+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Prompting & Context Engineering (https://www.datasculpting.ai/inference/forumdisplay.php?fid=15)
+--- Forum: Long Context / Memory (https://www.datasculpting.ai/inference/forumdisplay.php?fid=64)
+--- Thread: long context that stayed useful: anyone testing this? (/showthread.php?tid=1142)
Pages:
1
2
long context that stayed useful: anyone testing this? - nanobrook - 10-13-2026
This keeps coming up in conversation so I figured I'd start a thread.
The more I work with local setups, the more it seems like continuous batching decides everything downstream.
Wondering if this is just where the whole space is heading now.
For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.
Re: long context that stayed useful: anyone testing this? - quietarch - 10-13-2026
subscribed, curious where this lands.
Re: long context that stayed useful: anyone testing this? - tinycontext - 10-14-2026
Quote:subscribed, curious where this lands.
That matches what I was seeing. That matches what I saw on Threadripper box. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.
I'm happy to test one more variant if you want a second data point.
Re: long context that stayed useful: anyone testing this? - torchkite - 10-14-2026
I tested this tonight. I couldn't reproduce the slowdown until I pushed context much higher. The surprise was that memory pressure showed up before throughput collapsed.
Re: long context that stayed useful: anyone testing this? - ggufvault - 10-15-2026
Maybe, but I had the opposite result. I get why that seems right, but on B200 the bottleneck felt elsewhere. Could be a workload difference, but I wouldn't generalize too fast from one setup.
Re: long context that stayed useful: anyone testing this? - fusedhall - 10-16-2026
A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with FLUX and compare from there. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it. Turns out 'just one more test' was a lie.
Re: long context that stayed useful: anyone testing this? - emberquill - 10-17-2026
Maybe, but I had the opposite result. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model.
Re: long context that stayed useful: anyone testing this? - tinycontext - 10-17-2026
Did a small comparison this afternoon. On 5090 with PyTorch, I saw basically the same behavior. The main thing that moved was latency, not quality.
Re: long context that stayed useful: anyone testing this? - turboyard94 - 10-18-2026
That matches what I was seeing. I don't think your conclusion is far off. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.
Re: long context that stayed useful: anyone testing this? - quantghost - 10-18-2026
What worked for me was pretty simple. On 4090, 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.