Research on data-parallel integer processing with high-precision and high-performance

2016 Fiscal Year Final Research Report

• Project/Area Number: 26330144
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: High performance computing
• Research Institution: The University of Electro-Communications
• Principal Investigator: MURAO Hirokazu 電気通信大学, 大学院情報理工学研究科, 准教授 (60174265)
• Co-Investigator(Renkei-kenkyūsha): KIMURA Kinji 京都大学, 大学院情報学研究科, 特定准教授 (10447899) ; OSUMI Noriko (HYODO Noriko) サレジオ高等専門学校, 情報工学科, 准教授 (40712609)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 高精度整数計算 / 中国剰余定理 / GPGPU / マルチコア並列処理 / 陰関数描画 / WebGL / 数式処理 / 高性能計算

Outline of Final Research Achievements

Most of the current processors equips with data-parallel processing facility, designed mainly for numeric processing, such as SIMD instruction sets and GPU's SIMT architecture. This research project investigates appropriate computing methods to apply their computing power to a wide variety of integer processing from arithmetics of integer data, usually with arbitrary precision, to some kinds of applications such as curve plotting of implicit functions and optimal graph-path search problem. Integer processing actually investigated includes integer arithmetics represented by RNS or represented in mantissa of floating point numbers, bit-operations for GF(2), and graph structures. Our demonstration experiments indicate the practical effectiveness. Also, we were able to acquire skills and methodology for constructing platforms appropriate for our parallel-processing experiments.

Free Research Field

数式処理