Remove PPC perf counter support

This functionality is better accessed through tools like oprofile. Originally committed as revision 23808 to svn://svn.ffmpeg.org/ffmpeg/trunk

Måns Rullgård committed 16 years ago

Commit 2829ce41 parent a78819613 files · +0 −513Browse files

Unified · Split

13 files

configure +0 −4
doc/ffmpeg_powerpc_performance_evaluation_howto.txt +0 −172
ffmpeg.c +0 −5
libavcodec/ppc/dsputil_altivec.c +0 −42
libavcodec/ppc/dsputil_ppc.c +0 −79
libavcodec/ppc/dsputil_ppc.h +0 −154
libavcodec/ppc/fdct_altivec.c +0 −7
libavcodec/ppc/fft_altivec.c +0 −6
libavcodec/ppc/gmc_altivec.c +0 −8
libavcodec/ppc/h264_altivec.c +0 −1
libavcodec/ppc/h264_template_altivec.c +0 −16
libavcodec/ppc/idct_altivec.c +0 −14
libavcodec/ppc/mpegvideo_altivec.c +0 −5

configure+0−4

@@ -211,8 +211,6 @@ Advanced options (experts only): · 210 unmodified lines
211	211	--arch=ARCH select architecture [$arch]
212	212	--cpu=CPU select the minimum required CPU (affects
213	213	instruction selection, may crash on older CPUs)
214		− --enable-powerpc-perf enable performance report on PPC
215		− (requires enabling PMC)
216	214	--disable-asm disable all assembler optimizations
217	215	--disable-altivec disable AltiVec optimizations
218	216	--disable-amd3dnow disable 3DNow! optimizations
@@ -886,7 +884,6 @@ CONFIG_LIST=" · 667 unmodified lines
886	884	nonfree
887	885	pic
888	886	postproc
889		− powerpc_perf
890	887	rdft
891	888	runtime_cpudetect
892	889	shared
@@ -2772,7 +2769,6 @@ if enabled ppc; then · 1879 unmodified lines
2772	2769	echo "AltiVec enabled ${altivec-no}"
2773	2770	echo "PPC 4xx optimizations ${ppc4xx-no}"
2774	2771	echo "dcbzl available ${dcbzl-no}"
2775		− echo "performance report ${powerpc_perf-no}"
2776	2772	fi
2777	2773	if enabled sparc; then
2778	2774	echo "VIS enabled ${vis-no}"

doc/ffmpeg_powerpc_performance_evaluation_howto.txt+0−172

@@ -1,172 +0,0 @@
1		−FFmpeg & evaluating performance on the PowerPC Architecture HOWTO
2		−
3		−(c) 2003-2004 Romain Dolbeau <romain@dolbeau.org>
4		−
5		−
6		−
7		−I - Introduction
8		−
9		−The PowerPC architecture and its SIMD extension AltiVec offer some
10		−interesting tools to evaluate performance and improve the code.
11		−This document tries to explain how to use those tools with FFmpeg.
12		−
13		−The architecture itself offers two ways to evaluate the performance of
14		−a given piece of code:
15		−
16		−1) The Time Base Registers (TBL)
17		−2) The Performance Monitor Counter Registers (PMC)
18		−
19		−The first ones are always available, always active, but they're not very
20		−accurate: the registers increment by one every four bus cycles. On
21		−my 667 Mhz tiBook (ppc7450), this means once every twenty processor
22		−cycles. So we won't use that.
23		−
24		−The PMC are much more useful: not only can they report cycle-accurate
25		−timing, but they can also be used to monitor many other parameters,
26		−such as the number of AltiVec stalls for every kind of instruction,
27		−or instruction cache misses. The downside is that not all processors
28		−support the PMC (all G3, all G4 and the 970 do support them), and
29		−they're inactive by default - you need to activate them with a
30		−dedicated tool. Also, the number of available PMC depends on the
31		−procesor: the various 604 have 2, the various 75x (aka. G3) have 4,
32		−and the various 74xx (aka G4) have 6.
33		−
34		−WARNING: The PowerPC 970 is not very well documented, and its PMC
35		−registers are 64 bits wide. To properly notify the code, you must
36		−tune for the 970 (using --tune=970), or the code will assume 32 bit
37		−registers.
38		−
39		−
40		−II - Enabling FFmpeg PowerPC performance support
41		−
42		−This needs to be done by hand. First, you need to configure FFmpeg as
43		−usual, but add the "--powerpc-perf-enable" option. For instance:
44		−
45		−#####
46		−./configure --prefix=/usr/local/ffmpeg-svn --cc=gcc-3.3 --tune=7450 --powerpc-perf-enable
47		−#####
48		−
49		−This will configure FFmpeg to install inside /usr/local/ffmpeg-svn,
50		−compiling with gcc-3.3 (you should try to use this one or a newer
51		−gcc), and tuning for the PowerPC 7450 (i.e. the newer G4; as a rule of
52		−thumb, those at 550Mhz and more). It will also enable the PMC.
53		−
54		−You may also edit the file "config.h" to enable the following line:
55		−
56		−#####
57		−// #define ALTIVEC_USE_REFERENCE_C_CODE 1
58		−#####
59		−
60		−If you enable this line, then the code will not make use of AltiVec,
61		−but will use the reference C code instead. This is useful to compare
62		−performance between two versions of the code.
63		−
64		−Also, the number of enabled PMC is defined in "libavcodec/ppc/dsputil_ppc.h":
65		−
66		−#####
67		−#define POWERPC_NUM_PMC_ENABLED 4
68		−#####
69		−
70		−If you have a G4 CPU, you can enable all 6 PMC. DO NOT enable more
71		−PMC than available on your CPU!
72		−
73		−Then, simply compile FFmpeg as usual (make && make install).
74		−
75		−
76		−
77		−III - Using FFmpeg PowerPC performance support
78		−
79		−This FFmeg can be used exactly as usual. But before exiting, FFmpeg
80		−will dump a per-function report that looks like this:
81		−
82		−#####
83		−PowerPC performance report
84		− Values are from the PMC registers, and represent whatever the
85		− registers are set to record.
86		− Function "gmc1_altivec" (pmc1):
87		− min: 231
88		− max: 1339867
89		− avg: 558.25 (255302)
90		− Function "gmc1_altivec" (pmc2):
91		− min: 93
92		− max: 2164
93		− avg: 267.31 (255302)
94		− Function "gmc1_altivec" (pmc3):
95		− min: 72
96		− max: 1987
97		− avg: 276.20 (255302)
98		−(...)
99		−#####
100		−
101		−In this example, PMC1 was set to record CPU cycles, PMC2 was set to
102		−record AltiVec Permute Stall Cycles, and PMC3 was set to record AltiVec
103		−Issue Stalls.
104		−
105		−The function "gmc1_altivec" was monitored 255302 times, and the
106		−minimum execution time was 231 processor cycles. The max and average
107		−aren't much use, as it's very likely the OS interrupted execution for
108		−reasons of its own :-(
109		−
110		−With the exact same settings and source file, but using the reference C
111		−code we get:
112		−
113		−#####
114		−PowerPC performance report
115		− Values are from the PMC registers, and represent whatever the
116		− registers are set to record.
117		− Function "gmc1_altivec" (pmc1):
118		− min: 592
119		− max: 2532235
120		− avg: 962.88 (255302)
121		− Function "gmc1_altivec" (pmc2):
122		− min: 0
123		− max: 33
124		− avg: 0.00 (255302)
125		− Function "gmc1_altivec" (pmc3):
126		− min: 0
127		− max: 350
128		− avg: 0.03 (255302)
129		−(...)
130		−#####
131		−
132		−592 cycles, so the fastest AltiVec execution is about 2.5x faster than
133		−the fastest C execution in this example. It's not perfect but it's not
134		−bad (well I wrote this function so I can't say otherwise :-).
135		−
136		−Once you have that kind of report, you can try to improve things by
137		−finding what goes wrong and fixing it; in the example above, one
138		−should try to diminish the number of AltiVec stalls, as this may
139		−improve performance.
140		−
141		−
142		−
143		−IV) Enabling the PMC in Mac OS X
144		−
145		−This is easy. Use "Monster" and "monster". Those tools come from
146		−Apple's CHUD package, and can be found hidden in the developer web
147		−site & FTP site. "MONster" is the graphical application, use it to
148		−generate a config file specifying what each register should
149		−monitor. Then use the command-line application "monster" to use that
150		−config file, and enjoy the results.
151		−
152		−Note that "MONster" can be used for many other things, but it's
153		−documented by Apple, it's not my subject.
154		−
155		−If you are using CHUD 4.4.2 or later, you'll notice that MONster is
156		−no longer available. It's been superseeded by Shark, where
157		−configuration of PMCs is available as a plugin.
158		−
159		−
160		−
161		−V) Enabling the PMC on Linux
162		−
163		−On linux you may use oprofile from http://oprofile.sf.net, depending on the
164		−version and the cpu you may need to apply a patch[1] to access a set of the
165		−possibile counters from the userspace application. You can always define them
166		−using the kernel interface /dev/oprofile/* .
167		−
168		−[1] http://dev.gentoo.org/~lu_zero/development/oprofile-g4-20060423.patch
169		−
170		−--
171		−Romain Dolbeau <romain@dolbeau.org>
172		−Luca Barbato <lu_zero@gentoo.org>

