Data-Driven Offline Design of Hardware Accelerators

Home

Discover

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and more.Subscribe

CC BY

Source arXiv

Hardware Architecture

Machine Learning

Data-Driven Offline Design of Hardware Accelerators

Data-Driven Offline Optimization For Architecting Hardware Accelerators

We develop a data-driven offline optimization method for designing hardware accelerators, dubbed prime, that learns a conservative, robust estimate of the desired cost function, utilizes infeasible points, and optimizes the design against this estimate without any additional simulator queries during optimization.

Such an approach not only alleviates the need to run time-consuming simulation, but also enables data reuse and applies even when set of target applications or design constraints change.

We show that our approach outperforms state-of-the-art simulation-driven methods by about 1.54x and 1.20x, while considerably reducing the required total simulation time by 93% and 99%, respectively, while outperforming simulation-based methods by 1.26x.

Authors

Aviral Kumar, Amir Yazdanbakhsh, Milad Hashemi, Kevin Swersky, Sergey Levine

◐ Recommended Follows

◐ Latest Activity