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

Download:

Current browse context:

cs.DC

Change to browse by:

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 > Distributed, Parallel, and Cluster Computing

Title: Optimizing the Whole-life Cost in End-to-end CNN Acceleration

Abstract: The acceleration of CNNs has gained increasing atten-tion since their success in computer vision. With the heterogeneous functional layers that cannot be pro-cessed by the accelerators proposed for convolution layers only, modern end-to-end CNN acceleration so-lutions either transform the diverse computation into matrix/vector arithmetic, which loses data reuse op-portunities in convolution, or introduce dedicated functional unit to each kind of layer, which results in underutilization and high update expense. To enhance the whole-life cost efficiency, we need an acceleration solution that is efficient in processing CNN layers and has the generality to apply to all kinds of existing and emerging layers. To this end, we pro-pose GCONV Chain, a method to convert the entire CNN computation into a chain of standard general convolutions (GCONV) that can be efficiently pro-cessed by the existing CNN accelerators. This paper comprehensively analyzes the GCONV Chain model and proposes a full-stack implementation to support GCONV Chain. On one hand, the results on seven var-ious CNNs demonstrate that GCONV Chain improves the performance and energy efficiency of existing CNN accelerators by an average of 3.4x and 3.2x re-spectively. On the other hand, we show that GCONV Chain provides low whole-life costs for CNN accelera-tion, including both developer efforts and total cost of ownership for the users.
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
Cite as: arXiv:2104.05541 [cs.DC]
  (or arXiv:2104.05541v2 [cs.DC] for this version)

Submission history

From: Jiaqi Zhang [view email]
[v1] Mon, 12 Apr 2021 15:12:42 GMT (790kb)
[v2] Thu, 22 Jul 2021 00:46:39 GMT (790kb)

Link back to: arXiv, form interface, contact.