Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “GPUs”

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 325 records · Page 18

Evaluating performance and portability of high-level programming models: Julia, Python/Numba, and Kokkos on exascale nodes

We explore the performance and portability of the high-level programming models: the LLVM-based Julia and Python/Numba, and Kokkos on high-performance computing (HPC) nodes: AMD Epyc CPUs and MI250X graphical processing units (GPUs) on Frontier’s test bed Crusher system and Ampere’s Arm-based CPUs and NVIDIA’s A100 GPUs on the Wombat system at the Oak Ridge Leadership Computing Facilities. We compare the default performance of a hand-rolled dense matrix multiplication algorithm on CPUs against vendor-compiled C/OpenMP implementations, and on each GPU against CUDA and HIP. Rather than focusing on the kernel optimization per-se, we select this naive approach to resemble exploratory work in science and as a lower-bound for performance to isolate the effect of each programming model. Julia and Kokkos perform comparably with C/OpenMP on CPUs, while Julia implementations are competitive with CUDA and HIP on GPUs. Performance gaps are identified on NVIDIA A100 GPUs for Julia’s single precision and Kokkos, and for Python/Numba in all scenarios. We also comment on half-precision support, productivity, performance portability metrics, and platform readiness. We expect to contribute to the understanding and direction for high-level, high-productivity languages in HPC as the first-generation exascale systems are deployed.

Godoy, William↗

Parallelizing the QUDA Library for Multi-GPU Calculations in Lattice Quantum Chromodynamics

Graphics Processing Units (GPUs) are having a transformational effect on numerical lattice quantum chromodynamics (LQCD) calculations of importance in nuclear and particle physics. The QUDA library provides a package of mixed precision sparse matrix linear solvers for LQCD applications, supporting single GPUs based on NVIDIA's Compute Unified Device Architecture (CUDA). This library, interfaced to the QDP++/Chroma framework for LQCD calculations, is currently in production use on the "9g" cluster at the Jefferson Laboratory, enabling unprecedented price/performance for a range of problems in LQCD. Nevertheless, memory constraints on current GPU devices limit the problem sizes that can be tackled. In this contribution we describe the parallelization of the QUDA library onto multiple GPUs using MPI, including strategies for the overlapping of communication and computation. We report on both weak and strong scaling for up to 32 GPUs interconnected by InfiniBand, on which we sustain in excess of 4 Tflops.

43 PARTICLE ACCELERATORS↗

A Survey of Methods for Analyzing and Improving GPU Energy Efficiency

Recent years have witnessed a phenomenal growth in the computational capabilities and applications of GPUs. However, this trend has also led to dramatic increase in their power consumption. This paper surveys research works on analyzing and improving energy efficiency of GPUs. It also provides a classification of these techniques on the basis of their main research idea. Further, it attempts to synthesize research works which compare energy efficiency of GPUs with other computing systems, e.g. FPGAs and CPUs. The aim of this survey is to provide researchers with knowledge of state-of-the-art in GPU power management and motivate them to architect highly energy-efficient GPUs of tomorrow.

architecture techniques↗

Fast and Scalable Sparse Triangular Solver for Multi-GPU Based HPC Architectures

Designing efficient and scalable sparse linear algebra kernels on modern multi-GPU based HPC systems is a daunting task due to significant irregular memory references and workload imbalance across the GPUs. This is particularly the case for \textit{Sparse Triangular Solver (SpTRSV)} which introduces additional two-dimensional computation dependencies among subsequent computation steps. Dependency information is exchanged and shared among GPUs, thus warrant for efficient memory allocation, data partitioning, and workload distribution as well as fine-grained communication and synchronization support. In this work, we demonstrate that directly adopting unified memory can adversely affect the performance of SpTRSV on multi-GPU architectures, despite linking via fast interconnect like NVLinks and NVSwitches. Alternatively, we employ the latest NVSHMEM technology based on Partitioned Global Address Space programming model to enable efficient fine-grained communication and drastic synchronization overhead reduction. Furthermore, to handle workload imbalance, we propose a malleable task-pool execution model which can further enhance the utilization of GPUs. By applying these techniques, our experiments on the NVIDIA multi-GPU supernode V100-DGX-1 and DGX-2 systems demonstrate that our design can achieve on average 3.53x (up to 9.86x) speedup on a DGX-1 system and 3.66x (up to 9.64x) speedup on a DGX-2 system with 4-GPUs over the Unified-Memory design. The comprehensive sensitivity and scalability studies also show that the proposed zero-copy SpTRSV is able to fully utilize the computing and communication resources of the multi-GPU system.