ffmpeg.c+0−5

@@ -618,11 +618,6 @@ static int av_exit(int ret) · 617 unmodified lines
618	618
619	619	av_free(video_standard);
620	620
621		−#if CONFIG_POWERPC_PERF
622		− void powerpc_display_perf_report(void);
623		− powerpc_display_perf_report();
624		−#endif /* CONFIG_POWERPC_PERF */
625		−
626	621	for (i=0;i<AVMEDIA_TYPE_NB;i++)
627	622	av_free(avcodec_opts[i]);
628	623	av_free(avformat_opts);

libavcodec/ppc/dsputil_altivec.c+0−42

@@ -25,7 +25,6 @@ · 24 unmodified lines
25	25	#include <altivec.h>
26	26	#endif
27	27	#include "libavcodec/dsputil.h"
28		−#include "dsputil_ppc.h"
29	28	#include "util_altivec.h"
30	29	#include "types_altivec.h"
31	30	#include "dsputil_altivec.h"
@@ -610,7 +609,6 @@ static void add_bytes_altivec(uint8_t dst, uint8_t src, int w) { · 578 unmodified lines
610	609	/* next one assumes that ((line_size % 16) == 0) */
611	610	void put_pixels16_altivec(uint8_t block, const uint8_t pixels, int line_size, int h)
612	611	{
613		−POWERPC_PERF_DECLARE(altivec_put_pixels16_num, 1);
614	612	register vector unsigned char pixelsv1, pixelsv2;
615	613	register vector unsigned char pixelsv1B, pixelsv2B;
616	614	register vector unsigned char pixelsv1C, pixelsv2C;
@@ -622,7 +620,6 @@ POWERPC_PERF_DECLARE(altivec_put_pixels16_num, 1); · 5 unmodified lines
622	620	register int line_size_3 = line_size + line_size_2;
623	621	register int line_size_4 = line_size << 2;
624	622
625		−POWERPC_PERF_START_COUNT(altivec_put_pixels16_num, 1);
626	623	// hand-unrolling the loop by 4 gains about 15%
627	624	// mininum execution time goes from 74 to 60 cycles
628	625	// it's faster than -funroll-loops, but using
@@ -659,20 +656,16 @@ POWERPC_PERF_START_COUNT(altivec_put_pixels16_num, 1); · 30 unmodified lines
659	656	block +=line_size_4;
660	657	}
661	658	#endif
662		−POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_num, 1);
663	659	}
664	660
665	661	/* next one assumes that ((line_size % 16) == 0) */
666	662	#define op_avg(a,b) a = ( ((a)\|(b)) - ((((a)^(b))&0xFEFEFEFEUL)>>1) )
667	663	void avg_pixels16_altivec(uint8_t block, const uint8_t pixels, int line_size, int h)
668	664	{
669		−POWERPC_PERF_DECLARE(altivec_avg_pixels16_num, 1);
670	665	register vector unsigned char pixelsv1, pixelsv2, pixelsv, blockv;
671	666	register vector unsigned char perm = vec_lvsl(0, pixels);
672	667	int i;
673	668
674		−POWERPC_PERF_START_COUNT(altivec_avg_pixels16_num, 1);
675		−
676	669	for (i = 0; i < h; i++) {
677	670	pixelsv1 = vec_ld( 0, pixels);
678	671	pixelsv2 = vec_ld(16,pixels);
@@ -683,19 +676,14 @@ POWERPC_PERF_START_COUNT(altivec_avg_pixels16_num, 1); · 4 unmodified lines
683	676	pixels+=line_size;
684	677	block +=line_size;
685	678	}
686		−
687		−POWERPC_PERF_STOP_COUNT(altivec_avg_pixels16_num, 1);
688	679	}
689	680
690	681	/* next one assumes that ((line_size % 8) == 0) */
691	682	static void avg_pixels8_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
692	683	{
693		−POWERPC_PERF_DECLARE(altivec_avg_pixels8_num, 1);
694	684	register vector unsigned char pixelsv1, pixelsv2, pixelsv, blockv;
695	685	int i;
696	686
697		−POWERPC_PERF_START_COUNT(altivec_avg_pixels8_num, 1);
698		−
699	687	for (i = 0; i < h; i++) {
700	688	/* block is 8 bytes-aligned, so we're either in the
701	689	left block (16 bytes-aligned) or in the right block (not) */
@@ -719,14 +707,11 @@ POWERPC_PERF_START_COUNT(altivec_avg_pixels8_num, 1); · 17 unmodified lines
719	707	pixels += line_size;
720	708	block += line_size;
721	709	}
722		−
723		−POWERPC_PERF_STOP_COUNT(altivec_avg_pixels8_num, 1);
724	710	}
725	711
726	712	/* next one assumes that ((line_size % 8) == 0) */
727	713	static void put_pixels8_xy2_altivec(uint8_t block, const uint8_t pixels, int line_size, int h)
728	714	{
729		−POWERPC_PERF_DECLARE(altivec_put_pixels8_xy2_num, 1);
730	715	register int i;
731	716	register vector unsigned char pixelsv1, pixelsv2, pixelsavg;
732	717	register vector unsigned char blockv, temp1, temp2;
@@ -748,7 +733,6 @@ POWERPC_PERF_DECLARE(altivec_put_pixels8_xy2_num, 1); · 15 unmodified lines
748	733	(vector unsigned short)pixelsv2);
749	734	pixelssum1 = vec_add(pixelssum1, vctwo);
750	735
751		−POWERPC_PERF_START_COUNT(altivec_put_pixels8_xy2_num, 1);
752	736	for (i = 0; i < h ; i++) {
753	737	int rightside = ((unsigned long)block & 0x0000000F);
754	738	blockv = vec_ld(0, block);
@@ -782,14 +766,11 @@ POWERPC_PERF_START_COUNT(altivec_put_pixels8_xy2_num, 1); · 27 unmodified lines
782	766	block += line_size;
783	767	pixels += line_size;
784	768	}
785		−
786		−POWERPC_PERF_STOP_COUNT(altivec_put_pixels8_xy2_num, 1);
787	769	}
788	770
789	771	/* next one assumes that ((line_size % 8) == 0) */
790	772	static void put_no_rnd_pixels8_xy2_altivec(uint8_t block, const uint8_t pixels, int line_size, int h)
791	773	{
792		−POWERPC_PERF_DECLARE(altivec_put_no_rnd_pixels8_xy2_num, 1);
793	774	register int i;
794	775	register vector unsigned char pixelsv1, pixelsv2, pixelsavg;
795	776	register vector unsigned char blockv, temp1, temp2;
@@ -812,7 +793,6 @@ POWERPC_PERF_DECLARE(altivec_put_no_rnd_pixels8_xy2_num, 1); · 16 unmodified lines
812	793	(vector unsigned short)pixelsv2);
813	794	pixelssum1 = vec_add(pixelssum1, vcone);
814	795
815		−POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1);
816	796	for (i = 0; i < h ; i++) {
817	797	int rightside = ((unsigned long)block & 0x0000000F);
818	798	blockv = vec_ld(0, block);
