Le 18 août 2024 23:13:21 GMT+03:00, Ramiro Polla a
écrit :
> A53 A76
>pix_norm1_c: 519.2 231.5
>pix_norm1_neon: 195.0 ( 2.66x) 44.2 ( 5.24x)
>pix_sum_c: 344.5 242.2
>pix_sum_neon:119.0 ( 2.89x) 41.7 ( 5.81x)
>---
> libavcodec/
On Sun, Aug 18, 2024 at 10:43 PM Martin Storsjö wrote:
> On Sun, 18 Aug 2024, Ramiro Polla wrote:
>
> > A53 A76
> > pix_norm1_c: 519.2 231.5
> > pix_norm1_neon: 195.0 ( 2.66x) 44.2 ( 5.24x)
> > pix_sum_c: 344.5 242.2
> > pix_sum_neon:
On Fri, Aug 02, 2024 at 03:08:29PM -0700, Dale Curtis wrote:
> Found by fuzzer.
>
> Bug: https://crbug.com/356720789
> Signed-off-by: Dale Curtis
> ---
> libavformat/mov.c | 8 ++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
> mov.c |8 ++--
> 1 file changed, 6 insertions(+)
On Sun, 18 Aug 2024, Ramiro Polla wrote:
A76
pix_norm1_c:231.5
pix_norm1_neon: 44.2 ( 5.24x)
pix_norm1_dotprod: 20.7 (11.18x)
---
libavcodec/aarch64/mpegvideoencdsp_init.c | 10
libavcodec/aarch64/mpegvideoencdsp_neon.S | 28 +++
2 f
On Sun, 18 Aug 2024, Ramiro Polla wrote:
A53 A76
pix_norm1_c: 519.2 231.5
pix_norm1_neon: 195.0 ( 2.66x) 44.2 ( 5.24x)
pix_sum_c: 344.5 242.2
pix_sum_neon:119.0 ( 2.89x) 41.7 ( 5.81x)
---
Hmm, those speedups on the A53 look q
On Wed, Aug 14, 2024 at 2:41 PM Martin Storsjö wrote:
> On Tue, 6 Aug 2024, Ramiro Polla wrote:
> > checkasm --bench on a Raspberry Pi 5 Model B Rev 1.0:
> > yuv420p_gbrp_128_c: 1243.0
> > yuv420p_gbrp_128_neon: 453.5
> > yuv420p_gbrp_1920_c: 18165.5
> > yuv420p_gbrp_1920_neon: 6700.0
> > yuv422p_
On Sun, Aug 18, 2024 at 10:13 PM Ramiro Polla wrote:
>
>A53 A76
> pix_norm1_c: 519.2 231.5
> pix_norm1_neon: 195.0 ( 2.66x) 44.2 ( 5.24x)
> pix_sum_c: 344.5 242.2
> pix_sum_neon:119.0 ( 2.89x) 41.7 ( 5.81x)
This new patchset n
This commit also restricts w to 4, 8, or 16.
Intel(R) Core(TM) i5-5300U CPU @ 2.30GHz:
beforeafter
draw_edges_8_1724_4_c:45074.5 7144.7 ( 6.31x)
draw_edges_8_1724_8_c:41716.5 7216.0 ( 5.78x)
draw_edges_8_1724_16_c: 45282.7 16026.2 ( 2.83x)
draw_edge
The mmx memory copy code is not nearly as efficient as memcpy(), which
would make draw_edges_mmx much slower than draw_edges_8_c.
Intel(R) Core(TM) i5-5300U CPU @ 2.30GHz:
beforeafter
draw_edges_8_1724_4_mmx: 8697.2 8739.6 ( 1.00x)
draw_edges_8_1724_8_mmx:
Not only w == 8 and w == 16 are supported, but also w == 4.
