tools worth paying for

[tensorriver]

A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Llama and compare from there.

[modbench]

What worked for me was pretty simple. On 5090, the easiest win is often to back off the aggressive settings and see whether the problem survives. Once you know the plain path works, then add the faster path back in piece by piece.

[audiodesk]

worth checking the driver first.

[deepseeked]

Not sure I agree there. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model. Turns out 'just one more test' was a lie.

[promptfield]

Good point. The pattern around HBM constraints has been pretty similar for me. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[latentgarden]

I don't think that's the whole story. I've seen people blame small local stacks when the real issue was much more mundane. Not saying you're wrong, just that I wouldn't lock that in as the lesson yet.