Hi Burkhard,
You suggest multi-core brings highest gain.
I read in the blog on the parallel implementations of the video routines. So
we need to integrate these routines with the filters
we want to apply to the video? Right?
On Tue, Feb 15, 2011 at 10:57 PM, Burkhard Plaum <pl...@ipf.uni-stuttgart.de
> wrote:
> Hi,
>
> Am 15.02.2011 15:23, schrieb Vishnu Viswanath:
> > So slicing frames is not an option.
> > More on sending frame by frame?? any articles or resources to read?
>
> Hmm, no idea. I don't even think that there is a standardized
> way of doing that.
>
> The bottleneck is to send the uncompressed frames over a network. So the
> filtering has to be VERY CPU intensive until you gain something here. And
> also Gigabit ethernet is the minimum. Another issue might be, that the
> rendered frames can arrive out-of-order.
>
> I was thinking to implement a method to send video frames over pipes and
> sockets. With this, gmerlin applications can easily exchange A/V data.
> But it should be optimized (i.e. if sender and receiver are on the same
> machine,
> shared memory should be used instead etc.). I didn't have a serious need
> for that yet.
>
> I really think, that concentrating on multi-core machines brings the
> highest
> gain.
>
> > @Burkhard what did you mean by working scene based. i dint understand?
> > Thank You,
>
> This means to split the movie into independent scenes and send the scenes
> to
> the nodes.
>
> Burkhard
>
>
> ------------------------------------------------------------------------------
> The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE:
> Pinpoint memory and threading errors before they happen.
> Find and fix more than 250 security defects in the development cycle.
> Locate bottlenecks in serial and parallel code that limit performance.
> http://p.sf.net/sfu/intel-dev2devfeb
> _______________________________________________
> Gmerlin-general mailing list
> Gmerlin-general@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/gmerlin-general
>
--
Vishnu Viswanath
Excel 2010 Marketing In-charge
Computer Science & Engineering 2007-2011
Model Engineering College, Cochin
------------------------------------------------------------------------------
The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE:
Pinpoint memory and threading errors before they happen.
Find and fix more than 250 security defects in the development cycle.
Locate bottlenecks in serial and parallel code that limit performance.
http://p.sf.net/sfu/intel-dev2devfeb
_______________________________________________
Gmerlin-general mailing list
Gmerlin-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/gmerlin-general
