Martin Storsjö 4136405c86 aarch64: me_cmp: Don't do uaddlv once per iteration ... The max height is currently documented as 16; the max difference per pixel is 255, and a .8h element can easily contain 16*255, thus keep accumulating in two .8h vectors, and just do the final accumulationat the end. This should work for heights up to 256. This requires a minor register renumbering in ff_pix_abs16_xy2_neon. Before: Cortex A53 A72 A73 Graviton 3 pix_abs_0_0_neon: 97.7 47.0 37.5 22.7 pix_abs_0_1_neon: 154.0 59.0 52.0 25.0 pix_abs_0_3_neon: 179.7 96.7 87.5 41.2 After: pix_abs_0_0_neon: 96.0 39.2 31.2 22.0 pix_abs_0_1_neon: 150.7 59.7 46.2 23.7 pix_abs_0_3_neon: 175.7 83.7 81.7 38.2 Signed-off-by: Martin Storsjö <martin@martin.st>		2022-07-16 17:26:17 +03:00
..
aacpsdsp_init_aarch64.c
aacpsdsp_neon.S
asm-offsets.h
cabac.h
fft_init_aarch64.c
fft_neon.S	arm64: Fix wrong BTI landing pad	2022-04-26 10:26:49 +03:00
fmtconvert_init.c
fmtconvert_neon.S
h264chroma_init_aarch64.c
h264cmc_neon.S	configure: Use a separate config_components.h header for $ALL_COMPONENTS	2022-03-16 14:12:49 +02:00
h264dsp_init_aarch64.c
h264dsp_neon.S	aarch64: h264dsp: Fix incorrectly indented code	2022-02-11 10:49:12 +02:00
h264idct_neon.S	aarch64: Add Armv8.5-A BTI support	2021-11-16 13:43:56 +02:00
h264pred_init.c
h264pred_neon.S
h264qpel_init_aarch64.c
h264qpel_neon.S
hevcdsp_idct_neon.S
hevcdsp_init_aarch64.c	lavc/aarch64: add hevc sao edge 8x8	2022-05-25 08:04:46 +02:00
hevcdsp_sao_neon.S	lavc/aarch64: hevc_sao reschedule slightly	2022-05-26 08:10:41 +02:00
hpeldsp_init_aarch64.c
hpeldsp_neon.S
idct.h	avcodec/aarch64/idct: Add missing stddef	2022-02-21 13:10:04 +01:00
idctdsp_init_aarch64.c	avcodec/idctdsp: Arm 64-bit NEON block add and clamp fast paths	2022-04-01 10:03:34 +03:00
idctdsp_neon.S	avcodec/idctdsp: Arm 64-bit NEON block add and clamp fast paths	2022-04-01 10:03:34 +03:00
Makefile	lavc/aarch64: motion estimation functions in neon	2022-06-28 00:51:39 +03:00
mdct_neon.S	arm64: Add Armv8.3-A PAC support to assembly files	2022-03-09 15:04:25 +02:00
me_cmp_init_aarch64.c	lavc/aarch64: Add pix_abs16_x2 neon implementation	2022-07-13 23:25:22 +03:00
me_cmp_neon.S	aarch64: me_cmp: Don't do uaddlv once per iteration	2022-07-16 17:26:17 +03:00
mpegaudiodsp_init.c
mpegaudiodsp_neon.S
neon.S
neontest.c
opusdsp_init.c
opusdsp_neon.S
pixblockdsp_init_aarch64.c
pixblockdsp_neon.S
rv40dsp_init_aarch64.c
sbrdsp_init_aarch64.c
sbrdsp_neon.S
simple_idct_neon.S	aarch64: Use ret x<n> instead of br x<n> where possible	2021-11-16 13:43:56 +02:00
synth_filter_init.c
synth_filter_neon.S	arm64: Add Armv8.3-A PAC support to assembly files	2022-03-09 15:04:25 +02:00
vc1dsp_init_aarch64.c	avcodec/vc1: Arm 64-bit NEON unescape fast path	2022-04-01 10:03:34 +03:00
vc1dsp_neon.S	avcodec/vc1: Arm 64-bit NEON unescape fast path	2022-04-01 10:03:34 +03:00
videodsp.S
videodsp_init.c
vorbisdsp_init.c
vorbisdsp_neon.S
vp8dsp.h
vp8dsp_init_aarch64.c
vp8dsp_neon.S
vp9dsp_init.h
vp9dsp_init_10bpp_aarch64.c
vp9dsp_init_12bpp_aarch64.c
vp9dsp_init_16bpp_aarch64_template.c
vp9dsp_init_aarch64.c
vp9itxfm_16bpp_neon.S	aarch64: Use ret x<n> instead of br x<n> where possible	2021-11-16 13:43:56 +02:00
vp9itxfm_neon.S	aarch64: Use ret x<n> instead of br x<n> where possible	2021-11-16 13:43:56 +02:00
vp9lpf_16bpp_neon.S	aarch64: Use ret x<n> instead of br x<n> where possible	2021-11-16 13:43:56 +02:00
vp9lpf_neon.S	aarch64: Use ret x<n> instead of br x<n> where possible	2021-11-16 13:43:56 +02:00
vp9mc_16bpp_neon.S
vp9mc_aarch64.S
vp9mc_neon.S