• 0 Posts
  • 19 Comments
Joined 1 year ago
cake
Cake day: June 1st, 2023

help-circle









  • 7900 XTX recently got support for Stable Diffusion and LLM, on paper, it’s faster than 4090 RTX for FP16 computation, it does seem faster judging my experience using rented 4090 RTX on Runpod and my 7900 XTX GPU. 14 seconds (4090 RTX) vs 6 seconds (7900 XTX.)

    7900 XTX is an option if you want $1000 cheaper than 4090 RTX and have similar sized VRAM and having comparable performance to that of 4090 RTX.





  • Servers that run the MMO with dynamic persistence would likely need GPUs for keeping track and computing vectors/matrices for each player/asset.

    I always theorized that if Eve Online developers actually bothered, they could utilize their GPU on the servers to track all of the activity in space and they would easily support 10,000+ players on the same system without issue or time dilation (considering the capability of GPU today.)

    Assuming you have for an example, a GPU Server with 8 slots for GPU, and each GPU is a 7900 XTX and you use Vulkan Compute for it. You got 192 GB of vRAM and absurd amount of computation capability with ~1 PFlops on FP16 and 528 TFlops on FP32. Assuming you have 100,000 players on the same system, that GPU Server would be able to handle it with ease even with theoretical complication that can results from programming on GPU. You could store ~2MB worth of data about each player’s ship for 100,000 players on that GPU Server alone.

    Dynamic Persistence in MMO is basically Server with GPU that keep track of everything. (You can’t just defer GPU operations to the end-user, server have to be the one to do it.)