Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “BANDWIDTH”

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 1,333 records · Page 74

Access Patterns and Performance Behaviors of Multi-layer Supercomputer I/O Subsystems under Production Load

Scientific computing workloads at HPC facilities have been shifting from traditional numerical simulations to AI/ML applications for training and inference while processing and producing ever-increasing amounts of scientific data. To address the growing need for increased storage capacity, lower access latency, and higher bandwidth, emerging technologies such as non-volatile memory are integrated into supercomputer I/O subsystems. With these emerging trends, we need a better understanding of the multilayer supercomputer I/O systems and ways to use these subsystems efficiently. In this work, we study the I/O access patterns and performance characteristics of two representative supercomputer I/O subsystems. Through an extensive analysis of year-long I/O logs on each system, we report new observations in I/O reads and writes, unbalanced use of storage system layers, and new trends in user behaviors at the HPC I/O middleware stack.

Bez, JL↗

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↗

Tailor : Altering Skip Connections for Resource-Efficient Inference

Deep neural networks use skip connections to improve training convergence. However, these skip connections are costly in hardware, requiring extra buffers and increasing on- and off-chip memory utilization and bandwidth requirements. In this article, we show that skip connections can be optimized for hardware when tackled with a hardware-software codesign approach. We argue that while a network’s skip connections are needed for the network to learn, they can later be removed or shortened to provide a more hardware-efficient implementation with minimal to no accuracy loss. We introduceTailor, a codesign tool whose hardware-aware training algorithm gradually removes or shortens a fully trained network’s skip connections to lower the hardware cost.Tailorimproves resource utilization by up to 34% for block random access memories (BRAMs), 13% for flip-flops (FFs), and 16% for look-up tables (LUTs) for on-chip, dataflow-style architectures.Tailorincreases performance by 30% and reduces memory bandwidth by 45% for a two-dimensional processing element array architecture.

Computer Science↗

CommBench: Micro-Benchmarking Hierarchical Networks with Multi-GPU, Multi-NIC Nodes

Modern high-performance computing systems have multiple GPUs and network interface cards (NICs) per node. The resulting network architectures have multilevel hierarchies of subnetworks with different interconnect and software technologies. These systems offer multiple vendor-provided communication capabilities and library implementations (IPC, MPI, NCCL, RCCL, OneCCL) with APIs providing varying levels of performance across the different levels. Understanding this performance is currently difficult because of the wide range of architectures and programming models (CUDA, HIP, OneAPI). We present CommBench, a library with cross-system portability and a high-level API that enables developers to easily build microbenchmarks relevant to their use cases and gain insight into the performance (bandwidth & latency) of multiple implementation libraries on different networks. We demonstrate CommBench with three sets of microbenchmarks that profile the performance of six systems. Our experimental results reveal the effect of multiple NICs on optimizing the bandwidth across nodes and also present the performance characteristics of four available communication libraries within and across nodes of NVIDIA, AMD, and Intel GPU networks.

Hidayetoglu, Mert↗

Studying CPU and memory utilization of applications on Fujitsu A64FX and Nvidia Grace Superchip

ARM-based manycore CPU architectures are well-positioned to provide the rising memory throughput requirements of modern data intensive scientific applications in High Performance Computing (HPC). The Fujitsu A64FX CPU platform is based on the ARM v8.2A architecture, and is the processor of the flagship Japanese supercomputer - "Fugaku", which was previously ranked as the #1 supercomputer in the world according to the Top500 list. The Nvidia Grace superchip features 144 Neoverse V2 cores based on the ARMv9 architecture with 4x128b SVE2, providing exceptional computational power. The chip supports up to 480GB of memory, making it ideal for AI, machine learning, and scientific computing workloads. In this paper, we conduct a thorough performance exploration of a variety of parallel bandwidth-sensitive benchmarks and applications compiled with the native Fujitsu compiler on a Fugaku A64FX compute node and ARM (LLVM) Compiler on an NVIDIA Grace superchip compute node, engaging all the computational cores per cluster using OpenMP multithreading (assuming the cores can drive the available bandwidth). Our ultimate goals are to study the resource utilization of scientific applications and benchmarks on A64FX and Grace superchip, considering graph application scenarios ( GAP Benchmark suite) and eleven appli- cation proxies from the Rodinia heterogeneous benchmark suite (considering domains such as Data Mining, Bioinformatics, Fluid Dynamics, Pattern Recognition, etc.). Through exhaustive performance monitoring, we quantify the resource utilization of diverse OpenMP-based HPC applications on both the Fujitsu A64FX and the Nvidia Grace Superchip platforms.

