From: Krijnen, R. (Robbert) (robbert.krijnen++at++tno.nl)
Date: 02/28/2006 07:31:17
pfHi,
We finished testing the FX4500 with G-sync on Linux cards last week.Previously we tested with the FX3000G cards but they didn't work correctly.
On the FX4500 both genlocking and frame locking worked when using the JoinSwapGroupNV and BindSwapBarrierNV calls. The only problem we encountered is that after logging into the system we had to reconfigure the genlocking setup in the nvidia-settings panel. After that the test applications (generating different loads) work perfectly. Depending on the application which uses the most time the other application waits for the slowest.We tested both 81.74 and 81.78 driver versions. Older driver versions are broken.We also set both monitors to the same freq using a modeline.
Regards,
Robbert
-----Original Message-----From: owner-info-performer++at++performer.engr.sgi.com [mailto:owner-info-performer++at++performer.engr.sgi.com] On Behalf Of Hugues De KeyzerSent: dinsdag 28 februari 2006 9:38To: 윤재문Cc: info-performer++at++sgi.comSubject: [info-performer] Re: Issues about NVidia Frame-Locking
Thank you for the information about your experience. We finally figured out the problem. Actually, frame lock is NOT supported on Linux (although nVIDIA claims the opposite). Actually, some things work and others are broken, what makes the whole system useless until nVIDIA releases drivers with correct support.
Status for drivers 81.78 (2005-12-16):
What works: What nVIDIA calls "Frame Lock" (synchronization of the vertical retrace of several systems' displays): works, but changes the timings of the displays (hsync and vsync start, end and total), what makes it unusable on systems where timings need to be fixed, like domes where the picture is resampled to be distorted.
What doesn't work: What nVIDIA calls "Swap Sync" (synchronization of the buffer swaps of several systems): is broken. Some frames happen to swap concurrently sometimes, but that is all. No way to get this working (all displays have to wait for all the others to be ready and swap together, thus the whole group is limited by the slowest one for each frame).
What nVIDIA calls "Frame Sync" or "Genlock" (synchronization of a frame locked display group to an external house sync): not tested but seems to be working according to others experiences.
To have a set of displays working as one big display, swap sync is mandatory. Frame lock without swap sync is pointless, but frame lock is mandatory to have swap sync.
So, we're all waiting for nVIDIA, who doesn't seem to care at all (even if you buy 10 Quadro 4500G cards :)). Even the frame lock user manual is hidden somewhere on their registered developers restricted website, and explains only the Windows part (which is actually very similar to the Linux implementation).
We received several comments from Windows users, and the Windows implementation seems to be working perfectly indeed.
Thank you again for sharing your experience.
Regards,
Hugues De Keyzerde pinxi
On Thursday 23 February 2006 07:06, you wrote:> Hi. a remote Performer...> > I had a similar experience about implementing frame synchronization,> through NVidia graphic interface cards and OGL Performer.> > But, in my case, there are several differcences as belows.> 1) I implemented that in MS Windows environments, not GNU/LINUX.> 2) I used NVidia Quadro 3000G, not 4500G like you.> > I also used "wglJoinSwapGroupNV" and "wglBindSwpBarrierNV" for swap locking,> and got a successful result.> (Three channels have the same frame-rate, and frames are synchronized.)> > I refered the following articles.> 1) Genlock/Framelock and SDI (Ian Williams, NVidia)> 2) Frame Synchronization User's Guide (NVidia, Nov. 5, 2003)> > and also got a full source files from NVidia developer> 1) framelock_demo.h and framelock_demo.cpp> But, these files are for MS Windows-based development environments.> > Maybe you can get the same files of GNU/LINUX version from NVidia.> > I hope that this mail helps you, or you had solved the prob
lem already.> > Bye.>
----------------------------------------------------------------------- List Archives, Info, FAQ: http://www.sgi.com/software/performer/ Submissions: info-performer++at++sgi.com Admin. requests: info-performer-request++at++sgi.com-----------------------------------------------------------------------
This e-mail and its contents are subject to the DISCLAIMER at http://www.tno.nl/disclaimer/email.html
This archive was generated by hypermail 2b29 : Tue Feb 28 2006 - 07:59:43 PST