Search NASA⌕ Search

SEARCH · Search NASA

Results for “GPU accelerators”

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 703 records · Page 39

Machine learning at the Spallation Neutron Source accelerator and target

We describe the ongoing efforts to apply Machine Learning techniques to improve the performance of our accelerator and target. Specially, we are looking to minimize halo beam losses in the absence of a proper physics model, automatically detect and log anomalies in the target support systems such as cooling, and detect and prevent errant beam pulses in the linac. We also describe the infrastructure we use to acquire and stream data to the GPU cluster for training, our code development cycle, and edge computing for model inference. To minimize halo beam losses, we use a Reinforcement Learning technique tested on a virtual accelerator. The target anomaly detection is trained on archived data using incomplete physics models and is made part of the existing target reporting system. The errant beam prevention analyzes beam current and beam phase waveforms as well as accelerator configuration data to predict errant pulses. We also develop continual learning to adapt to changes in the accelerator.

Accelerator Physics↗

Asynchronous Iterative Solvers for Extreme-Scale Computing

The Asynchronous Iterative Solvers for Extreme-Scale Computing (AsyncIS) project aims to explore more efficient numerical algorithms by decreasing their overhead. AsyncIS does this by replacing the outer Krylov subspace solver with an asynchronous optimized Schwarz method, thereby removing the global synchronization and bulk synchronous operations typically used in numerical codes. AsyncIS—a U.S. Department of Energy (DOE)-funded collaboration between Georgia Tech, the University of Tennessee, Knoxville, Temple University, and Sandia National Laboratories—also focuses on the development and optimization of asynchronous preconditioners (i.e., preconditioners that are generated and/or applied in an asynchronous fashion). The novel preconditioning algorithms that provide fine-grained parallelism enable preconditioned Krylov solvers to run efficiently on large-scale distributed systems and manycore accelerators like GPUs.

97 MATHEMATICS AND COMPUTING↗

WE-G-18A-04: 3D Dictionary Learning Based Statistical Iterative Reconstruction for Low-Dose Cone Beam CT Imaging

Purpose: To develop a 3D dictionary learning based statistical reconstruction algorithm on graphic processing units (GPU), to improve the quality of low-dose cone beam CT (CBCT) imaging with high efficiency. Methods: A 3D dictionary containing 256 small volumes (atoms) of 3x3x3 voxels was trained from a high quality volume image. During reconstruction, we utilized a Cholesky decomposition based orthogonal matching pursuit algorithm to find a sparse representation on this dictionary basis of each patch in the reconstructed image, in order to regularize the image quality. To accelerate the time-consuming sparse coding in the 3D case, we implemented our algorithm in a parallel fashion by taking advantage of the tremendous computational power of GPU. Evaluations are performed based on a head-neck patient case. FDK reconstruction with full dataset of 364 projections is used as the reference. We compared the proposed 3D dictionary learning based method with a tight frame (TF) based one using a subset data of 121 projections. The image qualities under different resolutions in z-direction, with or without statistical weighting are also studied. Results: Compared to the TF-based CBCT reconstruction, our experiments indicated that 3D dictionary learning based CBCT reconstruction is able to recover finer structures, to remove more streaking artifacts, and is less susceptible to blocky artifacts. It is also observed that statistical reconstruction approach is sensitive to inconsistency between the forward and backward projection operations in parallel computing. Using high a spatial resolution along z direction helps improving the algorithm robustness. Conclusion: 3D dictionary learning based CBCT reconstruction algorithm is able to sense the structural information while suppressing noise, and hence to achieve high quality reconstruction. The GPU realization of the whole algorithm offers a significant efficiency enhancement, making this algorithm more feasible for potential clinical application. A high zresolution is preferred to stabilize statistical iterative reconstruction. This work was supported in part by NIH(1R01CA154747-01), NSFC((No. 61172163), Research Fund for the Doctoral Program of Higher Education of China (No. 20110201110011), China Scholarship Council.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Targeting GPUs with OpenMP directives on Summit: A simple and effective Fortran experience

We use OpenMP to target hardware accelerators (GPUs) on Summit, a newly deployed supercomputer at the Oak Ridge Leadership Computing Facility (OLCF), demonstrating simplified access to GPU devices for users of our astrophysics code GenASiS and useful speedup on a sample fluid dynamics problem. We modify our workhorse class for data storage to include members and methods that significantly streamline the persistent allocation of and association to GPU memory. Users offload computational kernels with OpenMP target directives that are rather similar to constructs already familiar from multi-core parallelization. In this initial example we ask, “With a given number of Summit nodes, how fast can we compute with and without GPUs?”, and find total wall time speedups of ~ 12X. We also find reasonable weak scaling up to 8000 GPUs (1334 Summit nodes). We make available the source code from this work at https://github.com/GenASiS/GenASiS_Basics.

