Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “application-specific 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.

34 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Power Converter Circuit Design Automation using Parallel Monte Carlo Tree Search

The tidal waves of modern electronic/electrical devices have led to increasing demands for ubiquitous application-specific power converters. A conventional manual design procedure of such power converters is computation- and labor-intensive, which involves selecting and connecting component devices, tuning component-wise parameters and control schemes, and iteratively evaluating and optimizing the design. To automate and speed up this design process, we propose an automatic framework that designs custom power converters from design specifications using Monte Carlo Tree Search. Specifically, the framework embraces the upper-confidence-bound-tree (UCT), a variant of Monte Carlo Tree Search, to automate topology space exploration with circuit design specification-encoded reward signals. Moreover, our UCT-based approach can exploit small offline data via the specially designed default policy and can run in parallel to accelerate topology space exploration. Further, it utilizes a hybrid circuit evaluation strategy to substantially reduce design evaluation costs. Empirically, we demonstrated that our framework could generate energy-efficient circuit topologies for various target voltage conversion ratios. Compared to existing automatic topology optimization strategies, the proposed method is much more computationally efficient --- the sequential version can generate topologies with the same quality while being up to 67% faster. Here, the parallelization schemes can further achieve high speedups compared to the sequential version.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2022

Building an Integrated Ecosystem of Computational and Observational Facilities to Accelerate Scientific Discovery

Future scientific discoveries will rely on flexible ecosystems that incorporate modern scientific instruments, high performance computing resources, parallel distributed data storage, and performant networks across multiple, independent facilities. In addition to connecting physical resources, such an ecosystem presents many challenges in logistics and accessibility, especially in orchestrating computations and experiments that span across leadership computing systems and experimental instruments. Past efforts have typically been application-specific or limited to interfaces for computing resources. This paper proposes a general framework for integrating computation resources and instrument operations, addressing challenges in code development/execution, data staging and collection, software stack, control mechanisms, resource authorization and governance, and hardware integration. We also describe a demonstration use case wherein a Bayesian optimization algorithm running on an edge computing resource guides a scanning probe microscope to autonomously and intelligently characterize a material sample. This science edge ecosystem framework will provide a blueprint for federating multi-institutional, disparate resources and orchestrating scientific workflows across them to enable next-generation discoveries.

Somnath, Suhas↗
DOE OSTI2023

Towards On-Chip Learning for Low Latency Reasoning with End-to-End Synthesis

The Software Defined Architectures (SODA) Synthesizer is an open-source compiler-based tool able to automatically generate domain-specialized systems targeting Application-Specific Integrated Circuits (ASICs) or Field Programmable Gate Arrays (FPGAs) starting from high-level programming. SODA is composed of a frontend, SODA-OPT, which leverages the multilevel intermediate representation (MLIR) framework to interface with productive programming tools (e.g., machine learning frame-works), identify kernels suitable for acceleration, and perform high-level optimizations, and of a state-of-the-art high-level synthesis backend, Bambu from the PandA framework, to generate custom accelerators. One specific application of the SODA Synthesizer is the generation of accelerators to enable ultra-low latency inference and control on autonomous systems for scientific discovery (e.g., electron microscopes, sensors in particle accelerators, etc.). This paper provides an overview of the flow in the context of the generation of accelerators for edge processing to be integrated in transmission electron microscopy (TEM) devices, focusing on use cases from precision material synthesis. We show the tool in action with an example of design space exploration for inference on reconfigurable devices with a conventional deep neural network model (LeNet). Finally, we discuss the research directions and opportunities enabled by SODA in the area of autonomous control for scientific experimental workflows.

Castellana, Vito G.↗
DOE OSTI2022

Bridging Python to Silicon: The SODA Toolchain

Systems performing scientific computing, data analysis, and machine learning tasks have a growing demand for application-specific accelerators that can provide high computational performance while meeting strict size and power requirements. However, the algorithms and applications that need to be accelerated are evolving at a rate that is incompatible with manual design processes based on hardware description languages. Agile hardware design tools based on compiler techniques can help by quickly producing an application-specific integrated circuit (ASIC) accelerator starting from a high-level algorithmic description. Here, we present the software-defined accelerator (SODA) synthesizer, a modular and open-source hardware compiler that provides automated end-to-end synthesis from high-level software frameworks to ASIC implementation, relying on multilevel representations to progressively lower and optimize the input code. Our approach does not require the application developer to write any register-transfer level code, and it is able to reach up to 364 giga floating point operations per second (GFLOPS)/W efficiency (32-bit precision) on typical convolutional neural network operators.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Ultrafast radiographic imaging and tracking: An overview of instruments, methods, data, and applications

Ultrafast radiographic imaging and tracking (U-RadIT) use state-of-the-art ionizing particle and light sources to experimentally study sub-nanosecond transients or dynamic processes in physics, chemistry, biology, geology, materials science and other fields. These processes are fundamental to modern technologies and applications, such as nuclear fusion energy, advanced manufacturing, communication, and green transportation, which often involve one mole or more atoms and elementary particles, and thus are challenging to compute by using the first principles of quantum physics or other forward models. One of the central problems in U-RadIT is to optimize information yield through, e.g. high-luminosity X-ray and particle sources, efficient imaging and tracking detectors, novel methods to collect data, and large-bandwidth online and offline data processing, regulated by the underlying physics, statistics, and computing power. We review and highlight recent progress in: (a.) Detectors such as high-speed complementary metal-oxide semiconductor (CMOS) cameras, hybrid pixelated array detectors integrated with Timepix4 and other application-specific integrated circuits (ASICs), and digital photon detectors; (b.) U-RadIT modalities such as dynamic phase contrast imaging, dynamic diffractive imaging, and four-dimensional (4D) particle tracking; (c.) U-RadIT data and algorithms such as neural networks and machine learning, and (d.) Applications in ultrafast dynamic material science using XFELs, synchrotrons and laser-driven sources. Hardware-centric approaches to U-RadIT optimization are constrained by detector material properties, low signal-to-noise ratio, high cost and long development cycles of critical hardware components such as ASICs. Interpretation of experimental data, including comparisons with forward models, is frequently hindered by sparse measurements, model and measurement uncertainties, and noise. Alternatively, U-RadIT make increasing use of data science and machine learning algorithms, including experimental implementations of compressed sensing. Machine learning and artificial intelligence approaches, refined by physics and materials information, may also contribute significantly to data interpretation, uncertainty quantification and U-RadIT optimization.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

Application-specific machine-learned interatomic potentials: exploring the trade-off between DFT convergence, MLIP expressivity, and computational cost

Machine-learned interatomic potentials (MLIPs) are revolutionizing computational materials science and chemistry by offering an efficient alternative to ab initio molecular dynamics (MD) simulations. However, fitting high-quality MLIPs remains a challenging, time-consuming, and computationally intensive task where numerous trade-offs have to be considered, e.g., How much and what kind of atomic configurations should be included in the training set? Which level of ab initio convergence should be used to generate the training set? Which loss function should be used for fitting the MLIP? Which machine learning architecture should be used to train the MLIP? The answers to these questions significantly impact both the computational cost of MLIP training and the accuracy and computational cost of subsequent MLIP MD simulations. In this study, we use a configurationally diverse beryllium dataset and quadratic spectral neighbor analysis potential. We demonstrate that joint optimization of energy versus force weights, training set selection strategies, and convergence settings of the ab initio reference simulations, as well as model complexity can lead to a significant reduction in the overall computational cost associated with training and evaluating MLIPs. This opens the door to computationally efficient generation of high-quality MLIPs for a range of applications which demand different accuracy versus training and evaluation cost trade-offs.

36 MATERIALS SCIENCE↗
DOE OSTI2021

A 24-Channel Digitizer With a JESD204B-Compliant Serial Interface for High-Speed Detectors

In this report a 24-channel application-specific integrated circuit (ASIC) for the readout of high-speed CMOS active pixel sensors for charged particle detection is presented. The chip comprises 24 preamplifiers, 24 distinct 12-bit, 25 MSPS Pipelined analog-to-digital converters (ADCs) with self-calibration, an internal phase-locked loop (PLL), and a 3 Gbps serial interface that conforms to the JESD204B standard. To simplify interfacing with a variety of sensors, the ASIC also includes an automatic offset calibration loop. The high level of integration of the ASIC reduces overall system cost and area, and exploiting the signal characteristics of the image sensor allows the ADC to be optimized for reduced power dissipation. The use of an integrated serializer and an industry standard protocol simplifies integration of the ASIC into a complete camera system. The ASIC, called the High-Speed Image Preprocessor Targeted for Electron Readout, or HIPSTER, with a die area of 64.26 mm 2 , is packaged in a 480-ball grid array (BGA) and is fabricated in 180-nm CMOS technology. HIPSTER achieves typical differential nonlinearity (DNL) < 0.55 LSB, input-referred thermal noise of $114.5 μV-rms, and a bit error rate (BER) of better than 10-14. The power dissipation is 98 mW/channel.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2023

ML-ACCEPT: Machine-Learning-enhanced Automated Circuit Configuration and Evaluation of Power Converters

This project, ML-ACCEPT, took a significant step toward achieving the first objective of the DIFFERENTIATE program: to help engineers to more rapidly and cost-effectively consider a wider range of more novel concepts before selecting an engineering-optimal architecture for high-fidelity detailed design and evaluation. Compared to existing methods, the ML-ACCEPT project investigated innovative technologies to make the design of power converters more cost-effective and time-efficient by i) integrating recent breakthroughs in ML, power electronics, simulation software, and optimization to research, develop, and developed a suite of ML-enhanced hypothesis generation tools for power converter design; and ii) facilitating the integration of the proposed software tools into existing power-converter design work-flows. Some existing tools for power-converter design have a certain level of intelligence to aid the design process, but to our knowledge none of them has ML-enhanced capability of automating the electrical circuit design process by selecting, configuring, and tailoring the individual components that enable available resources (e.g., high-voltage DC power) to be converted to the desired output (e.g., lower-voltage DC power with a desired voltage ripple), subject to application-specific (e.g., plasma generation and automotive applications) thermal and packaging considerations. The state-of-the-art circuit design of power converters is still heavily reliant on human experts to select the optimal topology and search for design parameters with human’s experience and intuitions, which can be very time-consuming, inefficient, and labor intensive. The team is composed of University of Michigan-Dearborn (UM-Dearborn), Lawrence Livermore National Laboratory (LLNL), Oak Ridge National Laboratory (ORNL), Modelon Inc., and lighthouse customers.

42 ENGINEERING↗
DOE OSTI2024

Neural network methods for radiation detectors and imaging

Recent advances in image data proccesing through deep learning allow for new optimization and performance-enhancement schemes for radiation detectors and imaging hardware. This enables radiation experiments, which includes photon sciences in synchrotron and X-ray free electron lasers as a subclass, through data-endowed artificial intelligence. We give an overview of data generation at photon sources, deep learning-based methods for image processing tasks, and hardware solutions for deep learning acceleration. Most existing deep learning approaches are trained offline, typically using large amounts of computational resources. However, once trained, DNNs can achieve fast inference speeds and can be deployed to edge devices. A new trend is edge computing with less energy consumption (hundreds of watts or less) and real-time analysis potential. While popularly used for edge computing, electronic-based hardware accelerators ranging from general purpose processors such as central processing units (CPUs) to application-specific integrated circuits (ASICs) are constantly reaching performance limits in latency, energy consumption, and other physical constraints. These limits give rise to next-generation analog neuromorhpic hardware platforms, such as optical neural networks (ONNs), for high parallel, low latency, and low energy computing to boost deep learning acceleration (LA-UR-23-32395).

edge computing↗
DOE OSTI2021

A Reconfigurable Neural Network ASIC for Detector Front-End Data Compression at the HL-LHC

Despite advances in the programmable logic capabilities of modern trigger systems, a significant bottleneck remains in the amount of data to be transported from the detector to off-detector logic where trigger decisions are made. We demonstrate that a neural network (NN) autoencoder model can be implemented in a radiation-tolerant application-specific integrated circuit (ASIC) to perform lossy data compression alleviating the data transmission problem while preserving critical information of the detector energy profile. For our application, we consider the high-granularity calorimeter from the Compact Muon Solenoid (CMS) experiment at the CERN Large Hadron Collider. The advantage of the machine learning approach is in the flexibility and configurability of the algorithm. By changing the NN weights, a unique data compression algorithm can be deployed for each sensor in different detector regions and changing detector or collider conditions. To meet area, performance, and power constraints, we perform quantization-aware training to create an optimized NN hardware implementation. The design is achieved through the use of high-level synthesis tools and the hls4ml framework and was processed through synthesis and physical layout flows based on a low-power (LP)-CMOS 65-nm technology node. The flow anticipates 200 Mrad of ionizing radiation to select gates and reports a total area of 3.6 mm 2 and consumes 95 mW of power. The simulated energy consumption per inference is 2.4 nJ. Furthermore, this is the first radiation-tolerant on-detector ASIC implementation of an NN that has been designed for particle physics applications.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2020

Changes in SSL Device Efficiency and Optical Performance Under Accelerated Aging Conditions

Lighting application efficiency (LAE) describes the efficient delivery of light from the light source to the lighted task and is viewed as a new frontier—increasing energy savings with solid-state lighting (SSL) technologies. The framework for LAE that is proposed by the U.S. Department of Energy (DOE) consists of four major elements: light source efficiency, optical delivery efficiency, spectral efficiency, and intensity effectiveness. This report focuses on a sampling of the available SSL products that can be broadly defined as having modified spectral output because the method of spectra modification has a significant impact on light source efficiency and long-term optical delivery and spectral efficiencies. This report focused on the changes in light source, optical delivery, and spectral efficiencies that occur during aging of SSL devices. An accelerated stress test (AST) regiment was developed for the devices under test (DUTs) examined in this report to provide insights into how the performance of SSL devices change with aging. The AST protocols demonstrate that the SSL products discussed in this report often reduce light source efficiency to achieve different spectral characteristics. In addition, aging of the optical components (e.g., lenses, solder masks) in SSL devices can produce increased light absorption, which negatively impacts optical delivery efficiency and light source efficiency. The modified spectral outputs of the selected SSL products discussed in this report come in a variety of form factors and achieve enhanced optical performance by using a variety of methods. All of the products examined in this study use mid-power light-emitting diodes (MP-LEDs), although the number, configuration, phosphor content, and light-emitting diode (LED) pump of the MP-LEDs differ. Product MS-1 is a 60-watt (W) replacement A19 lamp with a hermetically sealed glass globe, which contains an embedded optical filter to absorb green and yellow emissions, thereby creating a “sunlike” modified spectrum. Product MS-2 is a 60-W replacement A19 lamp with 30 MP-LEDs, and it uses a violet LED pump, along with green and red phosphor emissions, to produce a “healthy” spectrum. This spectrum omits blue emissions in an effort to reduce melanopic lux. Product MS-3 is an LED module consisting of 21 MP-LEDs that use a violet LED pump, along with blue, green, and red phosphors, to produce a “sunlike” spectrum. Products MS-4 and MS-5 are both 6-inch (in) downlights that use a manual switching mechanism so that users can select application-specific correlated color temperatures before installation. Products MS-4 and MS-5 both contain two LED primaries (2,700 Kelvin [K] and 5,000 K) for spectral tuning. Product MS-4 contains 12 MP-LEDs for each LED primary, and Product MS-5 contains 10 MP-LEDs for each LED primary. This report summarizes the overall findings from up to 8,000 hours (hrs) of AST on the lamp DUTs (Products MS-1 and MS-2); up to 5,000 hrs of AST on the LED light engine DUTs (Product MS-3); and up to 7,000 hrs of AST on the downlight DUTs (Products MS-4 and MS-5). The AST procedures used in this study included a room temperature operational life (RTOL) test, an operational life test conducted at 45 degrees Celsius (°C; 45OL) test, an operational life test conducted at 75°C (75OL), a wet high-temperature operational life test performed at 65°C and 90% relative humidity (6590), and a wet high-temperature operational life test performed at 75°C and 75% relative humidity (7575). The AST procedures used for Products MS-1 and MS-2 were RTOL, 45OL, and 6590. The AST procedures used for Products MS-3, MS-4, and MS-5 were RTOL, 75OL, and 7575. During the ASTs described herein, separate populations of each product (three DUTs in each population for Products MS-1, MS-2, MS-4, and MS-5; four DUTs in each population for Product MS-3) were subjected to power cycling of 1 hr on and 1 hr off. The key findings from this study include the following. Enhanced optical performance came at the cost of reduced light source efficiency for the lamps and light engines examined in this study. The optical filter used to produce a “sunlike” spectrum for Product MS-1 reduced light source efficiency by 26%, from 113 lumens per watt (lm/W) to 85 lm/W. Products MS-2 and MS-3 that used a violet-pumped LED to achieve “healthy” and “sunlike” enhanced optical performance, respectively, suffered the largest reduction in initial light source efficiency (49 lm/W for Product MS-2 and 68 lm/W for Products MS-3) perhaps because of the use of violet LED as the optical pump. Product MS-2 also had the poorest color fidelity because of undersaturation of blues and oversaturation of greens and yellows. Chromaticity maintenance of the products in this study was generally good, with parametric failure only occurring for one product at one AST condition (i.e., Product MS-3 at 7575). The chromaticity shift for Product MS-3 in 7575 test conditions resulted from reduced emissions of the broad green and red phosphors used to mimic sunlight in the 500–750 nanometer (nm) range. As a result of these phosphor emission losses, chromaticity shifted toward the more stable violet-pumped LED and blue phosphor. Although chromaticity maintenance was acceptable for most products tested, different chromaticity shift mechanisms were observed for temperature and humidity tests compared with temperature alone for Products MS-1, MS-2, and MS-3. For Product MS-1, a relative increase in green emissions was observed with humidity, which may indicate humidity-accelerated degradation of the optical filter at green wavelengths (optical delivery efficiency reduction) or degradation of the emitters in the red region (spectral efficiency reduction). The violet-pumped products (i.e., MS-2 and MS-3) exhibited different chromaticity shift behavior in the temperature-humidity environments, with Product MS-2 shifting generally yellow because of photo-induced oxidation of the plastic globe (i.e., a reduction of optical delivery efficiency) and Product MS-3 shifting toward violet and blue emitters because of faster loss of emission from green and red phosphors (i.e., a reduction in spectral efficiency). Because of an initial drop in power consumption, the light source efficiency of downlights (Products MS-4 and MS-5) increased at RTOL throughout the test duration. The light source efficiency initially increased during 75OL until a decay in luminous flux maintenance (LFM) dominated light source efficiency, whereas the light source efficiency decreased for the entire 7575 test duration. It was found that increasing the ambient environment of an SSL device from 25°C to 75°C decreases the LFM by 2.7 to 5.5, depending on design of the SSL device. Adding humidity to the system (75OL to 7575) was found to decrease LFM by another factor of approximately 3.0. A common location of failure was identified for the downlights (Products MS-4 and MS-5) operated in the 7575 environment. This failure abruptly occurred between 3,000 to 5,000 hrs of testing on the film capacitor of the electromagnetic interference (EMI) filter near the diode bridge for 11 of the 12 downlights. It is likely that this failure was caused by a combination of the high electrical voltage across the capacitor, the location of the capacitor near the heat sources (e.g., transformers, diode bridges, power resistors), and the high stress environment of 7575. Longer test times are needed to fully understand the LFM, luminous efficacy changes, and chromaticity shifts for some of the products. Because of the high reliability, light source efficiency, and spectral tuning capabilities of LEDs, the expectations of LAE for SSL products are greatly increased over traditional lighting products. The data gathered in this report begin to provide an understanding of the tradeoffs that current SSL products undergo when optimizing the different efficiency elements of LAE. As discussed in this report, the findings from the tests conducted show that the optimization of spectral efficiency can come at the cost of initial light source efficiency. Furthermore, the introduction of violet LEDs can promote long-term optical delivery efficiency degradation, and the introduction of optical filters or new phosphors can lead to unwanted spectral efficiency changes because the ratio of phosphor emitters changes with aging. The data presented in this report also identified a common failure location in 6-in downlights. The results regarding long-term behavior of the modified spectra devices studied provide valuable information about changes to the light source, spectral, and optical delivery efficiencies as the devices age. This information can be used to improve future SSL designs.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2025

The CMS Phase-2 Fast Beam Condition Monitor prototype test with beam

The Fast Beam Condition Monitor (FBCM) is a standalone luminometer for the High Luminosity LHC (HL-LHC) program of the CMS Experiment at CERN. The detector is under development and features a new, radiation-hard, front-end application-specific integrated circuit (ASIC) designed for beam monitoring applications. The achieved timing resolution of a few nanoseconds enables the measurement of both the luminosity and the beam-induced background. The ASIC, called FBCM23, features six channels with adjustable shaping times, enabling in-field fine-tuning. Each ASIC channel outputs a single binary asynchronous signal encoding time-of-arrival and time-over-threshold information. The FBCM is based on silicon-pad sensors, with two sensor designs presently being considered. This paper presents the results of tests of the FBCM detector prototype using both types of silicon sensors with hadron, muon, and electron beams. Irradiated FBCM23 ASICs and silicon-pad sensors were also tested to simulate the expected conditions near the end of the detector's lifetime in the HL-LHC radiation environment. Based on test results, direct bonding between the sensor and ASIC was chosen, and an optimal bias voltage and ASIC threshold for FBCM operation were proposed. The current design of the front-end test board was validated following the beam test and is now being used for the first front-end module, which is expected to be produced in summer 2025. These results represent a major step forward in validating the FBCM concept, first version of the firmware and establishing a reliable design path for the final detector.

Beam-line instrumentation (beam position and profi↗
DOE OSTI2025

High-throughput spin-bath characterization of spin defects in semiconductors

Detailed knowledge of the local environments of spin defects in semiconductors, such as nitrogenvacancy (NV) centers in diamond or divacancies in silicon carbide, is crucial for optimizing control and entanglement protocols in quantum sensing and information applications. However, at present a direct experimental characterization of individual defect environments is not scalable, as conventional spin-bath measurements are time consuming and difficult to automate. Achieving high-throughput characterization requires short experiments to probe the spin bath. However, with fewer and noisier measurements, the inverse problem of recovering spin-bath properties from measured data becomes ill posed, with multiple spin baths having a high likelihood of yielding the same data. In this work, we present a set of computational tools to resolve the ill-posed inverse problem of recovering the atomic positions and hyperfine couplings of random nuclei surrounding spin defects from sparse, noisy experimental coherence data, which can be obtained in hours. Here, we use a trans-dimensional Bayesian approach that incorporates ab initio data to yield full posterior distributions over nuclear spin environments, enabling robust recovery from limited data. We also provide practical tools and guidelines to determine the limits of detectability for hyperfine couplings under specific dynamical decoupling sequences and sampling conditions. In addition, we demonstrate how the tools developed here, in combination with ab initio simulations of spin baths, can guide the design of efficient experimental protocols for application-specific high-throughput screening. To showcase the utility of our approach, we apply it to design fast dynamical decoupling experiments to characterize the spin baths often individual NV centers in diamond. While the primary focus is on accelerating spin-bath characterization of spin defects, this Bayesian approach also lays the foundation for digital-twin studies of spin defects, where a virtual model of the spin-defect system evolves in real time with ongoing experimental measurements. Together, the set of tools we designed and applied paves the way for scalable deployment of spin defects in semiconductors for quantum sensing and information applications.

