Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “language”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,657 records · Page 92

From Text to Maps: LLM-Driven Extraction and Geotagging of Epidemiological Data

Epidemiological datasets are essential for public health analysis and decision-making, yet they remain scarce and often difficult to compile due to inconsistent data formats, language barriers, and evolving political boundaries. Traditional methods of creating such datasets involve extensive manual effort and are prone to errors in accurate location extraction. To address these challenges, we propose utilizing large language models (LLMs) to automate the extraction and geotagging of epidemiological data from textual documents. Our approach significantly reduces the manual effort required, limiting human intervention to validating a subset of records against text snippets and verifying the geotagging reasoning, as opposed to reviewing multiple entire documents manually to extract, clean, and geotag. Additionally, the LLMs identify information often overlooked by human annotators, further enhancing the dataset’s completeness. Our findings demonstrate that LLMs can be effectively used to semi-automate the extraction and geotagging of epidemiological data, offering several key advantages: (1) comprehensive information extraction with minimal risk of missing critical details; (2) minimal human intervention; (3) higher-resolution data with more precise geotagging; and (4) significantly reduced resource demands compared to traditional methods.

Harrod, Karly↗

Multilingual Connotation Frames: A Case Study on Social Media for Targeted Sentiment Analysis and Forecast

People around the globe respond to major real world events through social media. To study targeted public sentiments across many languages and geographic locations, we introduce multilingual connotation frames: an extension from English connotation frames of Rashkin et al. (2016) with 10 additional European languages, focusing on the implied sentiments among event participants engaged in a frame. As a case study, we present large scale analysis on targeted public sentiments using 1.2 million multilingual connotation frames extracted from Twitter. We rely on connotation frames to build models to forecast country-specific connotation dynamics – perspective change over time towards salient entities and events. Our results demonstrate that connotation dynamics can be accurately predicted up to half a week in advance.

connotation frames↗

Towards Digital and Performance-Based Supervisory HVAC Control Delivery

Upgrading supervisory HVAC control in commercial buildings is one of the most attractive decarbonization tools at our disposal. Modern controls are software programs and can in theory be deployed at scale and with a low up-front carbon "pulse". In practice, however, control delivery is a disjointed and inefficient process, dominated by manual handoffs of imprecise English language documents. A particularly high barrier exists between control implementation and building energy modeling (BEM) which results in control sequences typically not being tested for correctness or performance before implementation. Together with industry partners, DOE and the national labs are developing an ecosystem of tools and standards that can support fully digital performance-based control delivery workflows. This paper describes this ecosystem, which consists of three mutually supportive efforts. Semantic models of buildings and their systems enable automatic configuration and installation of control software. Platform-neutral control descriptions separate control algorithms from control platforms and enable the creation of libraries of reference control implementations. Dynamic whole-building energy-control simulation that can execute physically realistic control sequences makes it possible to test and evaluate the performance of control sequences and then directly compile them for installation and execution in control systems. In addition to digitizing and streamlining project-level control delivery, these standards and related software support benchmarking of control algorithms, both rule-based and optimization-based, and help to both advance the state of the art and to implement ratings and programs that encourage the adoption of high-performance control.

building controls↗

PDF Entity Annotation Tool (PEAT)

