We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.AR

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Hardware Architecture

Title: High Area/Energy Efficiency RRAM CNN Accelerator with Kernel-Reordering Weight Mapping Scheme Based on Pattern Pruning

Abstract: Resistive Random Access Memory (RRAM) is an emerging device for processing-in-memory (PIM) architecture to accelerate convolutional neural network (CNN). However, due to the highly coupled crossbar structure in the RRAM array, it is difficult to exploit the sparsity of the network in RRAM-based CNN accelerator. To optimize the weight mapping of sparse network in the RRAM array and achieve high area and energy efficiency, we propose a novel weight mapping scheme and corresponding RRAM-based CNN accelerator architecture based on pattern pruning and Operation Unit(OU) mechanism. Experimental results show that our work can achieve 4.16x-5.20x crossbar area efficiency, 1.98x-2.15x energy efficiency, and 1.15x-1.35x performance speedup in comparison with the traditional weight mapping method.
Comments: 6 pages, 7 figures
Subjects: Hardware Architecture (cs.AR)
ACM classes: C.1.3; C.3
Cite as: arXiv:2010.06156 [cs.AR]
  (or arXiv:2010.06156v1 [cs.AR] for this version)

Submission history

From: Songming Yu [view email]
[v1] Tue, 13 Oct 2020 04:08:33 GMT (422kb,D)

Link back to: arXiv, form interface, contact.