Search NASA⌕ Search

SEARCH · Search NASA

Results for “performances”

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 289 records · Page 16
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Summary of Geometric Parameters and Their Effects on Performance of U-10Mo Fuel Plates

A monolithic plate-type fuel system has been under development to convert high-performance test reactors from highly enriched uranium to low-enrichment uranium fuels and is now moving into the qualification phase, a predecessor to the timely conversion of the target reactors. To qualify this fuel system, the plates must meet safety standards and perform well in a reactor. The plates must maintain mechanical integrity, exhibit geometric stability, and have stable and predictable in-reactor behavior. The requirement to maintain mechanical integrity under normal operating conditions is primarily demonstrated by successful testing. However, each high-performance reactor employs a distinct design, resulting in distinct plate geometries, with unique features, attributes, irregularities, and tolerances. Due to the abundance of such distinct geometric varieties, a single “generic” plate geometry capturing all extremes is not achievable. It is also impractical to test each of these proposed designs in a reactor. This limitation necessitates cautious evaluations since the thermomechanical response of a plate with a certain geometry may not be representative for a plate with a significantly different geometry. To address concerns related to in-reactor plate performance, large set of sensitivity studies were performed. These parametric studies aimed to better understand irradiation performance, while evaluating the sensitivity of results to various modeling inputs, including geometric, operational, and material parameters. This work studied selected geometric parameters based on provided fuel specifications and performed a series of parametric simulations. The resulting temperature, displacement, and stress strains were comparatively evaluated to determine the effects of various geometric parameters. This draft provides a “high-level summary” of our parametric sensitivity studies and summarizes the key findings.

42 ENGINEERING↗
DOE OSTI2024

High–Performance NiCo 2 O 4 /Graphene Quantum Dots for Asymmetric and Symmetric Supercapacitors with Enhanced Energy Efficiency

For the sustainable growth of future generations, energy storage technologies like supercapacitors and batteries are becoming more and more common. However, reliable and high-performance materials’ design and development is the key for the widespread adoption of batteries and supercapacitors. Quantum dots with fascinating and unusual properties are expected to revolutionize future technologies. However, while the recent discovery of quantum dots honored with a Nobel prize in Chemistry, their benefits for the tenacious problem of energy are not realized yet. In this context, herein, chemical-composition tuning enabled exceptional performance of NiCo 2 O 4 (NCO)/graphene quantum dots (GQDs) is reported, which outperform the existing similar materials, in supercapacitors. A comprehensive study is performed on the synthesis, characterization, and electrochemical performance evaluation of highly functional NCO/GQDs in supercapacitors delivering enhanced energy efficiency. The high-performance, functional NCO/GQDs electrode materials are synthesized by the incorporation of GQDs into NCO. The effect of variable amount of GQDs on the energy performance characteristics of NCO/GQDs in supercapacitors is studied systematically. In-depth structural and chemical bonding analyses using X-ray diffraction (XRD) and Raman spectroscopic studies indicate that all the NCO/GQDs composites crystallize in the spinel cubic phase of NiCo 2 O 4 while graphene integration evident in all the NCO/GQDs. The scanning electron microscopy imaging analysis reveals homogeneously distributed spherical particles with a size distribution of 5–9 nm validating the formation of QDs. The high-resolution transmission electron microscopy analyses reveal that the NCOQDs are anchored on graphene sheets, which provide a high surface area of 42.27 m 2 g –1 and high mesoporosity for the composition of NCO/GQDs-10%. In addition to establishing reliable electrical connection to graphene sheets, the NCOQDs provide reliable 3D-conductive channels for rapid transport throughout the electrode as well as synergistic effects. Chemical-composition tuning, and optimization yields NCO/GQDs-10% to deliver the best specific capacitance of 3940 Fg –1 at 0.5 Ag –1 , where the electrodes retain ≈98% capacitance after 5000 cycles. The NCO/GQD-10%//AC asymmetric supercapacitor device demonstrates outstanding energy density and power density values of 118.04 Wh kg –1 and 798.76 W kg –1 , respectively. The NCO/GQDs-10%//NCO/GQDs-10% symmetric supercapacitor device delivers excellent energy and power density of 24.30 Wh kg –1 and 500 W kg –1 , respectively. These results demonstrate and conclude that NCO/GQDs are exceptional and prospective candidates for developing next-generation high-performance and sustainable energy storage devices.

