Etched Patent: Direct Memory-to-Compute Connection in AI Accelerators
AccBalanced · x · 2026-07-06
According to reposted content, AI chip startup Etched applied for patent US20240378175A1 in May 2023. The core idea is to remove the traditional switch between compute units and memory in AI accelerators, replacing it with a direct connection from memory channels to compute units.
This architectural change aims to reduce data transmission latency and improve memory bandwidth utilization. It represents a bottom-up chip design tailored specifically for AI inference workloads, diverging from general-purpose GPU architecture concepts.
More from Infra
- Alchemy adds effect/Workflow support across Durable Objects, Workflows and Lambda Durable Functions — samgoodwin89 · 2026-10-27
- Alchemy's Lambda Web Functions run plain Node web servers on Lambda, no OpenNext needed — samgoodwin89 · 2026-10-27
- Foldkit Adds SSR Support With a Few Lines of Config on Cloudflare and AWS — samgoodwin89 · 2026-10-27
- Alchemy Cuts Cloudflare Bindings CPU Time by Up to 166× Via Profiling — samgoodwin89 · 2026-10-27
- Importing all 438 AWS services now takes 13.9 MB and 0.14s, down from 227 MB and 1.61s — samgoodwin89 · 2026-10-27
- Alchemy makes all ~2,850 AWS bindings work from any cloud, not just S3/Lambda/DynamoDB — samgoodwin89 · 2026-10-27