@@ -846,14 +826,11 @@ POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1); · 27 unmodified lines
846	826	block += line_size;
847	827	pixels += line_size;
848	828	}
849		−
850		−POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1);
851	829	}
852	830
853	831	/* next one assumes that ((line_size % 16) == 0) */
854	832	static void put_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
855	833	{
856		−POWERPC_PERF_DECLARE(altivec_put_pixels16_xy2_num, 1);
857	834	register int i;
858	835	register vector unsigned char pixelsv1, pixelsv2, pixelsv3, pixelsv4;
859	836	register vector unsigned char blockv, temp1, temp2;
@@ -862,8 +839,6 @@ POWERPC_PERF_DECLARE(altivec_put_pixels16_xy2_num, 1); · 2 unmodified lines
862	839	register const vector unsigned char vczero = (const vector unsigned char)vec_splat_u8(0);
863	840	register const vector unsigned short vctwo = (const vector unsigned short)vec_splat_u16(2);
864	841
865		−POWERPC_PERF_START_COUNT(altivec_put_pixels16_xy2_num, 1);
866		−
867	842	temp1 = vec_ld(0, pixels);
868	843	temp2 = vec_ld(16, pixels);
869	844	pixelsv1 = vec_perm(temp1, temp2, vec_lvsl(0, pixels));
@@ -919,14 +894,11 @@ POWERPC_PERF_START_COUNT(altivec_put_pixels16_xy2_num, 1); · 49 unmodified lines
919	894	block += line_size;
920	895	pixels += line_size;
921	896	}
922		−
923		−POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_xy2_num, 1);
924	897	}
925	898
926	899	/* next one assumes that ((line_size % 16) == 0) */
927	900	static void put_no_rnd_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
928	901	{
929		−POWERPC_PERF_DECLARE(altivec_put_no_rnd_pixels16_xy2_num, 1);
930	902	register int i;
931	903	register vector unsigned char pixelsv1, pixelsv2, pixelsv3, pixelsv4;
932	904	register vector unsigned char blockv, temp1, temp2;
@@ -936,8 +908,6 @@ POWERPC_PERF_DECLARE(altivec_put_no_rnd_pixels16_xy2_num, 1); · 3 unmodified lines
936	908	register const vector unsigned short vcone = (const vector unsigned short)vec_splat_u16(1);
937	909	register const vector unsigned short vctwo = (const vector unsigned short)vec_splat_u16(2);
938	910
939		−POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
940		−
941	911	temp1 = vec_ld(0, pixels);
942	912	temp2 = vec_ld(16, pixels);
943	913	pixelsv1 = vec_perm(temp1, temp2, vec_lvsl(0, pixels));
@@ -993,18 +963,14 @@ POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1); · 49 unmodified lines
993	963	block += line_size;
994	964	pixels += line_size;
995	965	}
996		−
997		−POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
998	966	}
999	967
1000	968	static int hadamard8_diff8x8_altivec(/MpegEncContext/ void s, uint8_t dst, uint8_t *src, int stride, int h){
1001		−POWERPC_PERF_DECLARE(altivec_hadamard8_diff8x8_num, 1);
1002	969	int sum;
1003	970	register const vector unsigned char vzero =
1004	971	(const vector unsigned char)vec_splat_u8(0);
1005	972	register vector signed short temp0, temp1, temp2, temp3, temp4,
1006	973	temp5, temp6, temp7;
1007		−POWERPC_PERF_START_COUNT(altivec_hadamard8_diff8x8_num, 1);
1008	974	{
1009	975	register const vector signed short vprod1 =(const vector signed short)
1010	976	{ 1,-1, 1,-1, 1,-1, 1,-1 };
@@ -1100,7 +1066,6 @@ POWERPC_PERF_START_COUNT(altivec_hadamard8_diff8x8_num, 1); · 89 unmodified lines
1100	1066	vsum = vec_splat(vsum, 3);
1101	1067	vec_ste(vsum, 0, &sum);
1102	1068	}
1103		−POWERPC_PERF_STOP_COUNT(altivec_hadamard8_diff8x8_num, 1);
1104	1069	return sum;
1105	1070	}
1106	1071
@@ -1319,16 +1284,13 @@ static int hadamard8_diff16x8_altivec(/MpegEncContext/ void s, uint8_t dst, · 212 unmodified lines
1319	1284	}
1320	1285
1321	1286	static int hadamard8_diff16_altivec(/MpegEncContext/ void s, uint8_t dst, uint8_t *src, int stride, int h){
1322		−POWERPC_PERF_DECLARE(altivec_hadamard8_diff16_num, 1);
1323	1287	int score;
1324		−POWERPC_PERF_START_COUNT(altivec_hadamard8_diff16_num, 1);
1325	1288	score = hadamard8_diff16x8_altivec(s, dst, src, stride, 8);
1326	1289	if (h==16) {
1327	1290	dst += 8*stride;
1328	1291	src += 8*stride;
1329	1292	score += hadamard8_diff16x8_altivec(s, dst, src, stride, 8);
1330	1293	}
1331		−POWERPC_PERF_STOP_COUNT(altivec_hadamard8_diff16_num, 1);
1332	1294	return score;
1333	1295	}
1334	1296
@@ -1358,7 +1320,6 @@ static void vorbis_inverse_coupling_altivec(float mag, float ang, · 23 unmodified lines
1358	1320	/* next one assumes that ((line_size % 8) == 0) */
1359	1321	static void avg_pixels8_xy2_altivec(uint8_t block, const uint8_t pixels, int line_size, int h)
1360	1322	{
1361		−POWERPC_PERF_DECLARE(altivec_avg_pixels8_xy2_num, 1);
1362	1323	register int i;
1363	1324	register vector unsigned char pixelsv1, pixelsv2, pixelsavg;
1364	1325	register vector unsigned char blockv, temp1, temp2, blocktemp;
@@ -1383,7 +1344,6 @@ POWERPC_PERF_DECLARE(altivec_avg_pixels8_xy2_num, 1); · 18 unmodified lines
1383	1344	(vector unsigned short)pixelsv2);
1384	1345	pixelssum1 = vec_add(pixelssum1, vctwo);
1385	1346
1386		−POWERPC_PERF_START_COUNT(altivec_avg_pixels8_xy2_num, 1);
1387	1347	for (i = 0; i < h ; i++) {
1388	1348	int rightside = ((unsigned long)block & 0x0000000F);
1389	1349	blockv = vec_ld(0, block);
@@ -1418,8 +1378,6 @@ POWERPC_PERF_START_COUNT(altivec_avg_pixels8_xy2_num, 1); · 28 unmodified lines
1418	1378	block += line_size;
1419	1379	pixels += line_size;
1420	1380	}
1421		−
1422		−POWERPC_PERF_STOP_COUNT(altivec_avg_pixels8_xy2_num, 1);
1423	1381	}
1424	1382
1425	1383	void dsputil_init_altivec(DSPContext* c, AVCodecContext *avctx)

