Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Simulation of distributed computing systems”

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 1,405 records · Page 78
All sourcesNASA NTRSDOE OSTI
NASA NTRSDate not supplied

Tools for Performing SBG hyperspectral Observing System Simulation Experiment

One of NASA’s Decadal Survey mission, Surface Biology and Geology (SBG), will include a hyperspectral remote sensing imager, which has a very high spatial resolution and a wide spectral coverage (from UV to Near IR). Unprecedented large data volumes will be generated by the SBG hyperspectral instrument. Before the launch of the new satellite, an Observing System Simulation Experiment (OSSE) can be used to study different designs of the new satellite system. One of the key components in an OSSE study is a radiative transfer model (RTM) or forward model. In this presentation, we will describe a Principal Component-based Radiative Transfer Model (PCRTM) which is capable of simulating atmospheric (TOA) radiance or reflectance spectra from far IR to visible and UV spectral regions (50 wavenumber to 30000 wavenumber) quickly and accurately. Multiple scattering from multiple layers of clouds/aerosols are included in the model. The PCRTM has a very good accuracy relative to reference line-by-line radiative transfer models (LBLRTM), and it saves 3-4 orders of magnitude computational time relative to LBLRTM or MODTRAN. The PCRTM model has been successfully used to analyze large volumes of data from hyperspectral sensors such as AIRS, CrIS, and IASI. It has also been used to perform OSSE studies for the Climate Absolute Radiance and Refractivity Observatory (CLARREO) mission. Another useful tool for the OSSE is surface Bidirectional Reflectance Distribution Function (BRDF) database. It is very crucial for the SBG OSSE to include realistic BRDF spectra. Currently, most of the surface reflectance spectra such as those in the ECOSIS and ECOSTRESS are measured at specific observation geometries. We have developed a hyperspectral bidirectional reflectance (HSBR) model which combines Ross-Li BRDF model with the existing reflectance spectral libraries using a principal component analysis. This HSBR model can provide realistic BRDF spectra under various observation conditions. It can also be used to generate realistic BRDF spectra using measurements from multi-band imagers or spectrometers such as MODIS or VIIRS.

Xu Liu↗
NASA NTRS1994

Numerical Simulations of High-Speed Chemically Reacting Flow

The Essentially NonOscillatory (ENO) shock-capturing scheme for the solution of hyperbolic equations is extended to solve a system of coupled conservation equations governing two-dimensional, time-dependent, compressible chemically reacting flow with full chemistry. The thermodynamic properties of the mixture are modeled accurately, and stiff kinetic terms are separated from the fluid motion by a fractional step algorithm. The methodology is used to study the concept of shock-induced mixing and combustion, a process by which the interaction of a shock wave with a jet of low-density hydrogen fuel enhances mixing through streamwise vorticity generation. Test cases with and without chemical reaction are explored here. Our results indicate that, in the temperature range examined, vorticity generation as well as the distribution of atomic species do not change significantly with the introduction of a chemical reaction and subsequent heat release. The actual diffusion of hydrogen is also relatively unaffected by the reaction process. This suggests that the fluid mechanics of this problem may be successfully decoupled from the combustion processes, and that computation of the mixing problem (without combustion chemistry) can elucidate much of the important physical features of the flow.

Ton, V. T.↗
NASA NTRS1994

Numerical Simulations of High-Speed Chemically Reacting Flow

The essentially nonoscillatory (ENO) shock-capturing scheme for the solution of hyperbolic equations is extended to solve a system of coupled conservation equations governing two-dimensional, time-dependent, compressible chemically reacting flow with full chemistry. The thermodynamic properties of the mixture are modeled accurately, and stiff kinetic terms are separated from the fluid motion by a fractional step algorithm. The methodology is used to study the concept of shock-induced mixing and combustion, a process by which the interaction of a shock wave with a jet of low-density hydrogen fuel enhances mixing through streamwise vorticity generation. Test cases with and without chemical reaction are explored here. Our results indicate that, in the temperature range examined, vorticity generation as well as the distribution of atomic species do not change significantly with the introduction of a chemical reaction and subsequent heat release. The actual diffusion of hydrogen is also relatively unaffected by the reaction process. This suggests that the fluid mechanics of this problem may be successfully decoupled from the combustion processes, and that computation of the mixing problem (without combustion chemistry) can elucidate much of the important physical features of the flow.

