Search NASA⌕ Search

SEARCH · Search NASA

Results for “memory access optimization”

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 91 records · Page 5

Overcoming Pitfalls When Using HDF5 Compression

Compression of large datasets in S-NPP, JPSS and other HDF5 data files may substantially reduce their size, in turn reducing disk space requirements and file download time. However, mismatches between the layout of the files' datasets, the HDF5 instance's cache settings, choice of compression algorithm, and the access pattern of applications using the data can sometimes result in poor performance or exhausting machine resources when running the application. Whether designed in advance or modified in response to encountered problems, applications can be tuned to optimize efficiency of data access, avoid unnecessary repeated decompression, and reduce the amount of memory used. Examples will be given of problems that may be encountered, how to use available tools to diagnose or work around them, changing cache settings to conserve memory, and designing access strategies to avoid both performance and memory issues when creating or modifying applications.

performance↗

MDLoader: A Hybrid Model-Driven Data Loader for Distributed Graph Neural Network Training

Scalable data management is essential for processing large scientific dataset on HPC platforms for distributed deep learning. In-memory distributed storage is preferred for its speed, enabling rapid, random, and frequent data access required by stochastic optimizers. Processes use one-sided or collective communication to fetch remote data, with optimal performance depending on (i) dataset characteristics, (ii) training scale, and (iii) interconnection network. Empirical analysis shows collective communication excels with larger mini-batch sizes and/or fewer processes, whereas one-sided communication outperforms at larger scales. We propose MDLoader, a hybrid in-memory data loader for distributed graph neural network training. MDLoader features a model-driven performance estimator that dynamically selects between one-sided and collective communication at the beginning of training using Tree of Parzen Estimators (TPE). Evaluations on NERSC Perlmutter and OLCF Summit show MDLoader outperforms single-backend loaders by up to 2.83 × and predicts the suitable communication method with 96.3% (Perlmutter) and 94.3% (Summit) success rate.

Bae, Jonghyun↗

Critical Path-Based Thread Placement for NUMA Systems

Multicore multiprocessors use Non Uniform Memory Architecture (NUMA) to improve their scalability. However,NUMA introduces performance penalties due to remote memory accesses. Without efficiently managing data layout and thread mapping to cores, scientific applications, even if they are optimized for NUMA, may suffer performance loss. In this paper, we present an algorithm that optimizes the placement of OpenMP threads on NUMA processors. By collecting information from hardware counters and defining new metrics to capture the effects of thread placement, the algorithm reduces NUMA performance penalty by minimizing the critical path of OpenMP parallel regions and by avoiding local memory resource contention. We evaluate our algorithm with NPB benchmarks and achieve performance improvement between 8.13% and 25.68%, compared to the OS default scheduling.

99 GENERAL AND MISCELLANEOUS↗

OPER: Optimality-Guided Embedding Table Parallelization for Large-scale Recommendation Model

With the sharp increasing volume of user data, Deep Learning Recommendation Model (DLRM) becomes an indispensable infrastructure in large technology companies. However, large-scale DLRM on the multi-GPU platform is still inefficient due to unbalanced workload partitioning and intensive inter-GPU communication. To this end, we propose OPER, an OPtimality guided Embedding table placement for large-scale Recommendation model training and inference. OPER explores the potential of mitigating remote memory access latency in DLRM through fine-grained embedding table placement. Specifically, OPER proposes a theoretical modeling that builds up the relationship between EMT placement and the embedding communication latency in both training and inference. OPER proves the NP hardness of finding the optimal embedding table placement and proposes a heuristic algorithm that yields near optimal placement. OPER implements a SHMEM-based embedding table training system and a unified embedding index mapping to support fine-grained embedding table sharding and placement. Comprehensive experiments reveal that OPER achieves on average 3.4× and 5.1× speedup on training and inference respectively over state-of-the-art DLRM frameworks.

