Etched says inference means owning the full stack, not just writing kernels
AccBalanced · x · 2026-07-25
Etched says its inference stack goes far beyond kernel writing: the company claims ownership of the full stack, from KV cache management and sharding algorithms to workload-aware SerDes tuning.
It says it pushes the Pareto frontier daily, co-designs models with customers, works with chip designers, and continuously improves its model harness for recursive kernel development. The pitch also highlights topics like KV tree masking, diffusion speculators, latent compression, and large-scale scale-up domains.
More from coding & agent
- TRL adds Score Centering, a simple fix for RLHF training instability — lvwerra · 2026-10-27
- Precompute golden answers and match at runtime for exact wording control — colinmcnamara · 2026-10-27
- Dev argues there's still alpha in working with your coding agent interactively — alex_frantic · 2026-10-27
- Researchers launch AMA on patterns for safer AI agents used by citizen developers — Loose_Rip359 · 2026-10-27
- Gradio Adds In-Chat Human-in-the-Loop So ML Intern Pauses Before Training — Gradio · 2026-10-27
- Building Halo Infinite Forge maps with local LLMs, DCjson and Blender MCP — swagonflyyyy · 2026-10-27