Hello, A change in allocation has been requested: Requester: heinonen (Olle Heinonen) Project: Meso Title: Mesoscale computational materials science Description: In the first project, mesoscale coupled elastic systems, we are developing new methods and codes (within the MOOSE framework) as well as using the FEniCS code to more finite mesoscale systems in which elastic degrees of freedom are coupled to others, such as concentration or bound and free electric charges. Both the MOOSE and FEniCS codes use MPI and threading to parallelize and both have exhibited strong scaling up towards 100k cores. In these simulations we may use up to 512 cores for production runs. These codes both use PETSc, and MOOSE also uses LibMesh. We compile our own versions of these libraries. In the second project, oxygen defects, we are using a variety of first-principle electronic structure codes, principally SIESTA and FHI-aims, to calculate electronic properties and oxygen diffusion barriers (using nudged elastic band methods). For the heterostructures, we use the non-equilibrium transport code SMEAGOL to calculate the electronic transport. Our developer SIESTA/SMEAGOL version uses threading and MPI for parallelization and we have demonstrated scaling of up to ~64k cores. The FHI-aims code uses MPI only and is a very efficient codes. On blues, we expect to run up to 512 cores for the calculations; for much bigger systems, we are using Mira. We also have a collaboration with Subramanian Sankanayaranyan (CNM) who does MD simulations of interfaces that we feed into the electronic structure calculations. The third project, magnetization behavior, uses a home-grown code that uses OpenMP threading only, so this one runs only on up to 16 cores at a time per job but typically runs at close to 95% efficiency on 16 cores. The number of project members is five to six. Current: undetermined amount Justification: The main reason we need a larger (500k) allocation is that we have several rather heavy users in the project. The codes we use are FEnICs (a PETSc-based FEM code that scales very well), a development version of SIESTA/smeagol with proven scaling of up to ~16k cores, FHI-aims which is very efficient with its ELPA libarary, and a very efficient single-node multi-thread micromagnetic codes. Requested: 500000 A specific reason has been given: Because of heavy use we have already exhausted an initial 210k core-hour allocation for FY2014. We run efficient codes but have 3 - 4 heavy users and need a larger allocation. This needs to be approved and the final allocation amount decided upon. Thank You, The LCRC Accounts System