Ton, V. T.↗
DOE OSTI2023

Reinforcement Learning for Load-balanced Parallel Particle Tracing

We explore an online reinforcement learning (RL) paradigm to dynamically optimize parallel particle tracing performance in distributed-memory systems. Our method combines three novel components: (1) a work donation algorithm, (2) a high-order workload estimation model, and (3) a communication cost model. First, we design an RL-based work donation algorithm. Our algorithm monitors workloads of processes and creates RL agents to donate data blocks and particles from high-workload processes to low-workload processes to minimize program execution time. The agents learn the donation strategy on the fly based on reward and cost functions designed to consider processes' workload changes and data transfer costs of donation actions. Second, we propose a workload estimation model, helping RL agents estimate the workload distribution of processes in future computations. Third, we design a communication cost model that considers both block and particle data exchange costs, helping RL agents make effective decisions with minimized communication costs. We demonstrate that our algorithm adapts to different flow behaviors in large-scale fluid dynamics, ocean, and weather simulation data. Our algorithm improves parallel particle tracing performance in terms of parallel efficiency, load balance, and costs of I/O and communication for evaluations with up to 16,384 processors.

Distributed and parallel particle tracing↗
DOE OSTI2024

Addressing GPU memory limitations for Graph Neural Networks in High-Energy Physics applications

Introduction Reconstructing low-level particle tracks in neutrino physics can address some of the most fundamental questions about the universe. However, processing petabytes of raw data using deep learning techniques poses a challenging problem in the field of High Energy Physics (HEP). In the Exa.TrkX Project, an illustrative HEP application, preprocessed simulation data is fed into a state-of-art Graph Neural Network (GNN) model, accelerated by GPUs. However, limited GPU memory often leads to Out-of-Memory (OOM) exceptions during training, due to the large size of models and datasets. This problem is exacerbated when deploying models on High-Performance Computing (HPC) systems designed for large-scale applications. Methods We observe a high workload imbalance issue during GNN model training caused by the irregular sizes of input graph samples in HEP datasets, contributing to OOM exceptions. We aim to scale GNNs on HPC systems, by prioritizing workload balance in graph inputs while maintaining model accuracy. Our paper introduces diverse balancing strategies aimed at decreasing the maximum GPU memory footprint and avoiding the OOM exception, across various datasets. Results Our experiments showcase memory reduction of up to 32.14% compared to the baseline. We also demonstrate the proposed strategies can avoid OOM in application. Additionally, we create a distributed multi-GPU implementation using these samplers to demonstrate the scalability of these techniques on the HEP dataset. Discussion By assessing the performance of these strategies as data loading samplers across multiple datasets, we can gauge their effectiveness in both single-GPU and distributed environments. Our experiments, conducted on datasets of varying sizes and across multiple GPUs, broaden the applicability of our work to various GNN applications that handle input datasets with irregular graph sizes.

Lee, Claire Songhyun↗
NASA NTRS1998

Integrating Cache Performance Modeling and Tuning Support in Parallelization Tools

With the resurgence of distributed shared memory (DSM) systems based on cache-coherent Non Uniform Memory Access (ccNUMA) architectures and increasing disparity between memory and processors speeds, data locality overheads are becoming the greatest bottlenecks in the way of realizing potential high performance of these systems. While parallelization tools and compilers facilitate the users in porting their sequential applications to a DSM system, a lot of time and effort is needed to tune the memory performance of these applications to achieve reasonable speedup. In this paper, we show that integrating cache performance modeling and tuning support within a parallelization environment can alleviate this problem. The Cache Performance Modeling and Prediction Tool (CPMP), employs trace-driven simulation techniques without the overhead of generating and managing detailed address traces. CPMP predicts the cache performance impact of source code level "what-if" modifications in a program to assist a user in the tuning process. CPMP is built on top of a customized version of the Computer Aided Parallelization Tools (CAPTools) environment. Finally, we demonstrate how CPMP can be applied to tune a real Computational Fluid Dynamics (CFD) application.

Waheed, Abdul↗
NASA NTRS2012

The Gas Distribution in the Outer Regions of Galaxy Clusters

Aims. We present our analysis of a local (z = 0.04 - 0.2) sample of 31 galaxy clusters with the aim of measuring the density of the X-ray emitting gas in cluster outskirts. We compare our results with numerical simulations to set constraints on the azimuthal symmetry and gas clumping in the outer regions of galaxy clusters. Methods. We have exploited the large field-of-view and low instrumental background of ROSAT/PSPC to trace the density of the intracluster gas out to the virial radius, We stacked the density profiles to detect a signal beyond T200 and measured the typical density and scatter in cluster outskirts. We also computed the azimuthal scatter of the profiles with respect to the mean value to look for deviations from spherical symmetry. Finally, we compared our average density and scatter profiles with the results of numerical simulations. Results. As opposed to some recent Suzaku results, and confirming previous evidence from ROSAT and Chandra, we observe a steepening of the density profiles beyond approximately r(sub 500). Comparing our density profiles with simulations, we find that non-radiative runs predict density profiles that are too steep, whereas runs including additional physics and/ or treating gas clumping agree better with the observed gas distribution. We report high-confidence detection of a systematic difference between cool-core and non cool-core clusters beyond approximately 0.3r(sub 200), which we explain by a different distribution of the gas in the two classes. Beyond approximately r(sub 500), galaxy clusters deviate significantly from spherical symmetry, with only small differences between relaxed and disturbed systems. We find good agreement between the observed and predicted scatter profiles, but only when the 1% densest clumps are filtered out in the ENZO simulations. Conclusions. Comparing our results with numerical simulations, we find that non-radiative simulations fail to reproduce the gas distribution, even well outside cluster cores. Although their general behavior agrees more closely with the observations, simulations including cooling and star formation convert a large amount of gas into stars, which results in a low gas fraction with respect to the observations. Consequently, a detailed treatment of gas cooling, star formation, AGN feedback, and consideration of gas clumping is required to construct realistic models of the outer regions of clusters.

