Hi,
Le 4 oct. 2015 04:16, "Michael Niedermayer" <michae...@gmx.at> a écrit :
>
> On Sat, Oct 03, 2015 at 06:59:22PM +0200, Christophe Gisquet wrote:
> > The functions related to bitstream reading must use the natural zigzag
> > order, and not th
Hi,
2015-10-02 20:57 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> @@ -498,7 +498,7 @@ void dnxhd_unquantize_c(DNXHDEncContext *ctx, int16_t
> *block, int n,
> : ctx->cid_table->luma_weight;
>
> for (i = 1; i <= las
2015-10-02 21:00 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> This a 4:4:4 10 bits profile, where image size is not fixed by the
> profile, and which strays a bit outside the old frame header parsing
> code.
>
> Fixes ticket #4581 (DNxHR is
From: Jeremy James <jeremy.ja...@gmail.com>
Signed-off-by: Christophe Gisquet <christophe.gisq...@gmail.com>
---
libavcodec/dnxhddata.c | 7 +++
libavcodec/dnxhddec.c | 5 +++--
2 files changed, 10 insertions(+), 2 deletions(-)
diff --git a/libavcodec/dnxhddata.c b/libavcodec
Forward and inverse DCTs may not have the same permutation table,
resulting in a bitstream where the AC coefficients are encoded in
an unexpected order.
---
libavcodec/dnxhdenc.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/libavcodec/dnxhdenc.c
CIDs 1270 (10 & 12bits) to 1274, as produced by Avid Media Composer,
have been tested. I merged and adapted a bit Jeremy's work (but it's
mostly his code), hence the signing-off.
Christophe Gisquet (3):
isom: add support for DNxHR codec family
dnxhd: add CID 1270
dnxhddec: replace som
A series of 0 in a table can be confusing, and the corresponding checks
strange, so using a macro instead of that magic is more readable.
---
libavcodec/dnxhddata.c | 10 +-
libavcodec/dnxhddata.h | 3 +++
libavcodec/dnxhddec.c | 6 --
3 files changed, 12 insertions(+), 7
This a 4:4:4 10 bits profile, where image size is not fixed by the
profile, and which strays a bit outside the old frame header parsing
code.
Fixes ticket #4581 (DNxHR is not stricly supported, but that sequence is).
---
libavcodec/dnxhddata.c | 23 +
libavcodec/dnxhddec.c |
From: Jeremy James <jeremy.ja...@gmail.com>
These are DNxHR profiles with the following properties:
- Variable size in a profile, requiring variable-sized macroblock table
- Variable bitdepth, up to 12 bits
Signed-off-by: Christophe Gisquet <christophe.gisq...@gmail.com>
---
This is actually similar to DNxHD.
---
libavformat/isom.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavformat/isom.c b/libavformat/isom.c
index eff04ff..e55738b 100644
--- a/libavformat/isom.c
+++ b/libavformat/isom.c
@@ -240,6 +240,7 @@ const AVCodecTag ff_codec_movvideo_tags[] = {
crashes due to mishandling of code around execute2
The speed of slice-threading is equivalent to frame-threading, which
makes the later less attractive. Several decoders forgo implementing
it.
--
Christophe
From 354a8e016e1e10a4288534782104479aae3e62a9 Mon Sep 17 00:00:00 2001
From: Christophe Gisquet <
Hi,
2015-09-27 10:09 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> CID 1256 chroma table is the same as luma table for CID 1235 and 1256,
> which is consistent with the content being (normally) RGB.
Please ignore for now.
--
Nothing major, mostly related to error reporting.
Christophe Gisquet (7):
dnxhddec: remove unused qscale parameter
dnxhddec: simplify block parsing calls
dnxhddec: check and report bitstream errors
dnxhddec: use unsafe bitstream reader
dnxhddec: reindent/cosmetics
dnxhddata
Each line is padded by the format, and errors are now reported and
stop the decoding. Around 5% speedup.
---
libavcodec/dnxhddec.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavcodec/dnxhddec.c b/libavcodec/dnxhddec.c
index 8d30446..06d4780 100644
--- a/libavcodec/dnxhddec.c
+++
Fewer arguments, less duplicated code.