Wang, Zheng↗

Parameterized Micro-benchmarking: An Auto-tuning Approach for Complex Applications

Auto-tuning has emerged as an important practical method for creating highly optimized implementations of key computational kernels and applications. However, the growing complexity of architectures and applications is creating new challenges for auto-tuning. Complex applications can involve a prohibitively large search space that precludes empirical auto-tuning. Similarly, architectures are becoming increasingly complicated, making it hard to model performance. In this paper, we focus on the challenge to auto-tuning presented by applications with a large number of kernels and kernel instantiations. While these kernels may share a somewhat similar pattern, they differ considerably in problem sizes and the exact computation performed. We propose and evaluate a new approach to auto-tuning which we refer to as parameterized micro-benchmarking. It is an alternative to the two existing classes of approaches to auto-tuning: analytical model-based and empirical search-based. Particularly, we argue that the former may not be able to capture all the architectural features that impact performance, whereas the latter might be too expensive for an application that has several different kernels. In our approach, different expressions in the application, different possible implementations of each expression, and the key architectural features, are used to derive a simple micro-benchmark and a small parameter space. This allows us to learn the most significant features of the architecture that can impact the choice of implementation for each kernel. We have evaluated our approach in the context of GPU implementations of tensor contraction expressions encountered in excited state calculations in quantum chemistry. We have focused on two aspects of GPUs that affect tensor contraction execution: memory access patterns and kernel consolidation. Using our parameterized micro-benchmarking approach, we obtain a speedup of up to 2 over the version that used default optimizations, but no auto-tuning. We demonstrate that observations made from microbenchmarks match the behavior seen from real expressions. In the process, we make important observations about the memory hierarchy of two of the most recent NVIDIA GPUs, which can be used in other optimization frameworks as well.

auto-tuning↗

A Bandwidth-Optimized Multi-Core Architecture for Irregular Applications

This paper presents an architecture template for next-generation high performance computing systems specifically targeted to irregular applications. We start our work by considering that future generation interconnection and memory bandwidth full-system numbers are expected to grow by a factor of 10. In order to keep up with such a communication capacity, while still resorting to fine-grained multithreading as the main way to tolerate unpredictable memory access latencies of irregular applications, we show how overall performance scaling can benefit from the multi-core paradigm. At the same time, we also show how such an architecture template must be coupled with specific techniques in order to optimize bandwidth utilization and achieve the maximum scalability. We propose a technique based on memory references aggregation, together with the related hardware implementation, as one of such optimization techniques. We explore the proposed architecture template by focusing on the Cray XMT architecture and, using a dedicated simulation infrastructure, validate the performance of our template with two typical irregular applications. Our experimental results prove the benefits provided by both the multi-core approach and the bandwidth optimization reference aggregation technique.

bandwidth optimization↗

Efficient loading of reduced data ensembles produced at ORNL SNS/HFIR neutron time-of-flight facilities

We present algorithmic improvements to the loading operations of certain reduced data ensembles produced from neutron scattering experiments at Oak Ridge National Laboratory (ORNL) facilities. Ensembles from multiple measurements are required to cover a wide range of the phase space of a sample material of interest. They are stored using the standard NeXus schema on individual HDF5 files. This makes it a scalability challenge, as the number of experiments stored increases in a single ensemble file. The present work follows up on our previous efforts on data management algorithms, to address identified input output (I/O) bottlenecks in Mantid, an open-source data analysis framework used across several neutron science facilities around the world. We reuse an in-memory binary-tree metadata index that resembles data access patterns, to provide a scalable search and extraction mechanism. In addition, several memory operations are refactored and optimized for the current common use cases, ranging most frequently from 10 to 180, and up to 360 separate measurement configurations. Results from this work show consistent speed ups in wall-clock time on the Mantid LoadMD routine, ranging from 19% to 23% on average, on ORNL production computing systems. The latter depends on the complexity of the targeted instrument-specific data and the system I/O and compute variability for the shared computational resources available to users of ORNL’s Spallation Neutron Source (SNS) and the High Flux Isotope Reactor (HFIR) instruments. Nevertheless, we continue to highlight the need for more research to address reduction challenges as experimental data volumes, user time and processing costs increase.

