Hacker News new | ask | show | jobs
user: somnial
created: 2025-10-30
karma: 3

submissions:

0 points | 0 comments
0 points | 0 comments
Pushing memory bound CUDA kernels past the speed of light with data compression
2 points | 0 comments
Speculative KV coding: ~4× losslessly compressed KV cache using a small model
2 points | 0 comments
70x faster cold(ish) starts for SGLang
1 points | 0 comments
LLM powered data structures: A lock-free binary search tree
1 points | 0 comments
Parallel Primitives for Multi-Agent Workflows
1 points | 0 comments
Scheduling in LLM Inference
1 points | 0 comments
How fast can an LLM go?
2 points | 0 comments