Re: xf86-video-qxl performance

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Alexander Larsson píše v Pá 25. 05. 2012 v 09:07 +0200:
> On Wed, 2012-05-23 at 15:20 -0500, Jeremy White wrote:
> 
> > Also, as a crazy idea, has anyone considered implementing a pure
> > streaming video driver?  That is, what if we had a frame buffer driver,
> > and then a thread that fired 29 times a second to drive a theora or vp8
> > encoder, simply feeding the current frame at those intervals.
> 
> Its not crazy at all. In fact, all the times I've been pondering how to
> support modern (i.e. GPU-based) 3D acceleration to spice I've always
> ended up at this. Its not really practical to support GPUs like a set of
> remoted drawing operations, since:
> a) They are infinitely flexible (turing complete)
> b) Work on huge amounts on temporary data (textures, buffers, etc),
>    often multiple times the size of what the final rendered screen
>    size is.
> c) Its impossible (comparable to halting problem) to figure out
>    in general which part of the input sources some GPU code snippet
>    uses, or to calculate the bounding box of the result.
> 
> So, what I think we want for GPU accelerated spice is an
> eventually-losless video codec. By that I mean the fact that individual
> frames might be lossy (so it can be effective for video and games), but
> if the input is static we'd (quickly) converge on the lossless result
> (so that you can read your word documents).
> 
> If additionally we could separate each toplevel window as its own stream
> that would be even nicer, but It might be hard as these are composed to
> the final screen using the GPU.
> 

This should get priority when work on acceleration is started, not being
done as the last thing. Current CPU/bandwith/memory tradedoff is set too
much against CPU, making any but most powerful clients just too slow to
be usable with clients with software-rendered composite WMs (aka
gnome-shell) and per-window handling of the display is something that
can fix that in very efficent way.

David

> _______________________________________________
> Spice-devel mailing list
> Spice-devel@xxxxxxxxxxxxxxxxxxxxx
> http://lists.freedesktop.org/mailman/listinfo/spice-devel

-- 

David Jaša, RHCE

SPICE QE based in Brno
GPG Key:     22C33E24 
Fingerprint: 513A 060B D1B4 2A72 7F0D 0278 B125 CD00 22C3 3E24



_______________________________________________
Spice-devel mailing list
Spice-devel@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/spice-devel



[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux]     [Linux OMAP]     [Linux MIPS]     [ECOS]     [Asterisk Internet PBX]     [Linux API]     [Monitors]