Technology Jun 16, 2026 · 1 min read

GateGPT: 56k tokens per second Transformer (KV cache) on FPGA at 80 MHz

Article URL: https://twitter.com/fguzmanai/status/2065832668172845209 Comments URL: https://news.ycombinator.com/item?id=48557535 Points: 16 # Comments: 4

Hacker News

by laxmena

GateGPT: 56k tokens per second Transformer (KV cache) on FPGA at 80 MHz

Article URL: https://twitter.com/fguzmanai/status/2065832668172845209

Comments URL: https://news.ycombinator.com/item?id=48557535

Points: 16

# Comments: 4

Source

This article was originally published by Hacker News and written by laxmena.

Read original article on Hacker News

Back to Discover

GateGPT: 56k tokens per second Transformer (KV cache) on FPGA at 80 MHz

Reading List