25 ENERGY STORAGE↗
DOE OSTI2021

Instruction Roofline: An insightful visual performance model for GPUs

The Roofline performance model provides an intuitive approach to identify performance bottlenecks and guide performance optimization. However, the classic FLOP-centric approach is inappropriate for the emerging applications that perform more integer operations than floating point operations. In this article, we reintroduce our Instruction Roofline Model on NVIDIA GPUs and expand our evaluation of it. The Instruction Roofline incorporates instructions and memory transactions across all memory hierarchies together, and provides more performance insights than the FLOP-oriented Roofline Model, that is, instruction throughput, stride memory access patterns, bank conflicts, and thread predication. We use our Instruction Roofline methodology to analyze eight proxy applications: HPGMG from AMReX, Matrix Transpose benchmarks, ADEPT from MetaHipMer's sequence alignment phase, EXTENSION from MetaHipMer's local assembly phase, CUSP, cuSPARSE, cudaTensorCoreGemm, and cuBLAS. We demonstrate the ability of our methodology to understand various aspects of performance and performance bottlenecks on NVIDIA GPUs and motivate code optimizations.

Ding, N↗
DOE OSTI2024

Architecture and performance of Perlmutter's 35 PB ClusterStor E1000 all-flash file system

NERSC's newest system, Perlmutter, features a 35 PB all-flash Lustre file system built on HPE Cray ClusterStor E1000. Here, we present its architecture, early performance figures, and performance considerations unique to this architecture. We demonstrate the performance of E1000 OSSes through low-level Lustre tests that achieve over 90% of the theoretical bandwidth of the SSDs at the OST and LNet levels. We also show end-to-end performance for both traditional dimensions of I/O performance (peak bulk-synchronous bandwidth) and nonoptimal workloads endemic to production computing (small, incoherent I/Os at random offsets) and compare them to NERSC's previous system, Cori, to illustrate that Perlmutter achieves the performance of a burst buffer and the resilience of a scratch file system. Finally, we discuss performance considerations unique to all-flash Lustre and present ways in which users and HPC facilities can adjust their I/O patterns and operations to make optimal use of such architectures.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Benchmark of numerical modeling approaches on the systematic performance evaluation of wave energy converters

Different numerical modeling methods have been developed and applied to evaluate a variety of performance indicators of wave energy converters (WECs), including the power performance, structural loads, levelized cost of energy, etc. Based on the modeling fidelity, the commonly used numerical modeling approaches can be classified as linear modeling, weakly nonlinear modeling and fully nonlinear modeling approaches. Each method differs in accuracy and computational efficiency, making them suitable for different stages of WEC design. However, the selection of modeling approach could significantly impact evaluation outcomes. For instance, simplified linear models may underestimate structural loads or overestimate energy production in some operational conditions, potentially leading to less cost-effective designs. Given the widespread utilization of these models, it is essential to understand the uncertainties brought by them in performance evaluations. This work is dedicated to benchmarking different linear-potential-flow-based numerical models for evaluating the systematic performance of WECs. Three representative numerical modeling approaches are considered in this work, including linear frequency-domain modeling, statistically linearized spectral-domain modeling and Cummins equation-based nonlinear time-domain modeling. A generic point absorber WEC is considered as the research reference in this work, and different sea sites are taken into account. The numerical models are utilized to predict critical performance indicators, including power performance, the annual energy production, the capacity factor, the levelized cost of energy and the PTO fatigue loads. By comparing the results, this work identifies the uncertainties associated with different modeling approaches in evaluating WEC performance.