While different text mining approaches – including the use of Artificial Intelligence (AI) and other machine based methods - continue to expand at a rapid pace, the tools used by researchers to create the labeled datasets required for training, modeling, and evaluation remain rudimentary. Labeled datasets contain the target attributes the machine is going to learn; for example, training an algorithm to delineate between images of a car or truck would generally require a set of images with a quantitative description of the underlying features of each vehicle type. Development of labeled textual data that can be used to build natural language machine learning models for scientific literature is not currently integrated into existing manual workflows used by domain experts. Published literature is rich with important information, such as different types of embedded text, plots, and tables that can all be used as inputs to train ML/natural language processing (NLP) models, when extracted and prepared in machine readable formats. Currently, both normalized data extraction of use to domain experts and extraction to support development of ML/NLP models are labor intensive and cumbersome manual processes. Automatic extraction of data and information from formats such as PDFs that are optimized for layout and human readability, not machine readability. The PDF (Portable Document Format) Entity Annotation Tool (PEAT) was developed with the goal of allowing users to annotate publications within their current print format, while also allowing those annotations to be captured in a machine-readable format. One of the main issues with traditional annotation tools is that they require transforming the PDF into plain text to facilitate the annotation process. While doing so lessens the technical challenges of annotating data, the user loses all structure and provenance that was inherent in the underlying PDF. Also, textual data extraction from PDFs can be an error prone process. Challenges include identifying sequential blocks of text and a multitude of document formats (multiple columns, font encodings, etc.). As a result of these challenges, using existing tools for development of NLP/ML models directly from PDFs is difficult because the generated outputs are not interoperable. We created a system that allows annotations to be completed on the original PDF document structure, with no plain text extraction. The result is an application that allows for easier and more accurate annotations. In addition, by including a feature that grants the user the ability to easily create a schema, we have developed a system that can be used to annotate text for different domain-centric schemas of relevance to subject matter experts. Different knowledge domains require distinct schemas and annotation tags to support machine learning.

97 MATHEMATICS AND COMPUTING↗

Fiats: Functional inference and training for surrogates

Fiats provides a platform for research on the training and deployment of neural-network surrogate models for computational science. Fiats also supports exploring, advancing, and combining functional, object-oriented, and parallel programming patterns in Fortran 2023. As such, the Fiats name has dual expansions: “Functional Inference And Training for Surrogates” or “Fortran Inference And Training for Science.” Fiats inference and training procedures are pure and therefore satisfy a language constraint imposed on procedure invocations inside Fortran’s parallel loop construct: do concurrent. Furthermore, the Fiats training procedures are built around a do concurrent parallel reduction. Several compilers can automatically parallelize do concurrent on Central Processing Units (CPUs) or Graphics Processing Units (GPUs). Fiats thus aims to achieve performance portability through standard language mechanisms.

Rouson, Damian [Lawrence Berkeley National Laborat↗

TORCH Computational Reference Kernels - A Testbed for Computer Science Research

For decades, computer scientists have sought guidance on how to evolve architectures, languages, and programming models in order to improve application performance, efficiency, and productivity. Unfortunately, without overarching advice about future directions in these areas, individual guidance is inferred from the existing software/hardware ecosystem, and each discipline often conducts their research independently assuming all other technologies remain fixed. In today's rapidly evolving world of on-chip parallelism, isolated and iterative improvements to performance may miss superior solutions in the same way gradient descent optimization techniques may get stuck in local minima. To combat this, we present TORCH: A Testbed for Optimization ResearCH. These computational reference kernels define the core problems of interest in scientific computing without mandating a specific language, algorithm, programming model, or implementation. To compliment the kernel (problem) definitions, we provide a set of algorithmically-expressed verification tests that can be used to verify a hardware/software co-designed solution produces an acceptable answer. Finally, to provide some illumination as to how researchers have implemented solutions to these problems in the past, we provide a set of reference implementations in C and MATLAB.

97 MATHEMATICS AND COMPUTING↗

Automated design synthesis of robotic/human workcells for improved manufacturing system design in hazardous environments [Slides]