Xie, Chenhao↗

Ultrafast Error-bounded Lossy Compression for Scientific Datasets

Today's scientific high-performance computing applications and advanced instruments are producing vast volumes of data across a wide range of domains, which impose a serious burden on data transfer and storage. Error-bounded lossy compression has been developed and widely used in the scientific community because it not only can significantly reduce the data volumes but also can strictly control the data distortion based on the user-specified error bound. Existing lossy compressors, however, cannot offer ultrafast compression speed, which is highly demanded by numerous applications or use cases (such as in-memory compression and online instrument data compression). In this paper, we propose a novel ultrafast error-bounded lossy compressor that can obtain fairly high compression performance on both CPUs and GPUs and with reasonably high compression ratios. The key contributions are threefold. (1) We propose a generic error-bounded lossy compression framework---called SZx---that achieves ultrafast performance through its novel design comprising only lightweight operations such as bitwise and addition/subtraction operations, while still keeping a high compression ratio. (2) We implement SZx on both CPUs and GPUs and optimize the performance according to their architectures. (3) We perform a comprehensive evaluation with six real-world production-level scientific datasets on both CPUs and GPUs. Experiments show that SZx is 2~16x faster than the second-fastest existing error-bounded lossy compressor (either SZ or ZFP) on CPUs and GPUs, with respect to both compression and decompression.

Yu, Xiaodong↗

The Simple Cloud-Resolving E3SM Atmosphere Model Running on the Frontier Exascale System

We present an efficient and performance portable implementation of the Simple Cloud Resolving E3SM Atmosphere Model (SCREAM). SCREAM is a full featured atmospheric global circulation model with a nonhydrostatic dynamical core and state-of-the-art parameterizations for microphysics, moist turbulence and radiation. It has been written from scratch in C++ with the Kokkos library used to abstract the on-node execution model for both CPUs and GPUs. SCREAM is one of only a few global atmosphere models to be ported to GPUs. As far as we know, SCREAM is the first such model to run on both AMD GPUs and NVIDIA GPUs, as well as the first to run on nearly an entire Exascale system (Frontier). On Frontier, we obtained a record setting performance of 1.26 simulated years per day for a realistic cloud resolving simulation.

Taylor, Mark↗

Scalable Incremental Checkpointing using GPU-Accelerated De-Duplication

Writing large amounts of data concurrently to stable storage is a typical I/O pattern of many HPC workflows. This pattern introduces high I/O overheads and results in increased storage space utilization especially for workflows that need to capture the evolution of data structures with high frequency as checkpoints. In this context, many applications, such as graph pattern matching, perform sparse updates to large data structures between checkpoints. For these applications, incremental checkpointing techniques that save only the differences from one checkpoint to another can dramatically reduce the checkpoint sizes, I/O bottlenecks, and storage space utilization. However, such techniques are not without challenges: it is non-trivial to transparently determine what data has changed since a previous checkpoint and assemble the differences in a compact fashion that does not result in excessive metadata. State-of-art data reduction techniques (e.g., compression and de-duplication) have significant limitations when applied to modern HPC applications that leverage GPUs: slow at detecting the differences, generate a large amount of metadata to keep track of the differences, and ignore crucial spatiotemporal checkpoint data redundancy. This paper addresses these challenges by proposing a Merkle tree-based incremental checkpointing method to exploit GPUs' high memory bandwidth and massive parallelism. Experimental results at scale show a significant reduction of the I/O overhead and space utilization of checkpointing compared with state-of-the-art incremental checkpointing and compression techniques.

