Please wait, while we are loading the content...
Please wait, while we are loading the content...
Content Provider | ACM Digital Library |
---|---|
Author | Ferdman, Michael Milder, Peter Shen, Yongming |
Abstract | Convolutional neural networks (CNNs) are used to solve many challenging machine learning problems. These networks typically use convolutional layers for feature extraction and fully-connected layers to perform classification using those features. Significant interest in improving the performance of CNNs has led to the design of CNN accelerators to improve their evaluation throughput and efficiency. However, work on CNN accelerators has mostly concentrated on accelerating the computationally-intensive convolutional layers, while a major bottleneck of the existing designs arises due to the data-intensive fully-connected layers. Unfortunately, the leading approaches to reducing bandwidth of the fully-connected layers are limited by the storage capacity of the on-chip buffers. We observe that, in addition to the possibility of reducing CNN weight transfer bandwidth by adding more on-chip buffers, it is also possible to reduce the size of the on-chip buffers at the cost of CNN input transfer. Paradoxically, shrinking the size of the on-chip buffers costs significantly less input bandwidth than the weight bandwidth saved by adding more buffers. Leveraging these observations, we develop a design methodology for fully-connected layer accelerators that require substantially less off-chip bandwidth by balancing between the input and weight transfers. Using 160KB of BRAM enables the prior work to reduce off-chip bandwidth by 5x on the most bandwidth-intensive fully-connected layers of the popular AlexNet and VGGNet-E networks. With our newly proposed methodology, using the same 160KB of BRAM produces a design with 71x bandwidth reduction on the same networks. |
Starting Page | 293 |
Ending Page | 293 |
Page Count | 1 |
ISBN | 9781450343541 |
DOI | 10.1145/3020078.3021795 |
Language | English |
Publisher | Association for Computing Machinery (ACM) |
Publisher Date | 2017-02-22 |
Publisher Place | New York |
Access Restriction | Subscribed |
Subject Keyword | Bandwidth optimization Fpga accelerators Convolutional neural networks |
Content Type | Text |
Resource Type | Article |
National Digital Library of India (NDLI) is a virtual repository of learning resources which is not just a repository with search/browse facilities but provides a host of services for the learner community. It is sponsored and mentored by Ministry of Education, Government of India, through its National Mission on Education through Information and Communication Technology (NMEICT). Filtered and federated searching is employed to facilitate focused searching so that learners can find the right resource with least effort and in minimum time. NDLI provides user group-specific services such as Examination Preparatory for School and College students and job aspirants. Services for Researchers and general learners are also provided. NDLI is designed to hold content of any language and provides interface support for 10 most widely used Indian languages. It is built to provide support for all academic levels including researchers and life-long learners, all disciplines, all popular forms of access devices and differently-abled learners. It is designed to enable people to learn and prepare from best practices from all over the world and to facilitate researchers to perform inter-linked exploration from multiple sources. It is developed, operated and maintained from Indian Institute of Technology Kharagpur.
Learn more about this project from here.
NDLI is a conglomeration of freely available or institutionally contributed or donated or publisher managed contents. Almost all these contents are hosted and accessed from respective sources. The responsibility for authenticity, relevance, completeness, accuracy, reliability and suitability of these contents rests with the respective organization and NDLI has no responsibility or liability for these. Every effort is made to keep the NDLI portal up and running smoothly unless there are some unavoidable technical issues.
Ministry of Education, through its National Mission on Education through Information and Communication Technology (NMEICT), has sponsored and funded the National Digital Library of India (NDLI) project.
Sl. | Authority | Responsibilities | Communication Details |
---|---|---|---|
1 | Ministry of Education (GoI), Department of Higher Education |
Sanctioning Authority | https://www.education.gov.in/ict-initiatives |
2 | Indian Institute of Technology Kharagpur | Host Institute of the Project: The host institute of the project is responsible for providing infrastructure support and hosting the project | https://www.iitkgp.ac.in |
3 | National Digital Library of India Office, Indian Institute of Technology Kharagpur | The administrative and infrastructural headquarters of the project | Dr. B. Sutradhar bsutra@ndl.gov.in |
4 | Project PI / Joint PI | Principal Investigator and Joint Principal Investigators of the project |
Dr. B. Sutradhar bsutra@ndl.gov.in Prof. Saswat Chakrabarti will be added soon |
5 | Website/Portal (Helpdesk) | Queries regarding NDLI and its services | support@ndl.gov.in |
6 | Contents and Copyright Issues | Queries related to content curation and copyright issues | content@ndl.gov.in |
7 | National Digital Library of India Club (NDLI Club) | Queries related to NDLI Club formation, support, user awareness program, seminar/symposium, collaboration, social media, promotion, and outreach | clubsupport@ndl.gov.in |
8 | Digital Preservation Centre (DPC) | Assistance with digitizing and archiving copyright-free printed books | dpc@ndl.gov.in |
9 | IDR Setup or Support | Queries related to establishment and support of Institutional Digital Repository (IDR) and IDR workshops | idr@ndl.gov.in |
Loading...
|