times 4 dw 28
		35	times 4 dw 0
		36	times 4 dw 32
		37	times 4 dw 16
		38	times 4 dw 32
		39	times 4 dw 32
		40	times 4 dw 28
		41	times 4 dw 32
		42	times 4 dw 28
		43	rnd_rv40_1d_tbl: times 4 dw 0
		44	times 4 dw 2
		45	times 4 dw 4
		46	times 4 dw 2
		47	times 4 dw 4
		48	times 4 dw 3
		49	times 4 dw 4
		50	times 4 dw 3
		51	times 4 dw 0
		52	times 4 dw 4
		53	times 4 dw 2
		54	times 4 dw 4
		55	times 4 dw 4
		56	times 4 dw 3
		57	times 4 dw 4
		58	times 4 dw 3
		59
		60	cextern pw_3
		61	cextern pw_4
		62	cextern pw_8
		63	pw_28: times 8 dw 28
		64	cextern pw_32
		65	cextern pw_64
		66
		67	SECTION .text
		68
		69	%macro mv0_pixels_mc8 0
		70	lea r4, [r2*2 ]
		71	.next4rows:
		72	movq mm0, [r1 ]
		73	movq mm1, [r1+r2]
		74	add r1, r4
		75	CHROMAMC_AVG mm0, [r0 ]
		76	CHROMAMC_AVG mm1, [r0+r2]
		77	movq [r0 ], mm0
		78	movq [r0+r2], mm1
		79	add r0, r4
		80	movq mm0, [r1 ]
		81	movq mm1, [r1+r2]
		82	add r1, r4
		83	CHROMAMC_AVG mm0, [r0 ]
		84	CHROMAMC_AVG mm1, [r0+r2]
		85	movq [r0 ], mm0
		86	movq [r0+r2], mm1
		87	add r0, r4
		88	sub r3d, 4
		89	jne .next4rows
		90	%endmacro
		91
		92	%macro chroma_mc8_mmx_func 2-3
		93	%ifidn %2, rv40
		94	%ifdef PIC
		95	%define rnd_1d_rv40 r8
		96	%define rnd_2d_rv40 r8
		97	%define extra_regs 2
		98	%else ; no-PIC
		99	%define rnd_1d_rv40 rnd_rv40_1d_tbl
		100	%define rnd_2d_rv40 rnd_rv40_2d_tbl
		101	%define extra_regs 1
		102	%endif ; PIC
		103	%else
		104	%define extra_regs 0
		105	%endif ; rv40
		106	; put/avg_h264_chroma_mc8_(uint8_t dst /align 8/, uint8_t src /align 1*/,
		107	; int stride, int h, int mx, int my)
		108	cglobal %1_%2_chroma_mc8%3, 6, 7 + extra_regs, 0
		109	%if ARCH_X86_64
		110	movsxd r2, r2d
		111	%endif
		112	mov r6d, r5d
		113	or r6d, r4d
		114	jne .at_least_one_non_zero
		115	; mx == 0 AND my == 0 - no filter needed
		116	mv0_pixels_mc8
		117	REP_RET
		118
		119	.at_least_one_non_zero:
		120	%ifidn %2, rv40
		121	%if ARCH_X86_64
		122	mov r7, r5
		123	and r7, 6 ; &~1 for mx/my=[0,7]
		124	lea r7, [r7*4+r4]
		125	sar r7d, 1
		126	%define rnd_bias r7
		127	%define dest_reg r0
		128	%else ; x86-32
		129	mov r0, r5
		130	and r0, 6 ; &~1 for mx/my=[0,7]
		131	lea r0, [r0*4+r4]
		132	sar r0d, 1
		133	%define rnd_bias r0
		134	%define dest_reg r5
		135	%endif
		136	%else ; vc1, h264
		137	%define rnd_bias 0
		138	%define dest_reg r0
		139	%endif
		140
		141	test r5d, r5d
		142	mov r6, 1
		143	je .my_is_zero
		144	test r4d, r4d
		145	mov r6, r2 ; dxy = x ? 1 : stride
		146	jne .both_non_zero