libavcodec/ppc/dsputil_ppc.c+0−79

@@ -21,9 +21,6 @@ · 20 unmodified lines
21	21	*/
22	22
23	23	#include "libavcodec/dsputil.h"
24		−
25		−#include "dsputil_ppc.h"
26		−
27	24	#include "dsputil_altivec.h"
28	25
29	26	int mm_flags = 0;
@@ -39,63 +36,6 @@ int mm_support(void) · 9 unmodified lines
39	36	return result;
40	37	}
41	38
42		−#if CONFIG_POWERPC_PERF
43		−unsigned long long perfdata[POWERPC_NUM_PMC_ENABLED][powerpc_perf_total][powerpc_data_total];
44		−/* list below must match enum in dsputil_ppc.h */
45		−static unsigned char* perfname[] = {
46		− "ff_fft_calc_altivec",
47		− "gmc1_altivec",
48		− "dct_unquantize_h263_altivec",
49		− "fdct_altivec",
50		− "idct_add_altivec",
51		− "idct_put_altivec",
52		− "put_pixels16_altivec",
53		− "avg_pixels16_altivec",
54		− "avg_pixels8_altivec",
55		− "put_pixels8_xy2_altivec",
56		− "put_no_rnd_pixels8_xy2_altivec",
57		− "put_pixels16_xy2_altivec",
58		− "put_no_rnd_pixels16_xy2_altivec",
59		− "hadamard8_diff8x8_altivec",
60		− "hadamard8_diff16_altivec",
61		− "avg_pixels8_xy2_altivec",
62		− "clear_blocks_dcbz32_ppc",
63		− "clear_blocks_dcbz128_ppc",
64		− "put_h264_chroma_mc8_altivec",
65		− "avg_h264_chroma_mc8_altivec",
66		− "put_h264_qpel16_h_lowpass_altivec",
67		− "avg_h264_qpel16_h_lowpass_altivec",
68		− "put_h264_qpel16_v_lowpass_altivec",
69		− "avg_h264_qpel16_v_lowpass_altivec",
70		− "put_h264_qpel16_hv_lowpass_altivec",
71		− "avg_h264_qpel16_hv_lowpass_altivec",
72		− ""
73		−};
74		−#include <stdio.h>
75		−#endif
76		−
77		−#if CONFIG_POWERPC_PERF
78		−void powerpc_display_perf_report(void)
79		−{
80		− int i, j;
81		− av_log(NULL, AV_LOG_INFO, "PowerPC performance report\n Values are from the PMC registers, and represent whatever the registers are set to record.\n");
82		− for(i = 0 ; i < powerpc_perf_total ; i++) {
83		− for (j = 0; j < POWERPC_NUM_PMC_ENABLED ; j++) {
84		− if (perfdata[j][i][powerpc_data_num] != (unsigned long long)0)
85		− av_log(NULL, AV_LOG_INFO,
86		− " Function \"%s\" (pmc%d):\n\tmin: %"PRIu64"\n\tmax: %"PRIu64"\n\tavg: %1.2lf (%"PRIu64")\n",
87		− perfname[i],
88		− j+1,
89		− perfdata[j][i][powerpc_data_min],
90		− perfdata[j][i][powerpc_data_max],
91		− (double)perfdata[j][i][powerpc_data_sum] /
92		− (double)perfdata[j][i][powerpc_data_num],
93		− perfdata[j][i][powerpc_data_num]);
94		− }
95		− }
96		−}
97		−#endif /* CONFIG_POWERPC_PERF */
98		−
99	39	/* *** WARNING * WARNING * WARNING *** */
100	40	/*
101	41	clear_blocks_dcbz32_ppc will not work properly on PowerPC processors with a
@@ -118,10 +58,8 @@ and <http://developer.apple.com/technotes/tn/tn2086.html> · 16 unmodified lines
118	58	*/
119	59	static void clear_blocks_dcbz32_ppc(DCTELEM *blocks)
120	60	{
121		−POWERPC_PERF_DECLARE(powerpc_clear_blocks_dcbz32, 1);
122	61	register int misal = ((unsigned long)blocks & 0x00000010);
123	62	register int i = 0;
124		−POWERPC_PERF_START_COUNT(powerpc_clear_blocks_dcbz32, 1);
125	63	#if 1
126	64	if (misal) {
127	65	((unsigned long*)blocks)[0] = 0L;
@@ -143,7 +81,6 @@ POWERPC_PERF_START_COUNT(powerpc_clear_blocks_dcbz32, 1); · 15 unmodified lines
143	81	#else
144	82	memset(blocks, 0, sizeof(DCTELEM)664);
145	83	#endif
146		−POWERPC_PERF_STOP_COUNT(powerpc_clear_blocks_dcbz32, 1);
147	84	}
148	85
149	86	/* same as above, when dcbzl clear a whole 128B cache line
@@ -151,10 +88,8 @@ POWERPC_PERF_STOP_COUNT(powerpc_clear_blocks_dcbz32, 1); · 1 unmodified line
151	88	#if HAVE_DCBZL
152	89	static void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
153	90	{
154		−POWERPC_PERF_DECLARE(powerpc_clear_blocks_dcbz128, 1);
155	91	register int misal = ((unsigned long)blocks & 0x0000007f);
156	92	register int i = 0;
157		−POWERPC_PERF_START_COUNT(powerpc_clear_blocks_dcbz128, 1);
158	93	#if 1
159	94	if (misal) {
160	95	// we could probably also optimize this case,
@@ -169,7 +104,6 @@ POWERPC_PERF_START_COUNT(powerpc_clear_blocks_dcbz128, 1); · 8 unmodified lines
169	104	#else
170	105	memset(blocks, 0, sizeof(DCTELEM)664);
171	106	#endif
172		−POWERPC_PERF_STOP_COUNT(powerpc_clear_blocks_dcbz128, 1);
173	107	}
174	108	#else
175	109	static void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
@@ -279,19 +213,6 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx) · 103 unmodified lines
279	213	}
280	214	}
281	215
282		−#if CONFIG_POWERPC_PERF
283		− {
284		− int i, j;
285		− for (i = 0 ; i < powerpc_perf_total ; i++) {
286		− for (j = 0; j < POWERPC_NUM_PMC_ENABLED ; j++) {
287		− perfdata[j][i][powerpc_data_min] = 0xFFFFFFFFFFFFFFFFULL;
288		− perfdata[j][i][powerpc_data_max] = 0x0000000000000000ULL;
289		− perfdata[j][i][powerpc_data_sum] = 0x0000000000000000ULL;
290		− perfdata[j][i][powerpc_data_num] = 0x0000000000000000ULL;
291		− }
292		− }
293		− }
294		−#endif /* CONFIG_POWERPC_PERF */
295	216	}
296	217	#endif /* HAVE_ALTIVEC */
297	218	}

