Search NASA⌕ Search

SEARCH · Search NASA

Results for “FPGAs”

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 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Optimization with the OpenACC-to-FPGA framework on the Arria 10 and Stratix 10 FPGAs

The reconfigurable computing paradigm with field programmable gate arrays (FPGAs) has received renewed interest in the high-performance computing field due to FPGAs’ unique combination of performance and energy efficiency. However, difficulties in programming and optimizing FPGAs have prevented them from being widely accepted as general-purpose computing devices. In accelerator-based heterogeneous computing, portability across diverse heterogeneous devices is also an important issue, but the unique architectural features in FPGAs make this difficult to achieve. To address these issues, a directive-based, high-level FPGA programming and optimization framework was previously developed. In this work, developed optimizations were combined holistically using the directive-based approach to show that each individual benchmark requires a unique set of optimizations to maximize performance. We perform this exploration on Intel Arria 10 and Stratix 10 FPGAs. We also explored the relationships between performance, resource usages, and compilation times, and investigated implications for performance portability. Finally, we present an initial evaluation of a real-world proxy application, LULESH.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2012

Flexible Architecture for FPGAs in Embedded Systems

Commonly, field-programmable gate arrays (FPGAs) being developed in cPCI embedded systems include the bus interface in the FPGA. This complicates the development because the interface is complicated and requires a lot of development time and FPGA resources. In addition, flight qualification requires a substantial amount of time be devoted to just this interface. Another complication of putting the cPCI interface into the FPGA being developed is that configuration information loaded into the device by the cPCI microprocessor is lost when a new bit file is loaded, requiring cumbersome operations to return the system to an operational state. Finally, SRAM-based FPGAs are typically programmed via specialized cables and software, with programming files being loaded either directly into the FPGA, or into PROM devices. This can be cumbersome when doing FPGA development in an embedded environment, and does not have an easy path to flight. Currently, FPGAs used in space applications are usually programmed via multiple space-qualified PROM devices that are physically large and require extra circuitry (typically including a separate one-time programmable FPGA) to enable them to be used for this application. This technology adds a cPCI interface device with a simple, flexible, high-performance backend interface supporting multiple backend FPGAs. It includes a mechanism for programming the FPGAs directly via the microprocessor in the embedded system, eliminating specialized hardware, software, and PROM devices and their associated circuitry. It has a direct path to flight, and no extra hardware and minimal software are required to support reprogramming in flight. The device added is currently a small FPGA, but an advantage of this technology is that the design of the device does not change, regardless of the application in which it is being used. This means that it needs to be qualified for flight only once, and is suitable for one-time programmable devices or an application specific integrated circuit (ASIC). An application programming interface (API) further reduces the development time needed to use the interface device in a system.

Clark, Duane I.↗
NASA NTRS1999

Start Up Application Concerns with Field Programmable Gate Arrays (FPGAs)

This note is being published to improve the visibility of this subject, as we continue to see problems surface in designs, as well as to add additional information to the previously published note for design engineers. The original application note focused on designing systems with no single point failures using Actel Field Programmable Gate Arrays (FPGAs) for critical applications. Included in that note were the basic principles of operation of the Actel FPGA and a discussion of potential single-point failures. The note also discussed the issue of startup transients for that class of device. It is unfortunate that we continue to see some design problems using these devices. This note will focus on the startup properties of certain electronic components, in general, and current Actel FPGAs, in particular. Devices that are "power-on friendly" are currently being developed by Actel, as a variant of the new SX series of FPGAs. In the ideal world, electronic components would behave much differently than they do in the real world, The chain, of course, starts with the power supply. Ideally, the voltage will immediately rise to a stable V(sub cc) level, of course, it does not. Aside from practical design considerations, inrush current limits of certain capacitors must be observed and the power supply's output may be intentionally slew rate limited to prevent a large current spike on the system power bus. In any event, power supply rise time may range from less than I msec to 100 msec or more.