---
libavcodec/dnxhddec.c | 43 ---
1 file changed, 16 insertions(+), 27 deletions(-)
diff --git a/libavcodec/dnxhddec.c b/libavcodec/dnxhddec.c
index 34cf379..007eabe 100644
--- a/libavcodec/dnxhddec.c
+++
This only occur when an overrun in coefficient decoding is
detected.
---
libavcodec/dnxhddec.c | 48 ++--
1 file changed, 34 insertions(+), 14 deletions(-)
diff --git a/libavcodec/dnxhddec.c b/libavcodec/dnxhddec.c
index 007eabe..8d30446 100644
---
---
libavcodec/dnxhddec.c | 25 -
1 file changed, 12 insertions(+), 13 deletions(-)
diff --git a/libavcodec/dnxhddec.c b/libavcodec/dnxhddec.c
index fa93407..34cf379 100644
--- a/libavcodec/dnxhddec.c
+++ b/libavcodec/dnxhddec.c
@@ -63,7 +63,7 @@ typedef struct
CID 1256 chroma table is the same as luma table for CID 1235 and 1256,
which is consistent with the content being (normally) RGB.
CID 1244 weight tables are identica to CID 1244, but as this was not
tested against a real sample, it was not added.
---
libavcodec/dnxhddata.c | 16 +++-
---
libavcodec/dnxhddec.c | 32
1 file changed, 16 insertions(+), 16 deletions(-)
diff --git a/libavcodec/dnxhddec.c b/libavcodec/dnxhddec.c
index 06d4780..7b96004 100644
--- a/libavcodec/dnxhddec.c
+++ b/libavcodec/dnxhddec.c
@@ -64,18 +64,18 @@ typedef struct
., Baptiste Coudurier
* Copyright (c) 2011 MirriAd Ltd
+ * Copyright (c) 2015 Christophe Gisquet
*
* 10 bit support added by MirriAd Ltd, Joseph Artsimovich <jos...@mirriad.com>
+ * Slice multithreading and MB interlaced support added by Christophe Gisquet
*
* This file is part of
2015-09-28 11:21 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> Please ignore for now.
The specs explicitly says it is the same for luma and chroma. Here's a
patch that confidently reflects that.
--
Christophe
From bfbb0bdde4eee8d44d9c158e04b50fb4177c428f Mon Sep 17
Sent from a phone so please excuse terseness.
Le 20 sept. 2015 12:47, "Ronald S. Bultje" a écrit:
> if (more_data < 0) {
> s->tab_slice_address[ctb_addr_rs] = -1;
> +avpriv_atomic_int_set(>wpp_err, 1);
> +
r the following people:
> Michael Niedermayer
> Clément Bœsch
> James Almer
> Paul B Mahol
> Carl Eugen Hoyos
> Andreas Cadhalpun
> Ronald S. Bultje
> wm4
> Lukasz Marek
> Rostislav Pehlivanov
> Hendrik Leppkes
>
The sample position is made weird and non-nominal to force catching
such issues as default values or specialized operations hiding
issues in corner cases.
---
tests/fate/filter-video.mak | 4
tests/ref/fate/filter-scalechroma | 26 ++
2 files changed, 30
Hi,
2015-09-23 17:06 GMT+02:00 Mickaël Raulet :
> the fix looks weird to me. There is something else underlying.
The issue appeared in d249e682 (not the hash I reported, for some
reason). If you want to test, that commit is buggy (it causes
crashes): commit 7117547 is
There are 2 issues, listed in ticket #4876:
- new mb header syntax, where the firt bit indicates mbaff-like handling
- incorrectly scanned weight tables
Christophe Gisquet (1):
dnxhddec: decode and use interlace mb flag
Jeremy James (1):
dnxhddata: correct weight tables
libavcodec
Hi,
2015-09-25 22:36 GMT+02:00 Ronald S. Bultje :
> +pd_16: times 8 dd 16
In h264_intrapred_10bit.asm
> +pd_32: times 8 dd 32
In h264_idct_10bit.asm
And that's about anything remotely useful I have to say, as you were
careful about those duplications.
> +pmaxsw
Indicates a YCbCr->RGB transform at the block level. Although nothing
explicitly states it, this would assume the actual content is planar
RGB.
Currently unsupported, but the one sequence I found using it flagged
every mb that way, actually meaning the content was YCbCr, and thus
best left to the
It currently only applies to CID 1260, but this flag is dependent on
a higher-level flag located in the header.