libavcodec/ppc/dsputil_ppc.h+0−154

@@ -1,154 +0,0 @@
1		−/*
2		− * Copyright (c) 2003-2004 Romain Dolbeau <romain@dolbeau.org>
3		− *
4		− * This file is part of FFmpeg.
5		− *
6		− * FFmpeg is free software; you can redistribute it and/or
7		− * modify it under the terms of the GNU Lesser General Public
8		− * License as published by the Free Software Foundation; either
9		− * version 2.1 of the License, or (at your option) any later version.
10		− *
11		− * FFmpeg is distributed in the hope that it will be useful,
12		− * but WITHOUT ANY WARRANTY; without even the implied warranty of
13		− * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14		− * Lesser General Public License for more details.
15		− *
16		− * You should have received a copy of the GNU Lesser General Public
17		− * License along with FFmpeg; if not, write to the Free Software
18		− * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19		− */
20		−
21		−#ifndef AVCODEC_PPC_DSPUTIL_PPC_H
22		−#define AVCODEC_PPC_DSPUTIL_PPC_H
23		−
24		−#include "config.h"
25		−
26		−#if CONFIG_POWERPC_PERF
27		−void powerpc_display_perf_report(void);
28		−/* the 604* have 2, the G3* have 4, the G4s have 6,
29		− and the G5 are completely different (they MUST use
30		− ARCH_PPC64, and let's hope all future 64 bis PPC
31		− will use the same PMCs... */
32		−#define POWERPC_NUM_PMC_ENABLED 6
33		−/* if you add to the enum below, also add to the perfname array
34		− in dsputil_ppc.c */
35		−enum powerpc_perf_index {
36		− altivec_fft_num = 0,
37		− altivec_gmc1_num,
38		− altivec_dct_unquantize_h263_num,
39		− altivec_fdct,
40		− altivec_idct_add_num,
41		− altivec_idct_put_num,
42		− altivec_put_pixels16_num,
43		− altivec_avg_pixels16_num,
44		− altivec_avg_pixels8_num,
45		− altivec_put_pixels8_xy2_num,
46		− altivec_put_no_rnd_pixels8_xy2_num,
47		− altivec_put_pixels16_xy2_num,
48		− altivec_put_no_rnd_pixels16_xy2_num,
49		− altivec_hadamard8_diff8x8_num,
50		− altivec_hadamard8_diff16_num,
51		− altivec_avg_pixels8_xy2_num,
52		− powerpc_clear_blocks_dcbz32,
53		− powerpc_clear_blocks_dcbz128,
54		− altivec_put_h264_chroma_mc8_num,
55		− altivec_avg_h264_chroma_mc8_num,
56		− altivec_put_h264_qpel16_h_lowpass_num,
57		− altivec_avg_h264_qpel16_h_lowpass_num,
58		− altivec_put_h264_qpel16_v_lowpass_num,
59		− altivec_avg_h264_qpel16_v_lowpass_num,
60		− altivec_put_h264_qpel16_hv_lowpass_num,
61		− altivec_avg_h264_qpel16_hv_lowpass_num,
62		− powerpc_perf_total
63		−};
64		−enum powerpc_data_index {
65		− powerpc_data_min = 0,
66		− powerpc_data_max,
67		− powerpc_data_sum,
68		− powerpc_data_num,
69		− powerpc_data_total
70		−};
71		−extern unsigned long long perfdata[POWERPC_NUM_PMC_ENABLED][powerpc_perf_total][powerpc_data_total];
72		−
73		−#if !ARCH_PPC64
74		−#define POWERP_PMC_DATATYPE unsigned long
75		−#define POWERPC_GET_PMC1(a) __asm__ volatile("mfspr %0, 937" : "=r" (a))
76		−#define POWERPC_GET_PMC2(a) __asm__ volatile("mfspr %0, 938" : "=r" (a))
77		−#if (POWERPC_NUM_PMC_ENABLED > 2)
78		−#define POWERPC_GET_PMC3(a) __asm__ volatile("mfspr %0, 941" : "=r" (a))
79		−#define POWERPC_GET_PMC4(a) __asm__ volatile("mfspr %0, 942" : "=r" (a))
80		−#else
81		−#define POWERPC_GET_PMC3(a) do {} while (0)
82		−#define POWERPC_GET_PMC4(a) do {} while (0)
83		−#endif
84		−#if (POWERPC_NUM_PMC_ENABLED > 4)
85		−#define POWERPC_GET_PMC5(a) __asm__ volatile("mfspr %0, 929" : "=r" (a))
86		−#define POWERPC_GET_PMC6(a) __asm__ volatile("mfspr %0, 930" : "=r" (a))
87		−#else
88		−#define POWERPC_GET_PMC5(a) do {} while (0)
89		−#define POWERPC_GET_PMC6(a) do {} while (0)
90		−#endif
91		−#else /* ARCH_PPC64 */
92		−#define POWERP_PMC_DATATYPE unsigned long long
93		−#define POWERPC_GET_PMC1(a) __asm__ volatile("mfspr %0, 771" : "=r" (a))
94		−#define POWERPC_GET_PMC2(a) __asm__ volatile("mfspr %0, 772" : "=r" (a))
95		−#if (POWERPC_NUM_PMC_ENABLED > 2)
96		−#define POWERPC_GET_PMC3(a) __asm__ volatile("mfspr %0, 773" : "=r" (a))
97		−#define POWERPC_GET_PMC4(a) __asm__ volatile("mfspr %0, 774" : "=r" (a))
98		−#else
99		−#define POWERPC_GET_PMC3(a) do {} while (0)
100		−#define POWERPC_GET_PMC4(a) do {} while (0)
101		−#endif
102		−#if (POWERPC_NUM_PMC_ENABLED > 4)
103		−#define POWERPC_GET_PMC5(a) __asm__ volatile("mfspr %0, 775" : "=r" (a))
104		−#define POWERPC_GET_PMC6(a) __asm__ volatile("mfspr %0, 776" : "=r" (a))
105		−#else
106		−#define POWERPC_GET_PMC5(a) do {} while (0)
107		−#define POWERPC_GET_PMC6(a) do {} while (0)
108		−#endif
109		−#endif /* ARCH_PPC64 */
110		−#define POWERPC_PERF_DECLARE(a, cond) \
111		− POWERP_PMC_DATATYPE \
112		− pmc_start[POWERPC_NUM_PMC_ENABLED], \
113		− pmc_stop[POWERPC_NUM_PMC_ENABLED], \
114		− pmc_loop_index;
115		−#define POWERPC_PERF_START_COUNT(a, cond) do { \
116		− POWERPC_GET_PMC6(pmc_start[5]); \
117		− POWERPC_GET_PMC5(pmc_start[4]); \
118		− POWERPC_GET_PMC4(pmc_start[3]); \
119		− POWERPC_GET_PMC3(pmc_start[2]); \
120		− POWERPC_GET_PMC2(pmc_start[1]); \
121		− POWERPC_GET_PMC1(pmc_start[0]); \
122		− } while (0)
123		−#define POWERPC_PERF_STOP_COUNT(a, cond) do { \
124		− POWERPC_GET_PMC1(pmc_stop[0]); \
125		− POWERPC_GET_PMC2(pmc_stop[1]); \
126		− POWERPC_GET_PMC3(pmc_stop[2]); \
127		− POWERPC_GET_PMC4(pmc_stop[3]); \
128		− POWERPC_GET_PMC5(pmc_stop[4]); \
129		− POWERPC_GET_PMC6(pmc_stop[5]); \
130		− if (cond) { \
131		− for(pmc_loop_index = 0; \
132		− pmc_loop_index < POWERPC_NUM_PMC_ENABLED; \
133		− pmc_loop_index++) { \
134		− if (pmc_stop[pmc_loop_index] >= pmc_start[pmc_loop_index]) { \
135		− POWERP_PMC_DATATYPE diff = \
136		− pmc_stop[pmc_loop_index] - pmc_start[pmc_loop_index]; \
137		− if (diff < perfdata[pmc_loop_index][a][powerpc_data_min]) \
138		− perfdata[pmc_loop_index][a][powerpc_data_min] = diff; \
139		− if (diff > perfdata[pmc_loop_index][a][powerpc_data_max]) \
140		− perfdata[pmc_loop_index][a][powerpc_data_max] = diff; \
141		− perfdata[pmc_loop_index][a][powerpc_data_sum] += diff; \
142		− perfdata[pmc_loop_index][a][powerpc_data_num] ++; \
143		− } \
144		− } \
145		− } \
146		−} while (0)
147		−#else /* CONFIG_POWERPC_PERF */
148		−// those are needed to avoid empty statements.
149		−#define POWERPC_PERF_DECLARE(a, cond) int altivec_placeholder __attribute__ ((unused))
150		−#define POWERPC_PERF_START_COUNT(a, cond) do {} while (0)
151		−#define POWERPC_PERF_STOP_COUNT(a, cond) do {} while (0)
152		−#endif /* CONFIG_POWERPC_PERF */
153		−
154		−#endif /* AVCODEC_PPC_DSPUTIL_PPC_H */

