On 12/21/23 18:01, Andreas Rheinhardt wrote:
Signed-off-by: Andreas Rheinhardt
---
libavcodec/libjxlenc.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/libavcodec/libjxlenc.c b/libavcodec/libjxlenc.c
index d707f3a61b..055c9b7bb1 100644
--- a/libavcodec/libjxlenc.
The specification doesn't mention that clusters cannot have alphabet
sizes greater than 1 << bundle->log_alphabet_size, but the reference
implementation rejects these entropy streams as invalid, so we should
too. Refusing to do so can overflow a stack variable on line 556 that
should be large enoug
On an Intel Core i7 12700k:
decorrelate_ls_c: 814.3
decorrelate_ls_sse2: 165.8
decorrelate_ls_avx2: 101.3
decorrelate_sf_c: 1602.6
decorrelate_sf_sse4: 640.1
decorrelate_sf_avx2: 324.6
decorrelate_sm_c: 1564.8
decorrelate_sm_sse2: 379.3
decorrelate_sm_avx2: 203.3
decorrelate_sr_c: 785.3
decorrelat
On 12/22/2023 8:08 PM, Michael Niedermayer wrote:
On Thu, Dec 21, 2023 at 10:15:49PM -0300, James Almer wrote:
On an Intel Core i7 12700k:
decorrelate_ls_c: 814.3
decorrelate_ls_sse2: 165.8
decorrelate_ls_avx2: 101.3
decorrelate_sf_c: 1602.6
decorrelate_sf_sse4: 640.1
decorrelate_sf_avx2: 324.6
On Thu, Dec 21, 2023 at 10:15:49PM -0300, James Almer wrote:
> On an Intel Core i7 12700k:
>
> decorrelate_ls_c: 814.3
> decorrelate_ls_sse2: 165.8
> decorrelate_ls_avx2: 101.3
> decorrelate_sf_c: 1602.6
> decorrelate_sf_sse4: 640.1
> decorrelate_sf_avx2: 324.6
> decorrelate_sm_c: 1564.8
> decorre
On Fri, Dec 22, 2023, 11:41 Rémi Denis-Courmont wrote:
> Hello,
>
> The RISC-V board will be personally visiting the taylor to get a
> fashionable
> custom-made outfit. As a consequence, it will be taking a much deserved
> break
> from FATE service over the end of year holidays.
>
> --
>
On Fri, Dec 22, 2023 at 4:41 PM Ronald S. Bultje wrote:
> Hi Paul,
>
> On Fri, Dec 22, 2023 at 10:33 AM Paul B Mahol wrote:
>
> > There are some serious issues in this code.
> >
>
> While already very insightful, it would be enormously helpful if you could
> expand on this statement with more de
Fixes: out of array access:
Fixes: tickets/10745/poc12ffmpeg
Found-by: Li Zeyuan and Zeng Yunxiang.
Signed-off-by: Michael Niedermayer
---
libavfilter/vf_swaprect.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/libavfilter/vf_swaprect.c b/libavfilter/vf_swaprect.c
i
Signed-off-by: Michael Niedermayer
---
libavfilter/vf_swaprect.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/libavfilter/vf_swaprect.c b/libavfilter/vf_swaprect.c
index 8a983b605bf..20d8e233e45 100644
--- a/libavfilter/vf_swaprect.c
+++ b/libavfilter/vf_swaprect.c
@@ -
Signed-off-by: Michael Niedermayer
---
libavfilter/vf_swaprect.c | 5 +
1 file changed, 5 insertions(+)
diff --git a/libavfilter/vf_swaprect.c b/libavfilter/vf_swaprect.c
index 13ae149afdb..8a983b605bf 100644
--- a/libavfilter/vf_swaprect.c
+++ b/libavfilter/vf_swaprect.c
@@ -18,6 +18,7 @@
Fixes: out of array read
Fixes: tickets/10744/poc11ffmpeg
Found-by: Li Zeyuan and Zeng Yunxiang.
Signed-off-by: Michael Niedermayer
---
libavfilter/af_alimiter.c | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/libavfilter/af_alimiter.c b/libavfilter/af_alimiter.c
index
On Fri, Dec 22, 2023 at 7:20 AM Rémi Denis-Courmont wrote:
> >> > +checkasm_fail_func("%s",
> >> > + s == SIGFPE ? "fatal arithmetic error" :
> >> > + s == SIGILL ? "illegal instruction" :
> >> > + s == SIGBUS ?