Fatigue↗
DOE OSTI2025

Workflow for evaluating enzyme immobilization and performance for continuous flow manufacturing

Enzymes have shown promise in various industries due to their functional specificity, catalytic efficiency, and environmental sustainability. These biological catalysts can be a pivotal component of manufacturing pipelines like continuous flow chemistry. For this, there exists a need to robustly immobilize enzymes on solid supports and assess the effects of the solid supports on catalytic performance and stability. Here, we use an industrially relevant model enzyme, C. ensiformis (Jack bean) urease, to demonstrate immobilization and assess performance in the context of continuous flow manufacturing. Various immobilization strategies were screened focusing on immobilization efficiency, protocol simplicity, and urease biocatalyst kinetics. Based on this, CDI-agarose and NHS-agarose resins were identified as the best-performing immobilization strategies for urease. CDI-agarose-urease and NHS-agarose-urease were then scaled up and applied to a large-scale continuous flow reactor to evaluate product yields, operational stability, and long-term stability. These experiments identified differences in stability and performance depending on the immobilization method tested. This highlights the importance of screening immobilization methods and subsequent enzyme performance for each candidate biocatalyst used in manufacturing to promote optimal performance and stability. As such, this work provides a framework for evaluating enzyme biocatalyst immobilization approaches to improve performance and enable transition into industrial processes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Impact of reactor architecture and design parameters on the performance of microbial electrolysis cells revealed by the electrode potential slope analysis

Microbial electrolysis cells (MECs) are appealing for recovering the chemical energy contained in domestic and industrial liquid wastes as hydrogen gas. Despite several years of research in the field, there is still a lack of critical analysis of how the reactor architecture dictates the electrochemical performance of the cell. In this study, internal resistance and onset voltage from the electrode potential slope analysis (EPS) were used in combination with current density, hydrogen production rate, reactor packing density, electrode spacing, membrane type and composition from 23 different studies to identify the reactor design parameters that primarily govern electrochemical performance of MECs. Using anion exchange membranes resulted in smaller internal resistances (AEM R int = 41± 40 mΩ m 2 ) and larger current density (18 ± 14 A m −2 ) compared to single chamber reactors (SC R int = 68 ± 58 mΩ m 2 ; 22 ± 16 A m −2 ) or MECs with cation exchange membranes (CEM R int = 376 ± 280 mΩ m 2 ; 3.0 ± 2.1 A m −2 ). Higher electrochemical performance for AEM- and SC-MECs translated in larger hydrogen gas production rates (0.122 mL H 2 C −1 for AEM vs 0.117 mL H 2 C −1 for SC), but only when inhibitors against hydrogen scavengers were added in single chamber systems (0.080 mL H 2 C −1 for SC without inhibitors). Following membrane type and composition, maintaining a small electrode spacing was the most critical parameter to improve MEC performance, indicating that the low conductivity of the media primarily limit performance by increasing ohmic resistance. Here, reactor volume and electrode surface area negatively correlated with internal resistance and current density, indicating that better performance of scaled-up reactors can likely be obtained by stacking multiple smaller units rather than just increasing reactor size. Although challenges remain in the implementation of MECs for hydrogen production from liquid wastes, advances in electrochemical engineering of the reactors can facilitate scale up and performance prediction at scale.