libavcodec/ppc/fdct_altivec.c+0−7

@@ -24,7 +24,6 @@ · 23 unmodified lines
24	24	#endif
25	25	#include "libavutil/common.h"
26	26	#include "libavcodec/dsputil.h"
27		−#include "dsputil_ppc.h"
28	27	#include "dsputil_altivec.h"
29	28
30	29	#define vs16(v) ((vector signed short)(v))
@@ -198,7 +197,6 @@ static vector float fdctconsts[3] = { · 167 unmodified lines
198	197
199	198	void fdct_altivec(int16_t *block)
200	199	{
201		−POWERPC_PERF_DECLARE(altivec_fdct, 1);
202	200	vector signed short *bp;
203	201	vector float *cp;
204	202	vector float b00, b10, b20, b30, b40, b50, b60, b70;
@@ -206,9 +204,6 @@ POWERPC_PERF_DECLARE(altivec_fdct, 1); · 1 unmodified line
206	204	vector float mzero, cnst, cnsts0, cnsts1, cnsts2;
207	205	vector float x0, x1, x2, x3, x4, x5, x6, x7, x8;
208	206
209		− POWERPC_PERF_START_COUNT(altivec_fdct, 1);
210		−
211		−
212	207	/* setup constants {{{ */
213	208	/* mzero = -0.0 */
214	209	mzero = ((vector float)vec_splat_u32(-1));
@@ -487,8 +482,6 @@ POWERPC_PERF_DECLARE(altivec_fdct, 1); · 272 unmodified lines
487	482
488	483	#undef CTS
489	484	/* }}} */
490		−
491		−POWERPC_PERF_STOP_COUNT(altivec_fdct, 1);
492	485	}
493	486
494	487	/* vim:set foldmethod=marker foldlevel=0: */