Hello,
The RISC-V board will be personally visiting the taylor to get a fashionable
custom-made outfit. As a consequence, it will be taking a much deserved break
from FATE service over the end of year holidays.
--
Rémi Denis-Courmont
http://www.remlab.net/
__
Signed-off-by: Andreas Rheinhardt
---
libavcodec/d3d12va_av1.c | 2 +-
libavcodec/d3d12va_hevc.c | 6 +++---
libavcodec/d3d12va_vp9.c | 8
3 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/libavcodec/d3d12va_av1.c b/libavcodec/d3d12va_av1.c
index 6a3543353f..4a87169bfb 100
Le perjantaina 22. joulukuuta 2023, 18.13.51 EET Andreas Rheinhardt a écrit :
> Rémi Denis-Courmont:
> > Le perjantaina 22. joulukuuta 2023, 15.48.45 EET Andreas Rheinhardt a
écrit :
> >> Avoids relocations.
> >
> > This is a little bit misleading. It reduces the number of relocations
> > indeed,
decorrelate_sf_c: 259.2
decorrelate_sf_rvv_i32: 45.5
---
libavcodec/riscv/takdsp_init.c | 2 ++
libavcodec/riscv/takdsp_rvv.S | 21 +
2 files changed, 23 insertions(+)
diff --git a/libavcodec/riscv/takdsp_init.c b/libavcodec/riscv/takdsp_init.c
index 4312c8d99d..58be838
Rémi Denis-Courmont:
> Le perjantaina 22. joulukuuta 2023, 15.48.45 EET Andreas Rheinhardt a écrit :
>> Avoids relocations.
>
> This is a little bit misleading. It reduces the number of relocations indeed,
> but the data structures still end up in nonshareable .data.relro rather than
> .rodata d
Le perjantaina 22. joulukuuta 2023, 3.41.29 EET flow gg a écrit :
> It's at c908
>
> According to the benchmark results, if vlseg2e64 is used, the speed is
> almost as slow as C language (dcmul_add_rvv_f64: 86.2), if vsseg2e64 is
> used, it will be only a bit slower (dcmul_add_rvv_f64: 50.2).
Fai
Hi Paul,
On Fri, Dec 22, 2023 at 10:33 AM Paul B Mahol wrote:
> There are some serious issues in this code.
>
While already very insightful, it would be enormously helpful if you could
expand on this statement with more detail, such as a list of issues as they
exist in this code.
Ronald
__
Le perjantaina 22. joulukuuta 2023, 3.34.39 EET flow gg a écrit :
> func ff_decorrelate_sm_rvv, zve32x
> 1:
> vsetvli t0, a2, e32, m8, ta, ma
> vle32.v v8, (a1)
> sub a2, a2, t0
> vle32.v v0, (a0)
> vssra.vi v8, v8, 1
> vsub.vv v16, v0, v8
>
There are some serious issues in this code.
Also missing tests.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject
Signed-off-by: Andreas Rheinhardt
---
.mailmap | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/.mailmap b/.mailmap
index 7546cf0caf..dfef178cf0 100644
--- a/.mailmap
+++ b/.mailmap
@@ -17,7 +17,8 @@
-
+
+
rcombs
--
2.34.1
__
Le perjantaina 22. joulukuuta 2023, 15.48.45 EET Andreas Rheinhardt a écrit :
> Avoids relocations.
This is a little bit misleading. It reduces the number of relocations indeed,
but the data structures still end up in nonshareable .data.relro rather than
.rodata due to other remaining pointers.
This patch populates the third entry for HLS codec attribute using the
AAC profile.
The HLS specifications[1] require this digit to be the Object Type ID as
referred to in table 1.3 of ISO/IEC 14496-3:2009[2].
The numerical constants in the code refer to these OTIs minus one, as
documeted in comm
Quoting Andreas Rheinhardt (2023-12-22 14:48:45)
> Avoids relocations.
>
> Signed-off-by: Andreas Rheinhardt
> ---
Maybe mention that it's not needed after
acf63d5350adeae551d412db699f8ca03f7e76b9.
LGTM otherwise.