		147	.my_is_zero:
		148	; mx == 0 XOR my == 0 - 1 dimensional filter only
		149	or r4d, r5d ; x + y
		150
		151	%ifidn %2, rv40
		152	%ifdef PIC
		153	lea r8, [rnd_rv40_1d_tbl]
		154	%endif
		155	%if ARCH_X86_64 == 0
		156	mov r5, r0m
		157	%endif
		158	%endif
		159
		160	movd m5, r4d
		161	movq m4, [pw_8]
		162	movq m6, [rnd_1d_%2+rnd_bias*8] ; mm6 = rnd >> 3
		163	punpcklwd m5, m5
		164	punpckldq m5, m5 ; mm5 = B = x
		165	pxor m7, m7
		166	psubw m4, m5 ; mm4 = A = 8-x
		167
		168	.next1drow:
		169	movq m0, [r1 ] ; mm0 = src[0..7]
		170	movq m2, [r1+r6] ; mm1 = src[1..8]
		171
		172	movq m1, m0
		173	movq m3, m2
		174	punpcklbw m0, m7
		175	punpckhbw m1, m7
		176	punpcklbw m2, m7
		177	punpckhbw m3, m7
		178	pmullw m0, m4 ; [mm0,mm1] = A * src[0..7]
		179	pmullw m1, m4
		180	pmullw m2, m5 ; [mm2,mm3] = B * src[1..8]
		181	pmullw m3, m5
		182
		183	paddw m0, m6
		184	paddw m1, m6
		185	paddw m0, m2
		186	paddw m1, m3
		187	psrlw m0, 3
		188	psrlw m1, 3
		189	packuswb m0, m1
		190	CHROMAMC_AVG m0, [dest_reg]
		191	movq [dest_reg], m0 ; dst[0..7] = (A * src[0..7] + B * src[1..8] + (rnd >> 3)) >> 3
		192
		193	add dest_reg, r2
		194	add r1, r2
		195	dec r3d
		196	jne .next1drow
		197	REP_RET
		198
		199	.both_non_zero: ; general case, bilinear
		200	movd m4, r4d ; x
		201	movd m6, r5d ; y
		202	%ifidn %2, rv40
		203	%ifdef PIC
		204	lea r8, [rnd_rv40_2d_tbl]
		205	%endif
		206	%if ARCH_X86_64 == 0
		207	mov r5, r0m
		208	%endif
		209	%endif
		210	mov r6, rsp ; backup stack pointer
		211	and rsp, ~(mmsize-1) ; align stack
		212	sub rsp, 16 ; AA and DD
		213
		214	punpcklwd m4, m4
		215	punpcklwd m6, m6
		216	punpckldq m4, m4 ; mm4 = x words
		217	punpckldq m6, m6 ; mm6 = y words
		218	movq m5, m4
		219	pmullw m4, m6 ; mm4 = x * y
		220	psllw m5, 3
		221	psllw m6, 3
		222	movq m7, m5
		223	paddw m7, m6
		224	movq [rsp+8], m4 ; DD = x * y
		225	psubw m5, m4 ; mm5 = B = 8x - xy
		226	psubw m6, m4 ; mm6 = C = 8y - xy
		227	paddw m4, [pw_64]
		228	psubw m4, m7 ; mm4 = A = xy - (8x+8y) + 64
		229	pxor m7, m7
		230	movq [rsp ], m4
		231
		232	movq m0, [r1 ] ; mm0 = src[0..7]
		233	movq m1, [r1+1] ; mm1 = src[1..8]
		234	.next2drow:
		235	add r1, r2
		236
		237	movq m2, m0
		238	movq m3, m1
		239	punpckhbw m0, m7
		240	punpcklbw m1, m7
		241	punpcklbw m2, m7
		242	punpckhbw m3, m7
		243	pmullw m0, [rsp]
		244	pmullw m2, [rsp]
		245	pmullw m1, m5
		246	pmullw m3, m5
		247	paddw m2, m1 ; mm2 = A * src[0..3] + B * src[1..4]
		248	paddw m3, m0 ; mm3 = A * src[4..7] + B * src[5..8]
