I would love to hear more about what hardware and software solution you have or would use for a project like described below.
I’m working on a project for which we are looking for a solution to render four UHD (3840 x 2160) videos onto 12 monitors with a resolution of (1920 x 1080). After some discussion and research the most obvious solution would be to use 3 Quadro GPUs and a syncer as described on the scalable visualisations solutions page (see below). Development wise this is probably the most straightforward solution as we can use Mosaic to create one big desktop. I’m only not sure how this combined desktop would be used in software. E.g. would I simply create an OpenGL context and would OpenGL see one GPU instead of 3?
Another solution would be to use just one Quadro P4000 which according to the spec can do 4 * (4096 x 2160) @ 120fps. And then use a video splitter to split up the output to 12 * (1920 x 1080).
For both solutions I want to use the Video Codec SDK to decode h264 (we don’t use audio). Last week I did a test with a GTX 680 and the benchmark showed the same decoding speeds as nvidia notes in their application note. The pascal architecture is capable of decoding at 658fps (for 1920 x 1080) which looks promising, but of course I need to benchmark this.
What are your thoughts about the hardware and software for a project like this?
P4000 spec: https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/documents/Quadro-P4000-US-03Feb17.pdf
Application Note: http://studylib.net/doc/18850290/application-note---nvidia-developer