--
Anton Khirnov
___
ffmpeg-devel m
On 12/21/2023 5:05 PM, Paul B Mahol wrote:
On Thu, Dec 21, 2023 at 8:43 PM Tomas Härdin wrote:
ons 2023-12-20 klockan 20:11 +0100 skrev Michael Niedermayer:
On Wed, Dec 20, 2023 at 05:57:40PM +0100, Tomas Härdin wrote:
tis 2023-12-19 klockan 15:02 +0100 skrev Nicolas George:
[...]
[...] ,
Hi Martin,
Thanks for the review.
On 22/12/2023 12:15, Martin Storsjö wrote:
The change LGTM, but the wording here is slightly confusing IMO. The
problem isn't with using per se AVERROR, that's just a macro for
generating suitable integers, the issue is more about the fact that
we're returnin
On 12/22/2023 6:49 AM, Martin Storsjö wrote:
For memcpy and memcmp, we need to multiply by the element size,
otherwise we're copying and comparing only a fraction of the buffer.
For decorrelate_sr, the buffer p1 is the one that is mutated;
copy and check p1 instead of p2.
For decorrelate_sm, bo
On Fri, 22 Dec 2023, James Almer wrote:
Signed-off-by: James Almer
---
libavcodec/riscv/takdsp_init.c | 4 ++--
libavcodec/takdsp.c| 6 +++---
libavcodec/takdsp.h| 6 +++---
libavcodec/x86/takdsp_init.c | 12 ++--
tests/checkasm/takdsp.c| 6 +++---
5 fil
On Fri, 22 Dec 2023, James Almer wrote:
Signed-off-by: James Almer
---
Fixes broken logic as reported by Martin.
tests/checkasm/takdsp.c | 35 ---
1 file changed, 32 insertions(+), 3 deletions(-)
+static void test_decorrelate_sf(TAKDSPContext *s) {
+declar
Hi Frank,
On Fri, 22 Dec 2023, Frank Plowman wrote:
The gcrypt definition of `bn_new` used to use `AVERROR`, however it is
called in `dh_generate_key` and `ff_dh_init` which return pointers. As a
result, compiling with gcrypt and the ffrtmpcrypt protocol resulted in an
int-conversion warning. G
Signed-off-by: James Almer
---
libavcodec/riscv/takdsp_init.c | 4 ++--
libavcodec/takdsp.c| 6 +++---
libavcodec/takdsp.h| 6 +++---
libavcodec/x86/takdsp_init.c | 12 ++--
tests/checkasm/takdsp.c| 6 +++---
5 files changed, 17 insertions(+), 17 dele
On an Intel Core i7 12700k:
decorrelate_ls_c: 814.3
decorrelate_ls_sse2: 165.8
decorrelate_ls_avx2: 101.3
decorrelate_sf_c: 1602.6
decorrelate_sf_sse4: 640.1
decorrelate_sf_avx2: 324.6
decorrelate_sm_c: 1564.8
decorrelate_sm_sse2: 379.3
decorrelate_sm_avx2: 203.3
decorrelate_sr_c: 785.3
decorrelat
Signed-off-by: James Almer
---
Fixes broken logic as reported by Martin.
tests/checkasm/takdsp.c | 35 ---
1 file changed, 32 insertions(+), 3 deletions(-)
diff --git a/tests/checkasm/takdsp.c b/tests/checkasm/takdsp.c
index 3aecee1f72..78528b1c5d 100644
--- a/te
The gcrypt definition of `bn_new` used to use `AVERROR`, however it is
called in `dh_generate_key` and `ff_dh_init` which return pointers. As a
result, compiling with gcrypt and the ffrtmpcrypt protocol resulted in an
int-conversion warning. GCC 14 may upgrade these to errors [1].
