Presentation 2018-03-01
A C Description Approach for High Level Synthesis to Configure DNN Inference Circuit
Takuya Okamoto, Ryota Yamamoto, Shinya Honda,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Today, Deep Neural Network (DNN) is utilized in various ?elds. There is a demand for deep learning in the ?eld of embedded systems. On the other hand, DNN classi?er on the system requires memory-saving and real?time property. Therefore, we target to FPGA and con?gure a classi?er for deep learning on FPGA with high-level synthesis (HLS) tool. To satisfy the system requirement, we consider an e?ective C source code description for high level synthesis. Speci?cally, our proposed accelerators are following: pipelining, prefetching, packing and caching. Compared to the original implementation, the execution speed by each accelerator was 2.3 times, 2.8 times, 5.7 times, 3.3 times faster, respectively.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) FPGA / DNN / High Level Synthesis
Paper # VLD2017-116
Date of Issue 2018-02-21 (VLD)

Conference Information
Committee VLD / HWS
Conference Date 2018/2/28(3days)
Place (in Japanese) (See Japanese page)
Place (in English) Okinawa Seinen Kaikan
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Hiroyuki Ochi(Ritsumeikan Univ.)
Vice Chair Noriyuki Minegishi(Mitsubishi Electric)
Secretary Noriyuki Minegishi(Hiroshima City Univ.) / (NTT)
Assistant

Paper Information
Registration To Technical Committee on VLSI Design Technologies / Technical Committee on Hardware Security
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A C Description Approach for High Level Synthesis to Configure DNN Inference Circuit
Sub Title (in English)
Keyword(1) FPGA
Keyword(2) DNN
Keyword(3) High Level Synthesis
1st Author's Name Takuya Okamoto
1st Author's Affiliation Nagoya University(Nagoya Univ.)
2nd Author's Name Ryota Yamamoto
2nd Author's Affiliation Nagoya University(Nagoya Univ.)
3rd Author's Name Shinya Honda
3rd Author's Affiliation Nagoya University(Nagoya Univ.)
Date 2018-03-01
Paper # VLD2017-116
Volume (vol) vol.117
Number (no) VLD-455
Page pp.pp.163-168(VLD),
#Pages 6
Date of Issue 2018-02-21 (VLD)