They should work nicely together! RTX code runs fast on Pascal cards as well
Most importantly you can use the 1070Ti for your monitors and viewport performance and leave the 2080 only for GPU rendering
You will notice a difference in performance, VRAM usage and system interactivity. You should install the 1070ti in the first PCIe slot and connect your monitors to it.
And for final rendering you can use both cards for rendering, a frame per each card
Another 2080 Super is a better solution in future, could also use the 3 cards together in your setup if possible.
Yeah, that is what most people do..say if you have 4 cards, it is recommended to render a frame per each card…instead of having the 4 cards work on the same frame
Because in many cases the GPUs will be idling waiting for the scene to be loaded to the VRAM
you could do that with deadline or this tool for example https://vrscenegui.babylondreams.de/
Other render engines like RS or Cycles doesn’t support DR rendering of the same Frame on mutiple GPUs, It always has to be a frame per GPU(Vray is better obviosely giving users the option to choose)
so you won’t hear much about this issue
Wow I’ve never heard of it neither thought about it! It seems so logical Hmmm I assume it would also work with DR in case having 4 x GPU splitted into two different machines. Thanks so much for the links, everyday one can learn something new
I haven’t used DR for GPU rendering at all, for my old employer we had 2 machines with 11 GPUs each
For animation and in many cases, rendering of a single frame was usually faster than the amount of time Vray takes to load the scene and textures to VRAM
That is why you need to render a frame per GPU simultaneously, you don’t want your GPUs idling waiting for a frame to load into VRAM
this is also possible with commend line and Vray standalone, but easier with the tools I listed above.
At V-Ray for Rhino the animation tool Bongo has an useful function - if a frame rendering is started a small place holder file is saved to the disk. So, an other task know that this frame number is in work and the next frame can be started. This allow me to start two or three Rhino tasks and parallel rendering of a few frames. So, I get a nearly 100% GPU usage over the whole time. It’s a great time saver.
Let us take this as inspiration - it could be great if V-Ray would allow to load the next frame scene to the GPU during the current frame is calculated.
General we have wasted time for loading the scene and for saving the output. Saving the PNG and all channels needs a few seconds if the output is quite large like full HD. I have seen that the saving time is the same like the render time. Here a lot of time could be saved. Maybe after an image is calculated the data could be moved to CPU and RAM for the slow saving process.