CFD Review  
Serving the CFD Community with News, Articles, and Discussion
 
CFD Review

User Preferences
Site Sponsorship
Headline Feeds
Mobile Edition
Privacy Policy
Terms of Service
twitter

Submit a CFD Story

Site Sponsors
The Choice for CFD Meshing
Azore CFD
CFD Review

Tell a Friend
Help this site to grow by sending a friend an invitation to visit this site.

CFD News by Email
Did you know that you can get today's CFD Review headlines mailed to your inbox? Just log in and select Email Headlines Each Night on your User Preferences page.

 
NUMECA Partners With AllScale to Boost Runtime Efficiency
Posted Mon March 20, 2017 @11:14AM
Print version Email story Tweet story
News AllScale is an innovative programming approach for ExaScale, decoupling program specification parallelism from management tasks during runtime.

As strategic partner within the AllScale project, NUMECA will develop and extend the usage of it's FINE™/Open toolchain to exascale HPC.


Sponsor CFD Review

The AllScale parallel programming model is based on the nested recursive parallelism, focusing on the following developments:
  • Automated applications porting from small- to extreme scale architectures
  • Flexible tuning of program execution to fulfil trade-offs among execution time, energy and resource usage
  • Efficient management of hardware resources and associated parameters (e.g. clock speed)
  • Seamless Integration of resilience management measures to compensate for isolated hardware failures
  • Online performance monitoring and analysis

The above mentioned developments will provide an application independent, reusable codes, taking advantage of combining sophisticated, modular and customizable compiler and runtime system based solutions.

The request from the industry towards higher fidelity CFD and increased simulation reliability is more and more pressing. The use of unsteady LES (Large Eddy Simulation) or DNS (Direct Numerical Simulation) provides an appropriate approach to fulfil such imposed industry demand.

However, when applied to industrial cases on complex geometries, such LES/DNS simulations imply very large meshes, ranging from 1 to 10 billion mesh points, yielding problems of 50-100 billions+ degrees of freedom as well as thousands of unsteady time steps. To fulfil those requirements, the cost, in terms of CPU time, needs to be significantly reduced, and the best way to do it, is to increase the number of processors, towards ExaScale HPC (High Performance Computing).

AllScale will boost development productivity, portability and runtime efficiency. It will reduce energy needs, thus improving the resource efficiency utilisation of small to extreme scale parallel systems. The achieved outcomes are validated with applications from fluid dynamics, environmental hazard and space weather simulations; provided from industrial and academic consortium partners.

AllScale Objectives

  • Single Source to Any Scale
  • Facilitate Recursive Parallelism
  • Multi-Objective Optimization for time, energy and resource usage
  • Unified Runtime System
  • Mitigating Hardware Failures
  • Integrated Online/Offline Monitoring

DON'T MISS A DETAIL: Get continuous updates, directly from the AllScale developers, about the AllScale achievements and software releases by subscribing to the AllScale Twitter and Facebook accounts!

And be the first to be informed as the AllScale consortium is becoming an important player and technology developer in the ExaScale community.

[ Post Comment ]

1st AIAA Geometry and Mesh Generation Workshop | Power Electronics Cooling Calculations at Room Level  >

 

 
CFD Review Login
User name:

Password:

Create an Account

Related Links
  • AllScale
  • Facebook
  • FINE™/Open
  • Numeca
  • Twitter
  • More on News
  • This discussion has been archived. No new comments can be posted.

    Hain't we got all the fools in town on our side? And hain't that a big enough majority in any town? -- Mark Twain, "Huckleberry Finn" All content except comments
    ©2022, Viable Computing.

    [ home | submit story | search | polls | faq | preferences | privacy | terms of service | rss  ]