Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “performance tuning”

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 703 records · Page 39

Jet and electromagnetic tomography (JET) of extreme phases of matter in heavy-ion collisions

The Ohio State University (OSU) group contributed to the deliverables of the JET Collaboration three major products: 1. The code package iEBE-VISHNU for modeling the dynamical evolution of the soft medium created in relativistic heavy-ion collisions, from its creation all the way to final freeze-out using a hybrid approach that interfaces a free-streaming partonic pre-equilbrium stage with a (2+1)-dimensional viscous relativistic fluid dynamical stage for the quark-gluon plasma (QGP) phase and the microscopic hadron cascade UrQMD for the hadronic rescattering and freeze-out stage. Except for UrQMD, all dynamical evolution components and interfaces were developed at OSU and tested and implemented in collaboration with the Duke University group. 2. An electromagnetic radiation module for the calculation of thermal photon emission from the QGP and hadron resonance gas stages of a heavy-ion collision, with emission rates that have been corrected for viscous effects in the expanding medium consistent with the bulk evolution. The electromagnetic radiation module was developed under OSU leadership in collaboration with the McGill group and has been integrated in the iEBE-VISHNU code package. 3. An interface between the Monte Carlo jet shower evolution and hadronization codes developed by the Wayne State University (WSU), McGill and Texas A&M groups and the iEBE-VISHNU bulk evolution code, for performing jet quenching and jet shape modification studies in a realistically modeled evolving medium that was tuned to measured soft hadron data. Building on work performed at OSU for the theoretical framework used to describe the interaction of jets with the medium, initial work on the jet shower Monte Carlo was started at OSU and moved to WSU when OSU Visiting Assistant Professor Abhijit Majumder accepted a tenure track faculty position at WSU in September 2011. The jet-hydro interface was developed at OSU and WSU and tested and implemented in collaboration with the McGill, Texas A&M, and LBNL groups.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

High thermoelectric figure of merit and improved mechanical properties in melt quenched PbTe-Ge and PbTe-Ge{sub 1-x}Si{sub x} eutectic and hypereutectic composites.

We report the synthesis, microstructure, and transport properties of composite thermoelectric materials based on the eutectic phase relationship between PbTe and Ge. When quenched, these eutectic mixtures exhibit considerably stronger mechanical strength and reduced brittleness compared to PbTe itself, while at the same time they possess lower lattice thermal conductivity. Thermal conductivity measurements show values lower than expected based on the law of mixtures and multiphase composites. We find that the thermoelectric performance in these composites can be tuned through the use of hypereutectic compositions and alloying of Ge with Si. PbI{sub 2} was used as an n-type dopant, and precise control of the carrier concentration was achieved to optimize the electrical transport and thermoelectric properties. ZT values approaching 1.3 at 778 K have been obtained in samples of PbTe-Ge{sub 0.8}Si{sub 0.2}(5%), which represent an {approx}62% improvement over that of PbTe.

36 MATERIALS SCIENCE↗

Determination of muon momentum in the MicroBooNE LArTPC using an improved model of multiple Coulomb scattering

Here, we discuss a technique for measuring a charged particle's momentum by means of multiple Coulomb scattering (MCS) in the MicroBooNE liquid argon time projection chamber (LArTPC). This method does not require the full particle ionization track to be contained inside of the detector volume as other track momentum reconstruction methods do (range-based momentum reconstruction and calorimetric momentum reconstruction). We motivate use of this technique, describe a tuning of the underlying phenomenological formula, quantify its performance on fully contained beam-neutrino-induced muon tracks both in simulation and in data, and quantify its performance on exiting muon tracks in simulation. Using simulation, we have shown that the standard Highland formula should be re-tuned specifically for scattering in liquid argon, which significantly improves the bias and resolution of the momentum measurement. With the tuned formula, we find agreement between data and simulation for contained tracks, with a small bias in the momentum reconstruction and with resolutions that vary as a function of track length, improving from about 10% for the shortest (one meter long) tracks to 5% for longer (several meter) tracks. For simulated exiting muons with at least one meter of track contained, we find a similarly small bias, and a resolution which is less than 15% for muons with momentum below 2 GeV/c. Above 2 GeV/c, results are given as a first estimate of the MCS momentum measurement capabilities of MicroBooNE for high momentum exiting tracks.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Test Platform for Advanced Digital Control of Brushless DC Motors (MSFC Center Director's Discretionary Fund)