		249
		250	movq m0, [r1]
		251	movq m1, m0
		252	punpcklbw m0, m7
		253	punpckhbw m1, m7
		254	pmullw m0, m6
		255	pmullw m1, m6
		256	paddw m2, m0
		257	paddw m3, m1 ; [mm2,mm3] += C * src[0..7]
		258
		259	movq m1, [r1+1]
		260	movq m0, m1
		261	movq m4, m1
		262	punpcklbw m0, m7
		263	punpckhbw m4, m7
		264	pmullw m0, [rsp+8]
		265	pmullw m4, [rsp+8]
		266	paddw m2, m0
		267	paddw m3, m4 ; [mm2,mm3] += D * src[1..8]
		268	movq m0, [r1]
		269
		270	paddw m2, [rnd_2d_%2+rnd_bias*8]
		271	paddw m3, [rnd_2d_%2+rnd_bias*8]
		272	psrlw m2, 6
		273	psrlw m3, 6
		274	packuswb m2, m3
		275	CHROMAMC_AVG m2, [dest_reg]
		276	movq [dest_reg], m2 ; dst[0..7] = ([mm2,mm3] + rnd) >> 6
		277
		278	add dest_reg, r2
		279	dec r3d
		280	jne .next2drow
		281	mov rsp, r6 ; restore stack pointer
		282	RET
		283	%endmacro
		284
		285	%macro chroma_mc4_mmx_func 2
		286	%define extra_regs 0
		287	%ifidn %2, rv40
		288	%ifdef PIC
		289	%define extra_regs 1
		290	%endif ; PIC
		291	%endif ; rv40
		292	cglobal %1_%2_chroma_mc4, 6, 6 + extra_regs, 0
		293	%if ARCH_X86_64
		294	movsxd r2, r2d
		295	%endif
		296	pxor m7, m7
		297	movd m2, r4d ; x
		298	movd m3, r5d ; y
		299	movq m4, [pw_8]
		300	movq m5, [pw_8]
		301	punpcklwd m2, m2
		302	punpcklwd m3, m3
		303	punpcklwd m2, m2
		304	punpcklwd m3, m3
		305	psubw m4, m2
		306	psubw m5, m3
		307
		308	%ifidn %2, rv40
		309	%ifdef PIC
		310	lea r6, [rnd_rv40_2d_tbl]
		311	%define rnd_2d_rv40 r6
		312	%else
		313	%define rnd_2d_rv40 rnd_rv40_2d_tbl
		314	%endif
		315	and r5, 6 ; &~1 for mx/my=[0,7]
		316	lea r5, [r5*4+r4]
		317	sar r5d, 1
		318	%define rnd_bias r5
		319	%else ; vc1, h264
		320	%define rnd_bias 0
		321	%endif
		322
		323	movd m0, [r1 ]
		324	movd m6, [r1+1]
		325	add r1, r2
		326	punpcklbw m0, m7
		327	punpcklbw m6, m7
		328	pmullw m0, m4
		329	pmullw m6, m2
		330	paddw m6, m0
		331
		332	.next2rows:
		333	movd m0, [r1 ]
		334	movd m1, [r1+1]
		335	add r1, r2
		336	punpcklbw m0, m7
		337	punpcklbw m1, m7
		338	pmullw m0, m4
		339	pmullw m1, m2
		340	paddw m1, m0
		341	movq m0, m1
		342
		343	pmullw m6, m5
		344	pmullw m1, m3
		345	paddw m6, [rnd_2d_%2+rnd_bias*8]
		346	paddw m1, m6
		347	psrlw m1, 6
		348	packuswb m1, m1
		349	CHROMAMC_AVG4 m1, m6, [r0]
		350	movd [r0], m1
		351	add r0, r2
		352
		353	movd m6, [r1 ]
		354	movd m1, [r1+1]
		355	add r1, r2
		356	punpcklbw m6, m7
		357	punpcklbw m1, m7
		358	pmullw m6, m4
		359	pmullw m1, m2
		360	paddw m1, m6
		361	movq m6, m1
