Search NASA⌕ Search

SEARCH · Search NASA

Results for “bandwidth 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 253 records · Page 14
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

High efficiency Al/Sc-based multilayer coatings in the EUV wavelength range above 40 nanometers

In this Letter, we have developed new and highly efficient periodic multilayer mirrors Al/Sc, Al/Sc/SiC, and Mo/Al/Sc with optimized reflectance at wavelengths between 40 and 65 nm. We have reached record values in measured peak reflectance: 57.5% at 44.7 nm and 46.5% at 51 nm, with Al/Sc/SiC at near-normal incidence. Furthermore, to the best of our knowledge, we have achieved the largest reported bandwidth with Mo/Al/Sc at 57 nm and the narrowest bandwidth with Al/Sc at a 60 nm wavelength. These new and promising results demonstrate that Al/Sc-based multilayer coatings are excellent candidates for future generations of extreme ultraviolet (EUV) instruments for solar physics, EUV lasers, and attosecond science, in a wavelength range that has not been fully explored.

36 MATERIALS SCIENCE↗
DOE OSTI2012

Modeling and Validation of Performance Limitations for the Optimal Design of Interferometric and Intensity-Modulated Fiber Optic Displacement Sensors

Optical fiber sensors offer advantages over traditional electromechanical sensors, making them particularly well-suited for certain measurement applications. Generally speaking, optical fiber sensors respond to a desired measurand through modulation of an optical signal's intensity, phase, or wavelength. Practically, non-contacting fiber optic displacement sensors are limited to intensity-modulated and interferometric (or phase-modulated) methodologies. Intensity-modulated fiber optic displacement sensors relate target displacement to a power measurement. The simplest intensity-modulated sensor architectures are not robust to environmental and hardware fluctuations, since such variability may cause changes in the measured power level that falsely indicate target displacement. Differential intensity-modulated sensors have been implemented, offering robustness to such intensity fluctuations, and the speed of these sensors is limited only by the combined speed of the photodetection hardware and the data acquisition system (kHz-MHz). The primary disadvantages of intensity-modulated sensing are the relatively low accuracy (?m-mm for low-power sensors) and the lack of robustness, which consequently must be designed, often with great difficulty, into the sensor's architecture. White light interferometric displacement sensors, on the other hand, offer increased accuracy and robustness. Unlike their monochromatic-interferometer counterparts, white light interferometric sensors offer absolute, unambiguous displacement measurements over large displacement ranges (cm for low-power, 5 mW, sources), necessitating no initial calibration, and requiring no environmental or feedback control. The primary disadvantage of white light interferometric displacement sensors is that their utility in dynamic testing scenarios is limited, both by hardware bandwidth and by their inherent high-sensitivity to Doppler-effects. The decision of whether to use either an intensity-modulated interferometric sensor depends on an appropriate performance function (e.g., desired displacement range, accuracy, robustness, etc.). In this dissertation, the performance limitations of a bundled differential intensity-modulated displacement sensor are analyzed, where the bundling configuration has been designed to optimize performance. The performance limitations of a white light Fabry-Perot displacement sensor are also analyzed. Both these sensors are non-contacting, but they have access to different regions of the performance-space. Further, both these sensors have different degrees of sensitivity to experimental uncertainty. Made in conjunction with careful analysis, the decision of which sensor to deploy need not be an uninformed one.

42 ENGINEERING↗
DOE OSTI2012

Ideal-observer detectability in photon-counting differential phase-contrast imaging using a linear-systems approach

