Search

Your search keyword '"INOUE, HIROSHI"' showing total 112 results
Start Over Author "INOUE, HIROSHI" Publisher ieee
112 results on '"INOUE, HIROSHI"'

10. Multi-step LRU: SIMD-based Cache Replacement for Lower Overhead and Higher Precision

Author

Inoue, Hiroshi

Subjects
Networking and Internet Architecture (cs.NI), Performance (cs.PF), FOS: Computer and information sciences, Computer Science - Networking and Internet Architecture, Hardware_MEMORYSTRUCTURES, Computer Science - Performance, Computer Science - Databases, Databases (cs.DB)
Abstract

A key-value cache is a key component of many services to provide low-latency and high-throughput data accesses to a huge amount of data. To improve the end-to-end performance of such services, a key-value cache must achieve a high cache hit ratio with high throughput. In this paper, we propose a new cache replacement algorithm, multi-step LRU, which achieves high throughput by efficiently exploiting SIMD instructions without using per-item additional memory (LRU metadata) to record information such as the last access timestamp. For a small set of items that can fit within a vector register, SIMD-based LRU management without LRU metadata is known (in-vector LRU). It remembers the access history by reordering items in one vector using vector shuffle instruction. In-vector LRU alone cannot be used for a caching system since it can manage only few items. Set-associative cache is a straightforward way to build a large cache using in-vector LRU as a building block. However, a naive set-associative cache based on in-vector LRU has a poorer cache hit ratio than the original LRU although it can achieve a high throughput. Our multi-step LRU enhances naive set-associative cache based on in-vector LRU for improving cache accuracy by taking both access frequency and access recency of items into account while keeping the efficiency by SIMD instructions. Our results indicate that multi-step LRU outperforms the original LRU and GCLOCK algorithms in terms of both execution speed and cache hit ratio. Multi-step LRU improves the cache hit ratios over the original LRU by implicitly taking access frequency of items as well as access recency into account. The cache hit ratios of multi-step LRU are similar to those of ARC, which achieves a higher a cache hit ratio in a tradeoff for using more LRU metadata.

Published
2021

12. Static Eccentricity Fault Detection for PSH-type Induction Motors Considering High-order Air Gap Permeance Harmonics

Author

Lei Zhou, Inoue Hiroshi, Chungwei Lin, Bingnan Wang, and Masahito Miyoshi

Subjects
0209 industrial biotechnology, Rotor (electric), Computer science, 02 engineering and technology, Permeance, Fault (power engineering), 01 natural sciences, Fault detection and isolation, law.invention, 020901 industrial engineering & automation, law, Control theory, Harmonics, 0103 physical sciences, Harmonic, Air gap (plumbing), 010301 acoustics, Induction motor
Abstract

Diagnosis of static eccentricity (SE) fault for induction motors (IMs) is essential for the quality control of the machines, especially during their manufacturing process. Principal slot harmonic (PSH) type IMs have special combinations of rotor bar number and pole pair number, and it has been shown in previous works that conventional methods cannot effectively detect SE fault for these machines. Aiming at finding an effective approach for the SE fault detection for PSH-type IMs, this paper presents an analysis of SE-induced line currents in IMs with the higher-order harmonics of the air gap permeance considered. The analysis reveals that the second-order harmonic in the air gap permeance can induce SE-level-related signals in the line current of PSH-type three-phase IMs. The generation mechanism of the signature current signal is validated by simulations with an analytical IM model and a time-stepping finite element model. The signature signal in the motor's current discussed in this paper provides a new method for quantitative detection of SE fault for PSH-type IMs.

