10-18-2026, 10:38 PM
Quote:Tried to reproduce it just now. On 7900 XTX with Ollama, I saw basically the same behavior. The main thing that moved was latency, not quality.
What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at continuous batching. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.
still benchmarking everything