Inference › Inference Engines › TensorRT-LLM v
serving on Blackwell

Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Linear Mode
serving on Blackwell
batchcraft Offline
Member
***
Posts: 54
Threads: 8
Joined: Sep 2026
Reputation: 0
#1
10-16-2026, 10:10 PM
Thought I'd post a small recipe because a lot of threads ask similar things.

Maybe useful as a starting point: PyTorch on Threadripper box, keeping prefix caching in mind from the start.

Maybe useful as a starting point for newer folks.

I care more about stability than squeezing the last possible tokens/sec out of it. At this point the real bottleneck is my patience.
Find
Reply


Messages In This Thread
serving on Blackwell - by batchcraft - 10-16-2026, 10:10 PM
Re: serving on Blackwell - by pixelcrate82 - 10-17-2026, 12:44 AM
Re: serving on Blackwell - by rocmstack25 - 10-17-2026, 02:15 AM
Re: serving on Blackwell - by sysstation69 - 10-17-2026, 03:24 AM
Re: serving on Blackwell - by vectorport - 10-18-2026, 10:27 PM
Re: serving on Blackwell - by rasterbench - 10-19-2026, 04:49 AM
Re: serving on Blackwell - by pixelcrate82 - 10-17-2026, 04:01 PM
Re: serving on Blackwell - by thetarunner - 10-19-2026, 01:51 AM
Re: serving on Blackwell - by pixelcrate82 - 10-19-2026, 02:33 AM
Re: serving on Blackwell - by rasterbench - 10-19-2026, 03:47 AM
Re: serving on Blackwell - by visionyard23 - 10-19-2026, 04:23 AM
Re: serving on Blackwell - by kernelmike - 10-19-2026, 04:32 AM
Re: serving on Blackwell - by tensorfield88 - 10-19-2026, 05:18 AM

Forum Jump:


Users browsing this thread: 1 Guest(s)