A FY 2001 Center Director's Discretionary Fund task to develop a test platform for the development, implementation. and evaluation of adaptive and other advanced control techniques for brushless DC (BLDC) motor-driven mechanisms is described. Important applications for BLDC motor-driven mechanisms are the translation of specimens in microgravity experiments and electromechanical actuation of nozzle and fuel valves in propulsion systems. Motor-driven aerocontrol surfaces are also being utilized in developmental X vehicles. The experimental test platform employs a linear translation stage that is mounted vertically and driven by a BLDC motor. Control approaches are implemented on a digital signal processor-based controller for real-time, closed-loop control of the stage carriage position. The goal of the effort is to explore the application of advanced control approaches that can enhance the performance of a motor-driven actuator over the performance obtained using linear control approaches with fixed gains. Adaptive controllers utilizing an exact model knowledge controller and a self-tuning controller are implemented and the control system performance is illustrated through the presentation of experimental results.

Gwaltney, D. A.↗

Direct Measurements of the Mechanical Strength of Carbon Nanotube - Aluminum Interfaces

Interfacial load transfer plays a critical role in the bulk mechanical performance of nanofiber-reinforced metallic-matrix nanocomposites (MMNC). In this paper, we investigate the mechanical strength of interfaces in double-walled carbon nanotube (CNT)-reinforced aluminum (Al) nanocomposites by using in situ electron microscopy nanomechanical single-tube pull-out techniques. The nanomechanical measurements reveal the shear lag effect on the CNT-Al interface that is found to possess an average interfacial shear strength (IFSS) of about 28.7 MPa. The study also shows that thermal annealing results in substantially higher binding strength interfaces between CNTs and Al matrices. The average IFSS of CNT-Al interfaces that were thermally annealed at 400 °C is found to reach about 35.3 MPa, a 23% increase from that of the non-annealed interfaces. The maximum load bearing capacity of the annealed interfaces reaches about 304 nN, a 40.1% increase from that of the non-annealed ones (about 217 nN). The findings are useful to better understand the load transfer mechanism in CNT-reinforced MMNC and the tuning and optimization of the reinforcing performance through thermal processing.

Interfacial strength↗

Automatic Thread-Level Parallelization in the Chombo AMR Library

The increasing on-chip parallelism has some substantial implications for HPC applications. Currently, hybrid programming models (typically MPI+OpenMP) are employed for mapping software to the hardware in order to leverage the hardware?s architectural features. In this paper, we present an approach that automatically introduces thread level parallelism into Chombo, a parallel adaptive mesh refinement framework for finite difference type PDE solvers. In Chombo, core algorithms are specified in the ChomboFortran, a macro language extension to F77 that is part of the Chombo framework. This domain-specific language forms an already used target language for an automatic migration of the large number of existing algorithms into a hybrid MPI+OpenMP implementation. It also provides access to the auto-tuning methodology that enables tuning certain aspects of an algorithm to hardware characteristics. Performance measurements are presented for a few of the most relevant kernels with respect to a specific application benchmark using this technique as well as benchmark results for the entire application. The kernel benchmarks show that, using auto-tuning, up to a factor of 11 in performance was gained with 4 threads with respect to the serial reference implementation.

97 MATHEMATICS AND COMPUTING↗

Performance Engineering Research Center and RECOVERY. Performance Engineering Research Institution SciDAC-e Augmentation. Performance enhancement

This project concentrated on various ways to improve the measurement and tuning large-scale parallel applications. This project was supplement to the project DE-FC0206ER25763 (“Performance Engineering Research Center”). The research conducted during this project is summarized in this report. The complete details of the work are available in the ten publications listed at the end of the report. It also supported the Ph.D. studies of three students and one research scientist.

97 MATHEMATICS AND COMPUTING↗

Experiments in structural dynamics and control using a grid

Future spacecraft are being conceived that are highly flexible and of extreme size. The two features of flexibility and size pose new problems in control system design. Since large scale structures are not testable in ground based facilities, the decision on component placement must be made prior to full-scale tests on the spacecraft. Control law research is directed at solving problems of inadequate modelling knowledge prior to operation required to achieve peak performance. Another crucial problem addressed is accommodating failures in systems with smart components that are physically distributed on highly flexible structures. Parameter adaptive control is a method of promise that provides on-orbit tuning of the control system to improve performance by upgrading the mathematical model of the spacecraft during operation. Two specific questions are answered in this work. They are: What limits does on-line parameter identification with realistic sensors and actuators place on the ultimate achievable performance of a system in the highly flexible environment? Also, how well must the mathematical model used in on-board analytic redundancy be known and what are the reasonable expectations for advanced redundancy management schemes in the highly flexible and distributed component environment?