Tan, Nigel↗

Understanding GPU Memory Corruption at Extreme Scale: The Summit Case Study

GPU memory corruption and in particular double-bit errors (DBEs) remain one of the least understood aspects of HPC system reliability. Albeit rare, their occurrences always lead to job termination and can potentially cost thousands of node-hours, either from wasted computations or as the overhead from regular checkpointing needed to minimize the losses. As supercomputers and their components simultaneously grow in scale, density, failure rates, and environmental footprint, the efficiency of HPC operations becomes both an imperative and a challenge. We examine DBEs using system telemetry data and logs collected from the Summit supercomputer, equipped with 27,648 Tesla V100 GPUs with 2nd-generation high-bandwidth memory (HBM2). Using exploratory data analysis and statistical learning, we extract several insights about memory reliability in such GPUs. We find that GPUs with prior DBE occurrences are prone to experience them again due to otherwise harmless factors, correlate this phenomenon with GPU placement, and suggest manufacturing variability as a factor. On the general population of GPUs, we link DBEs to short- and long-term high power consumption modes while finding no significant correlation with higher temperatures. We also show that the workload type can be a factor in memory’s propensity to corruption.

Oles, Vlad↗

TrioSim: A Lightweight Simulator for Large-Scale DNN Workloads on Multi-GPU Systems

Deep Neural Networks (DNNs) have become increasingly capable of performing tasks ranging from image recognition to content generation. The training and inference of DNNs heavily rely on GPUs, as GPUs' massively parallel architecture delivers extremely high computing capability. With the growing complexity of DNNs and the size of training datasets, training DNNs with a large number of GPUs is becoming a prevalent strategy. Researchers have been exploring how to design software and hardware systems for GPU farms to achieve the best utilization, efficiency, and DNN accuracy during training or inference. However, when designing and deploying such systems, designers usually rely on testing on physical hardware platforms equipped with many GPUs, incurring high costs that are almost prohibitive for system designers to test different configurations and designs, even for highly resourceful companies. While an alternative solution is to test on GPU simulators, they are often too slow for these l

Li, Ying [William & Mary, Williamsburg, VA, USA] (↗

Scaling soft matter physics to thousands of graphics processing units in parallel

We describe a multi-graphics processing unit (GPU) implementation of the Ludwig application, which specialises in simulating a variety of complex fluids via lattice Boltzmann fluid dynamics coupled to additional physics describing complex fluid constituents. We describe our methodology in augmenting the original central processing unit (CPU) version with GPU functionality in a maintainable fashion. We present several optimisations that maximise performance on the GPU architecture through tuning for the GPU memory hierarchy. We describe how we implement particles within the fluid in such a way to avoid a major diversion of the CPU and GPU codebases, whilst minimising data transfer at each time step. We detail our halo-exchange communication phase for the code, which exploits overlapping to allow efficient parallel scaling to many GPUs. We present results showing that the application demonstrates excellent scaling to at least 8192 GPUs in parallel, the largest system tested at the time of writing. The GPU version (on NVIDIA K20X GPUs) is around 3.5–5 times faster that the CPU version (on fully utilised AMD Opteron 6274 16-core CPUs), comparing equal numbers of CPUs and GPUs.

Computer Science↗

SMC 2021 : Analyzing Resource Utilization and User Behavior on Titan Supercomputer