97 MATHEMATICS AND COMPUTING↗

GPU-friendly surface model for Monte-Carlo detector simulations

The demands for Monte-Carlo simulation are drastically increasing with the Large Hadron Collider’s high-luminosity upgrade, and are expected to exceed the currently available compute resources. At the same time, modern high-performance computing has adopted powerful hardware accelerators, particularly GPUs. The AdePT and Celeritas projects aim to address the demanding computational needs by leveraging these heterogeneous computing architectures. While both have successfully ported realistic detector simulations to GPUs using the VecGeom library, the complexity of geometry modeling emerged as a bottleneck. Thread divergence and high register usage were degrading the GPU performance. Therefore, a new, GPU-friendly surface-based model has been introduced in the VecGeom library that decomposes the divergent code of the 3D primitive solids into simpler and more balanced surface algorithms. In this work, we present the latest developments, focusing on the additions required to efficiently model complex setups like the CMS Phase-2 geometry. This includes memory reduction techniques, and adding accelerating structures for faster traversal.

Diederichs, Severin [CERN]↗

SU-C-17A-07: The Development of An MR Accelerator-Enabled Planning-To-Delivery Technique for Stereotactic Palliative Radiotherapy Treatment of Spinal Metastases

Purpose: To develop an MR accelerator-enabled online planning-todelivery technique for stereotactic palliative radiotherapy treatment of spinal metastases. The technical challenges include; automated stereotactic treatment planning, online MR-based dose calculation and MR guidance during treatment. Methods: Using the CT data of 20 patients previously treated at our institution, a class solution for automated treatment planning for spinal bone metastases was created. For accurate dose simulation right before treatment, we fused geometrically correct online MR data with pretreatment CT data of the target volume (TV). For target tracking during treatment, a dynamic T2-weighted TSE MR sequence was developed. An in house developed GPU based IMRT optimization and dose calculation algorithm was used for fast treatment planning and simulation. An automatically generated treatment plan developed with this treatment planning system was irradiated on a clinical 6 MV linear accelerator and evaluated using a Delta4 dosimeter. Results: The automated treatment planning method yielded clinically viable plans for all patients. The MR-CT fusion based dose calculation accuracy was within 2% as compared to calculations performed with original CT data. The dynamic T2-weighted TSE MR Sequence was able to provide an update of the anatomical location of the TV every 10 seconds. Dose calculation and optimization of the automatically generated treatment plans using only one GPU took on average 8 minutes. The Delta4 measurement of the irradiated plan agreed with the dose calculation with a 3%/3mm gamma pass rate of 86.4%. Conclusions: The development of an MR accelerator-enabled planning-todelivery technique for stereotactic palliative radiotherapy treatment of spinal metastases was presented. Future work will involve developing an intrafraction motion adaptation strategy, MR-only dose calculation, radiotherapy quality-assurance in a magnetic field, and streamlining the entire treatment process on an MR accelerator.

60 APPLIED LIFE SCIENCES↗

Towards Enhancing Coding Productivity for GPU Programming Using Static Graphs

The main contribution of this work is to increase the coding productivity of GPU programming by using the concept of Static Graphs. GPU capabilities have been increasing significantly in terms of performance and memory capacity. However, there are still some problems in terms of scalability and limitations to the amount of work that a GPU can perform at a time. To minimize the overhead associated with the launch of GPU kernels, as well as to maximize the use of GPU capacity, we have combined the new CUDA Graph API with the CUDA programming model (including CUDA math libraries) and the OpenACC programming model. We use as test cases two different, well-known and widely used problems in HPC and AI: the Conjugate Gradient method and the Particle Swarm Optimization. In the first test case (Conjugate Gradient) we focus on the integration of Static Graphs with CUDA. In this case, we are able to significantly outperform the NVIDIA reference code, reaching an acceleration of up to 11x thanks to a better implementation, which can benefit from the new CUDA Graph capabilities. In the second test case (Particle Swarm Optimization), we complement the OpenACC functionality with the use of CUDA Graph, achieving again accelerations of up to one order of magnitude, with average speedups ranging from 2x to 4x, and performance very close to a reference and optimized CUDA code. Our main target is to achieve a higher coding productivity model for GPU programming by using Static Graphs, which provides, in a very transparent way, a better exploitation of the GPU capacity. The combination of using Static Graphs with two of the current most important GPU programming models (CUDA and OpenACC) is able to reduce considerably the execution time w.r.t. the use of CUDA and OpenACC only, achieving accelerations of up to more than one order of magnitude. Finally, we propose an interface to incorporate the concept of Static Graphs into the OpenACC Specifications.

58 GEOSCIENCES↗

Evaluating LULESH Kernels on OpenCL FPGA

FPGAs are becoming promising heterogeneous computing components for high-performance computing. In this paper, we evaluate the resource utilizations, performance, and performance per watt of our implementations of the LULESH kernels in OpenCL on an Arria10-based FPGA platform. LULESH is a complex proxy application in the CORAL benchmark suite. We choose two representative kernels “CalcFBHourglassForceForElems” and “EvalEOSForElems” from the application in our study. Compared with the baseline implementations, our optimizations improve the performance by a factor of 1.65X and 2.96X for the two kernels on the FPGA, respectively. Using directives for accelerator programming, we also evaluate the performance of the kernels on an Intel Xeon 16-core CPU and an Nvidia K80 GPU. We find that the FPGA, constrained by the memory bandwidth, can perform 1.05X to 3.4X better than the CPU and GPU for small problem sizes. For the first kernel, the performance per watt on the FPGA is 1.59X and 7.1X higher than that on an Intel Xeon 16-core CPU and an Nvidia K80 GPU, respectively. For the second kernel, the performance per watt on the GPU is 1.82X higher than that on the FPGA. However, the performance per watt on the FPGA is 1.77X higher than that on the CPU.

FPGA↗

QCD tree amplitudes on modern GPUs: A case study for novel event generators

For more than a decade the current generation of CPU-based matrix element generators has provided hard scattering events with excellent flexibility and good efficiency.However, they are a bottleneck of current Monte Carlo event generator toolchains, and with the advent of the HL-LHC and more demanding precision requirements, faster matrix elements are needed, especially at intermediate to large jet multiplicities.We present first results of the new BlockGen family of matrix element algorithms, featuring GPU support and novel color treatments, and discuss the best choice to deliver the performance needed for the next generation of accelerated matrix element generators.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Aerodynamic optimization of supersonic compressor cascade using differential evolution on GPU

Differential Evolution (DE) is a powerful stochastic optimization method. Compared to gradient-based algorithms, DE is able to avoid local minima but requires at the same time more function evaluations. In turbomachinery applications, function evaluations are performed with time-consuming CFD simulation, which results in a long, non affordable, design cycle. Modern High Performance Computing systems, especially Graphic Processing Units (GPUs), are able to alleviate this inconvenience by accelerating the design evaluation itself. In this work we present a validated CFD Solver running on GPUs, able to accelerate the design evaluation and thus the entire design process. An achieved speedup of 20x to 30x enabled the DE algorithm to run on a high-end computer instead of a costly large cluster. The GPU-enhanced DE was used to optimize the aerodynamics of a supersonic compressor cascade, achieving an aerodynamic loss minimization of 20%.

97 MATHEMATICS AND COMPUTING↗

Quantum Monte Carlo Endstation for Petascale Computing

The major achievements enabled by QMC Endstation grant include * Performance improvement on clusters of x86 multi-core systems, especially on Cray XT systems * New and improved methods for the wavefunction optimizations * New forms of trial wavefunctions * Implementation of the full application on NVIDIA GPUs using CUDA The scaling studies of QMCPACK on large-scale systems show excellent parallel efficiency up to 216K cores on Jaguarpf (Cray XT5). The GPU implementation shows speedups of 10-15x over the CPU implementation on older generation of x86. We have implemented hybrid OpenMP/MPI scheme in QMC to take advantage of multi-core shared memory processors of petascale systems. Our hybrid scheme has several advantages over the standard MPI-only scheme. * Memory optimized: large read-only data to store one-body orbitals and other shared properties to represent the trial wave function and many-body Hamiltonian can be shared among threads, which reduces the memory footprint of a large-scale problem. * Cache optimized: the data associated with an active Walker are in cache during the compute-intensive drift-diffusion process and the operations on an Walker are optimized for cache reuse. Thread-local objects are used to ensure the data affinity to a thread. * Load balanced: Walkers in an ensemble are evenly distributed among threads and MPI tasks. The two-level parallelism reduces the population imbalance among MPI tasks and reduces the number of point-to-point communications of large messages (serialized objects) for the Walker exchange. * Communication optimized: the communication overhead, especially for the collective operations necessary to determine ET and measure the properties of an ensemble, is significantly lowered by using less MPI tasks. The multiple forms of parallelism afforded by QMC algorithms make them ideal candidates for acceleration in the many-core paradigm. We presented the results of our effort to port the QMCPACK simulation code to the NVIDIA CUDA GPU platform. We restructured the CPU algorithms to express additional parallelism, minimize GPU-CPU communication, and efficiently utilize the GPU memory hierarchy. Using mixed precision on GT200 GPUs and MPI for intercommunication and load balancing, we observe typical full-application speedups of approximately 10x to 15x relative to quad-core Xeon CPUs alone, while reproducing the double-precision CPU results within statistical error. We developed an all-electron quantum Monte Carlo (QMC) method for solids that does not rely on pseudopotentials, and used it to construct a primary ultra-high-pressure calibration based on the equation of state of cubic boron nitride. We computed the static contribution to the free energy with the QMC method and obtained the phonon contribution from density functional theory, yielding a high-accuracy calibration up to 900 GPa usable directly in experiment. We computed the anharmonic Raman frequency shift with QMC simulations as a function of pressure and temperature, allowing optical pressure calibration. In contrast to present experimental approaches, small systematic errors in the theoretical EOS do not increase with pressure, and no extrapolation is needed. This all-electron method is applicable to first-row solids, providing a new reference for ab initio calculations of solids and benchmarks for pseudopotential accuracy. We compared experimental and theoretical results on the momentum distribution and the quasiparticle renormalization factor in sodium. From an x-ray Compton-profile measurement of the valence-electron momentum density, we derived its discontinuity at the Fermi wavevector finding an accurate measure of the renormalization factor that we compared with quantum-Monte-Carlo and G0W0 calculations performed both on crystalline sodium and on the homogeneous electron gas. Our calculated results are in good agreement with the experiment. We have been studying the heat of formation for various Kubas complexes of molecular hydrogen on Ti(1,2)ethylene-nH2 using Diffusion Monte Carlo. This work has been started and is ongoing. We are studying systems involving 1 and 2 Ti bonding sites with up to 10 hydrogen molecules in numerous configurations. This work will establish a benchmark that will test the accuracy of density functional calculations and establish the feasibility of our methods for similar systems.

97 MATHEMATICS AND COMPUTING↗

Batched matrix computations on hardware accelerators based on GPUs

Scientific applications require solvers that work on many small size problems that are independent from each other. At the same time, the high-end hardware evolves rapidly and becomes ever more throughput-oriented and thus there is an increasing need for an effective approach to develop energy-efficient, high-performance codes for these small matrix problems that we call batched factorizations. The many applications that need this functionality could especially benefit from the use of GPUs, which currently are four to five times more energy efficient than multicore CPUs on important scientific workloads. This study, consequently, describes the development of the most common, one-sided factorizations, Cholesky, LU, and QR, for a set of small dense matrices. The algorithms we present together with their implementations are, by design, inherently parallel. In particular, our approach is based on representing the process as a sequence of batched BLAS routines that are executed entirely on a GPU. Importantly, this is unlike the LAPACK and the hybrid MAGMA factorization algorithms that work under drastically different assumptions of hardware design and efficiency of execution of the various computational kernels involved in the implementation. Thus, our approach is more efficient than what works for a combination of multicore CPUs and GPUs for the problems sizes of interest of the application use cases. The paradigm where upon a single chip (a GPU or a CPU) factorizes a single problem at a time is not at all efficient in our applications’ context. We illustrate all of these claims through a detailed performance analysis. With the help of profiling and tracing tools, we guide our development of batched factorizations to achieve up to two-fold speedup and three-fold better energy efficiency as compared against our highly optimized batched CPU implementations based on MKL library. Finally, the tested system featured two sockets of Intel Sandy Bridge CPUs and we compared with a batched LU factorizations featured in the CUBLAS library for GPUs, we achieve as high as 2.5× speedup on the NVIDIA K40 GPU.

97 MATHEMATICS AND COMPUTING↗

Evaluation of AC optimal power flow on graphical processing units

This paper investigates the performance of alternating current optimal power flow (ACOPF) on hardware accelerators such as graphical processing units (GPUs). We describe the strategies employed and the software used to port the ACOPF application to GPU. Through reorganizing the flow of fundamental calculations, restructuring data organization for the GPUs, and using portability libraries, maximum utilization of GPU is attempted. We present details of our efforts with representative results on 200, 500, and 2000-bus networks.

Abhyankar, Shrirang G.↗

Comparison of nested geometry treatments within GPU-based Monte Carlo neutron transport simulations of fission reactors

Monte Carlo (MC) neutron transport provides detailed estimates of radiological quantities within fission reactors. This involves tracking individual neutrons through a computational geometry. CPU-based MC codes use multiple polymorphic tracker types with different tracking algorithms to exploit the repeated configurations of reactors, but virtual function calls have high overhead on the GPU. The Shift MC code was modified to support GPU-based tracking with three strategies: dynamic polymorphism with virtual functions, static polymorphism, and a single tracker type with tree-based acceleration. On the Frontier supercomputer these methods achieve 77.8%, 91.2%, and 83.4%, respectively, of the tracking rate obtained using a specialized tracker optimized for rectilinear-grid-based reactors. This indicates that all three methods are suitable for typical reactor problems in which tracking does not dominate runtime. The flexibility of the single tracker method is highlighted with a hexagonal-grid microreactor problem, performed without hexagonal-grid-specific tracking routines, providing a 2.19× speedup over CPU execution.

97 MATHEMATICS AND COMPUTING↗

ComPort: Rigorous Testing Methods to Safeguard Software Porting (Final Technical Report)

This is a technical report from the lead institution – University of Utah, Kahlert School of Computing – funded under the Department of Energy, Office of Science, Office of Advanced Scientific Computing Research under award number DE-SC0022252. We summarize our work done over the three years of funding received. The relevant papers and software have already been uploaded at the DOE site.

97 MATHEMATICS AND COMPUTING↗

Failing in place for low-serviceability storage infrastructure using high-parity GPU-based RAID.

In order to provide large quantities of high-reliability disk-based storage, it has become necessary to aggregate disks into fault-tolerant groups based on the RAID methodology. Most RAID levels do provide some fault tolerance, but there are certain classes of applications that require increased levels of fault tolerance within an array. Some of these applications include embedded systems in harsh environments that have a low level of serviceability, or uninhabited data centers servicing cloud computing. When describing RAID reliability, the Mean Time To Data Loss (MTTDL) calculations will often assume that the time to replace a failed disk is relatively low, or even negligible compared to rebuild time. For platforms that are in remote areas collecting and processing data, it may be impossible to access the system to perform system maintenance for long periods. A disk may fail early in a platform's life, but not be replaceable for much longer than typical for RAID arrays. Service periods may be scheduled at intervals on the order of months, or the platform may not be serviced until the end of a mission in progress. Further, this platform may be subject to extreme conditions that can accelerate wear and tear on a disk, requiring even more protection from failures. We have created a high parity RAID implementation that uses a Graphics Processing Unit (GPU) to compute more than two blocks of parity information per stripe, allowing extra parity to eliminate or reduce the requirement for rebuilding data between service periods. While this type of controller is highly effective for RAID 6 systems, an important benefit is the ability to incorporate more parity into a RAID storage system. Such RAID levels, as yet unnamed, can tolerate the failure of three or more disks (depending on configuration) without data loss. While this RAID system certainly has applications in embedded systems running applications in the field, similar benefits can be obtained for servers that are engineered for storage density, with less regard for serviceability or maintainability. A storage brick can be designed to have a MTTDL that extends well beyond the useful lifetime of the hardware used, allowing the disk subsystem to require less service throughout the lifetime of a compute resource. This approach is similar to the Xiotech ISE. Such a design can be deliberately placed remotely (without frequent support) in order to provide colocation, or meet cost goals. For workloads where reliability is key, but conditions are sub-optimal for routine serviceability, a high-parity RAID can provide extra reliability in extraordinary situations. For example, for installations requiring very high Mean Time To Repair, the extra parity can eliminate certain problems with maintaining hot spares, increasing overall reliability. Furthermore, in situations where disk reliability is reduced because of harsh conditions, extra parity can guard against early data loss due to lowered Mean Time To Failure. If used through an iSCSI interface with a streaming workload, it is possible to gain all of these benefits without impacting performance.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