Godoy, William↗

Critical Path-Based Thread Placement for NUMA Systems

Multicore multiprocessors use a Non Uniform Memory Architecture (NUMA) to improve their scalability. However, NUMA introduces performance penalties due to remote memory accesses. Without efficiently managing data layout and thread mapping to cores, scientific applications, even if they are optimized for NUMA, may suffer performance loss. In this paper, we present algorithms and a runtime system that optimize the execution of OpenMP applications on NUMA architectures. By collecting information from hardware counters, the runtime system directs thread placement and reduces performance penalties by minimizing the critical path of OpenMP parallel regions. The runtime system uses a scalable algorithm that derives placement decisions with negligible overhead. We evaluate our algorithms and runtime system with four NPB applications implemented in OpenMP. On average the algorithms achieve between 8.13% and 25.68% performance improvement compared to the default Linux thread placement scheme. The algorithms miss the optimal thread placement in only 8.9% of the cases.

97 MATHEMATICS AND COMPUTING↗

VINE-A NUMERICAL CODE FOR SIMULATING ASTROPHYSICAL SYSTEMS USING PARTICLES. II. IMPLEMENTATION AND PERFORMANCE CHARACTERISTICS

We continue our presentation of VINE. In this paper, we begin with a description of relevant architectural properties of the serial and shared memory parallel computers on which VINE is intended to run, and describe their influences on the design of the code itself. We continue with a detailed description of a number of optimizations made to the layout of the particle data in memory and to our implementation of a binary tree used to access that data for use in gravitational force calculations and searches for smoothed particle hydrodynamics (SPH) neighbor particles. We describe the modifications to the code necessary to obtain forces efficiently from special purpose 'GRAPE' hardware, the interfaces required to allow transparent substitution of those forces in the code instead of those obtained from the tree, and the modifications necessary to use both tree and GRAPE together as a fused GRAPE/tree combination. We conclude with an extensive series of performance tests, which demonstrate that the code can be run efficiently and without modification in serial on small workstations or in parallel using the OpenMP compiler directives on large-scale, shared memory parallel machines. We analyze the effects of the code optimizations and estimate that they improve its overall performance by more than an order of magnitude over that obtained by many other tree codes. Scaled parallel performance of the gravity and SPH calculations, together the most costly components of most simulations, is nearly linear up to at least 120 processors on moderate sized test problems using the Origin 3000 architecture, and to the maximum machine sizes available to us on several other architectures. At similar accuracy, performance of VINE, used in GRAPE-tree mode, is approximately a factor 2 slower than that of VINE, used in host-only mode. Further optimizations of the GRAPE/host communications could improve the speed by as much as a factor of 3, but have not yet been implemented in VINE. Finally, we find that although parallel performance on small problems may reach a plateau beyond which more processors bring no additional speedup, performance never decreases, a factor important for running large simulations on many processors with individual time steps, where only a small fraction of the total particles require updates at any given moment.

79 ASTRONOMY AND ASTROPHYSICS↗

Threaded Multi-Core GEMM with MoA and Cache-Blocking: Preprint

A threaded multi-core implementation of the high performance dense linear algebra matrix-matrix multiply GEMM kernel is described. This kernel is widely implemented by vendors in the basic linear algebra subroutine BLAS library. The mathematics of arrays (MoA) paradigm due to Mullin (1988) results in contiguous memory accesses by employing outer-product forms. Our performance studies demonstrate that the MoA implementation of double precision DGEMM combined with optimal cache-blocking strategies results in at least a 25% performance gain on the Intel Xeon Skylake processor over the vendor supplied Intel MKL basic linear algebra libraries. Results are presented for the NREL Eagle supercomputer. The multi-core DGEMM achieves over 100 GigaFlops/sec with eight openMP threads.