Purpose: To provide a cascaded-systems framework based on the noise-power spectrum (NPS), modulation transfer function (MTF), and noise-equivalent number of quanta (NEQ) for quantitative evaluation of differential phase-contrast imaging (Talbot interferometry) in relation to conventional absorption contrast under equal-dose, equal-geometry, and, to some extent, equal-photon-economy constraints. The focus is a geometry for photon-counting mammography. Methods: Phase-contrast imaging is a promising technology that may emerge as an alternative or adjunct to conventional absorption contrast. In particular, phase contrast may increase the signal-difference-to-noise ratio compared to absorption contrast because the difference in phase shift between soft-tissue structures is often substantially larger than the absorption difference. We have developed a comprehensive cascaded-systems framework to investigate Talbot interferometry, which is a technique for differential phase-contrast imaging. Analytical expressions for the MTF and NPS were derived to calculate the NEQ and a task-specific ideal-observer detectability index under assumptions of linearity and shift invariance. Talbot interferometry was compared to absorption contrast at equal dose, and using either a plane wave or a spherical wave in a conceivable mammography geometry. The impact of source size and spectrum bandwidth was included in the framework, and the trade-off with photon economy was investigated in some detail. Wave-propagation simulations were used to verify the analytical expressions and to generate example images. Results: Talbot interferometry inherently detects the differential of the phase, which led to a maximum in NEQ at high spatial frequencies, whereas the absorption-contrast NEQ decreased monotonically with frequency. Further, phase contrast detects differences in density rather than atomic number, and the optimal imaging energy was found to be a factor of 1.7 higher than for absorption contrast. Talbot interferometry with a plane wave increased detectability for 0.1-mm tumor and glandular structures by a factor of 3-4 at equal dose, whereas absorption contrast was the preferred method for structures larger than {approx}0.5 mm. Microcalcifications are small, but differ from soft tissue in atomic number more than density, which is favored by absorption contrast, and Talbot interferometry was barely beneficial at all within the resolution limit of the system. Further, Talbot interferometry favored detection of 'sharp' as opposed to 'smooth' structures, and discrimination tasks by about 50% compared to detection tasks. The technique was relatively insensitive to spectrum bandwidth, whereas the projected source size was more important. If equal photon economy was added as a restriction, phase-contrast efficiency was reduced so that the benefit for detection tasks almost vanished compared to absorption contrast, but discrimination tasks were still improved close to a factor of 2 at the resolution limit. Conclusions: Cascaded-systems analysis enables comprehensive and intuitive evaluation of phase-contrast efficiency in relation to absorption contrast under requirements of equal dose, equal geometry, and equal photon economy. The benefit of Talbot interferometry was highly dependent on task, in particular detection versus discrimination tasks, and target size, shape, and material. Requiring equal photon economy weakened the benefit of Talbot interferometry in mammography.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2012

Modeling and Validation of Performance Limitations for the Optimal Design of Interferometric and Intensity-Modulated Fiber Optic Displacement Sensors [Dissertation]

Optical fiber sensors offer advantages over traditional electromechanical sensors, making them particularly well-suited for certain measurement applications. Generally speaking, optical fiber sensors respond to a desired measurand through modulation of an optical signal’s intensity, phase, or wavelength. Practically, non-contacting fiber optic displacement sensors are limited to intensity-modulated and interferometric (or phase-modulated) methodologies. Intensity-modulated fiber optic displacement sensors relate target displacement to a power measurement. The simplest intensity-modulated sensor architectures are not robust to environmental and hardware fluctuations, since such variability may cause changes in the measured power level that falsely indicate target displacement. Differential intensity-modulated sensors have been implemented, offering robustness to such intensity fluctuations, and the speed of these sensors is limited only by the combined speed of the photodetection hardware and the data acquisition system (kHz-GHz). The primary disadvantages of intensity-modulated sensing are the relatively low accuracy (μm-mm for low-power sensors) and the lack of robustness, which consequently must be designed, often with great difficulty, into the sensor’s architecture. White light interferometric displacement sensors, on the other hand, offer increased accuracy and robustness. Unlike their monochromatic-interferometer counterparts, white light interferometric sensors offer absolute, unambiguous displacement measurements over large displacement ranges (cm for low-power, 5 mW, sources), necessitating no initial calibration, and requiring no environmental or feedback control. The primary disadvantage of white light interferometric displacement sensors is that their utility in dynamic testing scenarios is limited, both by hardware bandwidth and by their inherent high-sensitivity to Doppler-effects. The decision of whether to use either an intensity-modulated interferometric sensor depends on an appropriate performance function (e.g., desired displacement range, accuracy, robustness, etc.). In this dissertation, the performance limitations of a bundled differential intensity-modulated displacement sensor are analyzed, where the bundling configuration has been designed to optimize performance. The performance limitations of a white light Fabry-Perót displacement sensor are also analyzed. Both these sensors are non-contacting, but they have access to different regions of the performance-space. Further, both these sensors have different degrees of sensitivity to experimental uncertainty. Made in conjunction with careful analysis, the decision of which sensor to deploy need not be an uninformed one.