Montgomery, R. C.↗

Development of a Sensorimotor Ground Analog from Astronaut Postflight Experience

Exploration class missions including Artemis, Gateway, and beyond will require a new level of autonomy around periods of gravitational transition, where sensorimotor disturbances are great. Because of this, there is a need to define sensorimotor assessment thresholds that indicate when performance in operational tasks might be impacted or unsafe. To define these thresholds, a Sensorimotor Adaptation Analog (SAA) was proposed that could induce varying levels of sensorimotor disorientation through combined vestibular, visual, and proprioceptive disruptions. The purpose of this study was to gather subjective feedback on the SAA from previously flown astronauts that would mimic their post-flight experience and functional performance immediately after landing (R+0-4hrs; high level) and post-landing (R+24-48hrs; low level). The SAA consisted of galvanic vestibular stimulation (GVS), visual disorientation goggles, and a weighted suit to alter proprioceptive feedback and replicate subjective heaviness. A random sum-of-sines profile between 0-1Hz was used for the GVS with peak amplitudes ranging from 1-4mA. The GVS was applied first at the low (2mA) and high (3mA) levels, followed by the weighted suit alone, then combined with the GVS at the low (20% body weight) and high levels (40% body weight). Last, the disorientation goggles were applied alongside both the GVS and weighted suit at the low (0.07-0.10+ blood alcohol content (BAC)) and high (0.12-0.15+ BAC) levels. Each element of the SAA could be increased or decreased depending on crew feedback to find the disorientation levels that best matched their experience. Five USOS astronauts (1 male, 4 female) who had previously flown (average time since flight: 377 days) participated. Crewmembers reported that GVS alone replicated ~80-90% of their post-flight performance with the weighted suit fine-tuning the experience to replicate an additional 5-10% of their experience. Crewmembers did not believe the disorientation goggles represented either the visual disruptions or illusory sensations that they experienced, nor did they impact performance in post-flight tasks similarly. The final SAA, resulting from crewmember feedback, includes the GVS at the levels described above and the weighted suit at 15% and 30% body weight. The disorientation goggles were removed from SAA. These results provided a more realistic SAA that can be used to define the sensorimotor assessment thresholds through ground subjects and potentially expanded for use in countermeasure testing and as a pre-flight training tool.

S. C. Moudy↗

Unified Beam Control System of J-PARC Linac

A beam control system has been developed for the J-PARC linac to achieve strict design goals of beam performance and beam loss suppression in the high-intensity beams. An emerging concept of unified beam control has been adopted in the system, where a unified data source, standardized control units, a unified online model, and uniform high-level application frameworks have been designed and implemented. For precise beam controls, various beam tuning algorithms have been developed and tested. Design goals of beam performance, beam stability, and beam loss suppression have been achieved in the linac.

43 PARTICLE ACCELERATORS↗

Sn/SnOx Core-Shell Nanospheres: Synthesis, Anode Performance in Li Ion Batteries, and Superconductivity

Sn/SnO{sub x} core?shell nanospheres have been synthesized via a modified polyol process. Their size can be readily controlled by tuning the usage of surface stabilizers and the temperature. Anode performance in Li ion batteries and their superconducting properties is detailed. As anode materials, 45 nm nanospheres outperform both larger and smaller ones. Thus, they exhibit a capacity of about 3443 mAh cm{sup -3} and retain about 88% of after 10 cycles. We propose a model based on the microstructural evolution to explain the size impact on nanosphere performance. Magnetic measurements indicate that the nanospheres become superconducting below the transition temperature T{sub C} = 3.7 K, which is similar to the value obtained in bulk tin. Although T{sub C} does not significantly change with the size of the Sn core, we determined that the critical field H{sub C} of nanospheres can be as much as a factor of 30 larger compared to the bulk value. Alternating current measurements demonstrated that a transition from conventional to filamentary superconducting structure occurs in Sn/SnO{sub x} particles as their size increases. The transition is determined by the relationship between the particle size and the magnetic field penetration depth.

