Show HN: AutoGPU – AI designs a real 7nm GPU, from Verilog to GDSII
Category: ai-ml
Tags: ai-agents, gpu-design, silicon-compiler, machine-learning, hardware-design
Score: 8.0/10 (Innovation: 9, Technical: 10, Documentation: 7, Utility: 6)
AutoGPU uses AI agents to autonomously design a 7nm GPU from Verilog to GDSII layout, including a 32x32 fp8 matmul accelerator with full sign-off toolchain and web-based die viewer. It represents a paradigm shift where AI designs silicon hardware end-to-end, potentially disrupting traditional chip design workflows.
Target audience: hardware engineers, ai researchers, chip designers, devops for EDA
Repository: https://github.com/npip99/autogpu · Python · 1 stars
View on Hacker News