Engineering(42)↗
DOE OSTI2015

Overview of the ultrasonic instrumentation research in the MYRRHA project

The Belgian Nuclear Research Centre SCK.CEN is in the process of developing MYRRHA, a new generation IV fast flux research reactor to replace the aging BR2. MYRRHA is conceptualized as an accelerator driven system cooled with lead bismuth eutectic mixture (LBE). As LBE is opaque to visual light, ultrasonic measurement techniques are employed as the main technology to provide feedback where needed. This paper we will give an overview of the R and D at SCK.CEN with respect to ultrasonic instrumentation in heavy liquid metals. High temperature ultrasonic transducers are deployed into the reactor to generate and receive the required ultrasonic signals. The ultrasonic waves are generated and sensed by means of a piezo-electric disc at the heart of the transducer. The acoustic properties of commonly used piezo-electric materials match rather well with the acoustic properties of heavy liquid metals, simplifying the design and construction of high bandwidth ultrasonic transducers for use in heavy liquid metals. The ultrasonic transducers will operate in a liquid metal environment, where radiation and high temperature limit the choice of materials for construction. Moreover, the high surface tension of the liquid metal hinders proper wetting of the transducer, required for optimal transmission and reception of the ultrasonic waves. In a first part of the paper, we will discuss the effect of these parameters on the performance of the overall ultrasonic system. In the second part of the paper, past, present and future ultrasonic experiments in LBE will be reviewed. We will show the results of an experiment where a transducer is scanned near the free surface of an LBE pool to render ultrasonic images of objects submerged in the heavy liquid metal. Additionally, the preliminary results of an ongoing experiment that measures the evolution of LBE wetting on different types of metals and various surface conditions will be reported. The evolution of wetting is an important parameter to assess the performance of ultrasonic systems that are submerged in LBE over longer time periods. In the last part we will briefly present the overall design of the two main ultrasonic systems that will be deployed in MYRRHA: the ultrasonic fuel identification system and the ultrasonic lost fuel localization system. (authors)

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2016

Composition and Realization of Source-to-Sink High-Performance Flows: File Systems, Storage, Hosts, LAN and WAN

