Inference › Hardware › GPUs v
dual GPU split inference

Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Linear Mode
dual GPU split inference
rocmriley Offline
Member
***
Posts: 211
Threads: 21
Joined: Sep 2026
Reputation: 0
#1
09-10-2026, 03:00 AM
Spent some time on this today and wanted to compare notes.

I've gone back and forth between MLX and vLLM for FLUX. The interesting difference for me has been quant drift.

What would you pick if you had to live with it for a few months?

I care more about stability than squeezing the last possible tokens/sec out of it.
7900 XTX / ROCm / somehow still optimistic
Find
Reply


Messages In This Thread
dual GPU split inference - by rocmriley - 09-10-2026, 03:00 AM
Re: dual GPU split inference - by batchwire - 09-10-2026, 04:26 AM
Re: dual GPU split inference - by minthall - 09-11-2026, 04:44 PM
Re: dual GPU split inference - by turboyard94 - 09-19-2026, 01:30 AM
Re: dual GPU split inference - by batchember - 09-11-2026, 05:28 PM
Re: dual GPU split inference - by cacheline - 09-13-2026, 05:45 AM
Re: dual GPU split inference - by cudapoint - 09-16-2026, 05:50 AM
Re: dual GPU split inference - by signaltrack44 - 09-17-2026, 06:41 AM
Re: dual GPU split inference - by thetarunner - 09-17-2026, 10:46 AM
Re: dual GPU split inference - by sanepeak48 - 09-17-2026, 11:05 AM
Re: dual GPU split inference - by sanevault - 09-17-2026, 12:12 PM
Re: dual GPU split inference - by phaseport36 - 09-17-2026, 12:57 PM

Forum Jump:


Users browsing this thread: 1 Guest(s)