
Indexed in
License and use
Extending RISC-V Processor Datapaths with Multi-Grain Reconfigurable Overlays
Publicated to:Dcis 2022 - Proceedings Of The 37th Conference On Design Of Circuits And Integrated Systems. 137-142 - 2022-01-01 (), DOI: 10.1109/DCIS55711.2022.9970069
Authors: Vázquez, D; Rodríguez, A; Otero, A; de la Torre, E
Affiliations
Abstract
The use of accelerators to perform computing offloading from software to hardware in heterogeneous computing systems has already been around for some time. The most common approach is to have hardware accelerators loosely-coupled with the main processor of the system to allow parallel execution. However, this can negatively affect performance in certain application scenarios. A different approach, which is gaining momentum nowadays thanks to the mainstream adoption of RISC-V, is to create accelerators that combine a processor core with dedicated and tightly-coupled hardware modules that are transparently managed via custom ISA extensions. These modules, which are often application-specific, become part of the processor datapath and cannot be changed after deployment. In this paper, a novel reconfigurable RISC-V processor is presented. Its datapath has been modified to include a spatially-configured CGRA overlay that is accessible through custom ISA extensions. The functionality of the overlay can be adapted at run time through reconfiguration. The proposed architecture is accompanied by a toolchain capable of automatically transforming loop-based computing-intensive sections of the application code into configuration data controlling the operations executed in the overlay. Experimental results show a performance improvement of up to 13x compared to alternative software-based and memory-mapped hardware-based implementations.
Keywords
Quality index
Bibliometric impact. Analysis of the contribution and dissemination channel
From a relative perspective, and based on the normalized impact indicator calculated from the Field Citation Ratio (FCR) of the Dimensions source, it yields a value of: 1.32, which indicates that, compared to works in the same discipline and in the same year of publication, it ranks as a work cited above average. (source consulted: Dimensions Jul 2025)
Specifically, and according to different indexing agencies, this work has accumulated citations as of 2025-07-20, the following number of citations:
- Scopus: 3
Impact and social visibility
Leadership analysis of institutional authors
There is a significant leadership presence as some of the institution’s authors appear as the first or last signer, detailed as follows: First Author (VAZQUEZ IGLESIAS, DANIEL) and Last Author (TORRE ARNANZ, EDUARDO DE LA).