benchmarking, Performance Analysis, High performan↗

The Scalable Checkpoint/Restart Library

The Scalable Checkpoint/Restart (SCR) library provides an interface that codes may use to worite our and read in application-level checkpoints in a scalable fashion. In the current implementation, checkpoint files are cached in local storage (hard disk or RAM disk) on the compute nodes. This technique provides scalable aggregate bandwidth and uses storage resources that are fully dedicated to the job. This approach addresses the two common drawbacks of checkpointing a large-scale application to a shared parallel file system, namely, limited bandwidth and file system contention. In fact, on current platforms, SCR scales linearly with the number of compute nodes. It has been benchmarked as high as 720GB/s on 1094 nodes of Atlas, which is nearly two orders of magnitude faster thanthe parallel file system.

ECP↗

Livermore Big Artificial Neural Network Toolkit

LBANN is a toolkit that is designed to train artificial neural networks efficiently on high performance computing architectures. It is optimized to take advantages of key High Performance Computing features to accelerate neural network training. Specifically it is optimized for low-latency, high bandwidth interconnects, node-local NVRAM, node-local GPU accelerators, and high bandwidth parallel file systems. It is built on top of the open source Elemental distributed-memory dense and spars-direct linear algebra and optimization library that is released under the BSD license. The algorithms contained within LBANN are drawn from the academic literature and implemented to work within a distributed-memory framework.

Essen, BrianVan↗

RMA-MT Benchmarks

The RMA-MT benchmarks are a suite of performance tests for multi-threaded Message Passing Interface (MPI) Remote Memory Access (RMA) implementations. The suite includes tests for bandwidth and latency with any number of threads. This suite was originally developed and release by Sandia National Laboratory. LANL modifications to the suite include a complete re-write of much of the source code, bi-directional bandwidth, and message rate tests.

Hjelm, Nathan↗

Bulk Data Mover

Bulk Data Mover (BDM) is a high-level data transfer management tool. BDM handles the issue of large variance in file sizes and a big portion of small files by managing the file transfers with optimized transfer queue and concurrency management algorithms. For example, climate simulation data sets are characterized by large volume of files with extreme variance in file sizes. The BDN achieves high performance using a variety of techniques, including multi-threaded concurrent transfer connections, data channel caching, load balancing over multiple transfer servers, and storage i/o pre-fetching. Logging information from the BDM is collected and analyzed to study the effectiveness of the transfer management algorithms. The BDM can accept a request composed of multiple files or an entire directory. The request also contains the target site and directory where the replicated files will reside. If a directory is provided at the source, then the BDM will replicate the structure of the source directory at the target site. The BDM is capable of transferring multiple files concurrently as well as using parallel TCP streams. The optimal level of concurrency or parallel streams depends on the bandwidth capacity of the storage systems at both ends of the transfer as well as achievable bandwidth of the wide-area network. Hardware req.-PC, MAC, Multi-platform & Workstation; Software req.: Compile/version-Java 1.50_x or above; Type of files: source code, executable modules, installation instructions other, user guide; URL: http://sdm.lbl.gov/bdm/

Natarajan, Vijaya↗

MLEC-Sim: A Simulator for Evaluating Multi-Level Erasure Coding

We present MLEC-Sim, a sophisticated simulator for Multi-Level Erasure Coding (MLEC), developed in approximately 13 KLOC. The simulator is engineered to analyze the impact of various system configurations and erasure coding policies on system durability and network overhead. It supports a comprehensive range of parameters including disk capacity, disk I/O bandwidth, failure rates, network bandwidth, and system scale, accommodating various erasure coding approaches such as Single-Level Erasure Coding (SLEC), Multi-Level Erasure Coding (MLEC), and Local Reconstruction Codes (LRC). MLEC-Sim provides support for multiple chunk placement policies, including clustered parity and declustered parity, and encompasses a variety of repair methods like Repair-ALL, Repair-FCO, Repair-HYB, and Repair-MIN. It is capable of simulating disk failures through a variety of means, including distribution-based or trace-based mechanisms, and can handle complex multi-level (de)clustered placements and repair processes. A key feature of MLEC-Sim is its adoption of the splitting simulation method for evaluating system durabilities at extremely high levels, which are challenging to assess with traditional simulation approaches. This feature allows for a detailed evaluation of system resilience under a range of conditions, aiding in the selection of appropriate erasure coding solutions for enhancing system durability. MLEC-Sim contributes to the field of data storage and reliability by providing a tool for the detailed evaluation of the durability and efficiency of erasure coding configurations, intended for use by researchers and practitioners in the design and optimization of storage systems.

