NARMADA: near-memory horizontal diffusion accelerator for scalable stencil computations

Gagandeep Singh, Dionysios Diamantopoulos, Christoph Hagleitner, Sander Stuijk, Henk Corporaal

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

Samenvatting

Real-world weather forecasting applications consist of compound stencil kernels that do not perform well on conventional architectures. This behavior is due to their complex data access patterns, limited data reusability, and low arithmetic intensity. To overcome these issues, we harness the potential of near-memory computing by offloading a horizontal diffusion kernel, which is a compound stencil kernel, from the COSMO weather prediction application to a reconfigurable fabric. We use a heterogeneous system that comprises a CPU and an FPGA with on-chip SRAM memory and on-board DRAM memory. By introducing a memory hierarchy tailored to the targeted application and using a coherent memory model, we move the computation close to the memory, which improves memory efficiency. Our hardware design on the FPGA uses high-level synthesis techniques and results in an accelerator with IBM CAPI 2.0 (Coherent Accelerator Processor Interface) technology. We evaluate it against a tuned software implementation running on an IBM POWER9 host system. The experimental results show that these kernels on an FPGA can outperform a complete 16-core POWER9 node (configured with 64 threads) by 3.3x. Moreover, our solution provides an 18x improvement in the active energy consumption.

Originele taal-2Engels
TitelProceedings - 29th International Conference on Field-Programmable Logic and Applications, FPL 2019
RedacteurenIoannis Sourdis, Christos-Savvas Bouganis, Carlos Alvarez, Leonel Antonio Toledo Diaz, Pedro Valero, Xavier Martorell
Plaats van productiePiscataway
UitgeverijInstitute of Electrical and Electronics Engineers
Pagina's263-269
Aantal pagina's7
ISBN van elektronische versie978-1-7281-4884-7
DOI's
StatusGepubliceerd - sep 2019
Evenement29th International Conferenceon Field-Programmable Logic and Applications, FPL 2019 - Barcelona, Spanje
Duur: 9 sep 201913 sep 2019

Congres

Congres29th International Conferenceon Field-Programmable Logic and Applications, FPL 2019
LandSpanje
StadBarcelona
Periode9/09/1913/09/19

    Vingerafdruk

Citeer dit

Singh, G., Diamantopoulos, D., Hagleitner, C., Stuijk, S., & Corporaal, H. (2019). NARMADA: near-memory horizontal diffusion accelerator for scalable stencil computations. In I. Sourdis, C-S. Bouganis, C. Alvarez, L. A. Toledo Diaz, P. Valero, & X. Martorell (editors), Proceedings - 29th International Conference on Field-Programmable Logic and Applications, FPL 2019 (blz. 263-269). [8892056] Piscataway: Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/FPL.2019.00050