Efficiency of Parallel Large-Scale Two-Layered MLP Training on Many-Core System

Author(s):  
Volodymyr Turchenko ◽  
Anatoly Sachenko
Keyword(s):  
Impact ◽  
2019 ◽  
Vol 2019 (10) ◽  
pp. 44-46
Author(s):  
Masato Edahiro ◽  
Masaki Gondo

The pace of technology's advancements is ever-increasing and intelligent systems, such as those found in robots and vehicles, have become larger and more complex. These intelligent systems have a heterogeneous structure, comprising a mixture of modules such as artificial intelligence (AI) and powertrain control modules that facilitate large-scale numerical calculation and real-time periodic processing functions. Information technology expert Professor Masato Edahiro, from the Graduate School of Informatics at the Nagoya University in Japan, explains that concurrent advances in semiconductor research have led to the miniaturisation of semiconductors, allowing a greater number of processors to be mounted on a single chip, increasing potential processing power. 'In addition to general-purpose processors such as CPUs, a mixture of multiple types of accelerators such as GPGPU and FPGA has evolved, producing a more complex and heterogeneous computer architecture,' he says. Edahiro and his partners have been working on the eMBP, a model-based parallelizer (MBP) that offers a mapping system as an efficient way of automatically generating parallel code for multi- and many-core systems. This ensures that once the hardware description is written, eMBP can bridge the gap between software and hardware to ensure that not only is an efficient ecosystem achieved for hardware vendors, but the need for different software vendors to adapt code for their particular platforms is also eliminated.


2010 ◽  
Vol 50 (1) ◽  
pp. 665
Author(s):  
Ally Oliver

A permit to work (PTW) system is a formal system used to control certain types of work that are identified as potentially hazardous. It is also a means of communication between facility management, plant supervisors and operators, and those who carry out the hazardous work. The essential features of a PTW system are: • Clear identification for who may authorise particular jobs, and who is responsible for specifying the necessary precautions; • Training and instruction in the issue and use of permits; and, • Monitoring and auditing to ensure that the system works as intended. PTW systems are the key to ensuring safe execution of activities at site, yet there are many approaches to how permit systems can, and should, work. Each approach has its own merits and weaknesses. Woodside recognised that, as part of its ongoing program to improve the safety of its workers, there existed significant scope for a new and better work management system. After many years of incremental evolution of the PTW and the fragmentation of the parent system as each facility developed its own variation, it was evident that a completely new system embracing modern technology would provide the best result, while simultaneously standardising Woodside with one common and centralised system. The divergence of the systems over time caused increasing difficulty in managing changes to the PTW system across all sites and in benchmarking to determine best practice. A centralised system would remove accountability from facilities for the development of the business rules, and instead ensure they focussed on compliance with the rules. The new system would adopt key learnings from the industry’s history and address root causes of past incidents. It would also enable the ability to adopt future learnings and become a conduit for rapid integration into the working practices on all sites. The Integrated Safe System of Work (iSSoW) developed by Woodside adopts best practices from permit systems worldwide and combines them with new innovative management features. The system is administered through a simple-to-use computer interface, with incorporation of many of the business rules into the software package. The iSSoW is now in place on all Woodside facilities (platforms, not-normally manned installations, FPSO’s and onshore plants). With nearly 4,000 users, the implementation has required careful coordination, and been supported by a comprehensive training programme. The system has been demonstrated to be both effective and efficient. Effectiveness—the improvement of safety performance—was the primary objective. The system has raised work party hazards awareness, and has resulted in significant improvements in working practices company-wide. Efficiency was a secondary goal, and is made possible through streamlining in the user-interface. The introduction of the new system complements Woodside’s work to develop an improved safety culture, and brings consistency across all sites and all shifts—essential features as our industry struggles to deal with the growing scarcity of skills and experience. The system is now being reviewed by organisations across many industry and service sectors in Australia, and has been implemented in the power industry. This paper discusses the attributes of the system, the many challenges associated with development and large-scale implementation of such a core system, and the additional opportunities the system presents. Using a case study of implementation of iSSoW onto the Woodside operational facilities, it highlights the critical success factors of introducing iSSoW on a company-wide basis.


2020 ◽  
Vol 1 (2) ◽  
pp. 101-123
Author(s):  
Hiroaki Shiokawa ◽  
Yasunori Futamura

This paper addressed the problem of finding clusters included in graph-structured data such as Web graphs, social networks, and others. Graph clustering is one of the fundamental techniques for understanding structures present in the complex graphs such as Web pages, social networks, and others. In the Web and data mining communities, the modularity-based graph clustering algorithm is successfully used in many applications. However, it is difficult for the modularity-based methods to find fine-grained clusters hidden in large-scale graphs; the methods fail to reproduce the ground truth. In this paper, we present a novel modularity-based algorithm, \textit{CAV}, that shows better clustering results than the traditional algorithm. The proposed algorithm employs a cohesiveness-aware vector partitioning into the graph spectral analysis to improve the clustering accuracy. Additionally, this paper also presents a novel efficient algorithm \textit{P-CAV} for further improving the clustering speed of CAV; P-CAV is an extension of CAV that utilizes the thread-based parallelization on a many-core CPU. Our extensive experiments on synthetic and public datasets demonstrate the performance superiority of our approaches over the state-of-the-art approaches.


Author(s):  
Anuja A. Tapase ◽  
Siddheshwar V. Patil ◽  
Dinesh B. Kulkarni

2020 ◽  
Vol 31 (5) ◽  
pp. 997-1008
Author(s):  
Teng Yu ◽  
Wenlai Zhao ◽  
Pan Liu ◽  
Vladimir Janjic ◽  
Xiaohan Yan ◽  
...  
Keyword(s):  

Author(s):  
Masaki Iwasawa ◽  
Daisuke Namekata ◽  
Ryo Sakamoto ◽  
Takashi Nakamura ◽  
Yasuyuki Kimura ◽  
...  

In this paper, we report the implementation and measured performance of our extreme-scale whole planetary ring simulation code on Sunway TaihuLight and two PEZY-SC2 systems: Shoubu System B and Gyoukou. The numerical algorithm is the parallel Barnes-Hut tree algorithm, which has been used in many large-scale astrophysical particle-based simulations. Our implementation is based on our FDPS framework. However, the extremely large numbers of cores of the systems used (10 M on TaihuLight and 16 M on Gyoukou) and their relatively poor memory and network bandwidth pose new challenges. We describe the new algorithms introduced to achieve high efficiency on machines with low memory bandwidth. The measured performance is 47.9, 10.6 PF, and 1.01PF on TaihuLight, Gyoukou and Shoubu System B (efficiency 40%, 23.5% and 35.5%). The current code is developed for the simulation of planetary rings, but most of the new algorithms are useful for other simulations, and are now available in the FDPS framework.


Sign in / Sign up

Export Citation Format

Share Document