Re: Server segmentation fault at address 0x306 using ParaView

[Jon TURNEY <jon.turney@dronecode.org.uk>]

On 04/01/2011 16:08, Ken Olum wrote:
>    From: Jon TURNEY <jon.turney@dronecode.org.uk>
>    Date: Sun, 02 Jan 2011 18:08:02 +0000
> 
>    I've updated them to fix the problem, and uploaded a Xserver snapshot at [1].
>    [1] ftp://cygwin.com/pub/cygwinx/XWin.20110102-git-29db9091c6ae4995.exe.bz2
> 
> This new snapshot works properly and does not crash.  Thanks very much!

Thanks for testing. I'll make an updated package with this change when I can.

>    I would suggest that for an application like ParaView, you will probably get
>    much better performance if you use the experimental hardware accelerated
>    OpenGL -wgl mode
> 
> I'll try this.  Should I prefer direct rendering from remote machines
> that support that and only use AIGLX when I would have indirect
> rendering anyway, or should I explicitly request indirect rendering in
> order to use AIGLX in all cases?  Thanks again.

There's a trade-off here between rendering performance (software rending vs.
hardware acceleration) and network latency (just sending the image vs. sending
lots of OpenGL commands (some of which might be synchronous, requiring a
round-trip for the response))

So, the short answer is, you should try both and see which performs best :-)

Provided your server isn't on the moon, I would guess that ParaView would
behave better with indirect but accelerated rendering, but that's just
speculation.

I'm not sure of the reasoning for mesa making direct using software rendering
the default for remote clients, I'm not sure exactly what kind of client it
helps with.

Possibly the UG should have some words discussing this issue.

-- 
Jon TURNEY
Volunteer Cygwin/X X Server maintainer

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://x.cygwin.com/docs/
FAQ:                   http://x.cygwin.com/docs/faq/