Electrochemistry↗
DOE OSTI2024

Performance evaluation of finned tube heat exchanger using curved wavy delta winglet vortex generators with circular perforations

Vortex generation is recognized as an effective passive approach to improve the heat transfer rate in fin and tube heat exchangers (FTHEs). The current study proposed innovative designs of curved wavy delta winglet vortex generators (CWDWVGs), both without and with circular perforations, to improve the heat transfer efficiency of FTHEs. There is potential to increase heat transfer performance further through various CWDWVG designs. Here, this study explores seven unique CWDWVG configurations, from 1-wave to 7-wave. A 3-D computational numerical model is utilized to evaluate the Thermo-hydraulic performance of FTHEs fitted with these different CWDWVG configurations across Reynolds numbers from 400 to 2000. This comparative analysis of the Thermo-hydraulic performance of FTHEs featuring four parallel circular tube layouts assesses configurations both with and without vortex generators (VGs) and various hole configurations. The evaluation of Thermo-hydraulic performance involves different parameters, including the London area goodness factor (LAGF), Colburn factor (j), friction factor (f), pressure drop (?P), and Nusselt number (Nu). Results demonstrate that the various CWDWVG configurations and the number of holes in them substantially affect the efficiency, as evaluated by the dimensionless Performance Evaluation Criteria (PEC). Notably, the 7-wave CWDWVGs surpassed other configurations, and integrating circular punched perforations further improved the thermal-hydraulic performance of FTHE. Specifically, the 7-wave CWDWVGs without holes demonstrated superior performance over other configurations, showing a significant increase in Nusselt number by 75.18% and 85.16% at Reynolds numbers of 2000 and 400, respectively, alongside an increase in pressure drop by 216.38% to 224.96%. Meanwhile, the 7-wave CWDWVGs with eight holes, in comparison to those without holes, exhibited a Nusselt number increase of 0.85%, a pressure drop decrease of 7.31%, and a reduction in the friction factor by 5.82%.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2020

Optimization of catalyst-coated membranes for enhancing performance in proton exchange membrane electrolyzer cells

To achieve large-scale application of proton exchange membrane electrolyzer cells (PEMECs) for hydrogen production, it is highly desirable to reduce the manufacturing cost while enhancing cell performance. In the PEMPECs, a catalyst-coated membrane (CCM) is the vital component where electrochemical reactions and mass transport mainly occur. The fabrication methods and catalyst layer (CL) structure can significantly affect the cell performance. Herein, for the first time, a comparative study of CCM fabrications with decal transfer and direct spray deposition methods have been conducted by both ex-situ materials characterization and in-situ performance testing in PEMECs. It is found CCMs that are fabricated with a direct spray deposition method display enhanced cell performance compared to CCMs fabricated with a decal transfer method, mainly due to the largely reduced ohmic resistance and improved mass transport. More importantly, cell performance can be greatly enhanced by simply regulating the Nafion ionomer content at the anode CL. The optimal Nafion ionomer content of 10 wt% gives the best cell performance at 80 °C with a low cell voltage of 1.887 V at 2 A cm -2 , outperforming the commercial CCM and most other previous publications. Our study provides a valuable guidance for fabrication and optimization of CCMs with significantly enhanced performance and reduced cost for practical application of the PEMECs.

08 HYDROGEN↗
DOE OSTI2022

Methane foam performance evaluation in water-wet unconsolidated porous media: A systematic experimental investigation at elevated pressure and temperature conditions

In the petroleum industry, enhanced oil recovery (EOR) techniques employ foam extensively to establish conformance control in heterogeneous and fractured reservoirs in order to increase the sweep efficiency. In such applications, foam performance evaluation under complex subsurface conditions is pivotal for the effective and optimized deployment of foam treatment. However, there is a scarcity of hydrocarbon gas foam generation and evaluation studies that examine the relationships between foam performance and the critical foam parameters at high-pressure and high-temperature conditions. Here, this study aims at methodically investigating the effects of several foam parameters on methane foam performance in water-wet proppant packs under harsh operating conditions. This is in relation to the technical needs of hydrocarbon foam injection into hydraulically-induced, propped fractures in unconventional oil reservoirs. To this end, a state-of-the-art experimental foam generation apparatus was designed, fabricated, and commissioned. We performed a large number of foam flow experiments on proppant packs using methane gas and different foaming agents at 3500 psi and 115°C. Anionic and amphoteric surfactants were employed to probe the effect of their ionic nature on foam performance. Foam performance sensitivities to various foam generation parameters and operating conditions, such as surfactant concentration, gas fraction, total injection rate, operating pressure, salinity, and proppant pack length were investigated. To this end, steady-state pressure drops across the proppant packs during foam generation and foam's apparent viscosity were measured to quantify the foam performance of surfactants. The results were then analyzed to determine optimum values of the foam parameters and the interplay between these parameters are discussed here. The systematic results achieved from this work are in agreement with the trends available in the literature and provide new insights into complexities of in situ foam generation in water-wet, unconsolidated porous media at extreme reservoir conditions.