cache-blocking↗

Petascale solvers for anisotropic PDEs in atmospheric modelling on GPU clusters

Memory bound applications such as solvers for large sparse systems of equations remain a challenge for GPUs. Fast solvers should be based on numerically efficient algorithms and implemented such that global memory access is minimised. To solve systems with trillions ( O ( 10 12 ) ) unknowns the code has to make efficient use of several million individual processor cores on large GPU clusters. We describe the multi-GPU implementation of two algorithmically optimal iterative solvers for anisotropic PDEs which are encountered in (semi-) implicit time stepping procedures in atmospheric modelling. In this application the condition number is large but independent of the grid resolution and both methods are asymptotically optimal, albeit with different absolute performance. In particular, an important constant in the discretisation is the CFL number; only the multigrid solver is robust to changes in this constant. We parallelise the solvers and adapt them to the specific features of GPU architectures, paying particular attention to efficient global memory access. We achieve a performance of up to 0.78 PFLOPs when solving an equation with 0.55 · 10 12 unknowns on 16384 GPUs; this corresponds to about 3% of the theoretical peak performance of the machine and we use more than 40% of the peak memory bandwidth with a Conjugate Gradient (CG) solver. Although the other solver, a geometric multigrid algorithm, has a slightly worse performance in terms of FLOPs per second, overall it is faster as it needs less iterations to converge; the multigrid algorithm can solve a linear PDE with half a trillion unknowns in about one second.

Computer Science↗

arco (Assembled Resource-Constrained Optimization) [SWR-26-030]

Arco (Assembled Resource-Constrained Optimization) is a memory-smart optimization DSL and solver for LP and MIP problems on constrained hardware. The software is an optimization framework built around a KDL-based domain-specific language and a CLI compiler/solver. You write optimization models in .kdl files, and the arco CLI compiles, validates, inspects, and solves them. Language bindings (Python today, more planned) provide programmatic access to the same engine. Built for harder optimization problems on constrained resources, Arco is intentional about every allocation, careful with stack and heap behavior, and relentless about minimizing memory usage so more systems can run real workloads. Arco is built primarily for internal use within our organization. You are welcome to try it, but we make no guarantees about API stability or robustness at this stage

Sanchez Perez, Pedro Andres [National Laboratory o↗

SU-E-T-258: Parallel Optimization of Beam Configurations for CyberKnife Treatments

Purpose: The CyberKnife delivers a large number of beams originating at different non-planar positions and with different orientation. We study how much the quality of treatment plans depends on the beams considered during plan optimization. Particularly, we evaluate a new approach to search for optimal treatment plans in parallel by running optimization steps concurrently. Methods: So far, no deterministic, complete and efficient method to select the optimal beam configuration for robotic SRS/SBRT is known. Considering a large candidate beam set increases the likelihood to achieve a good plan, but the optimization problem becomes large and impractical to solve. We have implemented an approach that parallelizes the search by solving multiple linear programming problems concurrently while iteratively resampling zero weighted beams. Each optimization problem contains the same set of constraints but different variables representing candidate beams. The search is synchronized by sharing the resulting basis variables among the parallel optimizations. We demonstrate the utility of the approach based on an actual spinal case with the objective to improve the coverage. Results: The objective function is falling and reaches a value of 5000 after 49, 31, 25 and 15 iterations for 1, 2, 4, and 8 parallel processes. This corresponds to approximately 97% coverage in 77%, 59%, and 36% of the mean number of iterations with one process for 2, 4, and 8 parallel processes, respectively. Overall, coverage increases from approximately 91.5% to approximately 98.5%. Conclusion: While on our current computer with uniform memory access the reduced number of iterations does not translate into a similar speedup, the approach illustrates how to effectively parallelize the search for the optimal beam configuration. The experimental results also indicate that for complex geometries the beam selection is critical for further plan optimization.

07 ISOTOPE AND RADIATION SOURCES↗

A Synthesis Methodology for Intelligent Memory Interfaces in Accelerator Systems

Domain-specific systems improve the performance of a specific set of applications compared to general-purpose processing systems by deploying custom hardware accelerators. These hardware accelerators are generated using high-level synthesis (HLS) tools. The HLS tools enable a comprehensive design space exploration to optimize the compute performance of the generated accelerators. However, they often ignore the challenges of implementing the accelerators in a system-on-chip, particularly how the accelerators access memory. Our work introduces a buffering system design that improves accelerators' memory accesses by intelligently employing burst transactions to prefetch useful data from external memory to on-chip local buffers. Our design is dynamic, parametric, and transparent to the accelerators generated by HLS tools. We derive the buffering system parameters using appropriate compiler-based analysis passes and memory channel latency constraints. The proposed buffering system design results in, on average, 8.8x performance improvements while lowering memory channel utilization on average by 53.2% for a set of PolyBench kernels.

