August 2014 - Allocations-admins - mailman3.cels.anl.gov

New LCRC report submitted for Li_trans_LiOH
by support＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2014/?project=Li_trans_LiOH

1 0

[LCRC Accounts] Yearly Allocation Request for radix
by accounts＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Pavan Balaji Project Name: radix Division: MCS Project title: Scalable Parallel System Software Associated funding: DoE core, SciDAC, NSF, FASTOS Other Systems: breadboard (unlimited) ALCF Mira/Cetus/Tukey (50,000,000 cpu hours) Science: Our mission is to develop the technologies required to dramatically increase the productivity of scientists developing applications for parallel supercomputers. The focus is fourfold: integration of parallel programming tools, reuse of parallel program components, development of scientific computing toolkits and portable libraries, and exploration of requirements of future parallel computers. Project description: The project deals with improvements to various system software techniques used by LCRC including programming models (MPI, PGAS, etc.), fault tolerance, I/O and data analytics, visualization, OS and low-level runtime and various others. Several of these projects are deeply related. For example, programming models and fault tolerance research cannot be done independently. Similarly, several aspects of OS and low-level runtime is closely related to I/O and data analytics and programming models research. Industry partnership: Project URL: http://www.mcs.anl.gov/group/pmrs/ Current FY Hours Used: undetermined amount New FY Requested allocation: 2000000 Q1: 500000 Q2: 500000 Q3: 500000 Q4: 500000 Justification: The project will enable scaling results for new techniques developed in these projects. Most of time will be used for scaling results as well as performance analysis at various scales. Most of the jobs are expected to be small (time-wise) but system scales would vary all the way from 1 node to 256 nodes, but there might be some jobs that would take a few hours of runtime (e.g., when we are doing application evaluations with our techniques). Storage requirements: Thank You, The LCRC Accounts System

1 0

New LCRC report submitted for Li_trans_LiOH
by support＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2014/?project=Li_trans_LiOH

1 0

New LCRC report submitted for Quantum_catalysis
by support＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2014/?project=Quantum_catal…

1 0

[LCRC Accounts] Yearly Allocation Request for ApsRenewalLattice
by accounts＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Michael Borland Project Name: ApsRenewalLattice Division: ASD Project title: Lattice Development and Beam Dynamics for APS Upgrade Associated funding: Department of Energy,Office of Science, Office of Basic Energy Sciences Other Systems: ASD "weed" cluster, operated by ASD/AOP group. weed has 56 Nehelem processors with a lustre filesystem and infiniband network. I typically have access to about 50% of the cluster full time. Science: As part of the APS Upgrade, the accelerator lattice (the location and powering of the magnets) may be completed replaced with a multi-bend achromat (MBA) configuration. This configuration promises dramatically lower emittance and dramatically higher x-ray brightness. If successful, it would make APS the brightest storage ring hard x-ray source in the world. The goal of this project is to design and validate an MBA lattice suitable for incorporation into the APS Upgrade. Project description: APS has developed a method of using multi-objective genetic algorithms (MOGA) for optimization of accelerator lattices. In past years, this method has been applied successfully using blues, fusion, intrepid, and a small APS cluster to development and evaluation of lattices for the APS Upgrade. In the coming year, we will contniue working on a multi-bend achromat (MBA) lattice design that promises to push APS x-ray brightness to world-leading levels. Significant progress has already been made and the results are used in on-going discussions of the project with DOE. Our approach is to create a series of lattices that gradually push performance up while simultaneously incorporating on-going constraints derived from on-going interactions with magnet and vacuum system designers. As such, the process is naturally iterative, and we have completed a series of iterations to date to conform to engineering limits. Further iterations are on-going, with interest in reducing cost also being relevant. In addition to MOGA, we will perform ensemble evaluation to assess the robustness of solutions and evaluate effects of insertion devices on the newly-developed solutions. A major review is anticipated in the second quarter of FY15. The codes being used for MOGA and ensemble evaluation have all been used in past years on fusion. These include Pelegant, a parallel accelerator simulation code used primarily for particle tracking and acceptance determination, as well as a Tcl script used for genetic optimization. The simulations are close to embarrassingly parallel and thus scale very well. In late FY14, we began a new effort to model the effects of the higher harmonic cavity (HHC), which plays a vital role in increasing the beam lifetime and reducing emittance blow-up. The simulations involve tracking beams consisting of 48 or more bunches of 100,000 or more particles each, and computing the collective electromagnetic interaction of these particles with the vacuum chamber and rf cavities. The code was optimized to improve efficiency, with 70% efficiency obtained on 256 cores for a "small" problem consisting of 48 bunches with 100,000 particles each. Efficiency increases for larger problems as computation increasingly dominates inter-process communication. These efforts will continue into FY15 and require significant disk space. Industry partnership: Project URL: Current FY Hours Used: undetermined amount New FY Requested allocation: 10000000 Q1: 3000000 Q2: 3000000 Q3: 2000000 Q4: 2000000 Justification: The code is already running efficiently on Blues. Scaling is very good for MOGA and ensemble evaluation because the code is close to embarrassingly parallel. The MOGA script runs many simultaneous, unconnected jobs, each using 2 nodes. Because each job uses only 32 cores, scaling is not relevant. Jobs are sized to take advantage of the number of cores provided, e.g., by tracking n*32 particles so that each core has an approximately equal workload. Efficiency for collective effect tracking was found to be good (>70%) for up to 256 cores for a small problem, and increases with problem size. Storage requirements: 3 TB, needed for storage of large data files generated by collective effects simulation. Thank You, The LCRC Accounts System

