Pete Beckman on Mira and Exascale
Argonne's Pete Beckman, director of the Exascale Technology Computing Institute (ETCi), talks about the IBM Bluegene Q supercomputer and the future of computing and Exascale technology.
SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Argonne's Pete Beckman, director of the Exascale Technology Computing Institute (ETCi), talks about the IBM Bluegene Q supercomputer and the future of computing and Exascale technology.
The Red Sky/Red Mesa supercomputing platform dramatically reduces the time required to simulate complex fuel models, from 4-6 months to just 4 weeks, allowing researchers to accelerate the pace at which they can address these complex problems. Its speed also reduces the need for laboratory and field testing, allowing for energy reduction far beyond data center walls.
Salil Mahajan applies ORNL's supercomputing capabilities to modeling the Earth's climate.
A new 3-D supercomputer model presents a new theory of how magnetic reconnection works in high-temperature plasmas. This Los Alamos National Laboratory research supports an upcoming NASA mission to study Earth's magnetosphere in greater detail than ever.
Modeling and simulation completed on ORNL supercomputers (Jaguar and Titan) improved highway fuel efficiency, taking a concept from design to market in 18 months.
Data is pouring into the hands of cancer researchers, thanks to improvements in imaging, models and understanding of genetics. But they’ll need a lot of help—and some powerful supercomputers—to translate this data into better, more personalized treatment for cancer patients. A new initiative called the Joint Design of Advanced Computing Solutions for Cancer, which taps four different national laboratories, is poised to help.
Lawrence Livermore’s supercomputers are playing a crucial role in advancing cancer research and treatment.
With Sequoia at Lawrence Livermore National Laboratory, researchers explore grand challenging problems and are generating results at scales never before achieved. Sequoia is the first computer to have more than one million processors and is one of the fastest supercomputers in the world.
A team of physicists and astrophysicists at Los Alamos National Laboratory, in collaboration with leading universities around the country, are using the Laboratory’s supercomputers to simulate the Big Bang nucleosynthesis and the early universe to unprecedented precision. These researchers developed a code, called BURST that describes the universe from a time of a few seconds after the Big Bang to several hundred thousand years later. BURST allows physicists to study the microscopic, quantum nature of fundamental particles — like nuclei and the ghostly, weakly interacting neutrinos — by simulating the universe at its largest, cosmological scale. BURST simultaneously describes all the particles present in the early universe as they develop, tracking their evolution, particularly the amounts of light nuclei fused in the cosmic soup.
The Hanford Site was quickly established to help end World War II, making history for producing the plutonium used in the world’s first nuclear weapons. Throughout the Cold War years, Hanford employees produced plutonium for most of the more than 60,000 weapons in the U.S. nuclear arsenal stockpile. Today, the once highly active nuclear reactors are shut down. And the mission at Hanford turned full-circle as scientists, engineers and specialists work to clean up our nation’s most contaminated nuclear site. PNNL Computational Geophysicist Tim Johnson is helping decision-makers understand the complexity and breadth of the contamination in soils at Hanford. Tim and others are applying remote, high-resolution geophysical imaging to determine the extent of contamination in the soil below the surface and understand the processes controlling its movement. They also provide real-time imaging of remediation processes that are working to limit the movement of contaminants below the surface and toward water resources. Geophysical imaging simply means that PNNL scientists are combining the techniques of geology, physics, mathematics and chemistry with supercomputer modeling to create three-dimensional images of the waste and its movement. These real-time, remote images are essential in reducing the uncertainty associated with cleanup costs and remediation technologies.
In this first episode, we sit down (on a chairlift) with Computer Scientist Geoff Fairchild to talk about how his team is using Los Alamos National Laboratory’s supercomputing capabilities to data mine social media and reveal trends relating to health and safety.
As modern supercomputers continue to be increasingly heterogeneous with diverse computational accelerators, software becomes a critical aspect necessary to fully exploit these accelerators while not increasing the effort and time required to make variable scientific discovery in the face of complicated programming environments introduced by these accelerators. This paper presents PCS, a productive computational science platform for cluster-scale heterogeneous computing. The PCS platform by design follows a basic principle that, multiple programming models in a unified programming environment, is essential to facilitate large-scale, accelerator-aware, heterogeneous computing for next- generation scientific applications. We propose a new scheme — Edge Neural Network — for managing regularity in large dataflow graphs using higher-order bulk synchronous operators, providing a foundation for formalizing computational models for cluster-wide dataflow programming of heterogeneous clusters.
As modern supercomputers continue to be increasingly heterogeneous with diverse computational accelerators, software becomes a critical aspect necessary to fully exploit these accelerators while not increasing the effort and time required to make variable scientific discovery in the face of complicated programming environments introduced by these accelerators. This paper presents PCS, a productive computational science platform for cluster-scale heterogeneous computing. The PCS platform by design follows a basic principle that, multiple programming models in a unified programming environment, is essential to facilitate large-scale, accelerator-aware, heterogeneous computing for next- generation scientific applications. We propose a new scheme - Edge Neural Network - for managing regularity in large dataflow graphs using higher-order bulk synchronous operators, providing a foundation for formalizing computational models for cluster-wide dataflow programming of heterogeneous clusters.
