Presentation 2007/6/7
The overview of the massively parallel processor based on the matrix architecture and the considerations of the image processing applications
Hiroyuki YAMASAKI, Hideyuki NODA, Tetsu NISHIJIMA, Kanako YOSHIDA, Takayuki GYOTEN, Tetsushi TANIZAKI, Katsuya MIZUMOTO, Masami NAKAJIMA, Yoshihiro OKUNO, Kazutami ARIMOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Recently, the installed applications in the digital devices has been remarkably progressed. Considering these backgrounds, we have developed the massively parallel processor based on the matrix architecture which realizes both high flexibilities and high performance ability while achieving small silicon area and low power consumption, which is attractive solution for Today's SoCs. This paper introduces the overview of the hardware architecture and the software development environment of this processor, and reports the results of the image processing applications. The evaluated performance of applying to 3x3 convolution filter processing is 147-times and 280-times higher than the conventional 32bit RISC CPUs for 16bit and 8bit data accuracy, respectively. The proposed processor is expected to have over 100-times higher performance than the conventional CPUs for various algorithms comparatively suitable for parallel processing, therefore, it offers promising possibilities to the software solutions realized in SoCs.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) massively parallel processing / SIMD / processor / image processing
Paper # SIS2007-4
Date of Issue

Conference Information
Committee SIS
Conference Date 2007/6/7(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Smart Info-Media Systems (SIS)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) The overview of the massively parallel processor based on the matrix architecture and the considerations of the image processing applications
Sub Title (in English)
Keyword(1) massively parallel processing
Keyword(2) SIMD
Keyword(3) processor
Keyword(4) image processing
1st Author's Name Hiroyuki YAMASAKI
1st Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.()
2nd Author's Name Hideyuki NODA
2nd Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
3rd Author's Name Tetsu NISHIJIMA
3rd Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
4th Author's Name Kanako YOSHIDA
4th Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
5th Author's Name Takayuki GYOTEN
5th Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
6th Author's Name Tetsushi TANIZAKI
6th Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
7th Author's Name Katsuya MIZUMOTO
7th Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
8th Author's Name Masami NAKAJIMA
8th Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
9th Author's Name Yoshihiro OKUNO
9th Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
10th Author's Name Kazutami ARIMOTO
10th Author's Affiliation System Solution Business Group, System Core Technology Div., Renesas Technology Corp.
Date 2007/6/7
Paper # SIS2007-4
Volume (vol) vol.107
Number (no) 93
Page pp.pp.-
#Pages 6
Date of Issue