Hello, A change in allocation has been requested: Requester: aithal (Shashi Aithal) Project: Nek5000 Title: Scalable CFD Algorithms Description: Nek5000 is a petascale fluid/thermal simulation code used for a host of applications including reactor thermal hydraulics, combustion, MHD, oceanography, spatiotemporal chaos, and vascular flow modeling. Nek5000 is a grid-based PDE (partial differential equation) solver in three space dimensions and time. The code has long been at the forefront of high-order methods and scalable algorithms. It was one of the first comprehensive software tools developed for distributed memory parallel platforms. It pioneered the development of high-order discretizations that, because of their ability to accurately capture interacations over a broad range of scales, are now deemed an essential ingredient of petascale simulation science and that are being adapted across virtually all disciplines (e.g., fusion, aeronautics, biofluids and geophysics). We have demonstrated > 80% parallel efficiency on BG/P (strong scaling) out to P=131,072 processors, using only 7300 points/processor at the largest scale. We note that we have access to a dedicated 128-core linux cluster (under a nuclear engineering project) and to resources on BG/P through INCITE and discretionary allocations. This request is to address mid-range capabilities -- particularly for those applications that would not fit on the 128-core cluster as discussed below. Our goal is to use Fusion to test the performance of mid-size problems (with 10,000 to 100000) elements with various compilers and compiler options. We also intend to set up a suite of large problems to ensure that various builds are working correctly. We also intend to run production cases for problems in nuclear engineering and conjugate heat-transfer. Current: undetermined amount Justification: Requested: 200000 A specific reason has been given: The extra time is required to test various mesh and parameter configurations for an LES calculation of the turbulent flow around a nuclear fuel rod at Reynolds number of 65,000. This calculation is important for the correct calculation of a problem with 37 pins in a hexagonal bundle, which will be performed on BG/P. The allocation request is for 200k cpu hours. Each calculation can run on as few as 64 cores, with each run lasting about 100 hours. About 30 such runs may be needed; however, fewer is definitely possible. This needs to be approved and the final allocation amount decided upon. Thank You, The LCRC Accounts System