This patch fixes
Fixes: out of array access
Fixes: tickets/10743/poc10ffmpeg
Found-by: Zeng Yunxiang and Li Zeyuan
Signed-off-by: Michael Niedermayer
---
libavfilter/vf_weave.c | 9 +++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/libavfilter/vf_weave.c b/libavfilter/vf_weave.c
index 84f3c
Fixes: out of array access
Fixes: Ticket10699
Fixes: poc5ffmpeg
Found-by: Zeng Yunxiang
Signed-off-by: Michael Niedermayer
---
libavfilter/edge_template.c | 29 +++--
1 file changed, 19 insertions(+), 10 deletions(-)
diff --git a/libavfilter/edge_template.c b/libavfilter
tests/checkasm/checkasm: C LSX LASX
put_hevc_qpel_uni_h4_8_c: 5.7 1.2
put_hevc_qpel_uni_h6_8_c: 12.22.7
put_hevc_qpel_uni_h8_8_c: 21.53.2
put_hevc_qpel_uni_h12_8_c: 47.29.2 7.2
put_hevc_qpel_uni_h16_8_c: 87.011.7
tests/checkasm/checkasm: C LSX LASX
put_hevc_epel_uni_w_hv4_8_c: 9.5 2.2
put_hevc_epel_uni_w_hv6_8_c: 18.55.0 3.7
put_hevc_epel_uni_w_hv8_8_c: 30.76.0 4.5
put_hevc_epel_uni_w_hv12_8_c: 63.714.010.7
put_hevc_epel_uni_w_hv16_8_c:
tests/checkasm/checkasm: C LSX LASX
put_hevc_pel_uni_w_pixels4_8_c:2.7 1.0
put_hevc_pel_uni_w_pixels6_8_c:6.2 2.0 1.5
put_hevc_pel_uni_w_pixels8_8_c:10.72.5 1.7
put_hevc_pel_uni_w_pixels12_8_c: 23.05.5 5.0
put_hevc_pel_uni_w_pixels16_8_c
tests/checkasm/checkasm: C LSX LASX
put_hevc_qpel_uni_w_h4_8_c:6.5 1.7 1.2
put_hevc_qpel_uni_w_h6_8_c:14.54.5 3.7
put_hevc_qpel_uni_w_h8_8_c:24.55.7 4.5
put_hevc_qpel_uni_w_h12_8_c: 54.717.512.0
put_hevc_qpel_uni_w_h1
After this patch, the peformance of decoding H265 4K 30FPS 30Mbps
on 3A6000 with 8 threads improves 2fps (45fps-->47fsp).
---
libavcodec/loongarch/Makefile | 3 +-
libavcodec/loongarch/hevc_add_res.S | 162 ++
libavcodec/loongarch/hevcdsp_init_loongarch.
Hello, everyone! The hevc asm optimizatons are submitted, here is a
brief introduction. After the 6 patches, the speedup of decoding H265
4K 30FPS 30Mbps on 3A6000 with 8 threads is about 33%(42fps-->56fps).
Reviews are welcome, thanks for in advance.
[PATCH v1 1/6] avcodec/hevc: Add init for sao
Forgot to init c->sao_edge_filter[idx] when idx=0/1/2/3.
After this patch, the speedup of decoding H265 4K 30FPS
30Mbps on 3A6000 is about 7% (42fps==>45fps).
Change-Id: I521999b397fa72b931a23c165cf45f276440cdfb
---
libavcodec/loongarch/hevcdsp_init_loongarch.c | 4
1 file changed, 4 inserti
Quoting James Almer (2023-12-20 04:02:37)
> On 12/19/2023 9:10 AM, Anton Khirnov wrote:
> > Quoting James Almer (2023-12-19 13:09:05)
> >> On 12/19/2023 7:41 AM, Anton Khirnov wrote:
> >>> Quoting James Almer (2023-12-18 20:30:47)
> On 12/18/2023 4:19 PM, Anton Khirnov wrote:
> > ---
> >>>
Quoting Paul B Mahol (2023-12-21 12:53:58)
> On Thu, Dec 7, 2023 at 6:26 PM Paul B Mahol wrote:
>
> >
> >
> > On Wed, Dec 6, 2023 at 2:38 PM Nicolas George wrote:
> >
> >> James Almer (12023-12-06):
> >> > I honestly can't believe you're arguing this.
> >>
> >> Yet I do, so I suggest you think a
The allocated buffer of an AVFrame need not be height * linesize
big. In case there is padding between lines, the last line need
not have this padding. Pathological examples exist with height == 1
(where linesize is not really meaningful); non-pathological examples
are produced by the separatefield
On Fri, 22 Dec 2023, Wu, Tong1 wrote:
From: ffmpeg-devel On Behalf Of
Martin Storsjö
Sent: Friday, December 22, 2023 4:41 PM
To: FFmpeg development discussions and patches
Subject: Re: [FFmpeg-devel] [PATCH 1/2] configure: Remove a redundant check
for UWP mode
On Fri, 22 Dec 2023, Wu, Tong1 w
On Thu, 21 Dec 2023, James Almer wrote:
Signed-off-by: James Almer
---
tests/checkasm/takdsp.c | 36 +---
1 file changed, 33 insertions(+), 3 deletions(-)
@@ -113,6 +113,35 @@ static void test_decorrelate_sm(TAKDSPContext *s) {
report("decorrelate_sm");
}
For memcpy and memcmp, we need to multiply by the element size,
otherwise we're copying and comparing only a fraction of the buffer.
