Presentation 2015-12-03
Accuracy Analysis of Machine Learning based Performance Modeling for Microprocessors
Yoshihiro Tanaka, Takatsugu Ono, Koji Inoue,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) When designing a computer system, a system designer need to select an appropriate processor to satisfy design constraints if an application to be executed is determined in advance. It is quite costly to implement the application onto all of the processor candidates. One of the representative approaches to address the above problem is to estimate the application’s performance on each target processor by using performance models based on machine learning. Since its accuracy of depends on various kinds of factors such as quality of training data, machine learning algorithms and so on, it is necessary to choose an appropriate method for the modeling. In this paper, in order to reveal their requirements for accurate performance estimation, we analyze the accuracy of performance models by varying the architecture parameters used for training data and performance modeling method. The result shows that in order to realize accurate performance modeling, CPU clock frequency is necessary to be used in training data and a number of architecture parameters is required to be selected according to the performance modeling method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Performance Estimation / Machine Learning / Empirical Modeling
Paper # CPSY2015-74
Date of Issue 2015-11-24 (CPSY)

Conference Information
Committee VLD / DC / IPSJ-SLDM / CPSY / RECONF / ICD / CPM
Conference Date 2015/12/1(3days)
Place (in Japanese) (See Japanese page)
Place (in English) Nagasaki Kinro Fukushi Kaikan
Topics (in Japanese) (See Japanese page)
Topics (in English) Design Gaia 2015 -New Field of VLSI Design-
Chair Yusuke Matsunaga(Kyushu Univ.) / Nobuyasu Kanekawa(Hitachi) / Masahiro Fukui(Ritsumeikan Univ.) / Yasuhiko Nakashima(NAIST) / Minoru Watanabe(Shizuoka Univ.) / Minoru Fujishima(Hiroshima Univ.) / Satoru Noge(Numazu National College of Tech.)
Vice Chair Takashi Takenana(NEC) / Michiko Inoue(NAIST) / / Koji Nakano(Hiroshima Univ.) / Hidetsugu Irie(Univ. of Tokyo) / Masato Motomura(Hokkaido Univ.) / Yuichiro Shibata(Nagasaki Univ.) / Hideto Hidaka(Renesas) / Fumihiko Hirose(Yamagata Univ.)
Secretary Takashi Takenana(Ritsumeikan Univ.) / Michiko Inoue(Fujitsu Labs.) / (RTRI) / Koji Nakano(Kyoto Sangyo Univ.) / Hidetsugu Irie(Sharp) / Masato Motomura(Kitakyushu City Univ.) / Yuichiro Shibata(Toshiba) / Hideto Hidaka(Fujitsu Labs.) / Fumihiko Hirose(NII)
Assistant Ittetsu Taniguchi(Ritsumeikan Univ.) / / / Shinya Takameda(NAIST) / Takeshi Ohkawa(Utsunomiya Univ.) / Kazuya Tanikagawa(Hiroshima City Univ.) / Takefumi Miyoshi(e-trees.Japan) / Makoto Takamiya(Univ. of Tokyo) / Hiroe Iwasaki(NTT) / Takashi Hashimoto(Panasonic) / Hiroyuki Ito(Tokyo Inst. of Tech.) / Pham Konkuha(Univ. of Electro-Comm.) / Takashi Sakamoto(NTT) / Yuichi Nakamura(Toyohashi Univ. of Tech.)

Paper Information
Registration To Technical Committee on VLSI Design Technologies / Technical Committee on Dependable Computing / Special Interest Group on System and LSI Design Methodology / Technical Committee on Computer Systems / Technical Committee on Reconfigurable Systems / Technical Committee on Integrated Circuits and Devices / Technical Committee on Component Parts and Materials
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Accuracy Analysis of Machine Learning based Performance Modeling for Microprocessors
Sub Title (in English)
Keyword(1) Performance Estimation
Keyword(2) Machine Learning
Keyword(3) Empirical Modeling
1st Author's Name Yoshihiro Tanaka
1st Author's Affiliation Kyushu University(Kyushu Univ.)
2nd Author's Name Takatsugu Ono
2nd Author's Affiliation Kyushu University(Kyushu Univ.)
3rd Author's Name Koji Inoue
3rd Author's Affiliation Kyushu University(Kyushu Univ.)
Date 2015-12-03
Paper # CPSY2015-74
Volume (vol) vol.115
Number (no) CPSY-342
Page pp.pp.75-80(CPSY),
#Pages 6
Date of Issue 2015-11-24 (CPSY)