https://github.com/mubashir1osmani/m4
build custom asics and fpga's using llms.
https://github.com/mubashir1osmani/m4
ai chip-design computer-use gpu hardware-designs llm
Last synced: 10 months ago
JSON representation
build custom asics and fpga's using llms.
- Host: GitHub
- URL: https://github.com/mubashir1osmani/m4
- Owner: mubashir1osmani
- Created: 2024-12-07T22:46:59.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-01-28T04:50:35.000Z (over 1 year ago)
- Last Synced: 2025-01-28T05:26:30.958Z (over 1 year ago)
- Topics: ai, chip-design, computer-use, gpu, hardware-designs, llm
- Language: Python
- Homepage:
- Size: 50.5 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
## idea
utilize agents to create chips and custom FPGAs and ASICs
## implementation
workflow: system prompt -> user specifications for chip -> generates verilog code / simulation of how it works -> use modelsim for testing <- reiterate
additional resources for model to train on: [Verilog/VHDL](https://github.com/klyone/opencores-ip)
## long term vision
after gaining feedback if this project has potential, can turn into a specialized product for EDA and chip design.
involves:
- pretraining on a corpus of hardware design data.
- adding task specific capabilites like:
- multi-modal input/output (eg: HDL, waveform simulation)
- integration with GNN for layout optimization
this project will be fully open source for this is a great way to learn and build something new for the future