Адрес e-mail:

Cеминар "Суперкомпьютерные технологии в науке, образовании и промышленности"

Семинар пройдет в большом конференц-зале НИВЦ МГУ ( http://srcc.msu.ru/nivc/about/arrive.html ) на 3-м этаже. Вход на семинар свободный, регистрироваться и присылать ФИО для пропусков не нужно. На вахте нужно будет сказать, что идете на семинар в большой конференц-зал.

Начало семинара в 16:30. Еще раз обращаем внимание на нестандартный день (среда) и место (НИВЦ МГУ) проведения семинара!

Подробная информация о семинаре - на сайте  http://agora.guru.ru/sct   

"The XPRESS Track to Exascale System Development"

The continuing growth of computing performance enabled through advances in semiconductor (Moore's Law) and communication technology is obstructed by encroaching limitations of power consumption and reliability. Further progress may demand possibly dramatic changes in system structure, both hardware and software, and programming models and methods. While the field of HPC has undergone such paradigm shifts before, the most recent was two decades ago. Now early symptoms of change are manifest in such structures as multicore processors and GPU accelerators as exemplified by Titan, the premier system on the Top-500 list. This presentation will discuss the technical strategy of the DOE XPRESS Project recently initiated under the US DOE X-Stack Program to develop concepts and elements of future programming models and the system software stack. XPRESS is developing two new software components, the HPX runtime system and the LXK operating system to provide a dynamic execution environment capable of unprecedented scalability. These are based on the experimental ParalleX execution model. Also described is a low-level interface, XPI, also being formulated to serve as a source-to-source compiler target from high-level languages to the runtime system software as well as a low-level application programming interface for early experimental code development. The goal of these complementing system software components is to greatly enhance computing node efficiency and scalability while improving both power and reliability of exascale system operation.

 

 

 

 

Если вы заметили в тексте ошибку, выделите её и нажмите Ctrl+Enter.

МФТИ в социальных сетях

soc-vk soc-fb soc-tw soc-li soc-li soc-yt
Яндекс.Метрика