Y Hacker News new | ask | show | jobs

user: charles_irl created: 2024-08-06 karma: 370 Building useful technology out of large neural networks. https://modal.com submissions: 0 points | 0 comments 0 points | 0 comments 0 points | 0 comments 0 points | 0 comments 0 points | 0 comments Cutting inference cold starts by 40x with LP, FUSE, C/R, and CUDA-checkpoint 91 points | 18 comments How to Achieve Serverless GPUs 8 points | 0 comments 0 points | 0 comments 0 points | 0 comments 0 points | 0 comments Three types of LLM workloads and how to serve them 75 points | 5 comments Host overhead is killing your inference efficiency 3 points | 0 comments 0 points | 0 comments Quantized Float Exposed 2 points | 1 comments Against SQL (2021) 82 points | 77 comments Length-extension attacks are still a thing 2 points | 1 comments The future of Python web services looks GIL-free 3 points | 0 comments Lexical differential highlighting instead of syntax highlighting 2 points | 0 comments CReact – JSX for the Cloud 1 points | 0 comments QUIC and the end of TCP sockets 62 points | 82 comments In C++ modules globally unique module names seem to be unavoidable 2 points | 0 comments Stupid jj Tricks 3 points | 0 comments 0 points | 0 comments 0 points | 0 comments We reverse-engineered Flash Attention 4 5 points | 0 comments A Tour of eBPF in the Linux Kernel: Observability, Security and Networking 2 points | 0 comments Categorical Foundations for Cute Layouts 39 points | 6 comments Pocket Casts, You Altered the Deal, So I Will Alter Your App 12 points | 3 comments Modal Notebooks: How we built a cloud GPU notebook that boots in seconds 4 points | 0 comments Public static void main(String[] args) is dead 210 points | 203 comments