libavcodec/ppc/fft_altivec.c+0−6

@@ -21,7 +21,6 @@ · 20 unmodified lines
21	21	* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
22	22	*/
23	23	#include "libavcodec/fft.h"
24		−#include "dsputil_ppc.h"
25	24	#include "util_altivec.h"
26	25	#include "dsputil_altivec.h"
27	26
@@ -38,7 +37,6 @@ · 10 unmodified lines
38	37	*/
39	38	static void ff_fft_calc_altivec(FFTContext s, FFTComplex z)
40	39	{
41		−POWERPC_PERF_DECLARE(altivec_fft_num, s->nbits >= 6);
42	40	register const vector float vczero = (const vector float)vec_splat_u32(0.);
43	41
44	42	int ln = s->nbits;
@@ -48,8 +46,6 @@ POWERPC_PERF_DECLARE(altivec_fft_num, s->nbits >= 6); · 3 unmodified lines
48	46	FFTComplex cptr, cptr1;
49	47	int k;
50	48
51		−POWERPC_PERF_START_COUNT(altivec_fft_num, s->nbits >= 6);
52		−
53	49	np = 1 << ln;
54	50
55	51	{
@@ -132,8 +128,6 @@ POWERPC_PERF_START_COUNT(altivec_fft_num, s->nbits >= 6); · 76 unmodified lines
132	128	nblocks = nblocks >> 1;
133	129	nloops = nloops << 1;
134	130	} while (nblocks != 0);
135		−
136		−POWERPC_PERF_STOP_COUNT(altivec_fft_num, s->nbits >= 6);
137	131	}
138	132
139	133	av_cold void ff_fft_init_altivec(FFTContext *s)

libavcodec/ppc/gmc_altivec.c+0−8

@@ -21,7 +21,6 @@ · 20 unmodified lines
21	21	*/
22	22
23	23	#include "libavcodec/dsputil.h"
24		−#include "dsputil_ppc.h"
25	24	#include "util_altivec.h"
26	25	#include "types_altivec.h"
27	26	#include "dsputil_altivec.h"
@@ -30,10 +29,8 @@ · 2 unmodified lines
30	29	altivec-enhanced gmc1. ATM this code assume stride is a multiple of 8,
31	30	to preserve proper dst alignment.
32	31	*/
33		−#define GMC1_PERF_COND (h==8)
34	32	void gmc1_altivec(uint8_t dst / align 8 /, uint8_t src /* align1 */, int stride, int h, int x16, int y16, int rounder)
35	33	{
36		−POWERPC_PERF_DECLARE(altivec_gmc1_num, GMC1_PERF_COND);
37	34	const DECLARE_ALIGNED(16, unsigned short, rounder_a) = rounder;
38	35	const DECLARE_ALIGNED(16, unsigned short, ABCD)[8] =
39	36	{
@@ -51,9 +48,6 @@ POWERPC_PERF_DECLARE(altivec_gmc1_num, GMC1_PERF_COND); · 11 unmodified lines
51	48	unsigned long dst_odd = (unsigned long)dst & 0x0000000F;
52	49	unsigned long src_really_odd = (unsigned long)src & 0x0000000F;
53	50
54		−
55		−POWERPC_PERF_START_COUNT(altivec_gmc1_num, GMC1_PERF_COND);
56		−
57	51	tempA = vec_ld(0, (unsigned short*)ABCD);
58	52	Av = vec_splat(tempA, 0);
59	53	Bv = vec_splat(tempA, 1);
@@ -133,6 +127,4 @@ POWERPC_PERF_START_COUNT(altivec_gmc1_num, GMC1_PERF_COND); · 73 unmodified lines
133	127	dst += stride;
134	128	src += stride;
135	129	}
136		−
137		−POWERPC_PERF_STOP_COUNT(altivec_gmc1_num, GMC1_PERF_COND);
138	130	}

libavcodec/ppc/h264_altivec.c+0−1

@@ -22,7 +22,6 @@ · 21 unmodified lines
22	22	#include "libavcodec/h264data.h"
23	23	#include "libavcodec/h264dsp.h"
24	24
25		−#include "dsputil_ppc.h"
26	25	#include "dsputil_altivec.h"
27	26	#include "util_altivec.h"
28	27	#include "types_altivec.h"

libavcodec/ppc/h264_template_altivec.c+0−16

@@ -77,7 +77,6 @@ · 76 unmodified lines
77	77
78	78	static void PREFIX_h264_chroma_mc8_altivec(uint8_t * dst, uint8_t * src,
79	79	int stride, int h, int x, int y) {
80		− POWERPC_PERF_DECLARE(PREFIX_h264_chroma_mc8_num, 1);
81	80	DECLARE_ALIGNED(16, signed int, ABCD)[4] =
82	81	{((8 - x) * (8 - y)),
83	82	(( x) * (8 - y)),
@@ -103,8 +102,6 @@ static void PREFIX_h264_chroma_mc8_altivec(uint8_t * dst, uint8_t * src, · 19 unmodified lines
103	102	vec_s16 vsrc2ssH, vsrc3ssH, psum;
104	103	vec_u8 vdst, ppsum, vfdst, fsum;
105	104
106		− POWERPC_PERF_START_COUNT(PREFIX_h264_chroma_mc8_num, 1);
107		−
108	105	if (((unsigned long)dst) % 16 == 0) {
109	106	fperm = (vec_u8){0x10, 0x11, 0x12, 0x13,
110	107	0x14, 0x15, 0x16, 0x17,
@@ -203,7 +200,6 @@ static void PREFIX_h264_chroma_mc8_altivec(uint8_t * dst, uint8_t * src, · 92 unmodified lines
203	200	}
204	201	}
205	202	}
206		− POWERPC_PERF_STOP_COUNT(PREFIX_h264_chroma_mc8_num, 1);
207	203	}
208	204
209	205	/* this code assume that stride % 16 == 0 */
@@ -295,7 +291,6 @@ static void PREFIX_no_rnd_vc1_chroma_mc8_altivec(uint8_t * dst, uint8_t * src, i · 85 unmodified lines
295	291
296	292	/* this code assume stride % 16 == 0 */
297	293	static void PREFIX_h264_qpel16_h_lowpass_altivec(uint8_t * dst, uint8_t * src, int dstStride, int srcStride) {
298		− POWERPC_PERF_DECLARE(PREFIX_h264_qpel16_h_lowpass_num, 1);
299	294	register int i;
300	295
301	296	LOAD_ZERO;
@@ -323,8 +318,6 @@ static void PREFIX_h264_qpel16_h_lowpass_altivec(uint8_t * dst, uint8_t * src, i · 21 unmodified lines
323	318
324	319	vec_u8 sum, vdst, fsum;
325	320
326		− POWERPC_PERF_START_COUNT(PREFIX_h264_qpel16_h_lowpass_num, 1);
327		−
328	321	for (i = 0 ; i < 16 ; i ++) {
329	322	vec_u8 srcR1 = vec_ld(-2, src);
330	323	vec_u8 srcR2 = vec_ld(14, src);
@@ -433,13 +426,10 @@ static void PREFIX_h264_qpel16_h_lowpass_altivec(uint8_t * dst, uint8_t * src, i · 102 unmodified lines
433	426	src += srcStride;
434	427	dst += dstStride;
435	428	}
436		− POWERPC_PERF_STOP_COUNT(PREFIX_h264_qpel16_h_lowpass_num, 1);
437	429	}
438	430
439	431	/* this code assume stride % 16 == 0 */
440	432	static void PREFIX_h264_qpel16_v_lowpass_altivec(uint8_t * dst, uint8_t * src, int dstStride, int srcStride) {
441		− POWERPC_PERF_DECLARE(PREFIX_h264_qpel16_v_lowpass_num, 1);
442		−
443	433	register int i;
444	434
445	435	LOAD_ZERO;
@@ -490,8 +480,6 @@ static void PREFIX_h264_qpel16_v_lowpass_altivec(uint8_t * dst, uint8_t * src, i · 44 unmodified lines
490	480
491	481	vec_u8 sum, vdst, fsum, srcP3a, srcP3b, srcP3;
492	482
493		− POWERPC_PERF_START_COUNT(PREFIX_h264_qpel16_v_lowpass_num, 1);
494		−
495	483	for (i = 0 ; i < 16 ; i++) {
496	484	srcP3a = vec_ld(0, srcbis += srcStride);
497	485	srcP3b = vec_ld(16, srcbis);
@@ -544,12 +532,10 @@ static void PREFIX_h264_qpel16_v_lowpass_altivec(uint8_t * dst, uint8_t * src, i · 46 unmodified lines
544	532
545	533	dst += dstStride;
546	534	}
547		− POWERPC_PERF_STOP_COUNT(PREFIX_h264_qpel16_v_lowpass_num, 1);
548	535	}
549	536
550	537	/* this code assume stride % 16 == 0 and tmp is properly aligned */
551	538	static void PREFIX_h264_qpel16_hv_lowpass_altivec(uint8_t * dst, int16_t * tmp, uint8_t * src, int dstStride, int tmpStride, int srcStride) {
552		− POWERPC_PERF_DECLARE(PREFIX_h264_qpel16_hv_lowpass_num, 1);
553	539	register int i;
554	540	LOAD_ZERO;
555	541	const vec_u8 permM2 = vec_lvsl(-2, src);
@@ -589,7 +575,6 @@ static void PREFIX_h264_qpel16_hv_lowpass_altivec(uint8_t * dst, int16_t * tmp, · 33 unmodified lines
589	575	vec_u8 fsum, sumv, sum, vdst;
590	576	vec_s16 ssume, ssumo;
591	577
592		− POWERPC_PERF_START_COUNT(PREFIX_h264_qpel16_hv_lowpass_num, 1);
593	578	src -= (2 * srcStride);
594	579	for (i = 0 ; i < 21 ; i ++) {
595	580	vec_u8 srcM2, srcM1, srcP0, srcP1, srcP2, srcP3;
@@ -779,5 +764,4 @@ static void PREFIX_h264_qpel16_hv_lowpass_altivec(uint8_t * dst, int16_t * tmp, · 183 unmodified lines
779	764
780	765	dst += dstStride;
781	766	}
782		− POWERPC_PERF_STOP_COUNT(PREFIX_h264_qpel16_hv_lowpass_num, 1);
783	767	}

