SC21 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

RIKEN CGRA: Data-Driven Architecture as an Extension of Multicore CPU for Future HPC

Authors: Boma Adhi (RIKEN Center for Computational Science (R-CCS)); Takuya Kojima (Keio University, Tokyo); Yiyu Tan (RIKEN Center for Computational Science (R-CCS)); Artur Podobas (KTH Royal Institute of Technology, Sweden); and Kentaro Sano (RIKEN Center for Computational Science (R-CCS))

Abstract: A coarse-grained reconfigurable array (CGRA) is currently gaining momentum to enter the HPC market as deep-learning accelerators due to its potential for performance scalability and energy efficiency. Traditionally, the usage of CGRA was limited in embedded systems to provide extra computing performance with low power consumption. This poster describes our research journey to find what it takes for CGRA to become an ideal HPC accelerator.

Best Poster Finalist (BP): no

Poster: PDF
Poster summary: PDF

Back to Poster Archive Listing