---
libavcodec/dnxhddec.c | 7 ++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/libavcodec/dnxhddec.c b/libavcodec/dnxhddec.c
index 0722aa0..82a0d3c 100644
---
Around 3x speedup with 4 threads. Maybe more mb lines should be
batched per thread, but that's good enough for a first try.
---
libavcodec/dnxhddec.c | 221 ++
1 file changed, 133 insertions(+), 88 deletions(-)
diff --git a/libavcodec/dnxhddec.c
there's actually change in the flag
value, it is useless. It raises again the question of colorspace
conversion in a codec when various libraries exist.
Christophe Gisquet (4):
dnxhddec: implement slice multithreading
dnxhddec: indicate colorspace
dnxhddec: proper rule for interlaced mb flag
It is supposed to only old BT.709 colorspaces.
---
libavcodec/dnxhddec.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavcodec/dnxhddec.c b/libavcodec/dnxhddec.c
index d9762e4..0722aa0 100644
--- a/libavcodec/dnxhddec.c
+++ b/libavcodec/dnxhddec.c
@@ -82,6 +82,7 @@ static av_cold int
This bit is 1 in some samples, and seems to coincide with interlaced
mbs and CID1260. 2008 specs do not know about it, and maintain qscale
is 11 bits. This looks oversized, but may help larger bitdepths.
Currently, it leads to an obviously incorrect qscale value, meaning
its syntax is shifted by
orrect chroma table for CID 1256.
Fixes last issue from ticket #4876.
Found-by: Christophe Gisquet <christophe.gisq...@gmail.com>
Signed-off-by: Christophe Gisquet <christophe.gisq...@gmail.com>
---
libavcodec/dnxhddata.c | 103 +++--
1 file chan
Hi,
2015-10-05 10:57 GMT+02:00 tim nicholson :
> I thought JJ had got hold of a copy of the specs...
He was quick to notice the 2014 ones do not cover DNxHR. But the
bitstream doesn't seem to depart that much from the old stuff, and
ffmpeg's output after this
Hi,
2015-10-04 4:14 GMT+02:00 Michael Niedermayer :
> ./ffmpeg -f lavfi -i testsrc=s=1440x1080 -vframes 1 -pix_fmt yuv422p -vcodec
> dnxhd -vb 80M -flags +ildct file.mov
btw this afaik uses CID 1260 which has a particular feature,
MBAFF-like. While we do handle it since a few
Hi,
2015-12-16 4:24 GMT+01:00 James Almer :
> Ping for patchset.
I'll probably have a look on Sunday. But I don't expect my review to
be very worthwhile, so several people can probably do a better and
faster one.
--
Christophe
___
Hi,
2016-01-06 18:32 GMT+01:00 foo86 :
> dca dca2libdcadec
> System 1: 0:11.90 0:11.16 0:19.73
> System 2: 0:57.00 0:55.23 1:21.33
> System 3: 7:41.31 7:00.84 13:16.70
Just to be sure, because I won't check myself: is
Hi,
2016-01-07 12:48 GMT+01:00 foo86 :
> bench dca pcm_f32le
> bench dca2 pcm_f32le
> bench libdcadec pcm_s32le
OK, that was mostly out of curiosity, as "dca2" has benefits
surpassing such issues anyway. And the improvement is 10% for where it
matters (raspberry).
Hi,
2016-01-07 12:11 GMT+01:00 Hendrik Leppkes :
> +static void p010LEToY_c(uint8_t *dst, const uint8_t *src, const uint8_t
> *unused1,
> +const uint8_t *unused2, int width, uint32_t *unused)
> +{
> +int i;
> +for (i = 0; i < width; i++) {
> +
2016-01-08 17:41 GMT+01:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> On a side note, do you intend to do an avx version?
Just did that, need a tail processing, and no gain on a(n) Haswell
over a 7 minute sample.
The code for reference, over your set of patches.
--
Christop
Hi,
2016-01-08 20:16 GMT+01:00 James Almer :
> Then honestly i don't think it's worth it. Maybe if we could change the
> alignment to 32 bytes and making len multiple of 8 or 16, but not sure
> how feasible is that.
There could be a more specialized version, but of doubtful
Hi,
2015-12-16 4:24 GMT+01:00 James Almer :
> On 12/10/2015 8:02 PM, James Almer wrote:
>> Signed-off-by: James Almer
>> ---
>> libavcodec/x86/hevc_sao_10bit.asm | 142
>> +++---
>> 1 file changed, 57 insertions(+), 85
Hi,
here's I think a list of things left to do. I remember saste doing it
on some occasions.
