Still Tearing in Watchout / Sync Cards

[DavidPatrick]

This is a multi-part question:

 

I occasionally get tearing on stills in Watchout. It doesn't seem to be consistent, and I can't quite nail down why. It's almost always big stills...8k, 9k wide. I've tried different file formats and nothing seems to be the silver bullet.

 

So the first part of the question is, what part of the computer does the heavy lifting of fading a still? Is that the GPU or CPU? And would additional system RAM or video ram help that?

 

The second and related question is using sync cards between machines. We have traditionally used AMD cards with the s400 sync modules and had good success. However, we have been moving to nVidia Quadro cards (M4000 and P4000) as they benchmark at better performance and have much more flexibility in the software. We have the appropriate Sync I and Sync II modules. 

 

In my current scenario with two display machines using these nVidia cards and sync modules, I have set them in various configurations: to be external sync on both cards, external sync on one card feeding sync the second, and internal sync on one feeding the second. In all cases, I still intermittently get tearing between machines. It doesn't seem to happen on the videos, but does on stills. Curious what's happening on the back end which could cause this. 

 

And finally, can someone elaborate on what the "sync chain master" checkbox does under the hood? Does Watchout actually talk to both the s400 and Sync I / II modules? If you are using external sync to each module, does this checkbox still matter, as there's not really a "master"? I have also found when I make an adjustment on one display machine (say re-launch Watchout), sync struggles until I re-launch on the second machine. Is this expected behavior?

 

Just trying to understand the technology better so I can fix things quicker in the middle of the night on a show! 

 

Thanks!

[Mike Fahl]

You're not saying whether tearing is within a single display or across displays. Hetre I'm going to assume within a display.

 

The only reason I can think of for tearing in stills within a single display is that vsync is disabled in the graphics card driver, or that the display device does something stupid. 

 

I've tried different file formats

 

File formats shouldn't make a difference, since all images are decoded before you run the show, and cached in an internal format, which is independent of the original image file format.

 

So the first part of the question is, what part of the computer does the heavy lifting of fading a still? Is that the GPU or CPU?

 

GPU.

 

And would additional system RAM or video ram help that?

 

Probably not.

 

We have the appropriate Sync I and Sync II modules.

 

This could indicate you're talking about tearing across display outputs. In which case it more comes down to the behavior of those ync cards, that can sometiumes b rather finicky.

 

I still intermittently get tearing between machines

 

OK, so you're likely talking tearing between computers. Some of my answers above may still help, though.

 

And finally, can someone elaborate on what the "sync chain master" checkbox does under the hood? 

 

You should set up one of the cards (the first in the sync chain) to act as the master. The remaining ones are slaved to this master. The master may get external sync if desired, but doesn't have to. Others should not need external sync, but will be synced from the master. Check the "sync chain master" checkbox for the display that acts as the sync master in the chain. I believe this setup is rather important for successful operation.

 

Hope this helps. It's been a while for me, but I believe my recollection here is correct.

 

Mike

[DavidPatrick]

Thanks Mike. Sorry for my lack of detail. Good info to know.

 

We have had lots of other issues on this show, so I can't entirely isolate the cause. However, after some reworking of the sync signal and EDIDs on the graphics card, this seems to be resolved. I'm going to chalk this one up to weird refresh rates in the chain.