Resource utilization statistics of submitted jobs on a supercomputer can help us understand how users from various scientific domains use HPC platforms and better design a job scheduler. We explore to generate insight regarding workload distribution and usage pattern domains from job scheduler trace, GPU failure information, and project-specific information collected from Titan supercomputer. Furthermore, we want to know how the scheduler performance varies over time and how the users' scheduling behavior changes following a system failure. These observations have the potential to provide valuable insight, which is helpful to prepare for system failures. These practices will help us develop and apply novel machine learning algorithms in understanding system behavior, requirement, and better scheduling of HPC systems. There are two datasets, RUR and GPU. RUR: This dataset is the job scheduler traces collected from the Titan supercomputerfrom 01/01/2015 to 07/31/2019 (2015.csv - 2019.csv). These were collected usingResource Utilization Report (RUR), a Cray-developed resource-usage data collectionand reporting system. It contains the usage information of its critical resources (CPU,Memory, GPU, and I/O) of each running job on Titan during that period [2]. ProjectAreas: Every job is associated with a project ID. TheProjectAreas.csvdatasetprovides a mapping of the project ID to its domain science. GPU: There have been some hardware-related issues in the GPUs in Titan that caused some GPUs to fail, sometimes irrecoverably during some job runs. This dataset provides information regarding these failures during the execution of the submitted jobs. GPUs on Titan are uniquely identified by a serial number (SN), and they are installed in a location. A GPU can be installed in a location, then removed from that location following a failure, and then re-installed in a different location after fixing the problem. If the failure can't be recovered, the GPU might be removed entirely from Titan. There are two prominent types of failures that resulted in the removal of GPUs from Titan: Double Bit Error (DBE) and Out of the Bus (OTB). The dataset (gc_full.csv) has the following fields: 1. SN : Serial number of a GPU 2. location : The location where it is installed 3. insert : The time when it was inserted into that location 4. remove : The time when it was removed from that location 5. duration : Amount of time the GPU spent in this location 6. out : If the device was taken out entirely w/o a re-installment into a new location. 7. event : If the GPU was taken out entirely, the reason for its removal. To learn more about this dataset, please refer to the git repositoryhttps://github.com/olcf/TitanGPULifeand the related publication [1]. References [1] George Ostrouchov, Don Maxwell, Rizwan A Ashraf, Christian Engelmann, MallikarjunShankar, and James H Rogers. Gpu lifetimes on titan supercomputer: Survival analysisand reliability. InSC20: International Conference for High Performance Computing,Networking, Storage and Analysis, pages 1-14. IEEE, 2020. [2] Feiyi Wang, Sarp Oral, Satyabrata Sen, and Neena Imam. Learning from five-yearresource-utilization data of titan system. In2019 IEEE International Conference onCluster Computing (CLUSTER), pages 1-6. IEEE, 2019.

42 ENGINEERING↗

SMC 2021 Data Challenge: Analyzing Resource Utilization and User Behavior on Titan Supercomputer

Resource utilization statistics of submitted jobs on a supercomputer can help us understand how users from various scientific domains use HPC platforms and better design a job scheduler. We explore to generate insight regarding workload distribution and usage pattern domains from job scheduler trace, GPU failure information, and project-specific information collected from Titan supercomputer. Furthermore, we want to know how the scheduler performance varies over time and how the users' scheduling behavior changes following a system failure. These observations have the potential to provide valuable insight, which is helpful to prepare for system failures. These practices will help us develop and apply novel machine learning algorithms in understanding system behavior, requirement, and better scheduling of HPC systems. There are two datasets, RUR and GPU: RUR dataset is the job scheduler traces collected from the Titan supercomputer from 01/01/2015 to 07/31/2019 (2015.csv - 2019.csv). These were collected using resource Utilization Report (RUR), a Cray-developed resource-usage data collection and reporting system. It contains the usage information of its critical resources (CPU, Memory, GPU, and I/O) of each running job on Titan during that period (https://ieeexplore.ieee.org/abstract/document/8891001). It includes ProjectAreas as additional information, every job is associated with a project ID. TheProjectAreas.csv dataset provides a mapping of the project ID to its domain science. GPU dataset has information regarding GPU failure on Titan. There have been some hardware-related issues in the GPUs in Titan that caused some GPUs to fail, sometimes irrecoverably during some job runs. This dataset provides information regarding these failures during the execution of the submitted jobs. GPUs on Titan are uniquely identified by a serial number (SN), and they are installed in a location. A GPU can be installed in a location, then removed from that location following a failure, and then re-installed in a different location after fixing the problem. If the failure can't be recovered, the GPU might be removed entirely from Titan. There are two prominent types of failures that resulted in the removal of GPUs from Titan: Double Bit Error (DBE) and Out of the Bus (OTB). The dataset (gc_full.csv) has seven attributes, we provided a short description of these attributes in the ReadMe file. To learn more about this dataset, please refer to the git repository https://github.com/olcf/TitanGPULife and the related publication (https://ieeexplore.ieee.org/abstract/document/9355319).

