I suggest that we approve this request Ray On 5/2/15, 1:35 PM, "[email protected]" <[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: Misun Min Applicant's institution: ANL Applicant's division: MCS Project Name: NEKCEM Project title: NEKCEM Associated funding: DOE ASCR, DOE CESAR Other Systems: OLCF Titan, ALCF BG/P, ALCF BG/Q Vesta Science: NEKCEM is a high-order spectral-element (SE) discretization basis code for solving nanoscale solar cells, plasmonic nanoparticle devices, and quantum electrodynamic systems, with scalable CPU performance up to 262,144 CPU cores and scalable GPU performance up to 16,384 GPUs.

Project description: NEKCEM is a high-order spectral-element (SE) discretization basis code for solving nanoscale solar cells, plasmonic nanoparticle devices, and quantum electrodynamic systems, with scalable CPU performance up to 262,144 CPU cores and scalable GPU performance up to 16,384 GPUs.

We have been developing a newly tuned gather-scatter kernel of NekCEM that can be portable for GPU runs at large scale and recently have completed scalability tests on OLCF Titan using 16,384 GPUs with GPUDirect and GPUDirect-disabled communication, compared to CPU-only runs on Titan using the same number of nodes (262,144 MPI ranks). We are now ready to perform intensive tests of the newly tuned communication kernel on different systems for CPU runs.

Especially, NekCEM's quantum mechanical solver is a newly-developed solver based on density matrix formalism that is designed for solving multi-state multi-quantum dot system models at exascale. Our algorithms use a compressed-sparse-row format allowing any quantum systems to be defined in a reduced dense form for the Hamiltonian, work scaling as O(N^2). Using this solver, Argonne scientists and university collaborators were able to explore a new physical scheme for creating significant bipartite entanglements at arbitrary time using repeated pulse in the hybrid quantum/plasmonic nanoparticle systems, not been studied before.

We also developed a spectral-element (SE) based drift-diffusion solver in time domain. Efficient timestepping algorithms are developed by treating nonlinear drift term explicitly and diffusion term implicitly, including projection and SE multigrid preconditioners. Full 3D simulations were performed for a potassium ion channel in biological membrane, validating other groups’ works that are based on dimension-reduction approach.

Industry partnership: We request 50,000 allocation hours.

Using LCRC computational resources, we'll test our newly tuned gather-scatter communication kernel for scalability on CPUs, quantum density matrix solver for large-scale quantum systems, drift-diffusion solver for nanowire solar cells.

Project URL: http://nekcem.mcs.anl.gov/ Requested allocation: 60000 Q1: 0 Q2: 0 Q3: 20000 Q4: 40000 Justification: Storage requirements:

The requester has used undetermined amount 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