Yes, 150K seems right for a start. Ray On 4/12/10 7:15 PM, "John Valdes" <[email protected]> wrote:
Any decision on this project? The request came in the day after the allocation meeting. If we approve, we need to decide:
Amount: Contact: Cat/Subcat: Priority:
The request sounds good to me; I would say 150K, either Ray or Steve Pieper as POC, P:Phy, and 3.
John
On Fri, Apr 02, 2010 at 04:09:21PM -0500, [email protected] wrote:
Hello,
A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee.
Applicant's name: Joe Bernstein Applicant's institution: ANL Applicant's division: HEP Project Name: Cosmology Project title: Harnessing Computational Cosmology Simulations to Investigate Cosmic Evolution Associated funding: None external to ANL sources Other Systems: ANL Blue Gene/P (renewal in preparation for ~300000 CPU-hrs over 4 months) Science: The evidence for dark energy represents one of the greatest mysteries of modern science. The proposed research will probe the implications of dark energy via analysis of computational cosmology simulations. Several independent observational methods provide evidence pointing to the conclusion that 95% of the current energy density of the Universe exists as part of the "dark sector" that is different from normal, i.e., "baryonic," matter. In particular, the dark sector is thought to be comprised of 22% "dark matter," which interacts with baryonic matter ostensibly only through gravity, and 73% "dark energy," which experiences gravity as a repulsive force. The nature of both dark sector components are essentially complete unknowns. It is presently an exciting time to be involved in cosmology because planned astronomical surveys will effectively result in dark sector probes becoming systematics-limited, making numerical simulations crucial to placing precision constraints on the dark sector. In addition, particle physics theorists predict that the Large Hadron Collider (LHC) and direct dark matter experiments may soon discover a dark matter candidate particle. ANL HEP staff and collaborating U. Illinois faculty are members of the Dark Energy Survey, which will provide the next generation of dark energy and (indirect) dark matter data. Local scientific simulation expertise and high-performance computing further make ANL ideally-suited to pursue answers to questions about the dark sector. The focus of the proposed research is the production and interpretation of computational cosmology simulations in order to elucidate the nature of the dark sector. Project description: The "Cosmology" project on fusion.lcrc.anl.gov (hereafter, "fusion") is in support of the one of the main research efforts of the PI as an ANL Computational Postdoctoral Fellow. The number of project members will be 3-5. We will make use of the requested number of CPU-hrs (275K) in efforts to generate initial conditions and verify scaling results for cosmology simulations run on the ANL Blue Gene/P ("BG/P"). Toward this end, we will harness piecewise-parabolic method (PPM) hydrodynamics and N-body gravity solvers, MPI for inter-processor communication, and HDF5 to implement parallel I/O. We will utilize both the C and FORTRAN90 programming languages.
Paul Ricker (U. Illinois) is a computational cosmology expert working with Boyana Norris (MCS) on BG/P cosmology studies. The PI of this proposal recently transitioned from a classification as an ANL Postdoctoral Appointee to an ANL Computational Postdoctoral Fellow. His new duties include collaborating on these efforts with the goal of submitting a 2011 DOE INCITE proposal. The ultimate target of the BG/P simulation effort is runs requiring PPM+N-body simulations with a 1 Gpc (1 pc = 3.09x10^16 m) box size, 2048^3 grid cells, and 10^11 dark matter particles. We anticipate employing fusion to generate initial conditions for a suite of a few dozen BG/P simulations. Generating such initial conditions is memory intensive and cannot currently be done on the BG/P. These requirements also push the limits of existing codes. Scaling the cosmology runs to the target level represents a marked increase relative to machines used to date. Verifying scaling results, on a traditional Linux cluster such as fusion, will constitute an important part of our BG/P studies. We are highly confident of fusion's suitability for the proposed simulations because such runs have been shown to scale to upwards of 4K CPUs on other Linux clusters.
The scaling studies will require approximately 10 runs each for strong and weak scaling for a total of ~20 runs. Based on Prof. Ricker's experience with PPM+N-body simulations performed at Oak Ridge National Laboratory, we estimate that a run with 2048^3 grid cells will cost approximately 400 CPU-hrs per timestep and require ~30 time steps for a total of 12K CPU-hrs. Assuming reasonable scaling, this implies a total cost of approximately 240K CPU-hrs for the scaling study. Generating initial conditions will require a significantly smaller cost by comparison. Thus, an allocation of 275K CPU-hrs will enable significant research progress. Project URL: Requested allocation: 275000 Justification:
The requester has used 0 hours of their initial startup project.
In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see:
https://wiki.lcrc.anl.gov/wiki/Processes/Categories
Once the Allocation committee has approved the project, please go to the Project Management page to create it:
https://accounts.lcrc.anl.gov/projects.php
Thank You, The LCRC Accounts System _______________________________________________ allocations-admins mailing list [email protected] https://lists.lcrc.anl.gov/mailman/listinfo/allocations-admins
allocations-admins mailing list [email protected] https://lists.lcrc.anl.gov/mailman/listinfo/allocations-admins
----------------------------------------- Ray Bair Computing, Environment, and Life Sciences Argonne National Laboratory and the University of Chicago Building 240, Room 4126 9700 South Cass Avenue Argonne, IL 60439 email: rbair(at)anl.gov Phone: (630)252-5751