A number of Department of Energy (DOE) science applications, involving exascale computing systems and large experimental facilities, are expected to generate large volumes of data, in the range of petabytes to exabytes, which will be transported over wide-area networks for the purpose of storage, visualization, and analysis. To support such capabilities, significant progress has been made in various components including the deployment of 100 Gbps networks with future 1 Tbps bandwidth, increases in end-host capabilities with multiple cores and buses, capacity improvements in large disk arrays, and deployment of parallel file systems such as Lustre and GPFS. High-performance source-to-sink data flows must be composed of these component systems, which requires significant optimizations of the storage-to-host data and execution paths to match the edge and long-haul network connections. In particular, end systems are currently supported by 10-40 Gbps Network Interface Cards (NIC) and 8-32 Gbps storage Host Channel Adapters (HCAs), which carry the individual flows that collectively must reach network speeds of 100 Gbps and higher. Indeed, such data flows must be synthesized using multicore, multibus hosts connected to high-performance storage systems on one side and to the network on the other side. Current experimental results show that the constituent flows must be optimally composed and preserved from storage systems, across the hosts and the networks with minimal interference. Furthermore, such a capability must be made available transparently to the science users without placing undue demands on them to account for the details of underlying systems and networks. And, this task is expected to become even more complex in the future due to the increasing sophistication of hosts, storage systems, and networks that constitute the high-performance flows. The objectives of this proposal are to (1) develop and test the component technologies and their synthesis methods to achieve source-to-sink high-performance flows, and (2) develop tools that provide these capabilities through simple interfaces to users and applications. In terms of the former, we propose to develop (1) optimization methods that align and transition multiple storage flows to multiple network flows on multicore, multibus hosts; and (2) edge and long-haul network path realization and maintenance using advanced provisioning methods including OSCARS and OpenFlow. We also propose synthesis methods that combine these individual technologies to compose high-performance flows using a collection of constituent storage-network flows, and realize them across the storage and local network connections as well as long-haul connections. We propose to develop automated user tools that profile the hosts, storage systems, and network connections; compose the source-to-sink complex flows; and set up and maintain the needed network connections. These solutions will be tested using (1) 100 Gbps connection(s) between Oak Ridge National Laboratory (ORNL) and Argonne National Laboratory (ANL) with storage systems supported by Lustre and GPFS file systems with an asymmetric connection to University of Memphis (UM); (2) ORNL testbed with multicore and multibus hosts, switches with OpenFlow capabilities, and network emulators; and (3) 100 Gbps connections from ESnet and their Openflow testbed, and other experimental connections. This proposal brings together the expertise and facilities of the two national laboratories, ORNL and ANL, and UM. It also represents a collaboration between DOE and the Department of Defense (DOD) projects at ORNL by sharing technical expertise and personnel costs, and leveraging the existing DOD Extreme Scale Systems Center (ESSC) facilities at ORNL.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

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 multi-threaded Huffman encoder. Experiments show that our solution can improve the encoding throughput by up to 5.0x and 6.8x on NVIDIA RTX 5000 and V100, respectively, over the state-of-the-art GPU Huffman encoder, and by up to 3.3x over the multi-thread encoder on two 28-core Xeon Platinum 8280 CPUs.

Tian, Jiannan↗
DOE OSTI2021

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↗
DOE OSTI2013

TOPS2 - Towards Optimal Petascale Simulations

Our goal in this project was to design scalable numerical algorithms needed by SciDAC applications that adapt to use evolving hardware resources as efficiently as possible. Our primary challenge is minimizing communication costs, where communication means moving data either between levels of a memory hierarchy (L1 cache to L2 cache to main memory etc.) or between processors over a network. Floating point rates are improving exponentially faster than bandwidth, which is improving exponentially faster than latency. So our goal is to minimize communication. We describe our progress in this area, both for direct and iterative linear algebra. In both areas we have (1) identified lower bounds on the amount of communication (measured both by the number of words moved and the number of messages) required to perform these algorithms, (2) analyzed existing algorithms, which by and large do not attain these lower bounds, and (3) identified or invented new algorithms that do attain them, and evaluated their speedups, which can be quite large.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2014

Simulation at Extreme-Scale: Co-Design Thinking and Practices

The current trend of computer architecture evolving towards exaflop/s is the fast increasing floating point performance (the so-called “free” flops) accompanied by much slowly improving the bandwidth of memory and network. Numerical simulation would undergo the challenge posed by the unbalanced increase in the compute power and the capability of data movement. In this paper, after reviewing the challenges of hardware and software in moving towards exascale computing, we present co-design thinking for selecting, optimizing, and developing a numerical algorithm and a simulation tool to meet the challenge of simulation at extreme scale. Examples are presented to demonstrate the new way of thinking and its effectiveness on the emerging architecture.

Computer Science↗
DOE OSTI2023

Ultra: Underwater Laser Telecommunications & Remote Access (Final Report)

High speed wireless communication has proven elusive in subsea environments due to the inherent bandwidth limitations of acoustics and range limitations of other transmission modalities. A truly connected subsea system necessitates a high-speed, resilient architecture that can enable the integration of new sensor technologies and edge analytics and allow closed-loop monitoring and control of subsea operations for integrity monitoring and optimization. Like terrestrial Internet of Things applications, the realization of this “digital subsea” vision requires the application of high speed, point-to-point wireless technologies to complement rather than replace “hard-wired” communications such as optical fiber or acoustic systems. This work addresses the development of ULTRA (Underwater LASER Telemetry and Remote Access), an ultra-long range underwater laser communications system for use in critical points of the subsea communications architecture to increase reliability, operational flexibility, and reduce communication system maintenance associated with physical subsea connections.

02 PETROLEUM↗
DOE OSTI2015

Design and development of laser damage resistant low group delay dispersion optical coatings for high reflection at 45o incidence P polarization and wavelengths from 800 to 1000 nm.

This paper describes the modelling and design development of an optical coating that is suitable for broad bandwidth high reflection (BBHR) at 45° angle of incidence (AOI), P polarization (Ppol) and fs-class laser pulses whose frequencies correspond to wavelengths from 800 to 1000 nm, and that can eventually be produced uniformly on meter-class optical substrates. The coating design process was guided by specifications of not only high reflection but also high laser-induced damage threshold (LIDT) as well as low group delay dispersion (GDD) for reflected light over the broad, 200 nm bandwidth in order to minimize temporal broadening of the fs pulses upon reflection. The coating is based on TiO 2 /SiO 2 layer pairs by means of e-beam evaporation with ion-assisted deposition (IAD). We used OptiLayer Thin Film Software to explore coating designs with a limited optimization process starting from TiO 2 /SiO 2 layer pairs with layer thicknesses in an opposite “chirp” arrangement. This approach proved to be successful, leading to a design with R > 99.5% from 801 – 999 nm and GDD < 20 fs 2 from 843 – 949 nm (45° AOI, Ppol). The GDD behaves in a smooth way that lends itself to compensation of GDD effects. Also, the electric field intensities are favorable to high LIDT in that they quench rapidly into the outer coating layers or are of moderate strength, or they are located in the higher band gap SiO 2 layers.

36 MATERIALS SCIENCE↗
DOE OSTI2012

Advanced X-Band Test Accelerator for High Brightness Electron and Gamma Ray Beams

In support of Compton scattering gamma-ray source efforts at LLNL, a multi-bunch test stand is being developed to investigate accelerator optimization for future upgrades. This test stand will enable work to explore the science and technology paths required to boost the current 10 Hz monoenergetic gamma-ray (MEGa-Ray) technology to an effective repetition rate exceeding 1 kHz, potentially increasing the average gamma-ray brightness by two orders of magnitude. Multiple bunches must be of exceedingly high quality to produce narrow-bandwidth gamma-rays. Modeling efforts will be presented, along with plans for a multi-bunch test stand at LLNL. The test stand will consist of a 5.5 cell X-band rf photoinjector, single accelerator section, and beam diagnostics. The photoinjector will be a high gradient standing wave structure, featuring a dual feed racetrack coupler. The accelerator will increase the electron energy so that the emittance can be measured using quadrupole scanning techniques. Multi-bunch diagnostics will be developed so that the beam quality can be measured and compared with theory. Design will be presented with modeling simulations, and layout plans.

43 PARTICLE ACCELERATORS↗
DOE OSTI2022

A flexible neutron spectrometer concept with a new ultra-high field steady-state vertical-bore magnet

The proposed facility explores materials under ultra-high magnetic fields. By combining the power of high fields to tune materials and of neutron scattering to probe the resulting changes down to the atomic scale, this facility will enable transformative progress in the study of quantum materials and is named for the “TITAN” subset of Greek gods to reflect this transformation. TITAN will offer DC magnetic fields up to at least 20 T. Exploiting the record brightness and bandwidth of the Second Target Station at the Spallation Neutron Source, TITAN will probe atomic-scale responses through high efficiency neutron spectroscopy up to 80 meV energy transfer, high resolution diffraction, and small angle neutron scattering. Focusing neutron optics will maximize flux on accurately positioned samples, while radial collimation and optimized shielding and detection strategies will minimize backgrounds.

