We Hit 100% GPU Utilization–and Then Made It 3× Faster by Not Using It
https://www.daft.ai/blog/embedding-millions-of-text-documents-with-qwen3
#HackerNews #GPU #Utilization #AI #Optimization #Performance #Boost
Kingman
Wenn Variabilität auf 100 % Auslastung trifft, explodieren die Wartezeiten – selbst kleine Störungen bringen das System zum Stillstand. Nicht die Variabilität ist das Problem, sondern Systeme ohne Puffer: Wer ohne Slack plant, plant den Stillstand.
https://no-bullshit-agile.de/wip/
#Kingman #Variability #Utilization
5/x
We Hit 100% GPU Utilization–and Then Made It 3× Faster by Not Using It
https://www.daft.ai/blog/embedding-millions-of-text-documents-with-qwen3
#HackerNews #GPU #Utilization #AI #Optimization #Performance #Boost
'I paid for the whole GPU, I am going to use the whole GPU'
https://modal.com/blog/gpu-utilization-guide
#HackerNews #GPU #Utilization #GPU #Computing #Tech #Insights #Modal #Blog
Going to #KubeCon? Then you should check out #HannahTaub's talk "The Node Tetris Rabbit Hole: Why Your #Binpacking Might Be Underperforming". You'll learn #Kubernetes cluster #utilization tips & tricks as well as some great #costefficiency techniques. Since I'm lucky enough to work with her I've gotten to see the work she'll be talking about first hand and it's impressive.