linux-kernel - Re: [PATCH v2 00/12] media: rkvdec: Add support for VDPU381 and VDPU383

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <3a9f613bfcc560281d7287ca2e411dadcac745b6.camel@ndufresne.ca>
Date: Thu, 18 Sep 2025 09:31:30 -0400
From: Nicolas Dufresne <nicolas@...fresne.ca>
To: Diederik de Haas <didi.debian@...ow.org>, Detlev Casanova
	 <detlev.casanova@...labora.com>, linux-kernel@...r.kernel.org
Cc: Mauro Carvalho Chehab <mchehab@...nel.org>, Heiko Stuebner
 <heiko@...ech.de>, 	linux-media@...r.kernel.org,
 linux-rockchip@...ts.infradead.org, 	linux-arm-kernel@...ts.infradead.org,
 kernel@...labora.com
Subject: Re: [PATCH v2 00/12] media: rkvdec: Add support for VDPU381 and
 VDPU383

Hi,

Le mercredi 17 septembre 2025 à 19:34 +0200, Diederik de Haas a écrit :
> Hi Detlev,
> 
> On Fri Aug 8, 2025 at 10:03 PM CEST, Detlev Casanova wrote:
> > These variants are found respectively in the RK3588 and RK3576 SoCs.
> > This patch only adds support for H264 and H265 in both variants.
> 
> I tested this on my Rock 5B (rk3588) with (ffmpeg and) mpv on Debian
> Forky with sway and I was quite impressed with the results :-)
> 
> In my earlier testing I found that the classic BBB video made for a
> really good test video and it played REALLY well. The 1080p version in
> both x264 and 8-bit x265 played without frame drops, although when
> pressing 'I' to show video info over the video it dropped some frames.
> But that never resulted in any visible artifacts.
> On the 2160p version I did see quite a number of dropped frames, but
> didn't notice any visual artifacts. Maybe that it got displayed on a
> 1080p monitor had something to do with it?

Sounds like we have "correctness" with ok performance, so that is enough for me
to consider the patches. There is few things that are not implemented that could
explain lower performance compare to the downstream stuff, such as the lack of
AFBC. It will hopefully come later, giving nice boost when doing GPU streaming.

> 
> 10-bit encoded x265 files (still) resulted in a completely blue output,
> but I have strong reasons to believe that's due to 'missing pieces' (ie
> NV15 support) in the rest of the display pipeline.

Its still quite early days for NV15 (and its derivatives) upstream of
everything, so I would not block on that. It passes the tests when we read back
the pixels from CPU I believe.

> 
> The displayed video seemed sharper and better then in my other tests.
> I thought I did see an artifact around 5:50 when a big rock almost fell
> on the little animal. Looking further, it appears to be an artifact in
> the original video :-O (I saw it also on my AMD GPU on my amd64 system
> and also when using software decoding).
> The fact I only noticed it when testing this patch set is saying a lot
> ... in a positive way! So I'm happy to provide my

That's it, your video watching experience is now ruined for the rest of your
life :-D

> 
> Tested-by: Diederik de Haas <didi.debian@...ow.org>  # Rock 5B

thanks a lot for testing,
Nicolas

Download attachment "signature.asc" of type "application/pgp-signature" (229 bytes)