25 ENERGY STORAGE↗

Photovoltaic properties and morphology of organic solar cells based on liquid-crystal semiconducting polymer with additive

Bulk heterojunction organic solar cell based on liquid crystal semiconducting polymers of poly[9,9-dioctylfluorene-co-bithiophene] (F8T2) as p-type semiconductors and fullerenes (C{sub 60}) and [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) as electron donor and acceptor has been fabricated and characterized for improving photovoltaic and optical properties. The photovoltaic performance including current voltage curves in the dark and illumination of the F8T2/C{sub 60} conventional and inverted bulk heterojunction solar cells were investigated. Relationship between the photovoltaic properties and morphological behavior was focused on tuning for optimization of photo-voltaic performance under annealing condition near glass transition temperature. Additive-effect of diiodooctane (DIO) and poly(3-hexylthiophene-2,5-diyl) (P3HT) on the photovoltaic performance and optical properties was investigated. Mechanism of the photovoltaic properties of the conventional and inverted solar cells will be discussed by the experimental results.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Active analog tuning of the phase of light in the visible regime by bismuth-based metamaterials

The active and analog tuning of the phase of light by metamaterials is needed to boost the switching performance of photonic devices. However, demonstrations of this type of tuning in the pivotal visible spectral region are still scarce. Herein, we report the active analog tuning of the phase of visible light reflected by a bismuth (Bi)-based metamaterial, enabled by a reversible solid-liquid transition. This metamaterial, fabricated by following a lithography-free approach, consists of twodimensional assemblies of polydisperse plasmonic Bi nanostructures embedded in a refractory and transparent aluminum oxide matrix. The analog tuning of the phase is achieved by the controlled heating of the metamaterial to melt a fraction of the nanostructures. A maximum tuning of 320° (1.8 π) is observed upon the complete melting of the nanostructures at 230°C. This tuning is reversible by cooling to 25°C. In addition, it presents a wide hysteretic character due to liquid Bi undercooling. This enables the phase achieved by this analog approach to remain stable over a broad temperature range upon cooling and until resolidification occurs around 100°C. Therefore, Bi-based metamaterials are endowed with analog optical memory capabilities, which are appealing for a wide range of applications, including optical data storage with enhanced information density or bistable photonic switching with a tunable “on” state.

36 MATERIALS SCIENCE↗

High Temperature Steam Electrolysis Process Performance and Cost Estimates

Technology readiness levels (TRLs) of electrolysis systems have dramatically increased in recent years as the interest in clean hydrogen production and decarbonization of transportation, industrial and other sectors increases across the globe. This is especially true of high temperature steam electrolysis (HTSE) / solid oxide electrolysis cell (SOEC) systems which show promise of much higher system efficiencies than other more developed electrolysis technologies. This possibility of higher efficiencies of HTSE / SOEC systems has been previously assumed to be theoretically possible but in recent years it has become less theoretical and more realistic as an increasing amount of suppliers complete lab and pilot tests showing very promising results. Research in the areas of manufacturing techniques, material selection, electrode and electrolyte compositions, and balance of plant size and integration continues at a fast pace as an increasing number of suppliers both internationally and domestically become involved. The advantages of HTSE become more pronounced when HTSE is coupled with nuclear power plants (NPPs). This is because thermal energy produced by the nuclear reactor can be used in a series of heat transfer loops and heat exchangers to vaporize HTSE feedwater, which drastically improves the economics of the process. Idaho National Laboratory (INL) has been very involved in the research and modeling of HTSE systems for a number of years, in collaboration with other national laboratories, academia, and industry stakeholders both on the hydrogen production as well as the hydrogen demand side. The modeling completed over the years on a large variety of projects has led to a wealth of knowledge at INL including in the area of the technoeconomic assessment (TEA) of HTSE systems. TEAs include process modeling of the HTSE systems to calculate system energy requirements and equipment sizing, followed by estimation of capital and operating costs to enable calculation of the levelized cost of hydrogen (LCOH). The TEA work performed has produced incremental improvements and tuning of the methods, assumptions, models, and results of the analyses as well as providing some opportunities for validating these results. The purpose of this document is to record the current baseline HTSE analyses led by INL to show the current status of assumptions and costs of these systems. Given the rapid development of this technology, the variety of suppliers entering the space, and the increasing attention government and industry are giving to such systems, this document may be updated on a periodic basis with updated analysis and assumptions. This document compiles various analyses results and approaches completed over a period of years into a single document to be used as a baseline going forward. It represents what the INL HTSE analysis group assumes to be the internal best estimate of the current operation, costs, and landscape of the HTSE industry state of the art capability for current SOEC technology in an Nth-of-a-Kind (NOAK) plant, which in this study is defined as existence of the manufacturing capacity to support previous deployment of N = 100 count of 25 MWe modular HTSE blocks (with modular equipment component cost reductions specified as following a 95% learning curve). That said, This is a public document and as such so no proprietary data was used or included in this report. There may be HTSE suppliers that have performance specifications, and cost estimates, and test data that differ from the analysis presented in this document. This document is meant to be a best conservative estimate of the technology and not an absolute reference.

