[MPlayer-dev-eng] vulkan video output

sylvain.bertrand at gmail.com sylvain.bertrand at gmail.com
Sun Mar 4 21:46:52 EET 2018


On Sun, Mar 04, 2018 at 02:17:12PM +0100, wm4 wrote:
> On Sun, 4 Mar 2018 10:35:42 +0200
> Lauri Kasanen <cand at gmx.com> wrote:
> 
> > On Thu, 1 Mar 2018 15:36:59 +0000
> > sylvain.bertrand at gmail.com wrote:
> > 
> > > Hi,
> > > 
> > > Any work in progress on a vulkan video output?  
> > 
> > What would be the use of one? Vulkan was made to lower cpu usage on
> > lots of drawcalls and transfers. A media player does one drawcall and
> > uses very simple fragment shaders - it does not benefit from vulkan.
> > 
> > It does not run on more platforms either, vulkan requires much more
> > complex hardware than gl/gles.
> 
> Actually, Vulkan drivers are less complex than GL(ES) ones (or at least
> that was the intention when the API was designed). It's possible that
> Vulkan effectively requires recent hardware currently, but that's
> because Vulkan was designed for it. E.g. current hardware was used to
> determine the minimum capabilities a Vulkan driver should support. Also
> nobody is going to bother writing Vulkan drivers for old hardware.
> 
> OpenGL(ES) drivers on the other hand are a true nightmare of
> complexity, which can be seen on the numerous bugs implementations have.
> 
> Vulkan is still new, but there's no reason why it wouldn't be available
> on _more_ platform when the time comes. Vendors will get bored of
> writing complex and slow GL drivers.
> 
> Vulkan helps media players too, because media players need to deal with
> large quantities of data (video frames). Vulkan's manual memory
> management allows you to make more efficient choices and to avoid slow,
> emulated code paths.

Why you bother with obvious trolls? Don't feed them!

-- 
Sylvain


More information about the MPlayer-dev-eng mailing list