42 ENGINEERING↗

Cabana: A Performance Portable Library for Particle-Based Simulations

Particle-based simulations are ubiquitous throughout many fields of computational science and engineering, spanning the atomistic level with molecular dynamics (MD), to mesoscale particle-in-cell (PIC) simulations for solid mechanics, device-scale modeling with PIC methods for plasma physics, and massive N-body cosmology simulations of galaxy structures, with many other methods in between (Hockney & Eastwood, 1989). While these methods use particles to represent significantly different entities with completely different physical models, many low-level details are shared including performant algorithms for short- and/or long-range particle interactions, multi-node particle communication patterns, and other data management tasks such as particle sorting and neighbor list construction. Cabana is a performance portable library for particle-based simulations, developed as part of the Co-Design Center for Particle Applications (CoPA) within the Exascale Computing Project (ECP) (Alexander et al., 2020). The CoPA project and its full development scope, including ECP partner applications, algorithm development, and similar software libraries for quantum MD, is described in (Mniszewski et al., 2021). Cabana uses the Kokkos library for on-node parallelism (Edwards et al., 2014; Trott et al., 2022), enabling simulation on multi-core CPU and GPU architectures, and MPI for GPU-aware, multi-node communication. Cabana provides particle simulation capabilities on almost all current Kokkos backends, including serial execution, OpenMP (including OpenMP-Target for GPUs), CUDA (NVIDIA GPUs), HIP (AMD GPUs), and SYCL (Intel GPUs), providing a clear path for the coming generation of accelerator-based exascale hardware. Cabana builds on Kokkos by providing new particle data structures and particle algorithms resulting in a similar execution policy-based, node-level programming model that is intended to be used in addition to the core Kokkos library within an application. Cabana is designed as an application and physics agnostic, but particle-specific toolkit which can either be used to generate a new application, or to be used as needed in existing applications at various levels of invasiveness including through interfaces that wrap user memory in existing data structures.

97 MATHEMATICS AND COMPUTING↗

Efficient Support for Matrix Computations on Heterogeneous Multi-core and Multi-GPU Architectures

We present a new methodology for utilizing all CPU cores and all GPUs on a heterogeneous multicore and multi-GPU system to support matrix computations e ciently. Our approach is able to achieve the objectives of a high degree of parallelism, minimized synchronization, minimized communication, and load balancing. Our main idea is to treat the heterogeneous system as a distributed-memory machine, and to use a heterogeneous 1-D block cyclic distribution to allocate data to the host system and GPUs to minimize communication. We have designed heterogeneous algorithms with two di erent tile sizes (one for CPU cores and the other for GPUs) to cope with processor heterogeneity. We propose an auto-tuning method to determine the best tile sizes to attain both high performance and load balancing. We have also implemented a new runtime system and applied it to the Cholesky and QR factorizations. Our experiments on a compute node with two Intel Westmere hexa-core CPUs and three Nvidia Fermi GPUs demonstrate good weak scalability, strong scalability, load balance, and e ciency of our approach.

97 MATHEMATICS AND COMPUTING↗

A Study of Geodesic Distance Kernel on an Integrated GPU

Graphics processing units (GPUs) are commonly used for graphics and general-purpose (GPGPU) computation. Currently, NVIDIA’s GPUs, through its CUDA framework for GPGPU programming, are ubiquitous for highperformance computing. Another class of GPUs, with a central processing unit (CPU) and a GPU integrated on the same chip, is commonly used in laptops, desktop computers, and low-power servers. While they are not designed to outperform discrete GPUs due to the power, area, and thermal constrains [1], there is a need to better understand the performance of a processor with an integrated GPU for floating-point intensive applications. The study helps us better understand the characteristics of hardware and software development tools, and the benefit of offloading such applications to an integrated GPU.