08 HYDROGEN↗

High Temperature Steam Electrolysis Process Performance and Cost Estimates - DOE Hydrogen Program AMR Presentation

Technology readiness levels (TRLs) of electrolysis systems have dramatically increased in recent years as the interest in clean hydrogen production and decarbonization of transportation, industrial and other sectors increases across the globe. This is especially true of high temperature steam electrolysis (HTSE) / solid oxide electrolysis cell (SOEC) systems which show promise of much higher system efficiencies than other more developed electrolysis technologies. This possibility of higher efficiencies of HTSE / SOEC systems has been previously assumed to be theoretically possible but in recent years it has become less theoretical and more realistic as an increasing amount of suppliers complete lab and pilot tests showing very promising results. Research in the areas of manufacturing techniques, material selection, electrode and electrolyte compositions, and balance of plant size and integration continues at a fast pace as an increasing number of suppliers both internationally and domestically become involved. The advantages of HTSE become more pronounced when HTSE is coupled with nuclear power plants (NPPs). This is because thermal energy produced by the nuclear reactor can be used in a series of heat transfer loops and heat exchangers to vaporize HTSE feedwater, which drastically improves the economics of the process. Idaho National Laboratory (INL) has been very involved in the research and modeling of HTSE systems for a number of years, in collaboration with other national laboratories, academia, and industry stakeholders both on the hydrogen production as well as the hydrogen demand side. The modeling completed over the years on a large variety of projects has led to a wealth of knowledge at INL including in the area of the technoeconomic assessment (TEA) of HTSE systems. TEAs include process modeling of the HTSE systems to calculate system energy requirements and equipment sizing, followed by estimation of capital and operating costs to enable calculation of the levelized cost of hydrogen (LCOH). The TEA work performed has produced incremental improvements and tuning of the methods, assumptions, models, and results of the analyses as well as providing some opportunities for validating these results. The purpose of this document is to record the current baseline HTSE analyses led by INL to show the current status of assumptions and costs of these systems. Given the rapid development of this technology, the variety of suppliers entering the space, and the increasing attention government and industry are giving to such systems, this document may be updated on a periodic basis with updated analysis and assumptions. This document compiles various analyses results and approaches completed over a period of years into a single document to be used as a baseline going forward. It represents what the INL HTSE analysis group assumes to be the internal best estimate of the current operation, costs, and landscape of the HTSE industry state of the art capability for current SOEC technology in an Nth-of-a-Kind (NOAK) plant, which in this study is defined as existence of the manufacturing capacity to support previous deployment of N = 100 count of 25 MWe modular HTSE blocks (with modular equipment component cost reductions specified as following a 95% learning curve). That said, this is a public document and as such so no proprietary data was used or included in this report. There may be HTSE suppliers that have performance specifications, and cost estimates, and test data that differ from the analysis presented in this document. This document is meant to be a best conservative estimate of the technology and not an absolute reference.

08 HYDROGEN↗

Metal–support interactions in metal oxide-supported atomic, cluster, and nanoparticle catalysis