1 0

[LCRC Accounts] Yearly Allocation Request for Nek5000
by accounts＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Shashi Aithal Project Name: Nek5000 Division: MCS Project title: Scalable CFD Algorithms Associated funding: Applied Math Research Program, Nuclear Energy Advanced Modeling and Simulation Other Systems: ALCF Intrepid (BG/P), 128-core Linux cluster, Cosmea Science: Our mission is to develop and deploy scalable algorithms for simulation-based science applications that rely on numerical solution of partial differential equations. Our primary vehicle for algorithm development is the fluid/thermal simulation code, Nek5000, and its off-shoots, NekCEM (for computational electromagnetics) and NekLBM (for lattice Boltzmann methods). There are approximately 30 research groups worldwide currently using Nek5000. We use Nek5000---and the science applications brought to us through our user community---as vehicles for the development and test of new numerical algorithms that include stabilized high-order methods, scalable linear solvers, advanced timestepping methods, and so forth. We emphasize that these algorithms cannot be developed in isolation from applications, as many of the relevant issues do not arise at small scale or in simple model problems devoid of complex physics and/or geometry. Thus, to address algorithmic issues at the forefront of science-based simulation we require a code that is capable of solving problems _at_scale_ and that is sufficiently flexible to allow adaptation to new physics. Project 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 interactions 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 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 Blues 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. This year we intend to test various large-element cases which have been run with LES models using 3 different uRANS models that have been implemented in Nek5000. These runs typically need about 100-150K core-hours per case. We aim to study 4-6 such cases with the allocation this year. Industry partnership: Project URL: http://www.mcs.anl.gov/~fischer/sem1b Current FY Hours Used: undetermined amount New FY Requested allocation: 480000 Q1: 120000 Q2: 120000 Q3: 120000 Q4: 120000 Justification: Storage requirements: Thank You, The LCRC Accounts System

1 0

[LCRC Accounts] Yearly Allocation Request for PETSc-Foam
by accounts＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Shashi Aithal Project Name: PETSc-Foam Division: CELS Project title: Investigation of robust, scalable solvers for OpenFoam Associated funding: PETSc Other Systems: Science: OpenFOAM (OF) is a multi-physics open-source solver code capable of modeling a wide-range of problems of interest to various divisions at ANL. One of the strengths of OpenFOAM is that new solvers and utilities can be created by its users. In this project, we intend to investigate fast, robust, scalable solvers for problems involving moving boundaries and combustion. The goal is to study various methodologies for load-balancing and solvers for large-scale simulations. Project description: OF is parallel, finite-volume CFD code which will be used for this project. Scaling studies on dsmcFoam and dieselFoam have shown good scalability. Small DSMC cases (1 million cells) have shown good scalability up to 192 cores. DieselFoam has shown good scalability up to 512 cores with about 500K cells. We intend to use OpenFoam 2.2-x as the computational tool. Various algorithms for efficient domain decomposition and solution methods for combusting mixtures will be investigated for optimal performance and robustness. New mesh-motion strategies, turbulence models and higher resolution studies are planned. It is seen that about 50000 cells/core is required for good linear scaling with OF with the problems we are working on. Typical simulations with about 1.5 million cells take about 1000 core hours. This year we plan to run some larger simulations (about 15 million cells) along with smaller problems to test various mesh-motion strategies and LES models. We anticipate about 15-20 smaller runs (1000 core-hours) and about 1-2 10,000 core-hour jobs/month, totalling about 40,000 core-hours/month or 480000 core-hours/year. Industry partnership: Project URL: Current FY Hours Used: undetermined amount New FY Requested allocation: 480000 Q1: 120000 Q2: 120000 Q3: 120000 Q4: 120000 Justification: Storage requirements: Thank You, The LCRC Accounts System

1 0

[LCRC Accounts] Yearly Allocation Request for PETSc-Foam
by accounts＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Shashi Aithal Project Name: PETSc-Foam Division: CELS Project title: Investigation of robust, scalable solvers for OpenFoam Associated funding: PETSc Other Systems: Science: OpenFOAM (OF) is a multi-physics open-source solver code capable of modeling a wide-range of problems of interest to various divisions at ANL. One of the strengths of OpenFOAM is that new solvers and utilities can be created by its users. In this project, we intend to investigate fast, robust, scalable solvers for problems involving moving boundaries and combustion. The goal is to study various methodologies for load-balancing and solvers for large-scale simulations. Project description: OF is parallel, finite-volume CFD code which will be used for this project. Scaling studies on dsmcFoam and dieselFoam have shown good scalability. Small DSMC cases (1 million cells) have shown good scalability up to 192 cores. DieselFoam has shown good scalability up to 512 cores with about 500K cells. We intend to use OpenFoam 2.2-x as the computational tool. Various algorithms for efficient domain decomposition and solution methods for combusting mixtures will be investigated for optimal performance and robustness. New mesh-motion strategies, turbulence models and higher resolution studies are planned. It is seen that about 50000 cells/core is required for good linear scaling with OF with the problems we are working on. Typical simulations with about 1.5 million cells take about 1000 core hours. This year we plan to run some larger simulations (about 15 million cells) along with smaller problems to test various mesh-motion strategies and LES models. We anticipate about 15-20 smaller runs (1000 core-hours) and about 1-2 10,000 core-hour jobs/month, totalling about 40,000 core-hours/month or 480000 core-hours/year. Industry partnership: Project URL: Current FY Hours Used: undetermined amount New FY Requested allocation: 480000 Q1: 120000 Q2: 120000 Q3: 120000 Q4: 120000 Justification: Storage requirements: Thank You, The LCRC Accounts System

1 0

New LCRC report submitted for Li-MetalOxide
by support＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2014/?project=Li-MetalOxide

1 0

New LCRC report submitted for Li-MetalOxide
by support＠lcrc.anl.gov 29 Aug '14

29 Aug '14

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2014/?project=Li-MetalOxide

1 0