---
libavcodec/x86/mpegvideoencdsp_init.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/libavcodec/x86/mpegvideoencdsp_init.c
b/libavcodec/x86/mpegvideoencdsp_init.c
index ec174b15aa..503548e668 100644
--- a/libav
---
tests/checkasm/mpegvideoencdsp.c | 53
1 file changed, 53 insertions(+)
diff --git a/tests/checkasm/mpegvideoencdsp.c b/tests/checkasm/mpegvideoencdsp.c
index 00d21ba0ba..21954a4a65 100644
--- a/tests/checkasm/mpegvideoencdsp.c
+++ b/tests/checkasm/mpegvideoen
A76
pix_norm1_c:231.5
pix_norm1_neon: 44.2 ( 5.24x)
pix_norm1_dotprod: 20.7 (11.18x)
---
libavcodec/aarch64/mpegvideoencdsp_init.c | 10
libavcodec/aarch64/mpegvideoencdsp_neon.S | 28 +++
2 files changed, 38 insertions(+)
diff --g
A53 A76
pix_norm1_c: 519.2 231.5
pix_norm1_neon: 195.0 ( 2.66x) 44.2 ( 5.24x)
pix_sum_c: 344.5 242.2
pix_sum_neon:119.0 ( 2.89x) 41.7 ( 5.81x)
---
libavcodec/aarch64/Makefile | 2 +
libavcodec/aarch64/mpegvideoenc
---
tests/checkasm/Makefile | 1 +
tests/checkasm/checkasm.c| 3 ++
tests/checkasm/checkasm.h| 1 +
tests/checkasm/mpegvideoencdsp.c | 77
4 files changed, 82 insertions(+)
create mode 100644 tests/checkasm/mpegvideoencdsp.c
diff --git
The sentence was added as an advice, given the maturity of the xHE-AAC
implementation, but didn't deserve its own paragraph, nor its subject
deserved being left a referential footnote in the stead of article.
---
src/index | 3 ---
1 file changed, 3 deletions(-)
diff --git a/src/index b/src/index
On Wed, Aug 14, 2024 at 12:07:04PM -0300, James Almer wrote:
> On 8/14/2024 11:34 AM, Michael Niedermayer wrote:
> > Fixes: use of uninitialized value
> > Fixes:
> > 70929/clusterfuzz-testcase-minimized-ffmpeg_dem_IAMF_fuzzer-5931276639469568
> >
> > Found-by: continuous fuzzing process
> > http
On Fri, Aug 16, 2024 at 08:27:07PM -0300, James Almer wrote:
> On 8/16/2024 8:15 PM, Michael Niedermayer wrote:
> > Fixes: left shift of 1 by 31 places cannot be represented in type 'int'
> > Fixes:
> > 70726/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HEVC_fuzzer-6149928703819776
> >
> > F
On Fri, Aug 16, 2024 at 08:38:41PM -0300, James Almer wrote:
> On 8/16/2024 8:15 PM, Michael Niedermayer wrote:
> > Fixes: Assertion width > 0 && width <= 32 failed
> > Fixes:
> > 71012/clusterfuzz-testcase-minimized-ffmpeg_BSF_HEVC_METADATA_fuzzer-6073354744823808
> >
> > Found-by: continuous fu
On Tue, Aug 13, 2024 at 4:03 PM J. Dekker wrote:
> When collecting performance information from checkasm it is common
> to parse the output for use in graphs to compare vs different
> architectures.
>
> Signed-off-by: J. Dekker
> ---
When I redirect stdout to a csv file, the first two lines are:
Attached.
Together with previous patch "Less CPU use in hwaccel MJPEG decoding"
the hwaccel mjpeg decoding uses about 90% less cpu than before.
>From 5960e16ae7561c6c6ad982c90f4e6ea1d30df91b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Llu=C3=ADs=20Batlle=20i=20Rossell?=
Date: Sun, 18 Aug 2024 14:14:
Previously one could only convert the entire array to a string, not
access individual elements.
---
doc/APIchanges| 3 ++
libavutil/opt.c | 119 +++---
libavutil/opt.h | 40 ++
libavutil/tests/opt.c | 63 ++
---
libavutil/opt.h | 78 +++--
1 file changed, 75 insertions(+), 3 deletions(-)
diff --git a/libavutil/opt.h b/libavutil/opt.h
index 07e27a9208..23bc495158 100644
--- a/libavutil/opt.h
+++ b/libavutil/opt.h
@@ -240,26 +240,98 @@
* before the file is
The two functions are exactly the same.
---
libavutil/opt.c | 13 +
1 file changed, 1 insertion(+), 12 deletions(-)
diff --git a/libavutil/opt.c b/libavutil/opt.c
index 32a9e059e3..2cfc2d9c5a 100644
--- a/libavutil/opt.c
+++ b/libavutil/opt.c
@@ -1277,18 +1277,7 @@ int av_opt_get_imag
I wrote `ff_vvc_w_avg_8_rvv` by mimicking the h264 weight function.
Based on the test results for 49 different resolutions, most of them were
significantly slower.
Only 2x32 and 2x64 had similar performance, without noticeable speed
improvement.
I'm not sure about the reason. Some differences ar
24 matches
Mail list logo
