10-12-2026, 04:11 PM
worth checking the driver first.
queue depth watcher
|
MCP server discovery mess on GB200
|
|
10-13-2026, 04:51 PM
Quote:worth checking the driver first. My first guess would be this: Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at batch size. Not glamorous advice, but repeatable baselines save a lot of time. Turns out 'just one more test' was a lie.
10-14-2026, 04:54 PM
That matches what I was seeing. I don't think your conclusion is far off. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.
10-14-2026, 06:07 PM
Interesting, I hadn't tried that. I don't think your conclusion is far off.
10-17-2026, 06:40 PM
Appreciate the detail. I ended up in about the same place using MLX. The biggest change for me was simplifying the stack instead of adding one more optimization layer.
10-18-2026, 10:41 PM
A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Qwen and compare from there.
learning in public
|
|
« Next Oldest | Next Newest »
|