Please comment on whether you think I have pointed an actual action to
perform. Don't mind the details for now, it's just to get the train
going.
2016-05-30 10:49 GMT+02:00 Michael Niedermayer
Hi,
sorry if I'm or was confusing, I'm best-effort here.
2016-06-03 21:13 GMT+02:00 Michael Niedermayer :
> > FOUR.TWO)
[...]
> i want some assistent to help with dayly server admin duties
> most root admins we have help and contribute but are often busy
> raz recently
Hi,
2016-05-29 21:51 GMT+02:00 Paul B Mahol :
> +typedef struct Slice {
> +uint32_t start;
> +uint32_t size;
> +} Slice;
I'm not a security expert, but is there a reason for not using plain int there ?
> +typedef struct MagicYUVContext {
> +AVFrame*p;
>
2016-05-30 17:50 GMT+02:00 Paul B Mahol :
> On 5/30/16, Piotr Bandurski wrote:
>> Hi,
>>
>>> patch attached.
>>
>> Is decoding of interlaced video supported? Because I get here invalid
>> output.
>>
>> Also crash happens with this fuzzed file:
>>
>>
Hi,
2016-05-30 15:09 GMT+02:00 Paul B Mahol :
Hi,
2016-05-30 15:09 GMT+02:00 Paul B Mahol :
>> ffmpeg seems to have libavutil/qsort.h, but I don't even know how much
>> effort is needed to use it here.
>
> Changed, doesn't help but maybe will for other archs.
Hi,
2016-01-08 16:22 GMT+01:00 James Almer :
> -test lenq, lenq
> -jz .end
> shl lenq, 2
> add src0q, lenq
> add src1q, lenq
> @@ -377,5 +375,4 @@ cglobal butterflies_float, 3,3,3, src0, src1, len
> mova[src0q + lenq], m0
Hi,
2016-01-08 16:22 GMT+01:00 James Almer :
> INIT_XMM sse
> cglobal butterflies_float, 3,3,3, src0, src1, len
> -%if ARCH_X86_64
> -movsxdlenq, lend
> -%endif
> -shl lenq, 2
> +shl lend, 2
All the more ok since, afaik, only WIN64 actually
Hi,
2016-01-08 16:22 GMT+01:00 James Almer :
> +shl offsetd, 2
> +add v1q, offsetq
> +add v2q, offsetq
> neg offsetq
> -shl offsetq, 2
> -sub v1q, offsetq
> -sub v2q, offsetq
Lucky that we never had any crash then.
> -#define YUVRGB_TABLE_HEADROOM 256
> +#define YUVRGB_TABLE_HEADROOM 512
[...]
> -const int yoffs = fullRange ? 384 : 326;
> +const int yoffs = fullRange ? 896 : 838;
I think it's time to use that macro everywhere it is actually used without
showing up.
Best regards,
Christophe
Hi,
2016-01-14 9:41 GMT+01:00 Michael Niedermayer :
> there are 2 seperate tables, the headroom in them also differs
> curently, also the exact relation between input values and headroom
> needed in the 2nd table depends on the yuv-rgb coefficients
>
> i can replace the
Hi,
2016-02-08 8:15 GMT+01:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> At this point, I'd say the encoder would better use that prefix.
> That's what the attached patch does (rebased but not tested). You may
> consider building on top of it for that purpose.
Which I h
Hi,
2016-02-09 6:27 GMT+01:00 Mark Reid :
[...]
> I'm guessing its okay to
> include "libavcodec/dnxhddata.h" in something thats in libavformat
> then? movenc.c uses those prefixes too.
Argh, nice catch, you can't use the the content of ff_dnxhd_headers
there, which is the
Hi,
2016-02-08 2:31 GMT+01:00 Mark Reid :
> +static int dnxhd_is_prefix(uint64_t prefix)
> +{
> + if (prefix == DNXHD_HEADER_PREFIX||
> + prefix == DNXHD_HEADER_PREFIX444 ||
> + prefix == DNXHD_HEADER_PREFIXHR1 ||
> + prefix == DNXHD_HEADER_PREFIXHR2)
> +
Hi,
as a motus operandi for this review, I have no time for a proper one,
or at least not fitting with John's timeframe. I'll try to close as
many pending discussions, and would prefer if someone else completed
the review/validation/commit.