Bayesian methods↗
DOE OSTI2024

Towards Sustainable Post-Exascale Leadership Computing

As computing systems approach the limits of traditional silicon technology, the diminishing returns in performance per watt present a significant barrier to sustaining growth in HPC. From a large-scale scientific supercomputing facility point of view, we propose a multifaceted strategy toward specialized hardware and architectures that are optimized for energy efficiency in specific applications. We also emphasize the need for integrating energy-aware practices across all levels of HPC, from system design and software development to operational policies. We discuss strategic opportunities such as the adoption of application-specific accelerators, the development of energy-efficient algorithms, and the implementation of data-driven operational analytics. Our goal is to develop a comprehensive roadmap ensuring that future leadership systems at OLCF can meet scientific demands while operating within stringent energy budgets, thereby supporting sustainable computing growth.

Shin, Woong↗
DOE OSTI2022

Model Assumptions and Data Characteristics: Impacts on Domain Adaptation in Building Segmentation

Studies on domain adaptation (DA) for remote sensing (RS) imagery analysis lack consistency in selection and description of evaluation scenarios. Without properly characterizing datasets, model assumptions, and evaluation scenarios, it is difficult to objectively compare DA methods and reach conclusions about their suitability across different applications. With this motivation, this work seeks to empirically assess to which extent the interaction between data characteristics and model assumptions influences the effectiveness of DA methods. Using the widely explored task of building footprint segmentation as a case study, we perform a large-scale study across over 200 DA scenarios that include variations across view angles, areas observed, and sensors used for data acquisition. Rather than adopting different model architectures or optimization criteria, we contrast the performances of two DA methods based on adversarial learning that differ only in their assumptions about source and target domains. Informed by metadata and data characteristics unveiled using traditional computer vision (CV) techniques as well as pretrained deep models, we provide a detailed meta-analysis of experiments highlighting the importance of accurately considering data assumptions for DA in RS segmentation tasks. As demonstrated by a “cherry-picking” exercise, different claims regarding which model is best could be made by selecting different subsets of evaluation scenarios. While well-calibrated assumptions can be beneficial, mismatching assumptions can lead to negative biases in DA applications. Furthermore, this study intends to motivate the community toward more consistent evaluation protocols while providing recommendations and insights toward creating novel benchmark datasets, documenting data characteristics, application-specific knowledge, and model assumptions.

42 ENGINEERING↗
DOE OSTI2022

SODA Synthesizer: an Open-source, Multi-level, Modular, Extensible Compiler from High-level Frameworks to Silicon

The SODA Synthesizer is an open-source modular, end-to-end hardware compiler framework. The SODA frontend, developed in MLIR, performs system-level design, code partitioning, and high-level optimizations to prepare the specifications for the hardware synthesis. The backend is based on a state-of-the-art high-level synthesis tool, and generates the final hardware design. The backend can interface with logic synthesis tools for field programmable gate arrays or with commercial and open-source logic synthesis tools for application-specific integrated circuits. We discuss the opportunities and challenges in integrating with commercial and open-source tools both at the frontend and backend, and the unique opportunities that an open-source hardware design ecosystem provides.

Bohm Agostini, Nicolas↗