03 NATURAL GAS↗
DOE OSTI2023

Irradiation performance of a U-7Mo in Al-Si matrix dispersion full-size fuel plate assembly

The Korea Atomic Energy Research Institute (KAERI) is leading the Ki-Jang Research Reactor (KJRR) project with the intent to develop a new reactor for medical isotope production and other nuclear research purposes. The KJRR core is designed to use high density fuel system where uranium alloyed with 7 wt% molybdenum (U-7Mo) particles are dispersed in a matrix of aluminum alloyed with 5 wt% silicon (Al-5Si) and clad in aluminum alloy 6061 (Al-6061) to form fuel plates. KAERI developed a fabrication facility to construct KJRR fuel assemblies and partnered with the Idaho National Laboratory (INL) to irradiate a full-size fuel assembly, with 21 total fuel plates, in the Advanced Test Reactor (ATR). Irradiation testing and subsequent Post Irradiation Exam (PIE) campaigns were performed successfully over a multi-year project. Monte Carlo neutronic calculations, coupling with a depletion code, were performed based on ATR’s as-run power history which showed that the highest power plate (plate 20) reached 83.1 % end-of-life (EOL) local burnup based on initial 235 U content. Additionally, finite element thermal modeling was performed based as-run power history which showed a beginning-of-life (BOL) peak local heat flux of 184 W/cm 2 . No anomalous fuel performance was observed during the irradiation and target test conditions were achieved. PIE showed favorable performance of the fuel assembly regarding all important phenomena. This paper describes the KJRR fuel assembly irradiation conditions and PIE data to support the conclusion that it performed well, without evidence of unexpected or problematic fuel performance, within an irradiation test designed to bound the KJRR design environment.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2020

Impact of Selected High-Performance Fuel Blends on Three-Way Catalyst Light Off under Synthetic Spark-Ignition Engine-Exhaust Conditions

The U.S. Department of Energy funded Co-Optimization of Fuels and Engines initiative aims to simultaneously develop advanced engines along with high-performance fuels to reduce petroleum consumption. The engine exhaust of spark-ignited light-duty vehicles contains pollutants such as nitrogen oxides, carbon monoxide (CO), and non-methane organic gases. When operated above their “light-off” temperature, three-way catalysts (TWCs) efficiently control the emissions of these pollutants from the vehicle exhaust. However, below the catalyst light-off temperature, during cold start, the TWCs are not effective. Thus, the stringent environmental regulations necessitate cold-start compliance of advanced engines operating on novel fuels for commercialization. Exhaust composition strongly impacts the effectiveness of TWCs. Hence, ensuring that the high-performance fuels under consideration do not have detrimental effects on current emissions control technology is necessary. To mitigate cold-start emissions, a low light-off temperature of the fuel on the TWC is desirable. As real-world fuels are multicomponent blends, we conducted investigations into the light-off behavior of representative fuel mixtures on a three-mode redox-aged commercial TWC under synthetic engine-exhaust conditions. The high-performance fuels in this study included 10–30% volumetric levels of ethanol, isobutanol, diisobutylene, and an aromatic mixture. Each of these high-performance fuel components was mixed into a gasoline surrogate blendstock for oxygenate blending (BOB). Our results showed that aromatics and alkenes in the surrogate BOB inhibit low-temperature reactivity of alkanes, alcohols, and CO on the TWC and dominate the blend light-off behavior. All the high-performance fuel blends had a very similar light-off behavior to the surrogate gasoline BOB, indicating that blending up to 30% (by vol.) of high-performance blendstocks in a gasoline base fuel can potentially reduce greenhouse gas emissions through improved engine efficiency and petroleum displacement without jeopardizing the ability to meet emissions regulations. While some high-performance blendstocks demonstrated lower light-off temperatures than a surrogate gasoline blend, taking advantage of the higher catalytic reactivity of these blendstocks to reduce cold-start emissions would require reducing the aromatic content in petroleum-based market fuels.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2021

