High Performance Computing Working Group

From IMAG_Wiki

Contents 1 High Performance Computing Working Group 1.1 Goals and Objectives: 1.2 Participation in Working Group: 1.3 Upcoming Presentations 1.4 Past Presentations 1.5 Relevant Infrastructure: 1.6 Challenges and Opportunities: 1.7 References: 1.8 List of Participants:

High Performance Computing Working Group

Working Group leads: Pratul Agarwal, Yufang Jin

Goals and Objectives:

The goal of this working group is to initiate discussion, foster collaboration and data sharing including (but not limited) to:

1. Computational resources including high performance computing.
2. Multi-scale modeling and simulations on emerging computer architectures (e.g. GPUs).
3. Computational algorithms, libraries, tool-kits and software.
4. Infrastructure and strategies to handle big-data problems.

Participation in Working Group:

Evolved from Working Group 5.

Currently, feed-back is being sought from various members of the community regarding the need and availability of the computing resources.

Please send your responses by e-mail to the working group leads. Single word responses (yes/no) or responses to partial list are fully acceptable.

Topic: Computing Hardware Resources

1. Some researchers already have access to what they need (computers, university clusters, storage space) but not everybody. Do you need access to computing resources that are not currently available to your multi-scale modeling needs?
2. At last year meeting, questions was raised at the meeting about access to supercomputers. Would it be beneficial to coordinate access for MSM Consortium to avail group time on various hardware as needed? Particularly resources at the DOE and NSF centers?
3. GPUs and other non-standard computer hardware is being used to solve the multi-scale problems. Would you benefits (and also willing) share your experiences and code to benefit members who do not have expertise or human resources to develop the code on their own?
4. Would cloud computing benefit your investigations? Would you like to access cloud environment such Amazon EC2?

Topic: Computational Algorithms

1. Do you see the benefits of sharing tool-kits, code, data (?) associated with multi-scale modeling on HPC resources? Would you be willing to share your algorithms and code?
2. Do you have a wish-list of optimized libraries on the HPC environment, which will improve your productivity while using HPC resources? Could you specify which ones?
3. If someone wants to seek advice on solving computational problem, do you see the benefits of having a group/forum discussion? These could be simple problems such as what compiler or hardware or libraries to use to solve the problems or something more complex.

Topic: Collaborate, define the needs of the community

1. It is discussed that 20% of all computing time on NSF supercomputers and 10% of time on DOE supercomputers is used by NIH funded PIs. However, each PI has to write separate proposals (peer-reviewed) and only big users are typically able to get access. Would it benefit to coordinate the access to supercomputers, especially for small to medium sized or beginning users?
2. Do you see the benefit of organizing focused work-shops to discuss and exchange ideas related to HPC for multi-scale modeling? Either through sessions at existing conferences and meetings? Or perhaps organize new workshop?

Upcoming Presentations

Past Presentations

Relevant Infrastructure:

Challenges and Opportunities:

References:

List of Participants: