Search NASA⌕ Search

SEARCH · Search NASA

Results for “Distributed Computing Resources”

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.

At least 505 records · Page 28

DETECTING UNSPECIFIED STRUCTURE IN LOW-COUNT IMAGES

Unexpected structure in images of astronomical sources often presents itself upon visual inspection of the image, but such apparent structure may either correspond to true features in the source or be due to noise in the data. This paper presents a method for testing whether inferred structure in an image with Poisson noise represents a significant departure from a baseline (null) model of the image. To infer image structure, we conduct a Bayesian analysis of a full model that uses a multiscale component to allow flexible departures from the posited null model. As a test statistic, we use a tail probability of the posterior distribution under the full model. This choice of test statistic allows us to estimate a computationally efficient upper bound on a p-value that enables us to draw strong conclusions even when there are limited computational resources that can be devoted to simulations under the null model. We demonstrate the statistical performance of our method on simulated images. Applying our method to an X-ray image of the quasar 0730+257, we find significant evidence against the null model of a single point source and uniform background, lending support to the claim of an X-ray jet.

79 ASTRONOMY AND ASTROPHYSICS↗

Constant Overhead Entanglement Distillation via Scrambling

High-fidelity quantum entanglement enables key quantum networking capabilities such as secure communication and distributed quantum computing, but long-distance entanglement distribution is limited by noise and loss. Entanglement distillation protocols address this problem by extracting high-fidelity Bell pairs from multiple noisy ones. The primary objective is minimizing the resource overhead: the number of noisy input pairs needed to distill each high-fidelity output pair. While protocols achieving optimal overhead are known in theory, they often require complex decoding operations that make practical implementation challenging. We circumvent this challenge by introducing protocols that use quantum scrambling—the spreading of quantum information under chaotic dynamics—through random Clifford operations. Based on this scrambling mechanism, our protocol maintains asymptotically constant overhead, independent of the desired output error rate $\bar{𝜖}$ , and can be implemented with shallow quantum circuits of depth 𝑂⁡(poly log log⁡ $\bar{𝜖}$ −1 ) and memory 𝑂⁡(poly log⁡ $\bar{𝜖}$ −1 ). Our protocol remains effective even with noisy quantum gates. By incorporating error correction, our protocol achieves state-of-the-art performance: starting with pairs of 10% initial infidelity, we require only seven noisy inputs per output pair to distill a single Bell pair with infidelity $\bar{𝜖}$ =10 −12 , substantially outperforming existing schemes. We demonstrate the utility of our protocols for quantum repeater networks.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Distributed Real-Time Simulation and its Applications to Wind Energy Research

This paper presents a novel approach - distributed Real-Time Simulations (RTS) - for innovative wind energy and grid integration research. Distributed RTS is performed with Digital Real-Time Simulators (DRTS) in different geographical locations and connected over the Internet. Geographically distributed RTS supports - higher computation capability for RTS, leveraging distributed assets, novel testing techniques and hybridized smart grid testbeds. This paper discusses some of the basic aspects of distributed RTS with the communication delay as a probabilistic variable. To leverage hardware resources and enable flexible collaboration of national laboratories, the distributed RTS connection between the simulators at Idaho National Laboratory (INL) and National Wind Test Center (NWTC) at National Renewable Energy Laboratory (NREL) is established. The main objective of this distributed lab setup is to perform wind turbine characterization and assessment of wind integration studies in large scale grids. Several systematic steps based on integrating models are carried out to assess the impact of communication delay on distributed RTS.

17 WIND ENERGY↗

Parallel implementation and performance optimization of the configuration-interaction method

The configuration-interaction (CI) method, long a popular approach to describe quantum many-body systems, is cast as a very large sparse matrix eigenpair problem with matrices whose dimension can exceed one billion. Such formulations place high demands on memory capacity and memory bandwidth - - two quantities at a premium today. In this paper, we describe an efficient, scalable implementation, BIGSTICK, which, by factorizing both the basis and the interaction into two levels, can reconstruct the nonzero matrix elements on the fly, reduce the memory requirements by one or two orders of magnitude, and enable researchers to trade reduced resources for increased computational time. We optimize BIGSTICK on two leading HPC platforms - - the Cray XC30 and the IBM Blue Gene/Q. Specifically, we not only develop an empirically-driven load balancing strategy that can evenly distribute the matrix-vector multiplication across 256K threads, we also developed techniques that improve the performance of the Lanczos reorthogonalization. Combined, these optimizations improved performance by 1.3-8× depending on platform and configuration.

97 MATHEMATICS AND COMPUTING↗

Charliecloud

Charliecloud is a set of scripts to let users run a virtual cluster of virtual machines (VMs) on a desktop or supercomputer. Key functions include: 1. Creating (typically by installing an operating system from vendor media) and updating VM images; 2. Running a single VM; 3. Running multiple VMs in a virtual cluster. The virtual machines can talk to one another over the network and (in some cases) the outside world. This is accomplished by calling external programs such as QEMU and the Virtual Distributed Ethernet (VDE) suite. The goal is to let users have a virtual cluster containing nodes where they have privileged access, while isolating that privilege within the virtual cluster so it cannot affect the physical compute resources. Host configuration enforces security; this is not included in Charliecloud, though security guidelines are included in its documentation and Charliecloud is designed to facilitate such configuration. Charliecloud manages passing information from host computers into and out of the virtual machines, such as parameters of the virtual cluster, input data specified by the user, output data from virtual compute jobs, VM console display, and network connections (e.g., SSH or X11). Parameters for the virtual cluster (number of VMs, RAM and disk per VM, etc.) are specified by the user or gathered from the environment (e.g., SLURM environment variables). Example job scripts are included. These include computation examples (such as a "hello world" MPI job) as well as performance tests. They also include a security test script to verify that the virtual cluster is appropriately sandboxed. Tests include: 1. Pinging hosts inside and outside the virtual cluster to explore connectivity; 2. Port scans (again inside and outside) to see what services are available; 3. Sniffing tests to see what traffic is visible to running VMs; 4. IP address spoofing to test network functionality in this case; 5. File access tests to make sure host access permissions are enforced. This test script is not a comprehensive scanner and does not test for specific vulnerabilities. Importantly, no information about physical hosts or network topology is included in this script (or any of Charliecloud); while part of a sensible test, such information is specified by the user when the test is run. That is, one cannot learn anything about the LANL network or computing infrastructure by examining Charliecloud code.

Priedhorsky, Reid↗

Charliecloud

Charliecloud is a set of scripts to let users run a virtual cluster of virtual machines (VMs) on a desktop or supercomputer. Key functions include: 1. Creating (typically by installing an operating system from vendor media) and updating VM images; 2. Running a single VM; 3. Running multiple VMs in a virtual cluster. The virtual machines can talk to one another over the network and (in some cases) the outside world. This is accomplished by calling external programs such as QEMU and the Virtual Distributed Ethernet (VDE) suite. The goal is to let users have a virtual cluster containing nodes where they have privileged access, while isolating that privilege within the virtual cluster so it cannot affect the physical compute resources. Host configuration enforces security; this is not included in Charliecloud, though security guidelines are included in its documentation and Charliecloud is designed to facilitate such configuration. Charliecloud manages passing information from host computers into and out of the virtual machines, such as parameters of the virtual cluster, input data specified by the user, output data from virtual compute jobs, VM console display, and network connections (e.g., SSH or X11). Parameters for the virtual cluster (number of VMs, RAM and disk per VM, etc.) are specified by the user or gathered from the environment (e.g., SLURM environment variables). Example job scripts are included. These include computation examples (such as a "hello world" MPI job) as well as performance tests. They also include a security test script to verify that the virtual cluster is appropriately sandboxed. Tests include: 1. Pinging hosts inside and outside the virtual cluster to explore connectivity; 2. Port scans (again inside and outside) to see what services are available; 3. Sniffing tests to see what traffic is visible to running VMs; 4. IP address spoofing to test network functionality in this case; 5. File access tests to make sure host access permissions are enforced. This test script is not a comprehensive scanner and does not test for specific vulnerabilities. Importantly, no information about physical hosts or network topology is included in this script (or any of Charliecloud); while part of a sensible test, such information is specified by the user when the test is run. That is, one cannot learn anything about the LANL network or computing infrastructure by examining Charliecloud code.

Priedhorsky, Reid↗

Distributed Training and Optimization of Neural Networks

Deep learning models are yielding increasingly better performances thanks to multiple factors. To be successful, model may have large number of parameters or complex architectures and be trained on large dataset. This leads to large requirements on computing resource and turn around time, even more so when hyperparameter optimization is done (e.g. search over model architectures). While this is a challenge that goes beyond particle physics, we review the various ways to do the necessary computations in parallel, and put it in the context of high-energy physics.

Vlimant, Jean-Roch↗

ATLAS FTK Challenge: Simulation of a Billion-fold Hardware Parallelism

During the current LHC shutdown period the ATLAS experiment will upgrade the Trigger and Data Acquisition system to include a hardware tracker coprocessor: the Fast TracKer (FTK). The FTK receives data from the 80 million of channels of the ATLAS silicon detector, identifying charged tracks and reconstructing their parameters at a rate of up to 100 KHz and within 100 microseconds. To achieve this performance, the FTK system identifies candidate tracks utilizing the computing power of a custom ASIC chip with associative memory (AM) designed to perform “pattern matching” at very high speed; track parameters are then calculated using modern FPGAs. A detailed simulation of this massive system has been developed with the goal of supporting the hardware design and studying its impact in the ATLAS online event selection at high LHC luminosities. We present the issues related to emulating FTK on a general-purpose CPU platform, using ATLAS computing Grid resources, and the solutions developed in order to mitigate these problems and allow the studies required to support the system design, construction and installation.

Distributed Computing↗

COOLR: A New System for Dynamic Thermal-Aware Computing

Extreme-scale systems need to walk a fine line between the amount of cooling they receive and the thermal-induced performance and reliability degradation they can sustain. System managers are extremely motivated to battle cooling-energy cost: the largest line item in their total operating cost. On the other hand, pushing the nodes to peak performance in tightened cooling regimes places the burden on the dynamic thermal management (DTM) to protect the hardware. DTM schemes throttle performance to relieve heat accumulation within nodes when cooling cannot mitigate the problem. These interventions prevent fatal failures but introduce inevitable performance degradations and variations. Such variations can have drastic consequences on the performance of extreme-scale systems. Furthermore, distribution of heat across different system nodes depends not only on the amount of workload. Even if a strictly equal share of computational load is assigned to all nodes, there are physical attributes and topological features of each system that inherently cause uneven accumulation of heat. These can cause one subcomponent to trigger DTM prematurely and penalize the overall system. Our goal in this project is to create a holistic thermal-aware view of the system, capturing both inherent physical attributes and dynamic system state. We developed COOLR, a dynamic system with the ability to evaluate the interplay between management of computation, data, power dissipation, and thermal state. Thereby, COOLR will achieve higher overall performance at the same energy and cooling cost. First, we performed a systematic power and thermal modeling and of high performance computing architectures. Analytical and empirical techniques were blended to generate a light-weight thermal model of target systems. Second, we developed a thermal-aware OS and runtime system. Policies for managing resources and schedules for processes has been designed to co-optimize the thermal state and performance of the system. The main outcome of our proposed thermal-aware dynamic computing system are scheduling and allocation of processes and memory accesses leading to an appropriately “skewed” distribution of activity across the system. The ultimate result of this new arrangement resulted in reduced occurrences of hotspots and reduced episodes of DTM intervention, leading to improvements in performance. The overall thermal characterization and modeling methodology (thermal instrumentation techniques, model building, and systematic model reduction) resulting from this project can be applied to systems beyond our immediate scope and will be applicable to future scaling. The thermal-aware dynamic computing paradigm will impact the efficiency of extreme-scale systems. The gained “thermal slack” can be given back to further tighten the cooling budget, benefiting the management cost of next generation extreme-scale systems.

97 MATHEMATICS AND COMPUTING↗

Steps toward fault-tolerant quantum chemistry.

Developing quantum chemistry programs on the coming generation of exascale computers will be a difficult task. The programs will need to be fault-tolerant and minimize the use of global operations. This work explores the use a task-based model that uses a data-centric approach to allocate work to different processes as it applies to quantum chemistry. After introducing the key problems that appear when trying to parallelize a complicated quantum chemistry method such as coupled-cluster theory, we discuss the implications of that model as it pertains to the computational kernel of a coupled-cluster program - matrix multiplication. Also, we discuss the extensions that would required to build a full coupled-cluster program using the task-based model. Current programming models for high-performance computing are fault-intolerant and use global operations. Those properties are unsustainable as computers scale to millions of CPUs; instead one must recognize that these systems will be hierarchical in structure, prone to constant faults, and global operations will be infeasible. The FAST-OS HARE project is introducing a scale-free computing model to address these issues. This model is hierarchical and fault-tolerant by design, allows for the clean overlap of computation and communication, reducing the network load, does not require checkpointing, and avoids the complexity of many HPC runtimes. Development of an algorithm within this model requires a change in focus from imperative programming to a data-centric approach. Quantum chemistry (QC) algorithms, in particular electronic structure methods, are an ideal test bed for this computing model. These methods describe the distribution of electrons in a molecule, which determine the properties of the molecule. The computational cost of these methods is high, scaling quartically or higher in the size of the molecule, which is why QC applications are major users of HPC resources. The complexity of these algorithms means that MPI alone is insufficient to achieve parallel scaling; QC developers have been forced to use alternative approaches to achieve scalability and would be receptive to radical shifts in the programming paradigm. Initial work in adapting the simplest QC method, Hartree-Fock, to this the new programming model indicates that the approach is beneficial for QC applications. However, the advantages to being able to scale to exascale computers are greatest for the computationally most expensive algorithms; within QC these are the high-accuracy coupled-cluster (CC) methods. Parallel coupledcluster programs are available, however they are based on the conventional MPI paradigm. Much of the effort is spent handling the complicated data dependencies between the various processors, especially as the size of the problem becomes large. The current paradigm will not survive the move to exascale computers. Here we discuss the initial steps toward designing and implementing a CC method within this model. First, we introduce the general concepts behind a CC method, focusing on the aspects that make these methods difficult to parallelize with conventional techniques. Then we outline what is the computational core of the CC method - a matrix multiply - within the task-based approach that the FAST-OS project is designed to take advantage of. Finally we outline the general setup to implement the simplest CC method in this model, linearized CC doubles (LinCC).

99 GENERAL AND MISCELLANEOUS↗

PanDA: Production and Distributed Analysis System

The Production and Distributed Analysis (PanDA) system is a data-driven workload management system engineered to operate at the LHC data processing scale. The PanDA system provides a solution for scientific experiments to fully leverage their distributed heterogeneous resources, showcasing scalability, usability, flexibility, and robustness. The system has successfully proven itself through nearly two decades of steady operation in the ATLAS experiment, addressing the intricate requirements such as diverse resources distributed worldwide at about 200 sites, thousands of scientists analyzing the data remotely, the volume of processed data beyond the exabyte scale, dozens of scientific applications to support, and data processing over several billion hours of computing usage per year. PanDA’s flexibility and scalability make it suitable for the High Energy Physics community and wider science domains at the Exascale. Beyond High Energy Physics, PanDA’s relevance extends to other big data sciences, as evidenced by its adoption in the Vera C. Rubin Observatory and the sPHENIX experiment. As the significance of advanced workflows continues to grow, PanDA has transformed into a comprehensive ecosystem, effectively tackling challenges associated with emerging workflows and evolving computing technologies. The paper discusses PanDA’s prominent role in the scientific landscape, detailing its architecture, functionality, deployment strategies, project management approaches, results, and evolution into an ecosystem.

97 MATHEMATICS AND COMPUTING↗

The environmental impact, carbon emissions and sustainability of computing in the ATLAS experiment

ATLAS, a general-purpose experiment at the Large Hadron Collider (LHC), makes use of a large internationally-distributed computing infrastructure, including over 10 6 TB of managed data on disk and tape and almost one million simultaneously running CPU cores. Upgrades for the High-Luminosity LHC (HL-LHC) will increase the required computing resources by a factor of 3–4 by the beginning of the 2030s, and by an order of magnitude before the conclusion of data taking at the beginning of the 2040s. These resources are spread over around 100 computing sites worldwide. Efforts are underway within the experiment to evaluate and mitigate various aspects of the environmental impact of the sites, with the additional long-term goal of making recommendations to the sites that will significantly reduce the total expected environmental impact in the HL-LHC era. These efforts take several forms: building awareness in the experiment community, adjusting aspects of the computing policy, and modifications of data center configurations, either in ways that take advantage of particular features of ATLAS workloads or in generic ways that reduce the environmental impact of the computing resources. This paper describes the ongoing investigations and approaches that have already provided useful and actionable outcomes.

Aad, G. [CNRS/IN2P3] (ORCID:0000000266654934)↗

HydraGNN_Predictive_GFM_2024 - Ensemble of predictive graph foundation models for ground state atomistic materials modeling

We provide the ensemble of fifteen pre-trained graph foundation models (GFMs) for atomistic materials modeling applications. Each one of the fifteen GFMs has been trained on five open-source datasets that (once aggregated) amount to over 154 million atomistic structures, which cover over two-thirds of the natural elements of the periodic table and that comprises a broad set of organic and inorganic compounds. This vast set of atomistic structures comprises ground state configurations that are dynamically stable (i.e., equilibrated structures with atomic forces approximately close to zero values) as well as dynamically unstable structures (i.e., non-equilibrium structures with non-negligible non-zero values of atomic forces). The ensemble of datasets aggregated does NOT include excited states. The datasets have been curated to remove atomistic structures with spectral norm of the force tensor above 100 eV/angstrom. Moreover, a linear term of the energy was computed for each dataset using a linear regression model that uses the chemical concentration of each natural element as regressor. The linear term predicted by the linear regression model has been subtracted from each original energy value to perform a re-alignment of the energy values across different electronic structures approximation theories performed to generate the diverse multi-source, multi-fidelity datasets. The folder "ADIOS_files" contains the set of pre-processed datasets in Adaptable I/O System (ADIOS) format (https://www.exascaleproject.org/research-project/adios/) that have been used for the development and training of GFMs in this work. The "ADIOS_files" directory contains 6 sub-directories named as follows: - ANI1x-v3.bp - MPTrj-v3.bp - OC2020-20M-v3.bp - OC2020-v3.bp - OC2022-v3.bp - qm7x-v3.bp Each sub-directory contains the pre-processed datasets converted in Adaptable I/O System (ADIOS) format (https://www.exascaleproject.org/research-project/adios/) that have been used to the development, training, and performance testing of the ensemble go predictive graph foundation models. Each GFM was developed using HydraGNN (https://github.com/ORNL/HydraGNN) as underlying graph neural network (GNN) architecture. The multi-task learning (MTL) capability of HydraGNN was used to simultaneously train the GFMs on labeled values for direct predictions of energy (a total system property of an atomistic structure that measures the chemical stability) and atomic forces (an atomic level property of an atomistic structure that measures the dynamical stability). The hyper parameters of the GFM have been tuned using scalable hyperparameter optimization (HPO) algorithms implemented in the software DeepHyper (https://github.com/deephyper/deephyper). The pre-training of each HPO trial was performed using distributed data parallelism (DDP) to scale the training across 128 compute nodes of the exascale OLCF supercomputer Frontier. Each HPO trial was trained only for 10 epochs and an early stopping was performed to avoid wasting significant computational resources on GNN architectures that were clearly underperforming. For each HPO trial, the 'omnistat' tool developed by (AMD Research - Advanced Micro Device) was used to measure the total energy consumption in kWh. The ensemble of GFMs was obtained by selecting the fifteen best performing HPO trials. Four models have been selected for their clear advantage in accuracy, and these are the GFMs with IDs 229, 156, 147, 260. Additional eleven models have been selected based on judicious balance between accuracy and energy consumption needed for training, and these are the GFMs with IDs 165, 78, 137, 1, 175, 171, 181, 67, 179, 167, 351. Each selected GFM of the ensemble was continued to cumulate a total of at most 30 epochs. In some cases, the total number of epochs actually performed was les than 30 due to two combined factors: (1) the size of the GFM (i.e., the number of model parameters to train) and (2) the total wall-clock time for which the computational resources could be allocated on OLCF-Frontier. The "Ensemble_of_models" directory contains 15 sub-directories named as follows: - gfm_0.229 - gfm_0.156 - gfm_0.147 - gfm_0.260 - gfm_0.165 - gfm_0.78 - gfm_0.137 - gfm_0.1 - gfm_0.175 - gfm_0.171 - gfm_0.181 - gfm_0.67 - gfm_0.179 - gfm_0.167 - gfm_0.351 Each one of these sub-directories refers to one of the fifteen HPO trials that have been selected to continue the pre-training with at most 30 epochs. With each sub-directory associated with a specific HPO trial, the following files can be found: - config.json: file for argument parsing to develop and train an HydraGNN architecture - gfm_0.ID_epoch_N.pk: file with model parameters for HPO ID trial after N epochs of training The ensemble of fifteen GFM architectures was used for (1) ensemble averaging to stabilize the predictions of energy and atomic forces after pre-training for post-processing analysis and (2) ensemble uncertainty quantification (UQ). The code used to develop, pre-train, and load the pre-trained models for post-processing analysis is available on the ORNL-GitHub at the following link: https://github.com/ORNL/HydraGNN/tree/Predictive_GFM_2024

36 MATERIALS SCIENCE↗

Uniform Beam Simulation Technique for Beam Scans and Machine Learning Studies at Fermilab

Fermilab's neutrino facilities, including NuMI and the upcoming LBNF, use proton beams to produce positively and negatively charged pions and kaons. Detailed simulations are necessary to study particle interactions and beam propagation. To efficiently analyze beam scan effects, we propose a technique to generate multiple simulation samples with high statistics. These samples can be used to develop beamline simulation based machine learning applications. In this technique, we generate a uniformly distributed single simulation data sample. We calculate Gaussian weights for each beam configurations and apply them to post-processing measurements. In this poster, we demonstrate the proposed simulation technique. This technique reduces simulation time and computing resources significantly.

Wickremasinghe, Athula↗

ASCR Science Network Requirements

The Energy Sciences Network (ESnet) is the primary provider of network connectivity for the US Department of Energy Office of Science, the single largest supporter of basic research in the physical sciences in the United States. In support of the Office of Science programs, ESnet regularly updates and refreshes its understanding of the networking requirements of the instruments, facilities, scientists, and science programs that it serves. This focus has helped ESnet to be a highly successful enabler of scientific discovery for over 20 years. In April 2009 ESnet and the Office of Advanced Scientific Computing Research (ASCR), of the DOE Office of Science, organized a workshop to characterize the networking requirements of the programs funded by ASCR. The ASCR facilities anticipate significant increases in wide area bandwidth utilization, driven largely by the increased capabilities of computational resources and the wide scope of collaboration that is a hallmark of modern science. Many scientists move data sets between facilities for analysis, and in some cases (for example the Earth System Grid and the Open Science Grid), data distribution is an essential component of the use of ASCR facilities by scientists. Due to the projected growth in wide area data transfer needs, the ASCR supercomputer centers all expect to deploy and use 100 Gigabit per second networking technology for wide area connectivity as soon as that deployment is financially feasible. In addition to the network connectivity that ESnet provides, the ESnet Collaboration Services (ECS) are critical to several science communities. ESnet identity and trust services, such as the DOEGrids certificate authority, are widely used both by the supercomputer centers and by collaborations such as Open Science Grid (OSG) and the Earth System Grid (ESG). Ease of use is a key determinant of the scientific utility of network-based services. Therefore, a key enabling aspect for scientists beneficial use of high performance networks is a consistent, widely deployed, well-maintained toolset that is optimized for wide area, high-speed data transfer (e.g. GridFTP) that allows scientists to easily utilize the services and capabilities that the network provides. Network test and measurement is an important part of ensuring that these tools and network services are functioning correctly. One example of a tool in this area is the recently developed perfSONAR, which has already shown its usefulness in fault diagnosis during the recent deployment of high-performance data movers at NERSC and ORNL. On the other hand, it is clear that there is significant work to be done in the area of authentication and access control - there are currently compatibility problems and differing requirements between the authentication systems in use at different facilities, and the policies and mechanisms in use at different facilities are sometimes in conflict. Finally, long-term software maintenance was of concern for many attendees. Scientists rely heavily on a large deployed base of software that does not have secure programmatic funding. Software packages for which this is true include data transfer tools such as GridFTP as well as identity management and other software infrastructure that forms a critical part of the Open Science Grid and the Earth System Grid.

97 MATHEMATICS AND COMPUTING↗

Grids, virtualization, and clouds at Fermilab

Fermilab supports a scientific program that includes experiments and scientists located across the globe. To better serve this community, in 2004, the (then) Computing Division undertook the strategy of placing all of the High Throughput Computing (HTC) resources in a Campus Grid known as FermiGrid, supported by common shared services. In 2007, the FermiGrid Services group deployed a service infrastructure that utilized Xen virtualization, LVS network routing and MySQL circular replication to deliver highly available services that offered significant performance, reliability and serviceability improvements. This deployment was further enhanced through the deployment of a distributed redundant network core architecture and the physical distribution of the systems that host the virtual machines across multiple buildings on the Fermilab Campus. In 2010, building on the experience pioneered by FermiGrid in delivering production services in a virtual infrastructure, the Computing Sector commissioned the FermiCloud, General Physics Computing Facility and Virtual Services projects to serve as platforms for support of scientific computing (FermiCloud 6 GPCF) and core computing (Virtual Services). Lastly, this work will present the evolution of the Fermilab Campus Grid, Virtualization and Cloud Computing infrastructure together with plans for the future.

97 MATHEMATICS AND COMPUTING↗

An Asynchronous Task-based Fan-Both Sparse Cholesky Solver

Systems of linear equations arise at the heart of many scientific and engineering applications. Many of these linear systems are sparse; i.e., most of the elements in the coefficient matrix are zero. Direct methods based on matrix factorizations are sometimes needed to ensure accurate solutions. For example, accurate solution of sparse linear systems is needed in shift-invert Lanczos to compute interior eigenvalues. The performance and resource usage of sparse matrix factorizations are critical to time-to-solution and maximum problem size solvable on a given platform. In many applications, the coefficient matrices are symmetric, and exploiting symmetry will reduce both the amount of work and storage cost required for factorization. When the factorization is performed on large-scale distributed memory platforms, communication cost is critical to the performance of the algorithm. At the same time, network topologies have become increasingly complex, so that modern platforms exhibit a high level of performance variability. This makes scheduling of computations an intricate and performance-critical task. In this paper, we investigate the use of an asynchronous task paradigm, one-sided communication and dynamic scheduling in implementing sparse Cholesky factorization (symPACK) on large-scale distributed memory platforms. Our solver symPACK relies on efficient and flexible communication primitives provided by the UPC++ library. Performance evaluation shows good scalability and that symPACK outperforms state-of-the-art parallel distributed memory factorization packages, validating our approach on practical cases.

Jacquelin, Mathias↗