Wang, Meng↗

Detector Interface for Streaming, Control, and Open-source integration (DISCO) v1.0.0

This suite consists of a multi-package ecosystem featuring detector emulators, EPICS areaDetector drivers, and remote server frameworks designed for the Advanced Light Source (ALS). Engineered for high-bandwidth devices—including VFCCD, Timepix3, Timepix4, and related pixel detectors—the software simulates hardware, wraps vendor SDKs into remote-callable servers, and integrates with open-source control systems. Key Capabilities: Distributed SDK Architecture: Server packages wrap hardware-specific SDKs, allowing areaDetector drivers to execute remote framework calls. This isolates proprietary libraries from the EPICS IOC, enhancing stability and enabling distributed computing across beamline networks. Device Support: Custom drivers for VFCCD, the Timepix family, and similar sensors optimize the data path from hardware control to high-speed transport. Full-Stack Emulation: Sophisticated emulator packages allow end-to-end pipeline testing and software development without requiring physical hardware or beam time. Integrated Workflows: Supports high-bandwidth streaming for real-time analysis and robust, metadata-rich file-based workflows (e.g., HDF5/NeXus). By standardizing interfaces across heterogeneous hardware, this suite reduces technical debt. It provides the ALS with a scalable, open-source solution to manage massive data rates within a unified control environment.

Mahl, Johannes [Lawrence Berkeley National Laborat↗

A Global Climate Model Agent for High Spatial and Temporal Resolution Data

Fine cell granularity in modern climate models can produce terabytes of data in each snapshot, causing significant I/O overhead. To address this issue, a method of reducing the I/O latency of high-resolution climate models by identifying and selectively outputting regions of interest is presented. Working with a Global Cloud Resolving Model and running with up to 10240 processors on a Cray XE6, this method provides significant I/O bandwidth reduction depending on the frequency of writes and size of the region of interest. The implementation challenges of determining global parameters in a strictly core-localized model and properly formatting output files that only contain subsections of the global grid are addressed, as well as the overall bandwidth impact and benefits of the method. The gains in I/O throughput provided by this method allow dual output rates for high-resolution climate models: a low-frequency global snapshot as well as a high-frequency regional snapshot when events of particular interest occur.

global cloud resolving model↗

Collective input/output under memory constraints

Compared with current high-performance computing (HPC) systems, exascale systems are expected to have much less memory per node, which can significantly reduce necessary collective input/output (I/O) performance. In this study, we introduce a memory-conscious collective I/O strategy that takes into account memory capacity and bandwidth constraints. The new strategy restricts aggregation data traffic within disjointed subgroups, coordinates I/O accesses in intranode and internode layers, and determines I/O aggregators at run time considering memory consumption among processes. We have prototyped the design and evaluated it with commonly used benchmarks to verify its potential. The evaluation results demonstrate that this strategy holds promise in mitigating the memory pressure, alleviating the contention for memory bandwidth, and improving the I/O performance for projected extreme-scale systems. Given the importance of supporting increasingly data-intensive workloads and projected memory constraints on increasingly larger scale HPC systems, this new memory-conscious collective I/O can have a significant positive impact on scientific discovery productivity.

collective input/output↗

Epidemic failure detection and consensus for extreme parallelism

Future extreme-scale high-performance computing systems will be required to work under frequent component failures. The MPI Forum s User Level Failure Mitigation proposal has introduced an operation, MPI Comm shrink, to synchronize the alive processes on the list of failed processes, so that applications can continue to execute even in the presence of failures by adopting algorithm-based fault tolerance techniques. This MPI Comm shrink operation requires a failure detection and consensus algorithm. This paper presents three novel failure detection and consensus algorithms using Gossiping. The proposed algorithms were implemented and tested using the Extreme-scale Simulator. The results show that in all algorithms the number of Gossip cycles to achieve global consensus scales logarithmically with system size. The second algorithm also shows better scalability in terms of memory and network bandwidth usage and a perfect synchronization in achieving global consensus. The third approach is a three-phase distributed failure detection and consensus algorithm and provides consistency guarantees even in very large and extreme-scale systems while at the same time being memory and bandwidth efficient.