47 OTHER INSTRUMENTATION↗
DOE OSTI2014

Scaling Irregular Applications through Data Aggregation and Software Multithreading

Bioinformatics, data analytics, semantic databases, knowledge discovery are emerging high performance application areas that exploit dynamic, linked data structures such as graphs, unbalanced trees or unstructured grids. These data structures usually are very large, requiring significantly more memory than available on single shared memory systems. Additionally, these data structures are difficult to partition on distributed memory systems. They also present poor spatial and temporal locality, thus generating unpredictable memory and network accesses. The Partitioned Global Address Space (PGAS) programming model seems suitable for these applications, because it allows using a shared memory abstraction across distributed-memory clusters. However, current PGAS languages and libraries are built to target regular remote data accesses and block transfers. Furthermore, they usually rely on the Single Program Multiple Data (SPMD) parallel control model, which is not well suited to the fine grained, dynamic and unbalanced parallelism of irregular applications. In this paper we present {\bf GMT} (Global Memory and Threading library), a custom runtime library that enables efficient execution of irregular applications on commodity clusters. GMT integrates a PGAS data substrate with simple fork/join parallelism and provides automatic load balancing on a per node basis. It implements multi-level aggregation and lightweight multithreading to maximize memory and network bandwidth with fine-grained data accesses and tolerate long data access latencies. A key innovation in the GMT runtime is its thread specialization (workers, helpers and communication threads) that realize the overall functionality. We compare our approach with other PGAS models, such as UPC running using GASNet, and hand-optimized MPI code on a set of typical large-scale irregular applications, demonstrating speedups of an order of magnitude.

Aggregation↗
DOE OSTI2019

Checkpointing Strategies for Shared High-Performance Computing Platforms

Input/output (I/O) from various sources often contend for scarcely available bandwidth. For example, checkpoint/restart (CR) protocols can help to ensure application progress in failure-prone environments. However, CR I/O alongside an application's normal, requisite I/O can increase I/O contention and might negatively impact performance. In this work, we consider different aspects (system-level scheduling policies and hardware) that optimize the overall performance of concurrently executing CR-based applications that share I/O resources. We provide a theoretical model and derive a set of necessary constraints to minimize the global waste on a given platform. Our results demonstrate that Young/Daly's optimal checkpoint interval, despite providing a sensible metric for a single, undisturbed application, is not sufficient to optimally address resource contention at scale. We show that by combining optimal checkpointing periods with contention-aware system-level I/O scheduling strategies, we can significantly improve overall application performance and maximize the platform throughput. Finally, we evaluate how specialized hardware, namely burst buffers, may help to mitigate the I/O contention problem. Altogether, these results provide critical analysis and direct guidance on how to design efficient, CR ready, large -scale platforms without a large investment in the I/O subsystem.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Understanding HPC Benchmark Performance on Intel Broadwell and Cascade Lake Processors

Hardware platforms in high performance computing are constantly getting more complex to handle even when considering multicore CPUs alone. Numerous features and configuration options in the hardware and the software environment that are relevant for performance are not even known to most application users or developers. Microbenchmarks, i.e., simple codes that fathom a particular aspect of the hardware, can help to shed light on such issues, but only if they are well understood and if the results can be reconciled with known facts or performance models. The insight gained from microbenchmarks may then be applied to real applications for performance analysis or optimization. In this paper we investigate two modern Intel x86 server CPU architectures in depth: Broadwell EP and Cascade Lake SP. We highlight relevant hardware configuration settings that can have a decisive impact on code performance and show how to properly measure on-chip and off-chip data transfer bandwidths. The new victim L3 cache of Cascade Lake and its advanced replacement policy receive due attention. Finally we use DGEMM, sparse matrix-vector multiplication, and the HPCG benchmark to make a connection to relevant application scenarios.

97 MATHEMATICS AND COMPUTING↗