https://www.datasculpting.ai/inference/ Mon, 26 Oct 2026 11:30:24 +0000 MyBB https://www.datasculpting.ai/inference/showthread.php?tid=1410 Sat, 17 Oct 2026 22:21:27 +0000 paperrunner]]> https://www.datasculpting.ai/inference/showthread.php?tid=1410
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling long context.

Would appreciate practical advice rather than theory.]]>
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling long context.

Would appreciate practical advice rather than theory.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1392 Sat, 17 Oct 2026 18:42:08 +0000 southtinker]]> https://www.datasculpting.ai/inference/showthread.php?tid=1392
I'm helping a small team set up a local workflow. We don't need a giant stack, but we do care about tail latency.

I'm especially interested in mistakes worth avoiding.

I care more about stability than squeezing the last possible tokens/sec out of it.]]>
I'm helping a small team set up a local workflow. We don't need a giant stack, but we do care about tail latency.

I'm especially interested in mistakes worth avoiding.

I care more about stability than squeezing the last possible tokens/sec out of it.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1413 Sat, 17 Oct 2026 11:16:03 +0000 cozygrid12]]> https://www.datasculpting.ai/inference/showthread.php?tid=1413
After a few runs on 5090, I ended up with decent performance, but the setup only really made sense once I adjusted around long context.

Happy to post more details if anyone wants them.

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.]]>
After a few runs on 5090, I ended up with decent performance, but the setup only really made sense once I adjusted around long context.

Happy to post more details if anyone wants them.

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1208 Wed, 14 Oct 2026 22:59:21 +0000 benchplease]]> https://www.datasculpting.ai/inference/showthread.php?tid=1208
If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

Simple and boring is a valid answer.]]>
If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

Simple and boring is a valid answer.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1183 Tue, 13 Oct 2026 22:36:16 +0000 nanobrook]]> https://www.datasculpting.ai/inference/showthread.php?tid=1183
I tried this with Phi using vLLM on 7900 XTX. The headline result was better than I expected, but tail latency still ended up being the limiting factor.

Happy to post more details if anyone wants them.

This is mostly for local work rather than public serving.]]>
I tried this with Phi using vLLM on 7900 XTX. The headline result was better than I expected, but tail latency still ended up being the limiting factor.

Happy to post more details if anyone wants them.

This is mostly for local work rather than public serving.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1138 Mon, 12 Oct 2026 23:12:24 +0000 amberplane]]> https://www.datasculpting.ai/inference/showthread.php?tid=1138
I'm comparing two approaches here: keep it simple on 7900 XTX, 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? I keep promising myself this is the last rebuild.]]>
I'm comparing two approaches here: keep it simple on 7900 XTX, 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? I keep promising myself this is the last rebuild.]]> https://www.datasculpting.ai/inference/showthread.php?tid=985 Sun, 11 Oct 2026 00:09:26 +0000 plainworks]]> https://www.datasculpting.ai/inference/showthread.php?tid=985
This isn't revolutionary, but the combination of MLX and Threadripper box felt surprisingly usable once I tuned around MoE serving.

Maybe useful as a starting point for newer folks. The logs were somehow calmer than I was.]]>
This isn't revolutionary, but the combination of MLX and Threadripper box felt surprisingly usable once I tuned around MoE serving.

Maybe useful as a starting point for newer folks. The logs were somehow calmer than I was.]]> https://www.datasculpting.ai/inference/showthread.php?tid=866 Thu, 08 Oct 2026 05:49:24 +0000 vramarch]]> https://www.datasculpting.ai/inference/showthread.php?tid=866
Just sharing a setup that ended up working reasonably well: Mixtral, PyTorch, GB200, with a lot of attention on KV cache pressure.

Maybe useful as a starting point for newer folks.

This is mostly for local work rather than public serving.]]>
Just sharing a setup that ended up working reasonably well: Mixtral, PyTorch, GB200, with a lot of attention on KV cache pressure.

