Advanced Support Program (ASP)

Home > User Info > Advanced Support Program

The Advanced Support Program (ASP) offers Advanced Support for TeraGrid Applications (ASTA), as well as referral to RP-specific activities. Advanced support provides collaboration between experts on the TeraGrid staff and users of TeraGrid computational and data resources to identify and pursue opportunities to improve their codes. Leveraging deep understanding of resource capabilities and expertise in the computational challenges of engineering and the domain sciences, TeraGrid staff can provide specialized assistance to accelerate progress and to reach new goals.

Specifically, ASP associates one or more staff members at an agreed-upon level of effort with a research project for a sustained period of collaboration. This can include support to help gateway developers utilize TeraGrid resources. It can be requested with all types of resource allocations.

How to Apply

Using the the Partnerships Online Proposal System (POPS), submit a one-page justification for advanced support as an appendix to the standard proposal documents.

How to Write a Request

Suggested content for a one-page request:

1. What kind of computational assistance does the principal investigator (PI) expect from the TeraGrid staff (e.g. assistance in problem size scaling or performance improvement at fixed scale)? Possible areas for advanced support include the following:
   • Algorithmic change
   • Code modification
   • Computer science optimization (single processor performance improvement, parallel scaling improvement, parallel I/O performance improvement, memory usage improvement, benchmarking)
   • Development of parallel code from serial code/algorithm
   • Incorporation or implementation of new or advanced numerical methods and/or math libraries
   • Incorporation of new data management or storage
   • New grid computing work
   • New workflow implementation
   • Incorporation of new visualization methods
   • Innovative scheduling implementation
   • Integration of TeraGrid resources into a portal or Science Gateway (see the Science Gateways page for more about gateways)
2. What are the criteria for successful completion?
3. How can the work impact not only the PI, but a wider or new community of computational scientists?
   • Will the work result in development of a widely used library?
   • Does the work involve community codes or software?
   • Is the PI from a non-traditional HPC community (e.g., social science)?
4. What is the level of commitment from the PI's research team?
5. Who from the PI's team (graduate student, postdoc) will collaborate with the TeraGrid staff and be the point of contact?

Selection Process

All requests will be reviewed by the respective Allocation Committee. Based on reviewer recommendations, the TeraGrid user services staff will follow up with you to discuss a collaborative work plan. Requests for advanced support will not affect compute or data requests.

Advanced Support In Action

NanoHub Simulations

PI: Klimeck (Purdue, Materials Research)

PSC scientists Brown and Wang helped the NanoHub team to get their Nemo1D and Nemo3D codes running on Cray XT3. Indiana experts also enabled Klimeck to get excellent scaling of Nemo codes on Big Red. Working with PSC grid computing systems engineer Shelmire and experts at Purdue and the University of Wisconsin, the team was able to set up Science Gateway runs with the safeguards provided by the community shell system. As a result, these codes are now highly scalable, and available to the NanoHub user community of over 5,900 researchers, educators and students. Nemo3D currently runs in production mode.

Nemo Research

nanoHUB Portal

SCEC PetaShake: An Earthquake System Science Approach to Physics-Based Seismic Hazard Research

PI: Jordan (University of Southern California, Earth Sciences)

As a part of SCEC's general goal to develop a comprehensive understanding of earthquakes in Southern California, SDSC staff worked closely with SCEC researchers to achieve the largest and most realistic simulations ever of a magnitude 7.8 earthquake on a 230-km section of the southern San Andreas Fault.

These simulations captured ground motion frequencies as high as 1Hz. Using a close grid spacing of just 100 meters resulted in 14.4 billion grid points for the 600 by 300 km area, eight times more than in previous simulations. The simulation produced ~3TB of data, including part archived in the SCEC Digital Library at SDSC. These simulations, part of the new USGS-initiated Multi-Hazard Demonstration Project, contribute to understanding of earthquake processes, and the improved understanding may help save money and lives through more effective earthquake hazard mitigation.

The simulations were carried out on many of the TeraGrid machines such as SDSC's DataStar (both P655 and P690 nodes), SDSC's BlueGene Data, SDSC's and NCSA's IA-64 clusters, TACC's Lonestar, PSC's BigBen, etc. GPFS-WAN was used for many of these simulations where output data, produced from simulations running on one TeraGrid machine, was post-processed using a different TeraGrid machine. SDSC staff implemented new dynamic rupture features and checkpointing capabilities in the AWM-Olsen code, incorporated new MPI I/O routines, performed porting, benchmarking, and optimization of SCEC codes on TeraGrid machines, and enabled scaling of the SCEC AWM-Olsen code on 40,000 cores of IBM BG/L machine at the IBM Watson Lab.

SDSC-produced visualizations of the massive simulation data collections have proven vital tools for helping scientists understand the "big picture" of their results, in guiding the simulations, and in explaining the results to students and the public. The visualizations were performed on SDSC's DataStar resources, and have appeared in print and television news broadcasts. SDSC staff created the SCEC Map Service Portal, an intuitive Web interface that integrates over 10TB of normally hard-to-access earthquake simulation data with virtual globes. The system offers a range of visualization products for scientific investigation, collaboration, and data dissemination. SDSC staff also contributed to SCEC's LRAC proposal.

News about PetaShake

SCEC images