2016-01-22 19:33 GMT+01:00 John Cox
Hi,
2016-02-22 22:43 GMT+01:00 James Almer :
> +.loop:
> +%if cpuflag(avx)
> +cvtdq2ps m4, [lfeq]
> +shufpsm5, m4, m4, q0123
> +%elif cpuflag(sse2)
> +movu m4, [lfeq]
> +cvtdq2ps m4, m4
> +pshufdm5, m4, q0123
> +%endif
> +
> +.inner_loop:
>
Hi,
> +.inner_loop:
Given this precludes reusing m5, then I don't have anything more to
comment, and seems ok.
--
Christophe
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Hi,
2016-01-20 15:27 GMT+01:00 John Cox :
> The by22 code gained me an overall factor of two in the abs level decode
> - the gains do depend a lot on the quantity of residual - you gain a lot
> more on I-frames than you do otherwise as they tend to have much longer
>
Hi,
2016-01-19 13:46 GMT+01:00 John Cox :
> I've just done a fair bit of work on hevc_cabac decode for the Rasberry
> Pi2 and I think that the patch is generally applicable. Patch is
> attached but you may prefer to take it from git:
This work is certainly impressive, and
Hi,
2016-01-22 14:29 GMT+01:00 John Cox :
>>This is a big slowdown on Win64 and UHD-bluray like sequences, but
>>that can be switched off in that case.
>
> I'm a bit surprised that it generated a big slowdown - some cache must
> be running just on the edge, but yes if you
Hi,
2016-01-21 11:45 GMT+01:00 John Cox :
> Hi
>
> v2 of my hevc residual patch
I'll review the bit not related to significant coeffs first, because I
think it is the most performance-sensitive. Also there are bits that
could be moved to other patches, at least some are
Hi,
2016-02-14 0:43 GMT+01:00 Michael Niedermayer :
> i can test and commit the code, it seems everyone who wanted to
> comment did comment
Yes, nothing really worth postponing the patchset from my side, except
maybe cleaner splitting of the bypass stuff, as agreed by the
Hi,
2016-02-14 6:49 GMT+01:00 Timothy Gu :
> %if HAVE_MMX_INLINE
Isn't that macro meant for C code (and in config.asm without much of a purpose)?
I suspect it is not useful, but I haven't dug into that.
> +; XXX some of these macros are not used right now, but they will
Hi,
2016-03-10 19:57 GMT+01:00 Hendrik Leppkes :
> This patch (the relocations part) broke debugging mingw-w64 ffmpeg
> builds with gdb, you can't set breakpoints anymore when its applied.
That issue prevented me to do anything interesting for ffmpeg since
then, thinking it
Hi,
2016-03-11 8:57 GMT+01:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
>> It should either be reverted or made dependent on
>> --enable/disable-debug (I would favor the first, honestly, since its a
>> rather ugly hack in itself).
>
> At the very least,
Hi,
2016-03-19 19:08 GMT+01:00 Ismail Donmez <ism...@i10z.com>:
>> 2016-03-11 8:57 GMT+01:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
>>>> It should either be reverted or made dependent on
>>>> --enable/disable-debug (I would favor the first, h
ophe
From 87e4f2a42bdb5f733d104ffba7cf70f786b72a03 Mon Sep 17 00:00:00 2001
From: Christophe Gisquet <christophe.gisq...@gmail.com>
Date: Sat, 19 Mar 2016 14:45:23 +0100
Subject: [PATCH] mingw64: configure: disable pie with debug enabled
This breaks use of gdb.
---
configure | 8 ++--
1 file changed, 6 insertions(
Hi,
2016-04-12 22:53 GMT+02:00 Paul B Mahol :
> -LLAudDSPContext dsp; ///< accelerated
And later:
> +static int scalarproduct_and_madd_int(int *v1, const int *v2,
> + const int *v3,
> +
2016-04-29 10:50 GMT+02:00 Paul B Mahol :
> Should be OK if it doesn't break anything.
I'll resend the current state of this patchset for easier testing &
applying. Michael ran this under valgrind with nothing popping up, and
fate passes.
I think the remaining thing is: is the
Patch 2 is the squashing of several previous commits, as there were
no opinion on their contents nor the way to go.
