NVIDIA NVS 810 for new system

[massta]

Correct, mesh warping is for slices. If you look at my image of the Datapath vs Individual GPU Outputs, the Advanced Output of the Datapath has connecting boundaries because Resolume thinks its one display. If you slice and warp each of the quadrants, you can not pull your mesh across these boarders because the video will bleed into undesirable quadrants. This can't happen in the Individual GPU Outputs. I suppose you would have to add a mask if this happens.

[elgarf]

I've never had problems with keeping slices into datapath quadrant boundaries.
If slice is out of boundary then you're doing something wrong, but even in this case you can just add mask.

[massta]

That sounds like a solution then. I'll have to give it another go since I have an Fx4 here. Thanks for the advice.

[massta]

About to finally give this ago go tonight.

I'm still not sure masking will solve this issue because the masks would need to be only acting on individual slices and not the slices stacked below it.

[massta]

So I was correct. When trying to use every last pixel of a Datapath FX4, you are limited by the internal boundaries of the quad zones. If you are video mapping and stretch over into this area, you will need to be careful of the other slices/quadrants.

I believe the way out of this is to scale the image down so it stays within its quadrant after editing points for mapping. But then you need to scale the image at the projector using the zoom. Doing this you lose pixel density.

[lightbx]

If you need your content to reach that large of a scale for mapping purposes, then what you should actually be doing is downscaling the slice in the Input Selection, not upscaling the slice in Output Transformation. This will accomplish the same visual effect you are looking for, without having extend the slice outside of its quadrant in a FX4 / AJA HA5-4K / etc.

[massta]

Yes, that is what to do. But I think you meant to scale the layer/clip first, no? That way you can grab extra area around your content so when looking at the output tab, there will be more room to stretch/warp the mapping points.

But when grabbing exact slices from a large video to be mapped using three projectors with edge blend, it becomes difficult. That was my issue when posting. I believe the answer is to scale down the content inside the layer/composition and overshoot with the projector. Simply because you can't have a slice over another slice. You need room to keep them separated. No?

[isprod.se]

The problem with multiple GPU is not stability. The problem is getting it to work in the first place.

With the GTX980 or 1080 and NVS810 in particular, the problem is that the computer won't recognise the GTX when both are installed. Nvidia considers using these two cards together 'undefined behaviour'. So even Nvidia's own Control Panel won't show it as an available GPU. It will just show the NVS810. The result is that Resolume will start using the NVS810 as its main renderer, which does not pack enough punch to run 8 HD outputs.

The only way to make this work, is to first disable all the NVS810 outputs via Windows Display manager. Then the GTX becomes available, and you can start Resolume with the GTX as the main renderer. Then you can re-enable the outputs again via Display manager, and have all 8 outputs of the NVS810 available, plus the outputs of the GTX.

We haven't done any real performance tests, because we don't really recommend this setup. It seems to run okay enough, but it's just too much hassle to get it to work. Every time you want to start Resolume, you will have to go through this whole process of disabling and re-enabling. I don't wish it upon my worst enemies to have to do this during the stress of a mid show power failure.

Joris, old thread but hard to find answers so hope you might be able considering this old comment.
I'm using res7 as a media server.
X299 10980xe with rtx a5000, a have 24 sdi bi-dir but need more syncable outputs. (Datapath is already considered. This would be more cost, as it would require 2 units to give 8 more outputs. And even then an other gpu would be needed as 4 outs are in use from the rtx a.

Considering your old answer, is this above mentioned process still needed?
I was considering to add a nvs810 with rtx a (quadro card), would this not work out of the box using my rtx a as main/active gpu?

[Zoltán]

Hard to say. I don't think driver compatibility has improved.
If you have the card already, just try it

[isprod.se]

Hard to say. I don't think driver compatibility has improved.
If you have the card already, just try it

I dont want to spend money if it dont work :-/
But I'll be getting another rtx a next week, I will try with them both in 1 machine and write my findings.