Katz, Richard B.↗
DOE OSTI2020

FPGAs-as-a-Service Toolkit (FaaST)

Computing needs for high energy physics are already intensive and are expected to increase drastically in the coming years. In this context, heterogeneous computing, specifically as-a-service computing, has the potential for significant gains over traditional computing models. Although previous studies and packages in the field of heterogeneous computing have focused on GPUs as accelerators, FPGAs are an extremely promising option as well. A series of workflows are developed to establish the performance capabilities of FPGAs as a service. Multiple different devices and a range of algorithms for use in high energy physics are studied. For a small, dense network, the throughput can be improved by an order of magnitude with respect to GPUs as a service. For large convolutional networks, the throughput is found to be comparable to GPUs as a service. This work represents the first open-source FPGAs-as-a-service toolkit.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2021

A Survey: Handling Irregularities in Neural Network Acceleration with FPGAs

In the last decade, Artificial Intelligence (AI) through Deep Neural Networks (DNNs) has penetrated virtually every aspect of science, technology, and business. Many types of DNNs have been and continue to be developed, including Convolutional Neural Networks (CNNs), Recurrent Neural Net- works (RNNs), and Graph Neural Networks (GNNs). The overall problem for all of these Neural Networks (NNs) is that their target applications generally pose stringent constraints on latency and throughput, while also having strict accuracy requirements. There have been many previous efforts in creating hardware to accelerate NNs. The problem designers face is that optimal NN models typically have significant irregularities, making them hardware-unfriendly. In this paper, we first define the problems in NN acceleration by characterizing common irregularities in NN processing into 4 types; then we summarize the existing works that handle the four types of irregularities efficiently using hardware, especially FPGAs; finally, we provide a new vision of next-generation FPGA-based NN acceleration: that the emerging heterogeneity in the next-generation FPGAs is the key to achieving higher performance.

Geng, Tong↗
DOE OSTI2021

Toward Evaluating High-Level Synthesis Portability and Performance between Intel and Xilinx FPGAs

Offloading computation from a CPU to a hardware accelerator is becoming a more common solution for improving performance because traditional gains enabled by Moore’s law and Dennard scaling have slowed. GPUs are often used as hardware accelerators, but field-programmable gate arrays (FPGAs) are gaining traction. FPGAs are beneficial because they allow hardware specific to a particular application to be created. However, they are notoriously difficult to program. To this end, two of the main FPGA manufacturers, Intel and Xilinx, have created tools and frameworks that enable the use of higher level languages to design FPGA hardware. Although Xilinx kernels can be designed by using C/C++, both Intel and Xilinx support the use of OpenCL C to architect FPGA hardware. However, not much is known about the portability and performance between these two device families other than the fact that it is theoretically possible to synthesize a kernel meant for Intel to Xilinx and vice versa.In this work, we evaluate the portability and performance of Intel and Xilinx kernels. We use OpenCL C implementations of a subset of the Rodinia benchmarking suite that were designed for an Intel FPGA and make the necessary modifications to create synthesizable OpenCL C kernels for a Xilinx FPGA. We find that the difficulty of porting certain kernel optimizations varies, depending on the construct. Once the minimum amount of modifications is made to create synthesizable hardware for the Xilinx platform, more nontrivial work is needed to improve performance. However, we find that constructs that are known to be performant for an FPGA should improve performance regardless of the platform; the difficulty comes in deciding how to invoke certain kernel optimizations while also abiding by the constraints enforced by a given platform’s hardware compiler.

Cabrera, Anthony↗
DOE OSTI2022

Graph Neural Networks for Charged Particle Tracking on FPGAs

The determination of charged particle trajectories in collisions at the CERN Large Hadron Collider (LHC) is an important but challenging problem, especially in the high interaction density conditions expected during the future high-luminosity phase of the LHC (HL-LHC). Graph neural networks (GNNs) are a type of geometric deep learning algorithm that has successfully been applied to this task by embedding tracker data as a graph—nodes represent hits, while edges represent possible track segments—and classifying the edges as true or fake track segments. However, their study in hardware- or software-based trigger applications has been limited due to their large computational cost. In this paper, we introduce an automated translation workflow, integrated into a broader tool called hls4ml , for converting GNNs into firmware for field-programmable gate arrays (FPGAs). We use this translation tool to implement GNNs for charged particle tracking, trained using the TrackML challenge dataset, on FPGAs with designs targeting different graph sizes, task complexites, and latency/throughput requirements. This work could enable the inclusion of charged particle tracking GNNs at the trigger level for HL-LHC experiments.

