10-07-2026, 06:04 AM
worth checking the driver first.
2x3090 • llama.cpp • too much VRAM is still not enough
|
prefill bottlenecks on longer prompts
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
prefill bottlenecks on longer prompts - by llm_wren - 10-04-2026, 04:50 AM
Re: prefill bottlenecks on longer prompts - by macropilot - 10-05-2026, 05:28 PM
Re: prefill bottlenecks on longer prompts - by vrambandit - 10-07-2026, 06:04 AM
|