		362	pmullw m0, m5
		363	pmullw m1, m3
		364	paddw m0, [rnd_2d_%2+rnd_bias*8]
		365	paddw m1, m0
		366	psrlw m1, 6
		367	packuswb m1, m1
		368	CHROMAMC_AVG4 m1, m0, [r0]
		369	movd [r0], m1
		370	add r0, r2
		371	sub r3d, 2
		372	jnz .next2rows
		373	REP_RET
		374	%endmacro
		375
		376	%macro chroma_mc2_mmx_func 2
		377	cglobal %1_%2_chroma_mc2, 6, 7, 0
		378	%if ARCH_X86_64
		379	movsxd r2, r2d
		380	%endif
		381
		382	mov r6d, r4d
		383	shl r4d, 16
		384	sub r4d, r6d
		385	add r4d, 8
		386	imul r5d, r4d ; xy<<16 \| y(8-x)
		387	shl r4d, 3
		388	sub r4d, r5d ; x(8-y)<<16 \| (8-x)(8-y)
		389
		390	movd m5, r4d
		391	movd m6, r5d
		392	punpckldq m5, m5 ; mm5 = {A,B,A,B}
		393	punpckldq m6, m6 ; mm6 = {C,D,C,D}
		394	pxor m7, m7
		395	movd m2, [r1]
		396	punpcklbw m2, m7
		397	pshufw m2, m2, 0x94 ; mm0 = src[0,1,1,2]
		398
		399	.nextrow:
		400	add r1, r2
		401	movq m1, m2
		402	pmaddwd m1, m5 ; mm1 = A * src[0,1] + B * src[1,2]
		403	movd m0, [r1]
		404	punpcklbw m0, m7
		405	pshufw m0, m0, 0x94 ; mm0 = src[0,1,1,2]
		406	movq m2, m0
		407	pmaddwd m0, m6
		408	paddw m1, [rnd_2d_%2]
		409	paddw m1, m0 ; mm1 += C * src[0,1] + D * src[1,2]
		410	psrlw m1, 6
		411	packssdw m1, m7
		412	packuswb m1, m7
		413	CHROMAMC_AVG4 m1, m3, [r0]
		414	movd r5d, m1
		415	mov [r0], r5w
		416	add r0, r2
		417	sub r3d, 1
		418	jnz .nextrow
		419	REP_RET
		420	%endmacro
		421
		422	%define rnd_1d_h264 pw_4
		423	%define rnd_2d_h264 pw_32
		424	%define rnd_1d_vc1 pw_3
		425	%define rnd_2d_vc1 pw_28
		426
		427	%macro NOTHING 2-3
		428	%endmacro
		429	%macro DIRECT_AVG 2
		430	PAVGB %1, %2
		431	%endmacro
		432	%macro COPY_AVG 3
		433	movd %2, %3
		434	PAVGB %1, %2
		435	%endmacro
		436
		437	INIT_MMX mmx
		438	%define CHROMAMC_AVG NOTHING
		439	%define CHROMAMC_AVG4 NOTHING
		440	chroma_mc8_mmx_func put, h264, _rnd
		441	chroma_mc8_mmx_func put, vc1, _nornd
		442	chroma_mc8_mmx_func put, rv40
		443	chroma_mc4_mmx_func put, h264
		444	chroma_mc4_mmx_func put, rv40
		445
		446	INIT_MMX mmxext
		447	chroma_mc2_mmx_func put, h264
		448
		449	%define CHROMAMC_AVG DIRECT_AVG
		450	%define CHROMAMC_AVG4 COPY_AVG
		451	chroma_mc8_mmx_func avg, h264, _rnd
		452	chroma_mc8_mmx_func avg, vc1, _nornd
		453	chroma_mc8_mmx_func avg, rv40
		454	chroma_mc4_mmx_func avg, h264
		455	chroma_mc4_mmx_func avg, rv40
		456	chroma_mc2_mmx_func avg, h264
		457
		458	INIT_MMX 3dnow
		459	chroma_mc8_mmx_func avg, h264, _rnd
		460	chroma_mc8_mmx_func avg, vc1, _nornd
		461	chroma_mc8_mmx_func avg, rv40