Published
2021

13. RaPiD: AI Accelerator for Ultra-low Precision Training and Inference

Author

Venkataramani, Swagath, primary, Srinivasan, Vijayalakshmi, additional, Wang, Wei, additional, Sen, Sanchari, additional, Zhang, Jintao, additional, Agrawal, Ankur, additional, Kar, Monodeep, additional, Jain, Shubham, additional, Mannari, Alberto, additional, Tran, Hoang, additional, Li, Yulong, additional, Ogawa, Eri, additional, Ishizaki, Kazuaki, additional, Inoue, Hiroshi, additional, Schaal, Marcel, additional, Serrano, Mauricio, additional, Choi, Jungwook, additional, Sun, Xiao, additional, Wang, Naigang, additional, Chen, Chia-Yu, additional, Allain, Allison, additional, Bonano, James, additional, Cao, Nianzheng, additional, Casatuta, Robert, additional, Cohen, Matthew, additional, Fleischer, Bruce, additional, Guillorn, Michael, additional, Haynie, Howard, additional, Jung, Jinwook, additional, Kang, Mingu, additional, Kim, Kyu-hyoun, additional, Koswatta, Siyu, additional, Lee, Saekyu, additional, Lutz, Martin, additional, Mueller, Silvia, additional, Oh, Jinwook, additional, Ranjan, Ashish, additional, Ren, Zhibin, additional, Rider, Scot, additional, Schelm, Kerstin, additional, Scheuermann, Michael, additional, Silberman, Joel, additional, Yang, Jie, additional, Zalani, Vidhi, additional, Zhang, Xin, additional, Zhou, Ching, additional, Ziegler, Matt, additional, Shah, Vinay, additional, Ohara, Moriyoshi, additional, Lu, Pong-Fei, additional, Curran, Brian, additional, Shukla, Sunil, additional, Chang, Leland, additional, and Gopalakrishnan, Kailash, additional

Published
2021
Full Text
View/download PDF

17. Efficient AI System Design With Cross-Layer Approximate Computing.

Author

Venkataramani, Swagath, Sun, Xiao, Wang, Naigang, Chen, Chia-Yu, Choi, Jungwook, Kang, Mingu, Agarwal, Ankur, Oh, Jinwook, Jain, Shubham, Babinsky, Tina, Cao, Nianzheng, Fox, Thomas, Fleischer, Bruce, Gristede, George, Guillorn, Michael, Haynie, Howard, Inoue, Hiroshi, Ishizaki, Kazuaki, Klaiber, Michael, and Lo, Shih-Hsien

Subjects
SYSTEMS design, ARTIFICIAL intelligence, ARCHITECTURAL design, COMPUTER architecture, COMPILERS (Computer programs)
Abstract

Advances in deep neural networks (DNNs) and the availability of massive real-world data have enabled superhuman levels of accuracy on many AI tasks and ushered the explosive growth of AI workloads across the spectrum of computing devices. However, their superior accuracy comes at a high computational cost, which necessitates approaches beyond traditional computing paradigms to improve their operational efficiency. Leveraging the application-level insight of error resilience, we demonstrate how approximate computing (AxC) can significantly boost the efficiency of AI platforms and play a pivotal role in the broader adoption of AI-based applications and services. To this end, we present RaPiD, a multi-tera operations per second (TOPS) AI hardware accelerator core (fabricated at 14-nm technology) that we built from the ground-up using AxC techniques across the stack including algorithms, architecture, programmability, and hardware. We highlight the workload-guided systematic explorations of AxC techniques for AI, including custom number representations, quantization/pruning methodologies, mixed-precision architecture design, instruction sets, and compiler technologies with quality programmability, employed in the RaPiD accelerator. [ABSTRACT FROM AUTHOR]

Published
2020
Full Text
View/download PDF

22. Extracting current waveforms of appliances from mixed current waveform of their current waveforms and unidentified current waveforms

Author

Suzuki Yasunao, Inoue Hiroshi, and Fumihiko Ishiyama

Subjects
Engineering, business.industry, 020209 energy, Function generator, Hardware_PERFORMANCEANDRELIABILITY, 02 engineering and technology, Arbitrary waveform generator, Evaluation function, Power (physics), Hardware_INTEGRATEDCIRCUITS, 0202 electrical engineering, electronic engineering, information engineering, Electronic engineering, Waveform, Current (fluid), business, Hardware_LOGICDESIGN
Abstract

We propose a method of extracting current waveforms of appliances from the current waveform, which is a mixture of their current waveforms and unidentified current waveforms. The purpose of the method is for monitoring the load of individual appliances by analyzing the current waveform on a power distribution board. Since the mixed current waveform contains unidentified current waveforms, a special method is required. Therefore, we propose to apply our method, called the “inscribed fitting method” with an asymmetric evaluation function, for the extraction. We apply our method to the mixed current waveforms of appliances, showing that we can extract the current waveforms of appliances from the mixed current waveforms which contain unidentified current waveforms.