Effects of Ink Formulation on Construction of Catalyst Layers for High-Performance Polymer Electrolyte Membrane Fuel Cells

Rational design of catalyst layers in a membrane electrode assembly (MEA) is crucial for achieving high-performance polymer electrolyte membrane fuel cells. Establishing a clear understanding of the property (catalyst ink)–structure (catalyst layer)–performance (MEA) relationship lays the foundation for this rational design. Here, a synergistic approach was taken to correlate the ink formulation, the microstructure of catalyst layers, and the resulting MEA performance to establish such a property–structure–performance relationship. The solvent composition (n-PA/H 2 O mixtures) demonstrated a strong influence on the performance of the MEA fabricated with an 830-EW (Aquivion) ionomer, especially polarization losses of cell activation and mass transport. The performance differences were studied in terms of how the solvent composition affects the catalyst/ionomer interface, ionomer network, and pore structure of the resulting catalyst layers. The ionomer aggregates mainly covered the surface of catalyst aggregates acting as oxygen reduction reaction active sites, and the aggregate sizes of the ionomer and catalyst (revealed by ultrasmall angle X-ray scattering and cryo-transmission electron microscopy) were dictated by tuning the solvent composition, which in turn determined the catalyst/ionomer interface (available active sites). In n-PA/H 2 O mixtures with 50~90 wt % H 2 O, the catalyst agglomerates could be effectively broken up into small aggregates, leading to enhanced kinetic activities. The boiling point of the mixed solvents determined the pore structure of ultimate catalyst layers, as evidenced by mercury porosimetry and scanning electron microscopy. For mixed solvents with a higher boiling point, the catalyst–ionomer aggregates in the ink tend to agglomerate during the solvent evaporation process and finally form larger catalyst–ionomer aggregates in the ultimate catalyst layer, resulting in more secondary pores and thus lower mass transport resistance. Both the enlarged catalyst/ionomer interface and appropriate pore structure were achieved with the catalyst layer fabricated from an n-PA/H 2 O mixture with 90 wt % H 2 O, leading to the best MEA performance.

25 ENERGY STORAGE↗
DOE OSTI2023

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↗
DOE OSTI2022

Understanding Strong Scaling on GPUs Using Empirical Performance Saturation Size

The roofline model provides a concise overview of the maximum performance capabilities of a given computer system through a combination of peak memory bandwidth and compute performance rates. The increasing complexity of scheduling and cache in recent GPUs, however, has introduced complicated performance variability that is not captured by arithmetic intensity alone. This work examines the effect of problem size and GPU launch configurations on roofline performance for V100, A100, MI100, and MI250X graphics processing units. We introduce an extended roofline model that takes problem size into account, and find that strong scaling on GPUs can be characterized by saturation problem sizes as additional key metrics. Saturation problem sizes break up a plot of GPU performance vs. problem size into three distinct performance regimes– size-limited, cache-bound, and DRAM-bound. With our extended roofline model, we are able to provide a robust view of these performance regimes across recent GPU architectures.