Maybe useful as a starting point for newer folks.

This is mostly for local work rather than public serving.]]> https://www.datasculpting.ai/inference/showthread.php?tid=800 Mon, 05 Oct 2026 15:47:44 +0000 cachebench]]> https://www.datasculpting.ai/inference/showthread.php?tid=800
One thing I keep noticing is that people underestimate reranking until they actually try to run this stuff for more than one afternoon.

Wondering if this is just where the whole space is heading now.]]>
One thing I keep noticing is that people underestimate reranking until they actually try to run this stuff for more than one afternoon.

Wondering if this is just where the whole space is heading now.]]> https://www.datasculpting.ai/inference/showthread.php?tid=751 Sun, 04 Oct 2026 14:17:53 +0000 paperrunner]]> https://www.datasculpting.ai/inference/showthread.php?tid=751
My current setup is vLLM with 5090, and I keep running into embeddings.

If you've tested something similar, what did you keep and what did you stop bothering with?]]>
My current setup is vLLM with 5090, and I keep running into embeddings.

If you've tested something similar, what did you keep and what did you stop bothering with?]]> https://www.datasculpting.ai/inference/showthread.php?tid=668 Fri, 02 Oct 2026 08:56:10 +0000 turbogrid11]]> https://www.datasculpting.ai/inference/showthread.php?tid=668
It feels like the conversation has shifted hard toward inference and serving details lately, and honestly that makes sense once you start dealing with VRAM headroom.

Maybe obvious to some of you, but it clicked for me this week.]]>
It feels like the conversation has shifted hard toward inference and serving details lately, and honestly that makes sense once you start dealing with VRAM headroom.

Maybe obvious to some of you, but it clicked for me this week.]]> https://www.datasculpting.ai/inference/showthread.php?tid=651 Thu, 01 Oct 2026 11:53:25 +0000 pixelcrate82]]> https://www.datasculpting.ai/inference/showthread.php?tid=651
I'm helping a small team set up a local workflow. We don't need a giant stack, but we do care about prefill/decode split.

Simple and boring is a valid answer.]]>
I'm helping a small team set up a local workflow. We don't need a giant stack, but we do care about prefill/decode split.

Simple and boring is a valid answer.]]> https://www.datasculpting.ai/inference/showthread.php?tid=536 Mon, 28 Sep 2026 00:04:57 +0000 vramport]]> https://www.datasculpting.ai/inference/showthread.php?tid=536
I'm comparing two approaches here: keep it simple on 4090, or squeeze more out of the stack and deal with the complexity.

Interested in tradeoffs more than fan loyalty.

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.]]>
I'm comparing two approaches here: keep it simple on 4090, or squeeze more out of the stack and deal with the complexity.

Interested in tradeoffs more than fan loyalty.

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.]]> https://www.datasculpting.ai/inference/showthread.php?tid=485 Fri, 25 Sep 2026 20:07:03 +0000 pixelcrate82]]> https://www.datasculpting.ai/inference/showthread.php?tid=485
If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

Would rather learn from your scars than collect my own all at once.

My use case is pretty ordinary: local chat, some document work, and occasional batch jobs.]]>
If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

Would rather learn from your scars than collect my own all at once.

My use case is pretty ordinary: local chat, some document work, and occasional batch jobs.]]> https://www.datasculpting.ai/inference/showthread.php?tid=443 Thu, 24 Sep 2026 05:20:01 +0000 mangotinker73]]> https://www.datasculpting.ai/inference/showthread.php?tid=443
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling router policy.

Would appreciate practical advice rather than theory.

My use case is pretty ordinary: local chat, some document work, and occasional batch jobs.

The part that surprised me was how much the boring details mattered more than the model choice.]]>
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling router policy.

Would appreciate practical advice rather than theory.

My use case is pretty ordinary: local chat, some document work, and occasional batch jobs.

The part that surprised me was how much the boring details mattered more than the model choice.]]>