Heterogeneous Computing Elements: A Quantitative Evaluation

Maya Gokhale | 19-ERD-004

Executive Summary

We will develop a suite of tools to enable rapid prototyping and evaluation of hardware units suitable for insertion into multiple computing fabric locations, including near processor/cache, memory/storage, and network. With the surge of heterogeneous accelerator designs currently being proposed for future computational architectures, our project fills a compelling need for tools that provide quantitative evaluation of heterogeneous processing units.

Publications, Presentations, Etc.

Gokhale, M. 2018. "Design Space Exploration of Near Memory Accelerators." Proceedings of the International Symposium on Memory Systems (MEMSYS 2018), 218–220. doi: 10.1145/3240302.3240428. LLNL-CONF-751363.

––– . 2019. "Extreme Heterogeneity as the norm." LLNL-PRES-779422.