TTS that sounds less fake on 4090

[vectorfield]

Spent some time on this today and wanted to compare notes.

I'm comparing two approaches here: keep it simple on B200, or squeeze more out of the stack and deal with the complexity.

What would you pick if you had to live with it for a few months?

[tensorworks]

I'd approach it in a boring order. If you're on PyTorch, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.

[deltagrid]

Appreciate the detail. I ended up in about the same place using SGLang. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[vramarch]

Tried to reproduce it just now. The issue went away once I changed one setting and restarted the stack cleanly. The surprise was that memory pressure showed up before throughput collapsed.

[deltagrid]

I'd start with the simpler stack.

[macropoint]

A couple questions before I guess too hard. Do you see the same thing with a smaller batch or shorter prompt?

[visionforge]

What exact setup are you running? Are you testing this on GB200 or something else?

[tensorworks]

What exact setup are you running? Are you testing this on GB200 or something else?

My first guess would be this: On A6000, 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.