Eberius, David↗
DOE OSTI2022

Performance Debugging and Tuning of Flash-X with Data Analysis Tools

State-of-the-art multiphysics simulations running on large scale leadership computing platforms have many variables contributing to their performance and scaling behavior. We recently encountered an interesting performance anomaly in Flash-X, a multiphysics multicomponent simulation software, when characterizing its performance behavior on several large-scale HPC platforms. The anomaly was tracked down to the interaction between the use of dynamic allocation of scratch data and data locality in the cache hierarchy. In this paper we present the details of unexpected performance variability of Flash-X, its extensive analysis using the performance measurement tool TAU to collect the data and Python data analysis libraries to explore the data, and our insights from this experience. In this process, we discovered and removed or mitigated two additional performance limiting bottlenecks for performance tuning.

Huck, Kevin↗
DOE OSTI2022

Automatic Generation of High-Performance Convolution Kernels on ARM CPUs for Deep Learning

In this work, we present FastConv, a template-based code auto-generation open source library that can automatically generate high-performance deep learning convolution kernels of arbitrary matrices/tensors shapes. FastConv is based on the Winograd algorithm, which is reportedly the highest performing algorithm for the time-consuming convolution layers of convolutional neural networks. ARM CPUs cover a wide range designs and specifications, from embedded devices to HPC-grade CPUs. The leads to the dilemma of how to consistently optimize Winograd-based convolution solvers for convolution layers of different shapes. FastConv addresses this problem by using templates to auto-generate multiple shapes of tuned kernels variants suitable for skinny tall matrices. As a performance portable library, FastConv transparently searches for the best combination of kernel shapes, cache tiles, scheduling of loop orders, packing strategies, access patterns, and online/offline computations. Auto-tuning is used to search the parameter configuration space for the best performance for a given target architecture and problem size. The experiments with layer-wise evaluation on the VGG--16 model confirms a 1.25x performance gains is got by tuning the Winograd library. Integrated comparison results shows 1.02x to 1.40x, 1.14x to 2.17x, and 1.22x and 2.48x speedup is achieved over NNPACK, Arm NN, and FeatherCNN on the Kunpeng 920 beside few cases. Furthermore, problem size performance portability experiments with various convolution shapes shows that FastConv achieves 1.2x to 1.7x speedup and 2x to 22x speedup over NNPACK and ARM NN inference engine using Winograd on Kunpeng 920 . CPU performance portability evaluation on the VGG--16 show an average speedup over NNPACK of 1.42x, 1.21x, 1.26x, 1.37x, 2.26x, and 11.02x is observed on Kunpeng 920, Snapdragon 835, 855, 888, Apple M1, and AWS Graviton2, respectively.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Editors’ Choice—Examining Performance and Durability of Anion Exchange Membrane Fuel Cells with Novel Spirocyclic Anion Exchange Membranes

A series of spirocyclic copolymer membranes with varying ion exchange capacities (IECs) were investigated to probe the impact of polymer properties on in situ fuel cell performance and stability. In-situ electrochemical tests and post-mortem electron microscopy analysis of cross-sectioned membrane electrode assemblies (MEAs) have been combined with voltage loss breakdown analysis to evaluate the performance and degradation of different MEAs, and to probe the catalyst morphology and electrode structure at different stages of operation. Voltage loss breakdown results show that membrane degradation and kinetic losses played only a minor role in observed performance degradation and that performance losses were primarily related to increasing mass transport losses. From microscopy studies, carbon corrosion and Pt nanoparticle growth were identified at both the cathode and anode although more pronounced on the cathode resulting in significant structural changes. The membrane with the lowest IEC (1.3 mmolg −1 ) demonstrated the lowest peak power density ~ 1.16 W cm −2 , however, it showed the most stable performance (constant 0.6 A cm −2 hold) with ~ 5% degradation over 540 h. Isolation of performance losses and microscopic analysis of electrodes for anion exchange membrane fuel cells has not been reported previously, and these results help identify critical performance degradation concerns.

25 ENERGY STORAGE↗