The SSE4 one is the final version from its last thread.
The last patch in this set is new, and silences a warning that's only
meaningful for 16bits content.
Christophe Gisquet (5
This is done by actually handling the cascaded LMS data as if it
were int16_t, thus requiring switching at various locations the
computations.
---
libavcodec/wmalosslessdec.c | 146 +---
1 file changed, 84 insertions(+), 62 deletions(-)
diff --git
---
tests/fate/lossless-audio.mak | 5 -
tests/ref/fate/lossless-wma24-1 | 1 +
tests/ref/fate/lossless-wma24-2 | 1 +
3 files changed, 6 insertions(+), 1 deletion(-)
create mode 100644 tests/ref/fate/lossless-wma24-1
create mode 100644 tests/ref/fate/lossless-wma24-2
diff --git
16bits samples with CDLMS orders of 8 are currently unsupported, but have never
been encountered before.
However, 8 seems to be the most frequent, if not the only order used for 24bits.
In that case, the dsp functions are fine with handling order that are multiples
of 8, so silence the warning.
The unique user so far is wmalossless 24bits. The few samples tested show an
order of 8, so more unrolling or an avx2 version do not make sense.
Timings: 72 -> 49 cycles
---
libavcodec/x86/lossless_audiodsp.asm| 31 +--
libavcodec/x86/lossless_audiodsp_init.c | 7
The loops are guaranteed to be at least multiples of 8, so this
unrolling is safe but allows exploiting execution ports.
For int32 version: 72 -> 57c.
---
libavcodec/lossless_audiodsp.c | 12
1 file changed, 8 insertions(+), 4 deletions(-)
diff --git
Hi,
2016-05-20 1:55 GMT+02:00 Lukasz Marek :
> Is Derek revoked to commit or what? Couldn't he just commit this patch and
> leave? :P I was a problem for some people, but I see they still have
> problems. Let people with problems go away with they problems.
Sorry if
Hi,
2016-05-18 20:40 GMT+02:00 Michael Niedermayer :
> Please state clearly if you agree to the text or if not.
> we can extend and tune it later and do another vote if there are more
> suggestions
I agree to having a CoC.
This text is a first step, so I'm ok with it,
Hi,
2016-05-20 2:38 GMT+02:00 Timothy Gu :
>> > Note how it has a list of specific violations, instead of vague things like
>> > "Be excellent" that the FFmpeg one has.
>> > Note how it has a huge section on disciplinary procedures.
[...]
> I have to agree with Kieran here.
2016-05-13 11:48 GMT+02:00 foo86 :
> -unsigned int v = get_unary(gb, 1, 128);
> +unsigned int v = get_unary(gb, 1, get_bits_left(gb));
Not that the patch is not ok, but I have a few uneducated questions:
1) Given the get_bits_long(gb, k) afterwards, won't that code
2016-05-01 15:33 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> The loops are guaranteed to be at least multiples of 8, so this
> unrolling is safe but allows exploiting execution ports.
>
> For int32 version: 68 -> 58c.
Ping?
This was ok'ed by James irr
Hi,
2016-05-03 19:24 GMT+02:00 Hendrik Leppkes :
>> +// The reference decoder ignores it, and its typical length is 0
>> +memset(put_bits_ptr(pb), 0, s->prefix_bytes);
>> skip_put_bytes(pb, s->prefix_bytes);
>> +
>
> I don't suppose we have a function to just
Le 3 mai 2016 22:15, "Rostislav Pehlivanov" <atomnu...@gmail.com> a écrit :
>
> On 3 May 2016 at 19:16, Christophe Gisquet <christophe.gisq...@gmail.com>
> wrote:
> >
> >
> > Btw, afaik, the padding is 0xFF, so expecting 0 in the buffer there
&g
Hi,
2016-05-02 16:02 GMT+02:00 Michael Niedermayer :
>> +fate-lossless-wma24-rawtile: CMD = md5 -i
>> $(TARGET_SAMPLES)/lossless-audio/g2_24bit.wma -f s24le
>
> where can i find that file ?
> i assume i should upload it ?
Sorry, I thought we had discussed it in this
h raw pcm tiles then.
--
Christophe
From 584999fcce24585f989d2dc770e8c7c85aa19db7 Mon Sep 17 00:00:00 2001
From: Christophe Gisquet <christophe.gisq...@gmail.com>
Date: Mon, 18 Apr 2016 12:53:21 +0200
Subject: [PATCH 1/4] fate: wma: add lossless 24bits tests
Should evaluate coefficients and raw pcm t
2016-05-01 15:33 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> This is done by actually handling the "prev_values" in the cascaded LMS data
> as if it were int16_t, thus requiring switching at various locations the
> computations.
Patch update s
Hi,
2016-05-01 15:33 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> +fate-lossless-wma24-2: CMD = md5 -i
> $(TARGET_SAMPLES)/lossless-audio/Mega_Weird_Audio_Test_24bit.wma -f s24le
The recent fixes actually changed the crc for that file.
Is https://trac.ffmpeg.or
This is done by actually handling the "prev_values" in the cascaded LMS data
as if it were int16_t, thus requiring switching at various locations the
computations.
---
libavcodec/wmalosslessdec.c | 109 +++-
1 file changed, 58 insertions(+), 51 deletions(-)
The unique user so far is wmalossless 24bits. The few samples tested show an
order of 8, so more unrolling or an avx2 version do not make sense.
Timings: 68 -> 49 cycles
---
libavcodec/x86/lossless_audiodsp.asm| 33 +
libavcodec/x86/lossless_audiodsp_init.c |
Due to the changes to the cascaded LMS coefficients, most of the code
needed a rewrite.
In particular, the SSE4 madd32 code is no longer that similar to be
shared inside a macro.
Christophe Gisquet (4):
fate: wma: add lossless 24bits test
wmalossless: allow calling madd_int16
x86: lossless
The loops are guaranteed to be at least multiples of 8, so this
unrolling is safe but allows exploiting execution ports.
For int32 version: 68 -> 58c.
---
libavcodec/lossless_audiodsp.c | 12
1 file changed, 8 insertions(+), 4 deletions(-)
diff --git
---
tests/fate/vcodec.mak | 17 -
tests/ref/vsynth/vsynth1-vc2-420p | 4
tests/ref/vsynth/vsynth1-vc2-420p10 | 4
tests/ref/vsynth/vsynth1-vc2-420p12 | 4
tests/ref/vsynth/vsynth1-vc2-422p | 4
The slice prefix is 0 in the reference encoder and the decoder ignores it.
Writing 0 there seems like the best temporary solution.
The padding could have contained uninitialized data, but reference VC2
encoders put 0xFF there, hence the memset value.
Overall this allows producing bistreams with
Hi,
2016-05-06 2:19 GMT+02:00 Rostislav Pehlivanov :
> I plan to merge the fate tests as well tomorrow or on Saturday when I'll
> have time to quickly fix bugs which appear on platforms I haven't tested
> the encoder on. Hopefully none, but you never know.
Sure, makes sense.
Hi,
2016-05-04 3:06 GMT+02:00 Rostislav Pehlivanov :
> vc2hqencode is not the reference encoder, vc2-reference is. It's even worse
> though.
Sorry, I thought authoritative could mean "from the authors", so I
didn't mean it as "the" reference/"the authority". Just a good
---
tests/fate/vcodec.mak | 17 -
tests/ref/vsynth/vsynth1-vc2-420p | 4
tests/ref/vsynth/vsynth1-vc2-420p10 | 4
tests/ref/vsynth/vsynth1-vc2-420p12 | 4
tests/ref/vsynth/vsynth1-vc2-422p | 4
2016-05-03 19:06 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
> +memset(pb->buf_ptr, 0, pad_c);
Commit squashing fail, attached patch should fix that. This
unfortunately requires updating the fate tests as I generated them
from this squashing.
--
Chr
2016-05-03 19:06 GMT+02:00 Christophe Gisquet <christophe.gisq...@gmail.com>:
[SNIP]
Incorrect padding used (0 instead of 0xFF), fixed in that patch series.
--
Christophe
From 22ff25711062fb1ca30da1674fd622fd6f81c8e3 Mon Sep 17 00:00:00 2001
From: Christophe Gisquet <christ
The slice prefix is 0 in the reference encoder and the decoder ignores it.
Writing 0 there seems like the best temporary solution.
The padding could have contained uninitialized data, but its standardized value
is 0xFF, hence the memset value.
Overall this allows producing bistreams with no
401 - 500 of 564 matches
Mail list logo