Manufacturing tasks that are deemed too hazardous for workers require the use of automation, robotics, and/or other remote handling tools. The associated hazards may be radiological or non-radiological, and based on the characteristics of the environment and processing, a design may necessitate robotic labor, human labor, or both. There are also other factors such as cost, ergonomics, maintenance, and efficiency that also effect task allocation and other design choices. Handling the tradeoffs of these factors can be complex, and lack of experience can be an issue when trying to determine if and what feasible automation/robotics options exist. To address this problem, we utilize common engineering design approaches adapted more for manufacturing system design in hazardous environments. We limit our scope to the conceptual and embodiment design stages, specifically a computational algorithm for concept generation and early design evaluation. In regard to concept generation, we first develop the functional model or function structure for the process, using the common 'verb-noun' format for describing function. A common language or functional basis for manufacturing was developed and utilized to formalize function descriptions and guide rules for function decomposition. Potential components for embodiment are also grouped in terms of this functional language and are stored in a database. The properties of each component are given as quantitative and qualitative criteria. Operators are also rated for task-relevant criteria which are used to address task compatibility. Through the gathering of process requirements/constraints, construction of the component database, and development of the manufacturing basis and rule set, design knowledge is stored and available for computer use. Thus, once the higher level process functions are defined, the computer can automate the synthesis of new design concepts through alternating steps of embodiment and function structure updates/decomposition. In the process, criteria guide function allocation of components/operators and help ensure compatibility and feasibility. Through multiple function assignment options and varied function structures, multiple design concepts are created. All of the generated designs are then evaluated based on a number of relevant evaluation criteria: cost, dose, ergonomics, hazards, efficiency, etc. These criteria are computed using physical properties/parameters of each system based on the qualities an engineer would use to make evaluations. Nuclear processes such as oxide conversion and electrorefining are utilized to aid algorithm development and provide test cases for the completed program. Through our approach, we capture design knowledge related to manufacturing and other operations in hazardous environments to enable a computational program to automatically generate and evaluate system design concepts.

42 ENGINEERING↗

Renewable Energy Requirements for Future Building Codes: Options for Compliance

As the model energy codes are improved to reach efficiency levels 50 percent greater than current codes, use of on-site renewable energy generation is likely to become a code requirement. This requirement will be needed because traditional mechanisms for code improvement, including envelope, mechanical and lighting, have been pressed to the end of reasonable limits. Research has been conducted to determine the mechanism for implementing this requirement (Kaufman 2011). Kaufmann et al. determined that the most appropriate way to structure an on-site renewable requirement for commercial buildings is to define the requirement in terms of an installed power density per unit of roof area. This provides a mechanism that is suitable for the installation of photovoltaic (PV) systems on future buildings to offset electricity and reduce the total building energy load. Kaufmann et al. suggested that an appropriate maximum for the requirement in the commercial sector would be 4 W/ft 2 of roof area or 0.5 W/ft 2 of conditioned floor area. As with all code requirements, there must be an alternative compliance path for buildings that may not reasonably meet the renewables requirement. This might include conditions like shading (which makes rooftop PV arrays less effective), unusual architecture, undesirable roof pitch, unsuitable building orientation, or other issues. In the short term, alternative compliance paths including high performance mechanical equipment, dramatic envelope changes, or controls changes may be feasible. These options may be less expensive than many renewable systems, which will require careful balance of energy measures when setting the code requirement levels. As the stringency of the code continues to increase however, efficiency trade-offs will be maximized, requiring alternative compliance options to be focused solely on renewable electricity trade-offs or equivalent programs. One alternate compliance path includes purchase of Renewable Energy Credits (RECs). Each REC represents a specified amount of renewable electricity production and provides an offset of environmental externalities associated with non-renewable electricity production. The purpose of this paper is to explore the possible issues with RECs and comparable alternative compliance options. Existing codes have been examined to determine energy equivalence between the energy generation requirement and the RECs alternative over the life of the building. The price equivalence of the requirement and the alternative are determined to consider the economic drivers for a market decision. This research includes case studies that review how the few existing codes have incorporated RECs and some of the issues inherent with REC markets. Section 1 of the report reviews compliance options including RECs, green energy purchase programs, shared solar agreements and leases, and other options. Section 2 provides detailed case studies on codes that include RECs and community based alternative compliance methods. The methods the existing code requirements structure alternative compliance options like RECs are the focus of the case studies. Section 3 explores the possible structure of the renewable energy generation requirement in the context of energy and price equivalence. The price of RECs have shown high variation by market and over time which makes it critical to for code language to be updated frequently for a renewable energy generation requirement or the requirement will not remain price-equivalent over time. Section 4 of the report provides a maximum case estimate for impact to the PV market and the REC market based on the Kaufmann et al. proposed requirement levels. If all new buildings in the commercial sector complied with the requirement to install rooftop PV arrays, nearly 4,700 MW of solar would be installed in 2012, a major increase from EIA estimates of 640 MW of solar generation capacity installed in 2009. The residential sector could contribute roughly an additional 2,300 MW based on the same code requirement levels of 4 W/ft 2 of roof area. Section 5 of the report provides a basic framework for draft code language recommendations based on the analysis of the alternative compliance levels.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Summer Student Internship working on the GPU

This summer internship consisted of learning computer programming skills and languages including Thrust, CUDA, C++, C and Linux. The application of this programming is for use in GPU, graphics processing units, that are processors that perform math intensive calculations rapidly in three dimensions typically in advanced video games (“NVIDIA GeForce…”). Our goal is to apply this technology towards exa-scale computing on the LANL supercomputing infrastructure. My mentors use these machines to do computations. The completed algorithms include KD tree and Top K. In order to access the GPU, CUDA was used. CUDA is a parallel computer architecture for general computing on the GPUs. There are different means by which to access CUDA. The first approach uses libraries, like Thrust, Linear Algebra, and Signal and Image Processing. Another approach parallelizes loops in Fortran or C and in the last approach people develop custom parallel algorithms using familiar language (“What is CUDA”). The algorithms I worked on this summer use the first approach of using the thrust library of parallel algorithms and data structures.

97 MATHEMATICS AND COMPUTING↗

From OO to FPGA :

Consumer electronics today such as cell phones often have one or more low-power FPGAs to assist with energy-intensive operations in order to reduce overall energy consumption and increase battery life. However, current techniques for programming FPGAs require people to be specially trained to do so. Ideally, software engineers can more readily take advantage of the benefits FPGAs offer by being able to program them using their existing skills, a common one being object-oriented programming. However, traditional techniques for compiling object-oriented languages are at odds with today s FPGA tools, which support neither pointers nor complex data structures. Open until now is the problem of compiling an object-oriented language to an FPGA in a way that harnesses this potential for huge energy savings. In this paper, we present a new compilation technique that feeds into an existing FPGA tool chain and produces FPGAs with up to almost an order of magnitude in energy savings compared to a low-power microprocessor while still retaining comparable performance and area usage.

Kou, Stephen↗

Simulink/PARS Integration Support

The state of the art for signal processor hardware has far out-paced the development tools for placing applications on that hardware. In addition, signal processors are available in a variety of architectures, each uniquely capable of handling specific types of signal processing efficiently. With these processors becoming smaller and demanding less power, it has become possible to group multiple processors, a heterogeneous set of processors, into single systems. Different portions of the desired problem set can be assigned to different processor types as appropriate. As software development tools do not keep pace with these processors, especially when multiple processors of different types are used, a method is needed to enable software code portability among multiple processors and multiple types of processors along with their respective software environments. Sundance DSP, Inc. has developed a software toolkit called “PARS”, whose objective is to provide a framework that uses suites of tools provided by different vendors, along with modeling tools and a real time operating system, to build an application that spans different processor types. The software language used to express the behavior of the system is a very high level modeling language, “Simulink”, a MathWorks product. ORNL has used this toolkit to effectively implement several deliverables. This CRADA describes this collaboration between ORNL and Sundance DSP, Inc.

97 MATHEMATICS AND COMPUTING↗

Center for Technology for Advanced Scientific Component Software

The UO portion of the larger TASCS project was focused on the usability subproject identified in the original project proposal. The key usability issue that we tacked was that of supporting legacy code developers in migrating to a component-oriented design pattern and development model with minimal manual labor. It was observed during the lifetime of the TASCS (and previous CCA efforts) that more often than not, users would arrive with existing code that was developed previous to their exposure to component design methods. As such, they were faced with the task of both learning the CCA toolchain and at the same time, manually deconstructing and reassembling their existing code to fit the design constraints imposed by components. This was a common complaint (and occasional reason for a user to abandon components altogether), so our task was to remove this manual labor as much as possible to lessen the burden placed on the end-user when adopting components for existing codes. To accomplish this, we created a source-based static analysis tool that used code annotations to drive code generation and transformation operations. The use of code annotations is due to one of the key technical challenges facing this work programming languages are limited in the degree to which application-specific semantics can be represented in code. For example, data types are often ambiguous. The C pointer is the most common example cited in practice. Given a pointer to a location in memory, should it be interpreted as a singleton or an array. If it is to be interpreted as an array, how many dimensions does the array have? What are their extents? The annotation language that we designed and implemented addresses this ambiguity issue by allowing users to decorate their code in places where ambiguity exists in order to guide tools to interpret what the programmer really intends.

97 MATHEMATICS AND COMPUTING↗

Trend Analysis in Biomedical Texts via Vector Space Model Synonym Extraction

Accurate synonym extraction is a highly desired tool in computational semantics and natural language processing (NLP). Intrinsic to many current language models is the Vector Space Model (VSM) whereby linguistic properties are modeled by real vectors of small length compared to the document size. Here, several vector models are discussed. Utilizing the open source Word2Vec tool, temporal trends in biomedical abstracts from MEDLINE have been analyzed via a unique transformation matrix method. Additionally, accuracy of the word vector method and clustering of temporal trends are discussed.

59 BASIC BIOLOGICAL SCIENCES↗

Approximating Dose and Risk for Contaminants in Groundwater from the Underground Nuclear Test Areas of the Nevada National Security Site (NNSS)

As part of the Environmental Management Program at the Nevada National Security Site (NNSS), the Underground Test Area (UGTA) Activity investigates the potential impacts of radionuclides that were introduced into groundwater from the underground nuclear tests conducted near or below the NNSS water table between 1951 and 1992. Groundwater models are being used to simulate contaminant transport and forecast contaminant boundaries that encompass areas where the groundwater has a five percent or greater probability of containing contaminants above the Safe Drinking Water Act Maximum Contaminant Levels (SDWA MCLs) at any time during the next 1,000 years. Transport modeling conducted for the Frenchman Flat Corrective Action Unit (CAU) at the NNSS identified the beta/photon-emitting radionuclides tritium (3H), carbon-14 ( 14 C), chlorine-36 ( 36 Cl), technetium-99 ( 99 Tc), and iodine-129 ( 129 I) as having the greatest influence in defining the farthest extent of the modeled CAU contaminant boundary. These same radionuclides are assumed here as the contaminants of concern (COCs) for all underground nuclear tests at the NNSS because models are not yet complete for the other CAUs. Potential public exposure to the COCs will only occur and be of concern if the COCs migrate into the groundwater beneath public or private lands at levels that exceed either individual SDWA MCLs or dose and risk limits. Groundwater flow directions strongly suggest that any contaminant boundary predicted by contaminant fate and transport modeling to overlap public or private lands is more likely to occur to the west and/or southwest of the NNSS and the adjacent Nevada Test and Training Range (NTTR). Well-established, rural communities exist in these directions. Estimates of representative activity concentrations at the applicable SDWA MCL were developed for the five COCs. It is assumed that these COC concentrations may collectively occur at some public or private location in the future, but that situation does not exist today. These representative activity concentrations are evaluated with respect to conforming collectively to a modern annual committed effective dose (CED) and lifetime excess cancer morbidity risk for a hypothetical reasonably maximally exposed individual (RMEI). This approach goes beyond the SDWA MCL focus of the contaminant boundary because individual COC concentrations may comply with the SDWA MCL but not collectively meet the modern health-protection metrics and the SDWA language, especially if future modeling studies or monitoring activities show multiple radionuclides from different SDWA MCL categories to be COCs. For the drinking water exposure pathway alone, the annual committed effective dose (CED) for the RMEI from all five COCs that are collectively at estimated activity concentrations equal to their SDWA MCL is well below the U.S. Department of Energy health-protective CED limit of 100 millirem (mrem)/yr. This is consistent using both the NNSS unclassified, 1992 decay-corrected radionuclide atom inventory and the atom inventory based on radionuclides measured in groundwater obtained from the ALMENDRO cavity in 2009 to calculate the SDWA MCL activity concentrations in groundwater. The RMEI’s total lifetime excess cancer risk from the drinking water exposure pathway for both atom inventories is within the range of 1 × 10 -4 to ≤ 1 × 10 -6 , which is considered health protective according to modern SDWA MCL regulatory language. The biosphere exposure pathways are drinking water, garden produce, animal products, inadvertent soil ingestion, and indoor and outdoor air inhalation. The exposure parameters for communities west and southwest of the NNSS were developed when the Yucca Mountain high-level, nuclear-waste disposal facility was under consideration. For all biosphere exposure pathways, calculations of the annual CED and lifetime excess cancer morbidity risk for the RMEI revealed that: 1) The annual CED is well within health-protective guidance (<< 100 mrem CED /yr) for the COC activity concentrations at the SDWA MCL, regardless of the atom inventory used. 2) The calculated 70-year lifetime excess cancer morbidity risk (6 x 10 -5 ) is within the health-protective range when the five COC activity concentrations are derived using the NNSS 1992 atom inventory, but it is at the upper limit of the acceptable range (1 x 10 -4 ) using the ALMENDRO 2009 atom inventory. 3) Tritium (3H) is the principal COC for producing annual dose and lifetime excess cancer risk, regardless of the atom inventory used. 4) Overall, the drinking water ingestion pathway is the dominant exposure pathway contributing to the total annual CED and lifetime excess cancer risk, followed by eating locally grown produce and animal products. 5) When tritium completely decays (after about 100 years), the RMEI’s lifetime risk will fall well within the health-protective range (i.e., 1 × 10-4 to ≤ 1 × 10 -6 ) and 36Cl will then become the most important contributor to the RMEI’s total annual CED and lifetime excess cancer morbidity risk from eating local produce and animal products. In the event that radionuclide concentrations begin to approach SDWA MCLs, a reasonable risk-management strategy for keeping lifetime risk more in compliance with regulatory guidance would be to use local sources of groundwater that are below SDWA MCLs or to limit the consumption of local produce and animal products that have ingested COC-contaminated groundwater. The viability of the latter strategy increases where the annual CED due to 36Cl approaches that of 3H.The dose and risk values calculated here for an RMEI are specific to the assumption that the five COCs occur in groundwater beneath public or private lands at concentrations that are collectively at the SDWA MCL. Currently, these COCs are essentially absent from groundwater beneath public or private lands beyond the boundaries of the NNSS and NTTR other than at very low, naturally occurring concentrations. The analyses presented here can be readily applied to determine dose and risk for COC concentrations actually measured in future monitoring samples.

54 ENVIRONMENTAL SCIENCES↗

Plant model of KIPT neutron source facility simulator

Argonne National Laboratory (ANL) of the United States and Kharkov Institute of Physics and Technology (KIPT) of Ukraine are collaborating on constructing a neutron source facility at KIPT, Kharkov, Ukraine. The facility has 100-kW electron beam driving a subcritical assembly (SCA). The electron beam interacts with a natural uranium target or a tungsten target to generate neutrons, and deposits its power in the target zone. The total fission power generated in SCA is about 300 kW. Two primary cooling loops are designed to remove 100-kW and 300-kW from the target zone and the SCA, respectively. A secondary cooling system is coupled with the primary cooling system to dispose of the generated heat outside the facility buildings to the atmosphere. In addition, the electron accelerator has a low efficiency for generating the electron beam, which uses another secondary cooling loop to remove the generated heat from the accelerator primary cooling loop. One of the main functions the KIPT neutron source facility is to train young nuclear specialists; therefore, ANL has developed the KIPT Neutron Source Facility Simulator for this function. In this simulator, a Plant Control System and a Plant Protection System were developed to perform proper control and to provide automatic protection against unsafe and improper operation of the facility during the steady-state and the transient states using a facility plant model. This report focuses on describing the physics of the plant model and provides several test cases to demonstrate its capabilities. The plant facility model uses the PYTHON script language. It is consistent with the computer language of the plant control system. It is easy to integrate with the simulator without an additional interface, and it is able to simulate the transients of the cooling systems with system control variables changing on real-time.