Limaye, Ankur M. (ORCID:0000000194062584)↗

Accessible Content Optimization for Research Needs (ACORN)

ACORN employs a set of automated processes for informing and/or enforcing defined content schemas to create standardized and highly structured data. Because of its standardized data source, ACORN easily applies computer automation to generate communication assets such as PDFs, Powerpoint presentations, and web pages. Built using the memory-safe Rust programming language, ACORN is portable and accessible for use on any Windows, Mac, or Linux machine.

Wohlgemuth, JasonHoward [Oak Ridge National Labora↗

Optimizing Data Movement for GPU-Based In-Situ Workflow Using GPUDirect RDMA

The extreme-scale computing landscape is increasingly dominated by GPU-accelerated systems. At the same time, in-situ workflows that employ memory-to-memory inter-application data exchanges have emerged as an effective approach for leveraging these extreme-scale systems. In the case of GPUs, GPUDirect RDMA enables third-party devices, such as network interface cards, to access GPU memory directly and has been adopted for intra-application communications across GPUs. In this paper, we present an interoperable framework for GPU-based in-situ workflows that optimizes data movement using GPUDirect RDMA. Specifically, we analyze the characteristics of the possible data movement pathways between GPUs from an in-situ workflow perspective, and design a strategy that maximizes throughput. Furthermore, we implement this approach as an extension of the DataSpaces data staging service, and experimentally evaluate its performance and scalability on a current leadership GPU cluster. The performance results show that the proposed design reduces data-movement time by up to 53% and 40% for the sender and receiver, respectively, and maintains excellent scalability for up to 256 GPUs.

Zhang, Bo↗

BULKI-Store v0.3.2

BULKI-Store is a distributed object storage system optimized for high-performance computing environments. Built with a Rust core and Python bindings, it efficiently manages scientific and machine learning datasets across HPC clusters. The system employs a client-server architecture with MPI integration, enabling seamless scaling on supercomputers like Perlmutter. BULKI-Store's object-oriented approach provides intuitive data organization with rich metadata support, contrasting with traditional file-based solutions. Key optimizations include selective checkpoint loading, unified checkpoint files, and object chunking for large data transfers. For machine learning workloads, BULKI-Store offers advantages through fine-grained access patterns, dynamic data sharing between training instances, and reduced memory pressure. Memory management features include strategic Python GC calls, minimized data copies, and batch processing capabilities. The system leverages Rayon's thread pool for asynchronous data prefetching and supports multiple CPU architectures (ARM64, x86, AMD, RISC-V). By combining performance optimizations with developer-friendly APIs, BULKI-Store addresses the complex data management challenges of modern HPC applications while maintaining compatibility across heterogeneous computing environments.

Zhang, Wei [Lawrence Berkeley National Laboratory ↗

OpenMP Parallelization and Optimization of Graph-Based Machine Learning Algorithms

In this paper, we investigate the OpenMP parallelization and optimization of two novel data classification algorithms. The new algorithms are based on graph and PDE solution techniques and provide significant accuracy and performance advantages over traditional data classification algorithms in serial mode. The methods leverage the Nystrom extension to calculate eigenvalue/eigenvectors of the graph Laplacian and this is a self-contained module that can be used in conjunction with other graph-Laplacian based methods such as spectral clustering. We use performance tools to collect the hotspots and memory access of the serial codes and use OpenMP as the parallelization language to parallelize the most time-consuming parts. Where possible, we also use library routines. We then optimize the OpenMP implementations and detail the performance on traditional supercomputer nodes (in our case a Cray XC30), and test the optimization steps on emerging testbed systems based on Intel’s Knights Corner and Landing processors. We show both performance improvement and strong scaling behavior. Finally, a large number of optimization techniques and analyses are necessary before the algorithm reaches almost ideal scaling.

97 MATHEMATICS AND COMPUTING↗