10-12-2026, 04:47 AM
Good point. That matches what I saw on 4090. The biggest change for me was simplifying the stack instead of adding one more optimization layer. The logs were somehow calmer than I was.
|
70B still worth the pain with TensorRT-LLM
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
70B still worth the pain with TensorRT-LLM - by oxiderunner - 10-10-2026, 11:26 PM
Re: 70B still worth the pain with TensorRT-LLM - by kernelmike - 10-11-2026, 04:07 AM
Re: 70B still worth the pain with TensorRT-LLM - by cobalttinker - 10-11-2026, 04:59 AM
Re: 70B still worth the pain with TensorRT-LLM - by audionode8 - 10-11-2026, 07:23 AM
Re: 70B still worth the pain with TensorRT-LLM - by edgeriver - 10-11-2026, 03:49 PM
Re: 70B still worth the pain with TensorRT-LLM - by orbitbench6 - 10-12-2026, 04:32 AM
Re: 70B still worth the pain with TensorRT-LLM - by latentcraft - 10-12-2026, 04:47 AM
Re: 70B still worth the pain with TensorRT-LLM - by kernelmike - 10-13-2026, 05:21 PM
Re: 70B still worth the pain with TensorRT-LLM - by cudaline - 10-13-2026, 07:24 PM
Re: 70B still worth the pain with TensorRT-LLM - by fastbench - 10-13-2026, 08:41 PM
|
Users browsing this thread: 1 Guest(s)