Hey Hive Mind,
Working on engineering a hefty install in a very large commercial bar/club and would love your input and criticism. Got a lot to unpack here so feel free to roast my inexperience with this scale.
The led is all 1.9mm. The processor is an h15 with 8 output cards mapping 9 individual screens. One of these 9 screens is a super ultra wide wrap around perimeter that will take 3 4k inputs laced to not scale terribly. May also need to stitch 2 inputs on one more of these screens to keep it also from scaling too much. Also sending a 4k back into the network to be routed to TVs from resolume. Two of these screens are under the 1080p space. The remaining screens fit within a 4k space. So we’re looking at 10 4k outs and 2 1080p outs, sliced out and top left justified to make this happen. Total pixel push is around 37 million.
Their in house techs and myself are comfortable with resolume and prefer it. Is the best way to approach this a very hefty graphic card with three 8k decklink g2? I’m talking like Xeon processor and an RTX 6000 Ada with as much ram as needed and a motherboard with plenty of the correct pcie professionally built. Or am I better splitting this project across a couple servers? A lot of the time these screens will be receiving the same 4k content just sliced out differently. I’ve read it’s not ideal to use multiple gpus as the rendering only happens on the main, but if I were to put three gpus in a machine instead of the decklinks, will the two additional gpus not function kind of like a decklink creating extra outputs with more edid options? Would I lose resolume integrity this way over a decklink setup?
The RTX 5090 seems to benchmark higher than the 6000 Ada, but has 16gb less vram, which would you go with for a single gpu multi decklinks setup? 5090 is considerably cheaper but that’s the least important factor. The 5090 does not have sync capability in a multi gpu setup where Ada does. A6000 or 5000 Ada a contender in your mind?
If I’m triggering clips from one instance and routing them in advanced output as needed, is a black magic sync generator sufficient to avoid tearing if tied to all the decklink cards and the led processor with the decklinks and sync generator properly setup in their interfaces?
Sorry for the long post. Thanks in advance to any for their insight!
