Re: gEDA-user: PCB+GL Testers (please test)

[Peter Clifton <pcjc2@xxxxxxxxx>]

On Thu, 2010-11-04 at 16:43 +0100, Kai-Martin Knaak wrote:
> Peter Clifton wrote:
> 
> >> Using VBOs that gives 85.4FPS, so effectively no difference.
> >> Using arrays it gives an average of 93.1FPS, literally no difference!
> > 
> > Thanks for trying it. I'm embarassed to say your board manages about
> > 9fps on my machine, with every trick I can throw at it!
> 
> I tried Richards power too and got the same result as Peter: No more than 11 
> FPS, even with my new ATI card driven by the closed source fglrx. :-|
> Just like my own board, there is no much divfference compared to the before 
> pours branch.
> What makes his set-up faster by an order of magnitude? 


Shinier more expensive pixels?

Not sure.

Anyway, seems my VBO code is useless and slow compared to vertex arrays,
so you might like to try with "local=true".

I've got some changes stacked up here which I might push soon, which
split this into two different variables:

buffer->use_vbo = (true or false);
buffer->use_map = (true or false, MUST be false if use_vbo is false).

Having tried out various re-arrangements of code, my own system is
ALWAYS faster with use_vbo and use_map both false. Otherwise the system
spends a huge quantity of its time copying data between GPU and CPU
domains, or invalidating huge areas of cached ram in the GPU / CPU. Not
the desired effect.

I'm beginning to think VBOs only become a real win for more static
geometry. There seems to be a fair bit of overhead for the driver to
manage them otherwise. I still can't imagine why it would _ever_ be
slower than the simple arrays case, but something seems to be amiss.

I've got so lost as to what works well / works badly that I might have
to go back to basics to figure it all out.

Best wishes,

-- 
Peter Clifton

Electrical Engineering Division,
Engineering Department,
University of Cambridge,
9, JJ Thomson Avenue,
Cambridge
CB3 0FA

Tel: +44 (0)7729 980173 - (No signal in the lab!)
Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me)



_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user