Supercomputer simulations of plasma-based accelerators typically use a “moving window” to restrict the simulation area to a “small” region of interest that encompasses the laser beam and the portion of wakefield that accelerates the electron beam. In this simulation, a laser beam (red and blue disks) propagating through an under-dense plasma displaces electrons, creating a wake that supports very high electric fields, or “wakefield” (pale blue and yellow), that can be used to accelerate a short electron beam (white) to high-energy over a very short distance. While the simulation box is spatially much smaller, the number of time steps that are required to simulate the crossing of the laser through the plasma is still very large, typically over a million. (Credit: Jean-Luc Vay/Berkeley Lab)
Watch this animation to see how the ATHENA project is applying the power of U.S. Department of Energy supercomputers to inform efficiency improvements at transportation hubs like Dallas-Fort Worth Airport.
This talk provides an overview of the FY22 allocation process for obtaining computing time on the Eagle supercomputer. The overall process is outlined, with specific instructions on completing the request.
In order to run computer simulations efficiently on massively parallel computers with hundreds of thousands or millions of processors, care must be taken that the calculation is load balanced across the processors. Examining the workload of every processor leads to an unscalable algorithm, with run time at least as large as O(N), where N is the number of processors. We present a scalable load balancing algorithm, with run time 0(log(N)), that involves iterated processor-pair-wise balancing steps, ultimately leading to a globally balanced workload. We demonstrate scalability of the algorithm up to 2 million processors on the Sequoia supercomputer at Lawrence Livermore National Laboratory. (authors)
Over the past 15 years the architecture of high-performance computing systems has become increasingly heterogeneous with individual computing cores being subdivided into nodes and even auxiliary co-processing units such as graphics processing units or the Intel many integrated core co-processors. In fact, future computer platforms is moving toward a larger number of nodes and cores with lower memory available per node. Today, several production Monte Carlo codes use some types of hybrid parallelism strategy that makes use of shared-memory parallelism within a computing node and message-passing parallelism between nodes, which enables the codes to utilize available computing resources as much as possible, both among and within computing nodes on a cluster. The so-called hybrid parallelization is also possible with Reactor Monte Carlo Code RMC by activating both MPI and OpenMP parallel algorithms together. On HPC' which typically consists of several or several-ten thousand nodes equipping 16-32 computing intra-cores per node, simulation of RMC can be carried out in the hybrid parallelization with the MPI inter-node parallelization and the OpenMP intra-node parallelization. RMC has adopted a hybrid message-passing and shared-memory approach to parallelism, which enables the code to utilize available computing resources, both among and within computing nodes on a cluster. Shared-memory parallelism is especially valuable for large-model simulations because it enables multiple compute cores within a node to share common model information such as the geometry description, material, cross section and tally data, thus reducing the memory usage per processor. The calculation on the Tianhe-2 supercomputer took 6144 cores, resulting in a rate of 173,314,000 neutrons/minute for BEAVRS benchmark. While pin-wise full core H-M benchmark burnup calculations with millions of depletion regions was selected using MPI/OpenMP parallelism, the speed of 6,144 cores reach 1,843,200 burnable regions per minute. Keeping up with the continuing trend for increases in the number of computing resources within modern HPC systems is a significant challenge for teams that develop and maintain high-performance MC radiation transport solvers. order to address this challenge, RMC has adopted a hybrid message-passing and shared-memory approach to parallelism, which enables the code to utilize available computing resources, both among and within computing nodes on a cluster. Shared-memory parallelism is especially valuable for large-model simulations because it enables multiple compute cores within a node to share common model information such as the geometry description, material and composition definitions, and cross section data, thus reducing the memory usage per processor. The parallel performance has been studied by using two complex and representative benchmarks, H-M and BEAVRS benchmark. All simulations were run on Intel Xeon E5-2692 12 core, 2.2 GHz processors at MilkyWay-2 Super Computer with up to 12,288 available computing cores. While employing MPI parallelism, excellent parallel efficiency is achieved even above 12,000 processors and RMC reaches the parallel efficiency of 96% on 12,288 cores for the pin-by-pin computation of the BEAVRS benchmark. While BEAVRS criticality benchmark was selected using MPI/OpenMP parallelism, the speed of 6,144 cores reach 173,314,000 particles per minute. While pin-wise full core H-M benchmark burnup calculations with millions of depletion regions was selected using MPI/OpenMP parallelism, the speed of 6,144 cores reach 1,843,200 burnable regions per minute. Results from these studies demonstrate that RMC, using the hybrid parallel model described in this paper, is able to handle pin-wise full core burnup calculations with millions of depletion regions and scales well through thousands of processors. (authors)