Supported metal catalysts are essential to a plethora of processes in the chemical industry. The overall performance of these catalysts depends strongly on the interaction of adsorbates at the atomic level, which can be manipulated and controlled by the different constituents of the active material (i.e., support and active metal). The description of catalyst activity and the relationship between active constituent and the support, or metal–support interactions (MSI), in heterogeneous (thermo)catalysts is a complex phenomenon with multivariate (dependent and independent) contributions that are difficult to disentangle, both experimentally and theoretically. So-called “strong metal–support interactions” have been reported for several decades and summarized in excellent review articles. However, in recent years, there has been a proliferation of new findings related to atomically dispersed metal sites, metal oxide defects, and, for example, the generation and evolution of MSI under reaction conditions, which has led to the designation of (sub)classifications of MSI deserving to be critically and systematically evaluated. These include dynamic restructuring under alternating redox and reaction conditions, adsorbate-induced MSI, and evidence of strong interactions in oxide-supported metal oxide catalysts. Here, we review recent literature on MSI in oxide-supported metal particles to provide an up-to-date understanding of the underlying physicochemical principles that dominate the observed effects in supported metal atomic, cluster, and nanoparticle catalysts. Critical evaluation of different subclassifications of MSI is provided, along with discussions on the formation mechanisms, theoretical and characterization advances, and tuning strategies to manipulate catalytic reaction performance. We also provide a perspective on the future of the field, and we discuss the analysis of different MSI effects on catalysis quantitatively.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Implementation of A New Microwave Scattering Database and A Forward Model for Active Microwave Sensors in CRTM

Radiative transfer models are extensively used for the assimilation of satellite observations into NWP models as well as retrieving geophysical products from satellite measurements. CRTM is a community model developed by NOAA JCSDA and widely used for different purposes requiring RT calculations. CRTM requires bulk scattering lookup tables in order to perform all-sky RT calculations. However, the current CRTM lookup tables for microwave frequencies were generated based on the Mie theory by assuming spherical frozen particles. The scattering lookup tables generated using the DDA technique has shown to largely improve the RT scattering calculations in the MW region. This presentation targets (i) the implementation and validation of a DDA database that was originally developed for the ARTS RT model into CRTM, and (ii) developing the CRTM active sensor module that takes advantage of the backscattering coefficients computed using the DDA method. The DDA database only provides single scattering properties of different habits, while CRTM requires bulk scattering properties. The CRTM cloud coefficients were previously generated based on the effective radius for representing the size of the particles. However, effective radius is neither measurable nor provided by the NWP models, thus need to be estimated from other geophysical variables such as water content. Therefore, in addition to calculating the CRTM bulk scattering properties from the DDA single scattering database, the CRTM was also largely modified to use cloud water content (kg.m-3), instead of effective radius, for performing the interpolation over size/mass of the particles. CRTM already requires water content as input, thus no extra variables are required for performing scattering calculations using the new ARTS DDA database. The CRTM scattering modules search for effective radius in cloud coefficient files and will use the cloud water content if the effective radius dimension is not found in the cloud coefficient files. Figure 1 shows the CRTM simulated brightness temperatures computed using different cloud coefficients versus ATMS observed values over Hurricane Irma on September 7, 2017 at 18:00 UTC. We used all the cloud water content values included in ERA5 with default CRTM/DDA habits for water, rain, snow, ice, hail, and graupel. ERA5 does not provide separate water content values for ice, hail, and graupel, thus the ice water content values were divided between ice, hail, and graupel clouds similar to what was explained in the previous section. In channels with a frequency lower than 90 GHz, emission from water and rain clouds can compensate for cloud scattering so that cloud contaminated Tbs are larger than corresponding clear sky Tbs. The DDA simulations for channels 1-7 largely perform better than the Mie simulations. The DDA simulations show a mix of small negative and positive simulated minus observed values, while the Mie results show large negative biases. The weighting functions for some of the ATMS temperature sounding channels (channels 9-15) peak mostly above the clouds, therefore the measured Tbs become less sensitive to clouds so that the results of both Mie and DDA become very similar. The Mie lookup tables generate excessive scattering for channel 16, but not enough scattering for the water vapor channels. In the specific case of Hurricane Maria, the DDA lookup tables do not generate enough scattering for channel 16, but the DDA results are much more consistent with observations for water vapor channels than for channel 16. It should be noted that the results may vary if we use other habits to represent snow, hail, and graupel in the DDA simulations. Although these results clearly show the advantage of the DDA database over the Mie dataset, different error sources such as error in the observations, displacement of clouds in the ERA5 reanalysis, and also lack of convective clouds or in general errors in the input atmospheric and cloud profiles contribute to the differences between the simulated and observed values. Aside from the improvements in the simulations, a major advantage of the new dataset is a large number of habits that can be used to tune the data assimilation systems to perform well in different weather conditions.

Isaac Moradi↗

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↗