97 MATHEMATICS AND COMPUTING↗

Demystifying asynchronous I/O Interference in HPC applications

With increasing complexity of HPC workflows, data management services need to perform expensive I/O operations asynchronously in the background, aiming to overlap the I/O with the application runtime. However, this may cause interference due to competition for resources: CPU, memory/network bandwidth. The advent of multi-core architectures has exacerbated this problem, as many I/O operations are issued concurrently, thereby competing not only with the application but also among themselves. Furthermore, the interference patterns can dynamically change as a response to variations in application behavior and I/O subsystems (e.g. multiple users sharing a parallel file system). Without a thorough understanding, I/O operations may perform suboptimally, potentially even worse than in the blocking case. To fill this gap, here we investigate the causes and consequences of interference due to asynchronous I/O on HPC systems. Specifically, we focus on multi-core CPUs and memory bandwidth, isolating the interference due to each resource. Then, we perform an in-depth study to explain the interplay and contention in a variety of resource sharing scenarios such as varying priority and number of background I/O threads and different I/O strategies: sendfile, read/write, mmap/write underlining trade-offs. The insights from this study are important both to enable guided optimizations of existing background I/O, as well as to open new opportunities to design advanced asynchronous I/O strategies.

97 MATHEMATICS AND COMPUTING↗

Three-beam rotational coherent anti-Stokes Raman spectroscopy thermometry in scattering environments

Three-beam rotational coherent anti-Stokes Raman scattering (CARS) measurements performed in highly scattering environments are susceptible to contamination by two-beam CARS signals generated by the pump–probe and Stokes–probe interactions at the measurement volume. If this occurs, differences in the Raman excitation bandwidth between the two-beam and three-beam CARS signals can add significant errors to the spectral analysis. This interference to the best of our knowledge has not been acknowledged in previous three-beam rotational CARS experiments, but may introduce measurement errors up to 25% depending on the temperature, amount of scattering, and differences between the two-beam and three-beam Raman excitation bandwidths. Here, the presence of two-beam CARS signal contamination was experimentally verified using a femtosecond–picosecond rotational CARS instrument in two scattering environments: (1) a fireball generated by a laboratory-scale explosion that contained particulate matter, metal fragments, and soot, and (2) a flow of air and small liquid droplets. A polarization scheme is presented to overcome this interference. By rotating the pump and Stokes polarizations +55° and –55° from the probe, respectively, the two-beam and three-beam CARS signals are orthogonally polarized and can be separated using a polarization analyzer. Using this polarization arrangement, the Raman-resonant three-beam CARS signal amplitude is reduced by a factor of 2.3 compared to the case where all polarizations are parallel. This method is successfully demonstrated in both scattering environments. A theoretical model is presented, and the temperature measurement error is studied for different experimental conditions. The criteria for when this interference may be present are discussed.

42 ENGINEERING↗

Laser Doppler vibrometer employing active frequency feedback

We present a heterodyne Michelson interferometer for vibration measurement in which feedback is used to obviate the need to unwrap phase data. The Doppler shift of a vibrating target mirror is sensed interferometrically and compensated by means of a voltage-controlled oscillator driving an acousto-optic modulator. For frequencies within the servo bandwidth, the oscillator control voltage provides a direct measurement of the target velocity. Outside the servo bandwidth, phase-sensitive detection is used to evaluate high-frequency displacements. This approach is of great interest for the frequently-occurring situation where vibration amplitudes at low frequency exceed an optical wavelength, but knowledge of the vibration spectrum at high frequency is important as well.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Excess noise reduction by optical technique in amplitude-sensitive heterodyne interferometer for small differential phase detection

An amplitude-sensitive technique associated with a heterodyne interferometer for detecting small differential phase is reported. The excess noise with the amplitude-sensitive technique is reduced by optical subtraction instead of electronic subtraction. The differential phase introduced by the orthogonally polarized laser beams is converted to the amplitudes of two heterodyne interferometric signals, which presents amplitude and phase quadrature simultaneously. Thus the excess noise power and quantum noise power are both differential phase dependent. The advantages of differential and additive operations by optical technique and the real time differential phase determination without phase lock in are demonstrated experimentally. The theoretical signal-to-noise ratio (SNR) and minimum detectable differential phase are derived, which takes quantum noise and excess noise into consideration. The experimental results demonstrated the resolutions of differential phase detection closes to 10{sup -6} rad/{radical}Hz (10{sup -13} m/{radical}Hz) level over 100 kHz bandwidth and at 10{sup -8} rad/{radical}Hz (10{sup -15} m/{radical}Hz) level over 125 MHz bandwidth, respectively, under 2.5 mW incident power.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