Democratizing Hardware Design

The OpenROAD project attacks the barriers of Cost, Expertise and Uncertainty (i.e., Risk) that block the feasibility of hardware design in advanced technologies.

About OpenROAD

Problem: Hardware design requires too much effort, cost and time.
Challenge: Costs and the “expertise gap” block system designers’ access to advanced technology.
Objective:  Enable no-human-in-loop, 24-hour design to remove the barrier to hardware innovation

Foundations and Realization of Open, Accessible Design

Prof. Kahng & the OpenROAD team are aiming to develop open-source tools that achieve autonomous, 24-hour layout implementation.

PowerPoint & video presentation from 2021 ERI Summit

Our Goals

24-hour, No-Human-In-The-Loop layout design for SOC with no Power-Performance-Area (PPA) loss Tapeout-capable tools in source code form, with permissive licensing → seed future “Linux of EDA”

Impact

  • Create new “Base Technologies” that enable 24-hour, autonomous design
    • Extreme partitioning (bite-sized problems)
    • Parallel search and optimization
    • Machine learning: models of tools, designs
  • New paradigm for design tools and methods: autonomy first
  • Bring down barriers to democratize HW design

The Problem

Our Approach

  • No Humans: tools must adapt and self-tune, must never get stuck unexpectedly
  • 24 hours: extreme partitioning of problems
    • parallel search on cloud
    • machine learning for predictability
  • Mantra: Correctness and safety by construction
  • Mantra: Embrace freedom from choice
  • Mantra: Often, only one thread needs to succeed
@OpenROAD_EDA

- 16 days ago

The DAC-2022 #59DAC "Open-Source EDA and Benchmarking Summit" Birds-of-a-Feather session is July 12, 2022 7-10pm, Room 3000, Moscone West in SF. Link: https://t.co/9jxidCgFwq Please mark your calendar and fill out our planning survey: https://t.co/dC2PxWGiaf -- See you there!
h J R
@OpenROAD_EDA

- 16 days ago

OpenROAD is heavily used in the https://t.co/MFzRaKokYV repo of the TILOS AI Institute @tilos_ai . This effort aims to provide a public, baseline implementation of Google Brain's https://t.co/9Kk7LCVwFK (Morpheus) deep RL-based placement method. #GoogleResearch
h J R