97 MATHEMATICS AND COMPUTING↗

Performance-portable Binary Neutron Star Mergers with AthenaK

We introduce an extension to the AthenaK code for general-relativistic magnetohydrodynamics (GRMHD) in dynamical spacetimes using a 3+1 conservative Eulerian formulation. Like the fixed-spacetime GRMHD solver, we use standard finite-volume methods to evolve the fluid and a constrained-transport scheme to preserve the divergence-free constraint for the magnetic field. We also utilize a first-order flux correction (FOFC) scheme to reduce the need for an artificial atmosphere and optionally enforce a maximum principle to improve robustness. We demonstrate the accuracy of AthenaK using a set of standard tests in flat and curved spacetimes. Using a SANE accretion disk around a Kerr black hole, we compare the new solver to the existing solver for stationary spacetimes using the so-called "HARM-like" formulation. We find that both formulations converge to similar results. We also include the first published binary neutron star (BNS) mergers performed on graphical processing units (GPUs). Thanks to the FOFC scheme, our BNS mergers maintain a relative error of $\mathcal{O}$(10 –11 ) or better in baryon mass conservation up to collapse. Finally, we perform scaling tests of AthenaK on OLCF Frontier, where we show excellent weak scaling of ≥80% efficiency up to 32,768 GPUs and 74% up to 65,536 GPUs for a GRMHD problem in dynamical spacetimes with six levels of mesh refinement. AthenaK achieves an order-of-magnitude speedup using GPUs compared to CPUs, demonstrating that it is suitable for performing numerical relativity problems on modern exascale resources.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Hardware Architectures for Data-Intensive Computing Problems: A Case Study for String Matching

DNA analysis is an emerging application of high performance bioinformatic. Modern sequencing machinery are able to provide, in few hours, large input streams of data, which needs to be matched against exponentially growing databases of known fragments. The ability to recognize these patterns effectively and fastly may allow extending the scale and the reach of the investigations performed by biology scientists. Aho-Corasick is an exact, multiple pattern matching algorithm often at the base of this application. High performance systems are a promising platform to accelerate this algorithm, which is computationally intensive but also inherently parallel. Nowadays, high performance systems also include heterogeneous processing elements, such as Graphic Processing Units (GPUs), to further accelerate parallel algorithms. Unfortunately, the Aho-Corasick algorithm exhibits large performance variability, depending on the size of the input streams, on the number of patterns to search and on the number of matches, and poses significant challenges on current high performance software and hardware implementations. An adequate mapping of the algorithm on the target architecture, coping with the limit of the underlining hardware, is required to reach the desired high throughputs. In this paper, we discuss the implementation of the Aho-Corasick algorithm for GPU-accelerated high performance systems. We present an optimized implementation of Aho-Corasick for GPUs and discuss its tradeoffs on the Tesla T10 and he new Tesla T20 (codename Fermi) GPUs. We then integrate the optimized GPU code, respectively, in a MPI-based and in a pthreads-based load balancer to enable execution of the algorithm on clusters and large sharedmemory multiprocessors (SMPs) accelerated with multiple GPUs.

Tumeo, Antonino↗

Revisiting Huffman Coding: Toward Extreme Performance on Modern GPU Architectures

Today's high-performance computing (HPC) applications are producing vast volumes of data, which are challenging to store and transfer efficiently during the execution, such that data compression is becoming a critical technique to mitigate the storage burden and data movement cost. Huffman coding is arguably the most efficient Entropy coding algorithm in information theory, such that it could be found as a fundamental step in many modern compression algorithms such as DEFLATE. On the other hand, today's HPC applications are more and more relying on the accelerators such as GPU on supercomputers, while Huffman encoding suffers from low throughput on GPUs, resulting in a significant bottleneck in the entire data processing. In this paper, we propose and implement an efficient Huffman encoding approach based on modern GPU architectures, which addresses two key challenges: (1) how to parallelize the entire Huffman encoding algorithm, including codebook construction, and (2) how to fully utilize the high memory-bandwidth feature of modern GPU architectures. The detailed contribution is fourfold. (1) We develop an efficient parallel codebook construction on GPUs that scales effectively with the number of input symbols. (2) We propose a novel reduction based encoding scheme that can efficiently merge the codewords on GPUs. (3) We optimize the overall GPU performance by leveraging the state-of-the-art CUDA APIs such as Cooperative Groups. (4) We evaluate our Huffman encoder thoroughly using six real-world application datasets on two advanced GPUs and compare with our implemented multithreaded Huffman encoder. Experiments show that our solution can improve the encoding throughput by up to 5.0× and 6.8× on NVIDIA RTX 5000 and V100, respectively, over the state-of-the-art GPU Huffman encoder, and by up to 3.3× over the multithread encoder on two 28-core Xeon Platinum 8280 CPUs.

Tian, Jiannan↗