vector DB that stayed out of the way

[macronest]

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

I'm comparing two approaches here: keep it simple on dual 3090, or squeeze more out of the stack and deal with the complexity.

Trying to avoid rebuilding the whole thing twice.

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.

[bytesupply76]

I don't think that's the whole story. I get why that seems right, but on mini PC the bottleneck felt elsewhere. Could be a workload difference, but I wouldn't generalize too fast from one setup.

[routecraft]

That matches what I was seeing. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[oxidegrove40]

same here on my side.

[marbleyard32]

A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at NVFP4. Not glamorous advice, but repeatable baselines save a lot of time.

I'm happy to test one more variant if you want a second data point.

[papermixer4]

I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with DeepSeek and compare from there.

[benchplease]

I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with DeepSeek and compare from there.

I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with Mistral and compare from there.

[macrolab]

Appreciate the detail. The pattern around quant drift has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[minthall]

Appreciate the detail. I don't think your conclusion is far off.

[rasterhall]

A couple questions before I guess too hard. Are you testing this on dual 3090 or something else?