Elabd, Abdelrahman↗
NASA NTRS2004

Tradeoffs in Flight Design Upset Mitigation in State of the Art FPGAs: Hardened by Design vs. Design Level Hardening

This presentation compares and contrasts the effectiveness and the system/designer impacts of the two main approaches to upset hardening: the Actel approach (RTSX-S and RTAX-S) of low-level (inside each flip-flop) triplication and the Xilinx approach (Virtex and Virtex2) of design-level triplication of both functional blocks and voters. The effectiveness of these approaches is compared using measurements made in conjunction with each of the FPGAs' manufacturer: for Actel, published data [1] and for Xilinx, recent results from the Xilinx SEE Test Consortium (note that the author is an active and founding member). The impacts involve Actel advantages in the areas of transistor-utilization efficiency and minimizing designer involvement in the triplication while the Xilinx advantages relate to the ability to custom tailor upset hardness and the flexibility of re-configurability. Additionally, there are currently clear Xilinx advantages in available features such as the number of I/O's, logic cells, and RAM blocks as well as speed. However, the advantage of the Actel anti-fuses for configuration over the Xilinx SRAM cells is that the latter need additional functionality and external circuitry (PROMs and, at least a watchdog timer) for configuration and configuration scrubbing. Further, although effectively mitigated if done correctly, the proton upset-ability of the Xilinx FPGAs is a concern in severe proton-rich environments. Ultimately, both manufacturers' upset hardening is limited by SEFI (single-event functional interrupt) rates where it appears the Actel results are better although the Xilinx Virtex2-family result of about one SEFI in 65 device-years in solar-min GCR (the more intense part of the galactic cosmic-ray background) should be acceptable to most missions

radiation effects↗
NASA NTRS2008

Assessing and mitigating radiation effects in Xilinx FPGAs

The main purpose of this guideline is to document the current understanding of the effort and tradeoffs involved in using SRAM-based FPGAs with maximum achievable upset tolerance. This guideline summarizes the different testing strategies to be considered when assessing the SEE tolerance of SRAM-based FPGAs. For critical applications, in-beam testing of flight designs is strongly recommended in order to verify that the upset mitigation is working as well as the designer intended. At present, there are four manufacturers (Actel, Xilinx, Atmel, and Aeroflex) offering six FPGA types to the military market (i.e., some military spec screening and testing are specified) and four device types to the aerospace application market, which indicates that the manufacturers have added radiation parameter specifications. There is also a Honeywell and BAE Systems device that has been under development for several years, but it does not appear to be available yet; inquiries to Honeywell’s military/aerospace marketing arm on this product have been unanswered.

Allen, Greg↗
NASA NTRS2005

Qualification Strategies of Field Programmable Gate Arrays (FPGAs) for Space Application

This viewgraph document reviews the issue of using Field Programmable Gate Arrays (FPGAs) in Space Application, and the some of the strategies for qualifying the FPGA. Qualification and risk management of such complex systems requires new approaches. The paper presents a matrix approach to qualification has been presented that: - Complements historical specifications - Highlights the importance of device physics as a cornerstone to qualification. - Provides levels of risk management that expressly document trade offs. - Stresses the role of the FPGA vendor as team member in the development of modern spacecraft.

Mil Standard 883↗
DOE OSTI2020

Fast inference of Boosted Decision Trees in FPGAs for particle physics

We describe the implementation of Boosted Decision Trees in the hls4ml library, which allows the translation of a trained model into FPGA firmware through an automated conversion process. Thanks to its fully on-chip implementation, hls4ml performs inference of Boosted Decision Tree models with extremely low latency. With a typical latency less than 100 ns, this solution is suitable for FPGA-based real-time processing, such as in the Level-1 Trigger system of a collider experiment. These developments open up prospects for physicists to deploy BDTs in FPGAs for identifying the origin of jets, better reconstructing the energies of muons, and enabling better selection of rare signal processes.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2021