Published
2016

26. Inscribed fitting method with asymmetric evaluation function for disaggregating current waveforms

Author

Takashi Ohyama, Fumihiko Ishiyama, Toshio Watanabe, and Inoue Hiroshi

Subjects
Distribution board, Computer science, business.industry, Electrical engineering, Hardware_PERFORMANCEANDRELIABILITY, Computer Science::Human-Computer Interaction, Evaluation function, Power (physics), Power demand, Computer Science::Sound, Hardware_INTEGRATEDCIRCUITS, Electronic engineering, Waveform, Current (fluid), Focus (optics), business, Inscribed figure, Hardware_LOGICDESIGN
Abstract

We propose a method for disaggregating the current waveforms of appliances from a current waveform on a power distribution board. We focus on the case where some appliances have current waveforms that vary continuously and non-proportionally. Our method is a variation of the least squares method with an asymmetric evaluation function. We calculate a standard waveform for each appliance, and we inscribe the standard waveforms to the current waveform on the power distribution board. We apply our method to the aggregate current waveforms of three appliances and compare the results with the least squares method.

Published
2015

27. DeepTools: Compiler and Execution Runtime Extensions for RaPiD AI Accelerator.

Author

Venkataramani, Swagath, Choi, Jungwook, Srinivasan, Vijayalakshmi, Wang, Wei, Zhang, Jintao, Schaal, Marcel, Serrano, Mauricio J., Ishizaki, Kazuaki, Inoue, Hiroshi, Ogawa, Eri, Ohara, Moriyoshi, Chang, Leland, and Gopalakrishnan, Kailash

Subjects
ARTIFICIAL intelligence, COMPILERS (Computer programs), DEEP learning, MACHINE learning
Abstract

The ubiquitous adoption of systems specialized for AI requires bridging two seemingly conflicting challenges—the need to deliver extreme processing efficiencies while employing familiar programming interfaces, making them compelling even for non-expert users. We take a significant first step towards this goal and present an end-to-end software stack for the RaPiD AI accelerator developed by IBM Research. We present a set of software extensions, called Deeptools, that leverage and work within popular deep learning frameworks. DeepTools requires no additional user input and enables aggressive, accelerator-specific performance optimization akin to a full, custom framework. DeepTools has two key components: 1) a compiler runtime called DeepRT, which automatically identifies how best to execute a given DNN graph on RaPiD and constructs the requisite program binaries; and 2) an execution runtime called RaPiDLib, which triggers and manages the execution of compute and data-transfer operations on RaPiD. We integrate DeepTools with TensorFlow and map popular DNNs (AlexNet, VGG, ResNet, LSTM) to RaPiD. We demonstrate substantial improvement in performance over hand-tuned mappings. [ABSTRACT FROM AUTHOR]

Published
2019
Full Text
View/download PDF

33. Development of Home Energy Management System with Advice Function

Author

Inoue Hiroshi and Masaki Yamamoto

Subjects
Consumption (economics), Building management system, Computer science, Energy management, business.industry, media_common.quotation_subject, Energy consumption, Computer security, computer.software_genre, Energy conservation, Energy management system, Risk analysis (engineering), Electric power, Electricity, Function (engineering), business, computer, media_common
Abstract

First of all, this paper describes the present situation of the Japanese energy consumption trend and the necessity of home energy saving. Next, the feature of the present Home Energy Management System (HEMS) is described. Finally, this paper introduces our HEMS with advice function. The main feature of our HEMS is provision of suitable advice to user by analyzing his or her electric power consumption.

Published
2011

40. A new embedded die package — WFOP™

Author

Takahashi, Tomoko, primary, Inoue, Hiroshi, additional, Yada, Takahiro, additional, Hayashi, Naoki, additional, Imaizumi, Yukari, additional, Ikemoto, Yoshihiko, additional, Sawachi, Shigenori, additional, Furuno, Atsushi, additional, Yoshimitsu, Katsushi, additional, Ooida, Mitsuru, additional, Katsumata, Akio, additional, and Hiruta, Yoichi, additional

Published
2014
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results