Hi Marcus,
On Mon, 2006-02-13 at 15:11 +0100, Marcus Lindblom wrote:
> (I thought I had posted about this a few months back, but I can't
> remember getting a reply nor could find the post now that I searched for
> it. )
I don't remember sending one.
> I noticed that RenderAction always calls glFinish(), on each frame, in
> order to collect profiling data. While that is ok, I would really avoid
> doing that (i.e. waiting for all gl-commands to finish before preparing
> the next frame). (I've also read that it is not recommended by
> hw-manufacturers, since it destroys a lot of parallellism between cpu
> and gpu, which makes good sense.)
In general yes, they don't recommend it.
> I would expect it to not matter much for GPU-limited apps, but if you're
> pushing the system and want to run with as few threads as possible, then
> it could be hampering performance seriously.
Possible.
> I have compiled my own version of the OpenSG libs with that call
> commented out, but I just wanted to bring this to attention in case
> there are others with similar problems.
Have you seen any improvements? How much difference did it make?
The flexibility of OpenSG (any number of Viewports, any number of
Windows, all with separate Statistics, and maybe with shared
RenderActions) makes the usual approach of collecting multiple frames
and dividing infeasible.
Yours
Dirk
-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems? Stop! Download the new AJAX search engine that makes
searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users