Fast convolutional neural networks on FPGAs with hls4ml

We introduce an automated tool for deploying ultra low-latency, low-power deep neural networks with convolutional layers on field-programmable gate arrays (FPGAs). By extending the hls4ml library, we demonstrate an inference latency of 5 µs using convolutional architectures, targeting microsecond latency applications like those at the CERN Large Hadron Collider. Considering benchmark models trained on the Street View House Numbers Dataset, we demonstrate various methods for model compression in order to fit the computational constraints of a typical FPGA device used in trigger and data acquisition systems of particle detectors. In particular, we discuss pruning and quantization-aware training, and demonstrate how resource utilization can be significantly reduced with little to no loss in model accuracy. We show that the FPGA critical resource consumption can be reduced by 97% with zero loss in model accuracy, and by 99% when tolerating a 6% accuracy degradation.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2023

Ultra-low latency recurrent neural network inference on FPGAs for physics applications with hls4ml

Abstract Recurrent neural networks have been shown to be effective architectures for many tasks in high energy physics, and thus have been widely adopted. Their use in low-latency environments has, however, been limited as a result of the difficulties of implementing recurrent architectures on field-programmable gate arrays (FPGAs). In this paper we present an implementation of two types of recurrent neural network layers—long short-term memory and gated recurrent unit—within the hls4ml framework. We demonstrate that our implementation is capable of producing effective designs for both small and large models, and can be customized to meet specific design requirements for inference latencies and FPGA resources. We show the performance and synthesized designs for multiple neural networks, many of which are trained specifically for jet identification tasks at the CERN Large Hadron Collider.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Fast jet tagging with MLP-Mixers on FPGAs

We explore the innovative use of MLP-Mixer models for real-time jet tagging and establish their feasibility on resource-constrained hardware like FPGAs. MLP-Mixers excel in processing sequences of jet constituents, achieving state-of-the-art performance on datasets mimicking Large Hadron Collider conditions. By using advanced optimization techniques such as High-Granularity Quantization and Distributed Arithmetic, we achieve unprecedented efficiency. These models match or surpass the accuracy of previous architectures, reduce hardware resource usage by up to 97%, double the throughput, and half the latency. Additionally, non-permutation-invariant architectures enable smart feature prioritization and efficient FPGA deployment, setting a new benchmark for machine learning in real-time data processing at particle colliders.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

Deploying Simulations to FPGAs for HIL-Style Tester Development

Hardware-in-the-Loop (HIL) methodologies for test system development often require simulations capable of running at MHz speeds on FPGAs. The stringent memory and speed constraints necessitate compromises between model fidelity and execution speed. Numerically solving the underlying governing equations represents the highest accuracy, but slowest responding approach. By storing pre-determined results in look-up-tables (LUT), one can balance speed and accuracy.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

HGQ: High Granularity Quantization for Real-time Neural Networks on FPGAs

Neural networks with sub-microsecond inference latency are required by many critical applications. Targeting such applications deployed on FPGAs, we present High Granularity Quantization (HGQ), a quantization-aware training framework that optimizes parameter bit-widths through gradient descent. Unlike conventional methods, HGQ determines the optimal bit-width for each parameter independently, making it suitable for hardware platforms supporting heterogeneous arbitrary precision arithmetic. In our experiments, HGQ shows superior performance compared to existing network compression methods, achieving orders of magnitude reduction in resource consumption and latency while maintaining the accuracy on several benchmark tasks. These improvements enable the deployment of complex models previously infeasible due to resource or latency constraints. HGQ is open-source and is used for developing next-generation trigger systems at the CERN ATLAS and CMS experiments for particle physics, enabling the use of advanced machine learning models for real-time data selection with sub-microsecond latency.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
NASA NTRS2003

A Genetic Representation for Evolutionary Fault Recovery in Virtex FPGAs

Most evolutionary approaches to fault recovery in FPGAs focus on evolving alternative logic configurations as opposed to evolving the intra-cell routing. Since the majority of transistors in a typical FPGA are dedicated to interconnect, nearly 80% according to one estimate, evolutionary fault-recovery systems should benefit hy accommodating routing. In this paper, we propose an evolutionary fault-recovery system employing a genetic representation that takes into account both logic and routing configurations. Experiments were run using a software model of the Xilinx Virtex FPGA. We report that using four Virtex combinational logic blocks, we were able to evolve a 100% accurate quadrature decoder finite state machine in the presence of a stuck-at-zero fault.

Lohn, Jason↗