https://www.datasculpting.ai/inference/ Mon, 26 Oct 2026 11:30:27 +0000 MyBB https://www.datasculpting.ai/inference/showthread.php?tid=1462 Sun, 18 Oct 2026 20:44:00 +0000 paperrunner]]> https://www.datasculpting.ai/inference/showthread.php?tid=1462 https://www.datasculpting.ai/inference/showthread.php?tid=1430 Sat, 17 Oct 2026 15:58:41 +0000 cudaline]]> https://www.datasculpting.ai/inference/showthread.php?tid=1430
Short version: LM Studio behaved well for me, but only after I stopped ignoring workstation thermals.

Happy to post more details if anyone wants them.]]>
Short version: LM Studio behaved well for me, but only after I stopped ignoring workstation thermals.

Happy to post more details if anyone wants them.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1426 Sat, 17 Oct 2026 08:12:05 +0000 benchplease]]> https://www.datasculpting.ai/inference/showthread.php?tid=1426
I've gone back and forth between PyTorch and PyTorch for Qwen. The interesting difference for me has been speculative decoding.

Trying to avoid rebuilding the whole thing twice.]]>
I've gone back and forth between PyTorch and PyTorch for Qwen. The interesting difference for me has been speculative decoding.

Trying to avoid rebuilding the whole thing twice.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1156 Wed, 14 Oct 2026 02:56:53 +0000 vramscope]]> https://www.datasculpting.ai/inference/showthread.php?tid=1156
Two paths both seem viable, but they break down differently once token cost becomes important.

What would you pick if you had to live with it for a few months?]]>
Two paths both seem viable, but they break down differently once token cost becomes important.

What would you pick if you had to live with it for a few months?]]> https://www.datasculpting.ai/inference/showthread.php?tid=1190 Tue, 13 Oct 2026 18:56:08 +0000 llmarch51]]> https://www.datasculpting.ai/inference/showthread.php?tid=1190
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling VRAM headroom.

If you've tested something similar, what did you keep and what did you stop bothering with?]]>
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling VRAM headroom.

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=1136 Tue, 13 Oct 2026 04:23:12 +0000 cudaline]]> https://www.datasculpting.ai/inference/showthread.php?tid=1136
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling reranking.

If you've tested something similar, what did you keep and what did you stop bothering with?]]>
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling reranking.

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=1065 Mon, 12 Oct 2026 05:23:44 +0000 cachebench]]> https://www.datasculpting.ai/inference/showthread.php?tid=1065
I've gone back and forth between TensorRT-LLM and vLLM for Kimi. The interesting difference for me has been KV cache pressure.

Interested in tradeoffs more than fan loyalty.]]>
I've gone back and forth between TensorRT-LLM and vLLM for Kimi. The interesting difference for me has been KV cache pressure.

Interested in tradeoffs more than fan loyalty.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1050 Mon, 12 Oct 2026 04:01:27 +0000 benchplease]]> https://www.datasculpting.ai/inference/showthread.php?tid=1050
I'm mostly working with DeepSeek in SGLang on Threadripper box. The part I'm trying to understand is prefill/decode split.

Would appreciate practical advice rather than theory.]]>
I'm mostly working with DeepSeek in SGLang on Threadripper box. The part I'm trying to understand is prefill/decode split.

Would appreciate practical advice rather than theory.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1056 Mon, 12 Oct 2026 02:02:46 +0000 phasestream53]]> https://www.datasculpting.ai/inference/showthread.php?tid=1056
After a few runs on A6000, I ended up with decent performance, but the setup only really made sense once I adjusted around KV cache pressure.

Happy to post more details if anyone wants them.

I care more about stability than squeezing the last possible tokens/sec out of it.]]>
After a few runs on A6000, I ended up with decent performance, but the setup only really made sense once I adjusted around KV cache pressure.

Happy to post more details if anyone wants them.

I care more about stability than squeezing the last possible tokens/sec out of it.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1084 Sun, 11 Oct 2026 23:15:59 +0000 cacheworks]]> https://www.datasculpting.ai/inference/showthread.php?tid=1084
If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

I'm especially interested in mistakes worth avoiding.

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?

I'm especially interested in mistakes worth avoiding.

My use case is pretty ordinary: local chat, some document work, and occasional batch jobs.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1008 Sun, 11 Oct 2026 01:50:22 +0000 southden]]> https://www.datasculpting.ai/inference/showthread.php?tid=1008
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?

This is mostly for local work rather than public serving.]]>
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?

This is mostly for local work rather than public serving.]]> https://www.datasculpting.ai/inference/showthread.php?tid=1014 Sat, 10 Oct 2026 23:05:40 +0000 plainline]]> https://www.datasculpting.ai/inference/showthread.php?tid=1014
Maybe useful as a starting point: vLLM on 4090, keeping speculative decoding in mind from the start.

Feel free to tear it apart if you see something dumb.

I care more about stability than squeezing the last possible tokens/sec out of it.]]>
Maybe useful as a starting point: vLLM on 4090, keeping speculative decoding in mind from the start.

Feel free to tear it apart if you see something dumb.

I care more about stability than squeezing the last possible tokens/sec out of it.]]> https://www.datasculpting.ai/inference/showthread.php?tid=912 Fri, 09 Oct 2026 03:20:34 +0000 vrambandit]]> https://www.datasculpting.ai/inference/showthread.php?tid=912
I'm comparing two approaches here: keep it simple on A6000, 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 A6000, 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=813 Tue, 06 Oct 2026 15:46:52 +0000 rocmriley]]> https://www.datasculpting.ai/inference/showthread.php?tid=813
I'm helping a small team set up a local workflow. We don't need a giant stack, but we do care about tokens/sec.

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 tokens/sec.

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=773 Mon, 05 Oct 2026 22:59:30 +0000 llmarch51]]> https://www.datasculpting.ai/inference/showthread.php?tid=773
After a few runs on 5090, I ended up with decent performance, but the setup only really made sense once I adjusted around embeddings.

Happy to post more details if anyone wants them.]]>
After a few runs on 5090, I ended up with decent performance, but the setup only really made sense once I adjusted around embeddings.

Happy to post more details if anyone wants them.]]>