Stories by brrrrrm

Show HN: Binfer, an experimental LLM inference engine in TypeScript and CUDA

Five Times Faster

Bitwise Consistent On-Policy Reinforcement Learning with VLLM and TorchTitan

Should we apply old-school multi-core scheduling to GPUs?

Show HN: GT: experimental multiplexed distributed tensor framework

GT – Experimental multiplexing tensor framework for distributed GPU computing

MFU Is Poorly Approximating Billions of Dollars in Compute

Model Flop Utilization Beyond 6ND

SWE-bench verified agents may look at future repository state

Animators Say, 'That's All, Folks' (2000)

Chicken Tax

Up to 1.9X Higher Llama 3.1 Performance with Medusa

Show HN: Collaborative ASCII Drawing with Telnet

Show HN: Eye-Tracking for Immersive 3D

Show HN: Locally run a "blue text" bot (llama3)

PyTorch is live streaming their meeting on twitch

Is Elon specially promoting Mr Beast's recent video?

Drawer Drawer (draw by clicking drawers)

Show HN: Brr.js – Trying to make WebGPU a bit easier to use

[forgot to include URL, won't let me add it]

1
2
3
4