For decorrelate_sr, the buffer p1 is the one that is mutated;
copy and check p1 instead of p2.
For decorrelate_sm, both buffers are mutated, so copy and check
both
On Mon, 18 Dec 2023, flow gg wrote:
+static void test_decorrelate_sm(TAKDSPContext *s) {
+#define BUF_SIZE 1024
+declare_func(void, int32_t *, int32_t *, int);
+
+if (check_func(s->decorrelate_sm, "decorrelate_sm")) {
+LOCAL_ALIGNED_32(int32_t, p1, [BUF_SIZE]);
+LOCAL
On Mon, 18 Dec 2023, flow gg wrote:
+static void test_decorrelate_sr(TAKDSPContext *s) {
+#define BUF_SIZE 1024
+declare_func(void, int32_t *, int32_t *, int);
+
+if (check_func(s->decorrelate_sr, "decorrelate_sr")) {
+LOCAL_ALIGNED_32(int32_t, p1, [BUF_SIZE]);
+LOCAL
On Mon, 18 Dec 2023, flow gg wrote:
+static void test_decorrelate_ls(TAKDSPContext *s) {
+#define BUF_SIZE 1024
+declare_func(void, int32_t *, int32_t *, int);
+
+if (check_func(s->decorrelate_ls, "decorrelate_ls")) {
+LOCAL_ALIGNED_32(int32_t, p1, [BUF_SIZE]);
+LOCAL
On Fri, Dec 22, 2023 at 7:28 AM Anton Khirnov wrote:
> Hi Paul,
> Quoting Paul B Mahol (2023-12-21 21:05:18)
> > Say what serious feature you contributed ? - Nothing.
>
> this is a personal attach. Personal attacks are not allowed on this
> mailing list. Please refrain from them in the future.
>
Avoids casts all over the place; in this case, it also
replaces the unsafe cast uint8_t**->const uint8_t **
by the safe cast uint8_t**->const uint8_t * const*.
Signed-off-by: Andreas Rheinhardt
---
libavutil/hwcontext_d3d12va.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff
Le 21/12/2023 à 12:11, Tomas Härdin a écrit :
ons 2023-12-20 klockan 18:04 +0100 skrev Cédric Le Barz:
Le 19/12/2023 à 14:36, Tomas Härdin a écrit :
+for ( comp = 0; comp < component_count; comp++ ) {
+avio_write(pb, &sc->j2k_info.j2k_comp_desc[3*comp] , 3);
+}
Looks like this
>From: ffmpeg-devel On Behalf Of
>Martin Storsjö
>Sent: Friday, December 22, 2023 4:41 PM
>To: FFmpeg development discussions and patches de...@ffmpeg.org>
>Subject: Re: [FFmpeg-devel] [PATCH 1/2] configure: Remove a redundant check
>for UWP mode
>
>On Fri, 22 Dec 2023, Wu, Tong1 wrote:
>
>>
>>
On Fri, 22 Dec 2023, Wu, Tong1 wrote:
Subject: [FFmpeg-devel] [PATCH 1/2] configure: Remove a redundant check for
UWP mode
The check for UWP mode was duplicated from right above, in
d54127c41a81cf2078a3504f78e0e4232cfe11b7.
Also, instead of several lines with "enabled uwp && ...", make one
"
On Fri, 22 Dec 2023 at 08:05, Anton Khirnov wrote:
> Quoting Kieran Kunhya (2023-12-22 02:40:55)
> > On Fri, 22 Dec 2023, 00:36 Michael Niedermayer,
> > wrote:
> >
> > > On Thu, Dec 21, 2023 at 03:22:20AM +0100, Michael Niedermayer wrote:
> > > > Hi all
> > > >
> > > > My main development machie
Quoting Kieran Kunhya (2023-12-22 02:40:55)
> On Fri, 22 Dec 2023, 00:36 Michael Niedermayer,
> wrote:
>
> > On Thu, Dec 21, 2023 at 03:22:20AM +0100, Michael Niedermayer wrote:
> > > Hi all
> > >
> > > My main development machiene died today, so expect increased delays
> > > about things that in
60 matches
Mail list logo