libavcodec/ppc/idct_altivec.c+0−14

@@ -43,7 +43,6 @@ · 42 unmodified lines
43	43	#endif
44	44	#include "libavcodec/dsputil.h"
45	45	#include "types_altivec.h"
46		−#include "dsputil_ppc.h"
47	46	#include "dsputil_altivec.h"
48	47
49	48	#define IDCT_HALF \
@@ -161,13 +160,9 @@ static const vec_s16 constants[5] = { · 111 unmodified lines
161	160
162	161	void idct_put_altivec(uint8_t* dest, int stride, int16_t *blk)
163	162	{
164		−POWERPC_PERF_DECLARE(altivec_idct_put_num, 1);
165	163	vec_s16 block = (vec_s16)blk;
166	164	vec_u8 tmp;
167	165
168		−#if CONFIG_POWERPC_PERF
169		−POWERPC_PERF_START_COUNT(altivec_idct_put_num, 1);
170		−#endif
171	166	IDCT
172	167
173	168	#define COPY(dest,src) \
@@ -183,13 +178,10 @@ POWERPC_PERF_START_COUNT(altivec_idct_put_num, 1); · 9 unmodified lines
183	178	COPY (dest, vx5) dest += stride;
184	179	COPY (dest, vx6) dest += stride;
185	180	COPY (dest, vx7)
186		−
187		−POWERPC_PERF_STOP_COUNT(altivec_idct_put_num, 1);
188	181	}
189	182
190	183	void idct_add_altivec(uint8_t* dest, int stride, int16_t *blk)
191	184	{
192		−POWERPC_PERF_DECLARE(altivec_idct_add_num, 1);
193	185	vec_s16 block = (vec_s16)blk;
194	186	vec_u8 tmp;
195	187	vec_s16 tmp2, tmp3;
@@ -197,10 +189,6 @@ POWERPC_PERF_DECLARE(altivec_idct_add_num, 1); · 1 unmodified line
197	189	vec_u8 perm1;
198	190	vec_u8 p0, p1, p;
199	191
200		−#if CONFIG_POWERPC_PERF
201		−POWERPC_PERF_START_COUNT(altivec_idct_add_num, 1);
202		−#endif
203		−
204	192	IDCT
205	193
206	194	p0 = vec_lvsl (0, dest);
@@ -226,7 +214,5 @@ POWERPC_PERF_START_COUNT(altivec_idct_add_num, 1); · 19 unmodified lines
226	214	ADD (dest, vx5, perm1) dest += stride;
227	215	ADD (dest, vx6, perm0) dest += stride;
228	216	ADD (dest, vx7, perm1)
229		−
230		−POWERPC_PERF_STOP_COUNT(altivec_idct_add_num, 1);
231	217	}
232	218

libavcodec/ppc/mpegvideo_altivec.c+0−5

@@ -26,7 +26,6 @@ · 25 unmodified lines
26	26	#include "libavcodec/dsputil.h"
27	27	#include "libavcodec/mpegvideo.h"
28	28
29		−#include "dsputil_ppc.h"
30	29	#include "util_altivec.h"
31	30	#include "types_altivec.h"
32	31	#include "dsputil_altivec.h"
@@ -479,14 +478,11 @@ static int dct_quantize_altivec(MpegEncContext* s, · 446 unmodified lines
479	478	static void dct_unquantize_h263_altivec(MpegEncContext *s,
480	479	DCTELEM *block, int n, int qscale)
481	480	{
482		−POWERPC_PERF_DECLARE(altivec_dct_unquantize_h263_num, 1);
483	481	int i, level, qmul, qadd;
484	482	int nCoeffs;
485	483
486	484	assert(s->block_last_index[n]>=0);
487	485
488		−POWERPC_PERF_START_COUNT(altivec_dct_unquantize_h263_num, 1);
489		−
490	486	qadd = (qscale - 1) \| 1;
491	487	qmul = qscale << 1;
492	488
@@ -569,7 +565,6 @@ POWERPC_PERF_START_COUNT(altivec_dct_unquantize_h263_num, 1); · 76 unmodified lines
569	565	block[0] = backup_0;
570	566	}
571	567	}
572		−POWERPC_PERF_STOP_COUNT(altivec_dct_unquantize_h263_num, nCoeffs == 63);
573	568	}
574	569
575	570

Remove PPC perf counter support

Keyboard shortcuts

General

Issues and pull requests

Issues, pull requests and Inbox

Repository code