43 PARTICLE ACCELERATORS↗

Diagramming Transactive Building Business Cases: Using Principles of e3 Value to Document Valuation Studies

Energy management in buildings is becoming more transactive. Pacific Northwest National Laboratory and the U.S. Department of Energy Building Technologies Office recently defined innovative use cases wherein market-like mechanisms are used to manage energy within buildings, between buildings, and between buildings and third-party entities, such as power utilities. A next step toward defining a set of transactive use cases in the buildings domain is to carefully diagram the corresponding business cases to capture details of transactions among all stakeholders and their economic value propositions. The principles of e3-value diagramming are applied in this report toward creating business value diagrams. These principles are extended to be consistent with Universal Modeling Language use-case diagrams. Example diagrams are presented for a subset of buildings-domain use cases that were introduced in an earlier Pacific Northwest National Laboratory report. The diagrams are intended to clearly represent an understanding of the transactions through which individual entities accumulate value in their respective use cases, and the diagrams should therefore support economic valuation studies. The report reviews some of the foundational principles of e3 value and includes authors’ insights concerning the formulation of these diagrams using Universal Modeling Language as a more systematic modeling approach.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Evaluation of the Single-precision Floatingpoint Vector Add Kernel Using the Intel FPGA SDK for OpenCL

Open Computing Language (OpenCL) is a high-level language that enables software programmers to explore Field Programmable Gate Arrays (FPGAs) for application acceleration. The Intel FPGA software development kit (SDK) for OpenCL allows a user to specify applications at a high level and explore the performance of low-level hardware acceleration. In this report, we present the FPGA performance and power consumption results of the single-precision floating-point vector add OpenCL kernel using the Intel FPGA SDK for OpenCL on the Nallatech 385A FPGA board. The board features an Arria 10 FPGA. We evaluate the FPGA implementations using the compute unit duplication and kernel vectorization optimization techniques. On the Nallatech 385A FPGA board, the maximum compute kernel bandwidth we achieve is 25.8 GB/s, approximately 76% of the peak memory bandwidth. The power consumption of the FPGA device when running the kernels ranges from 29W to 42W.

97 MATHEMATICS AND COMPUTING↗

Evaluation of the OpenCL AES Kernel using the Intel FPGA SDK for OpenCL

The OpenCL standard is an open programming model for accelerating algorithms on heterogeneous computing system. OpenCL extends the C-based programming language for developing portable codes on different platforms such as CPU, Graphics processing units (GPUs), Digital Signal Processors (DSPs) and Field Programmable Gate Arrays (FPGAs). The Intel FPGA SDK for OpenCL is a suite of tools that allows developers to abstract away the complex FPGA-based development flow for a high-level software development flow. Users can focus on the design of hardware-accelerated kernel functions in OpenCL and then direct the tools to generate the low-level FPGA implementations. The approach makes the FPGA-based development more accessible to software users as the needs for hybrid computing using CPUs and FPGAs are increasing. It can also significantly reduce the hardware development time as users can evaluate different ideas with high-level language without deep FPGA domain knowledge. In this report, we evaluate the performance of the kernel using the Intel FPGA SDK for OpenCL and Nallatech 385A FPGA board. Compared to the M506 module, the board provides more hardware resources for a larger design exploration space. The kernel performance is measured with the compute kernel throughput, an upper bound to the FPGA throughput. The report presents the experimental results in details. The Appendix lists the kernel source code.

97 MATHEMATICS AND COMPUTING↗