		462	chroma_mc4_mmx_func avg, h264
		463	chroma_mc4_mmx_func avg, rv40
		464
		465	%macro chroma_mc8_ssse3_func 2-3
		466	cglobal %1_%2_chroma_mc8%3, 6, 7, 8
		467	%if ARCH_X86_64
		468	movsxd r2, r2d
		469	%endif
		470	mov r6d, r5d
		471	or r6d, r4d
		472	jne .at_least_one_non_zero
		473	; mx == 0 AND my == 0 - no filter needed
		474	mv0_pixels_mc8
		475	REP_RET
		476
		477	.at_least_one_non_zero:
		478	test r5d, r5d
		479	je .my_is_zero
		480	test r4d, r4d
		481	je .mx_is_zero
		482
		483	; general case, bilinear
		484	mov r6d, r4d
		485	shl r4d, 8
		486	sub r4, r6
		487	mov r6, 8
		488	add r4, 8 ; x*288+8 = x<<8 \| (8-x)
		489	sub r6d, r5d
		490	imul r6, r4 ; (8-y)(x255+8) = (8-y)x<<8 \| (8-y)(8-x)
		491	imul r4d, r5d ; y (x255+8) = y x<<8 \| y (8-x)
		492
		493	movd m7, r6d
		494	movd m6, r4d
		495	movdqa m5, [rnd_2d_%2]
		496	movq m0, [r1 ]
		497	movq m1, [r1+1]
		498	pshuflw m7, m7, 0
		499	pshuflw m6, m6, 0
		500	punpcklbw m0, m1
		501	movlhps m7, m7
		502	movlhps m6, m6
		503
		504	.next2rows:
		505	movq m1, [r1+r2*1 ]
		506	movq m2, [r1+r2*1+1]
		507	movq m3, [r1+r2*2 ]
		508	movq m4, [r1+r2*2+1]
		509	lea r1, [r1+r2*2]
		510	punpcklbw m1, m2
		511	movdqa m2, m1
		512	punpcklbw m3, m4
		513	movdqa m4, m3
		514	pmaddubsw m0, m7
		515	pmaddubsw m1, m6
		516	pmaddubsw m2, m7
		517	pmaddubsw m3, m6
		518	paddw m0, m5
		519	paddw m2, m5
		520	paddw m1, m0
		521	paddw m3, m2
		522	psrlw m1, 6
		523	movdqa m0, m4
		524	psrlw m3, 6
		525	%ifidn %1, avg
		526	movq m2, [r0 ]
		527	movhps m2, [r0+r2]
		528	%endif
		529	packuswb m1, m3
		530	CHROMAMC_AVG m1, m2
		531	movq [r0 ], m1
		532	movhps [r0+r2], m1
		533	sub r3d, 2
		534	lea r0, [r0+r2*2]
		535	jg .next2rows
		536	REP_RET
		537
		538	.my_is_zero:
		539	mov r5d, r4d
		540	shl r4d, 8
		541	add r4, 8
		542	sub r4, r5 ; 255*x+8 = x<<8 \| (8-x)
		543	movd m7, r4d
		544	movdqa m6, [rnd_1d_%2]
		545	pshuflw m7, m7, 0
		546	movlhps m7, m7
		547
		548	.next2xrows:
		549	movq m0, [r1 ]
		550	movq m1, [r1 +1]
		551	movq m2, [r1+r2 ]
		552	movq m3, [r1+r2+1]
		553	punpcklbw m0, m1
		554	punpcklbw m2, m3
		555	pmaddubsw m0, m7
		556	pmaddubsw m2, m7
		557	%ifidn %1, avg
		558	movq m4, [r0 ]
		559	movhps m4, [r0+r2]
		560	%endif
		561	paddw m0, m6
		562	paddw m2, m6
		563	psrlw m0, 3
		564	psrlw m2, 3
		565	packuswb m0, m2
		566	CHROMAMC_AVG m0, m4
		567	movq [r0 ], m0
		568	movhps [r0+r2], m0
		569	sub r3d, 2
		570	lea r0, [r0+r2*2]
		571	lea r1, [r1+r2*2]
		572	jg .next2xrows
		573	REP_RET
		574
		575	.mx_is_zero:
		576	mov r4d, r5d
		577	shl r5d, 8
		578	add r5, 8
		579	sub r5, r4 ; 255*y+8 = y<<8 \| (8-y)
		580	movd m7, r5d
		581	movdqa m6, [rnd_1d_%2]
		582	pshuflw m7, m7, 0
		583	movlhps m7, m7
		584
		585	.next2yrows:
		586	movq m0, [r1 ]
		587	movq m1, [r1+r2 ]
		588	movdqa m2, m1
		589	movq m3, [r1+r2*2]
		590	lea r1, [r1+r2*2]
		591	punpcklbw m0, m1
		592	punpcklbw m2, m3
		593	pmaddubsw m0, m7
		594	pmaddubsw m2, m7
		595	%ifidn %1, avg
		596	movq m4, [r0 ]
		597	movhps m4, [r0+r2]
		598	%endif
		599	paddw m0, m6
		600	paddw m2, m6
		601	psrlw m0, 3
		602	psrlw m2, 3
		603	packuswb m0, m2
		604	CHROMAMC_AVG m0, m4
		605	movq [r0 ], m0
		606	movhps [r0+r2], m0
		607	sub r3d, 2
		608	lea r0, [r0+r2*2]
		609	jg .next2yrows
		610	REP_RET
		611	%endmacro
		612
		613	%macro chroma_mc4_ssse3_func 2
		614	cglobal %1_%2_chroma_mc4, 6, 7, 0
		615	%if ARCH_X86_64
		616	movsxd r2, r2d
		617	%endif
		618	mov r6, r4
		619	shl r4d, 8
		620	sub r4d, r6d
		621	mov r6, 8
		622	add r4d, 8 ; x*288+8
		623	sub r6d, r5d
		624	imul r6d, r4d ; (8-y)(x255+8) = (8-y)x<<8 \| (8-y)(8-x)
		625	imul r4d, r5d ; y (x255+8) = y x<<8 \| y (8-x)
		626
		627	movd m7, r6d
		628	movd m6, r4d
		629	movq m5, [pw_32]
		630	movd m0, [r1 ]
		631	pshufw m7, m7, 0
		632	punpcklbw m0, [r1+1]
		633	pshufw m6, m6, 0
		634
		635	.next2rows:
		636	movd m1, [r1+r2*1 ]
		637	movd m3, [r1+r2*2 ]
		638	punpcklbw m1, [r1+r2*1+1]
		639	punpcklbw m3, [r1+r2*2+1]
		640	lea r1, [r1+r2*2]
		641	movq m2, m1
		642	movq m4, m3
		643	pmaddubsw m0, m7
		644	pmaddubsw m1, m6
		645	pmaddubsw m2, m7
		646	pmaddubsw m3, m6
		647	paddw m0, m5
		648	paddw m2, m5
		649	paddw m1, m0
		650	paddw m3, m2
		651	psrlw m1, 6
		652	movq m0, m4
		653	psrlw m3, 6
		654	packuswb m1, m1
		655	packuswb m3, m3
		656	CHROMAMC_AVG m1, [r0 ]
		657	CHROMAMC_AVG m3, [r0+r2]
		658	movd [r0 ], m1
		659	movd [r0+r2], m3
		660	sub r3d, 2
		661	lea r0, [r0+r2*2]
		662	jg .next2rows
		663	REP_RET
		664	%endmacro
		665
		666	%define CHROMAMC_AVG NOTHING
		667	INIT_XMM ssse3
		668	chroma_mc8_ssse3_func put, h264, _rnd
		669	chroma_mc8_ssse3_func put, vc1, _nornd
		670	INIT_MMX ssse3
		671	chroma_mc4_ssse3_func put, h264
		672
		673	%define CHROMAMC_AVG DIRECT_AVG
		674	INIT_XMM ssse3
		675	chroma_mc8_ssse3_func avg, h264, _rnd
		676	chroma_mc8_ssse3_func avg, vc1, _nornd
		677	INIT_MMX ssse3
		678	chroma_mc4_ssse3_func avg, h264

Subversion Repositories Kolibri OS

(root)/contrib/sdk/sources/ffmpeg/libavcodec/x86/h264_chromamc.asm – Rev 4349