Eckert, D.↗
DOE OSTI2021

Model Membrane Systems Used to Study Plasma Membrane Lipid Asymmetry

It is well known that the lipid distribution in the bilayer leaflets of mammalian plasma membranes (PMs) is not symmetric. Despite this, model membrane studies have largely relied on chemically symmetric model membranes for the study of lipid–lipid and lipid–protein interactions. This is primarily due to the difficulty in preparing stable, asymmetric model membranes that are amenable to biophysical studies. However, in the last 20 years, efforts have been made in producing more biologically faithful model membranes. Here, we review several recently developed experimental and computational techniques for the robust generation of asymmetric model membranes and highlight a new and particularly promising technique to study membrane asymmetry.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Developing Data-Driven Synthetic Infrastructure Models for Resilience Analysis

Research on infrastructure resilience has produced promising methods to simulate and optimize complex networks to improve performance. However, restrictions on sharing infrastructure models and the steep cost of developing and maintaining infrastructure models presents a roadblock to adoption. To overcome this limitation, this research focuses on methods to create data-driven infrastructure models that will help improve infrastructure resilience and security. The analysis couples incomplete utility data, geospatial data, machine learning, and synthetic network generation methods to rapidly develop and update infrastructure models. The methods are validated using realistic utility models and site-specific data, with a focus on Puerto Rico due to its unique infrastructure challenges and available data. This research highlights promising opportunities for the use of synthetic network generation and machine learning to create infrastructure models when very little data is available. Results demonstrate that hybrid methods, which combine sparse utility data with synthetic models, can enhance model accuracy, and machine learning can predict model attributes using training data from other models. However, the complexity of infrastructure systems means that even minor changes in network connectivity can significantly impact simulation results. Resilience analysis using synthetic infrastructure models shows that while some system behaviors are preserved, the magnitude of disruptions may not be accurately represented, indicating the need for more research and validation before using synthetic models for critical infrastructure investment decisions. The framework outlined in this report represents a significant advance to infrastructure model development and could be applied to additional domains and sites. Future research will continue to streamline and validate methods to help reduce roadblocks to resilience analysis.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2022

Capacitance of Carbon Nanotube/Graphene Composite Electrodes with [BMIM + ][BF$_4^–$]/Acetonitrile: Fixed Voltage Molecular Dynamics Simulations

The capacitance of nanoporous carbon electrode materials is dictated by the physical interactions with electrolyte ions and molecules at the accessible interior electrode surface. While significant progress has been made in designing and synthesizing carbon materials with well-defined and relatively homogeneous nanoporosity, the majority of materials remain heterogeneous, and the resulting properties reflect an average over this distribution. In this regard, computer simulations can be a valuable tool to predict or validate structure/property relationships by systematically investigating well-defined, model electrode morphologies. In this work, we utilize fixed-voltage molecular dynamics simulations to predict structure/capacitance relationships for five different model morphologies of carbon nanotube/graphene (CNT/G) composite electrodes with 1-butyl-3-methylimidazolium tetrafluoroborate/acetonitrile electrolyte. The CNT/G electrode models are inspired by experimental “layer-by-layer” syntheses and strike a balance between realism and computational tractability. Comparison between different model electrode architectures elucidates important structure/property relationships. We find that CNT/graphene contact points serve as “hot spots” with significantly enhanced charge separation relative to the rest of the electrode. Furthermore, we demonstrate a specific nanoconfinement motif that provides substantial 3–4$\times$ enhancement of local capacitance, resulting in a ~40% increase of the total electrode differential capacitance. Because the accessible surface area of the model CNT/G electrodes is precisely determined, a comparison of per-area capacitance across systems is unambiguous. Our results thus complement a prior computational demonstration of capacitance enhancement in nanoconfinement while elucidating additional interaction motifs at CNT/G electrochemical interfaces.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
NASA NTRS1996

Concurrent Probabilistic Simulation of High Temperature Composite Structural Response

A computational structural/material analysis and design tool which would meet industry's future demand for expedience and reduced cost is presented. This unique software 'GENOA' is dedicated to parallel and high speed analysis to perform probabilistic evaluation of high temperature composite response of aerospace systems. The development is based on detailed integration and modification of diverse fields of specialized analysis techniques and mathematical models to combine their latest innovative capabilities into a commercially viable software package. The technique is specifically designed to exploit the availability of processors to perform computationally intense probabilistic analysis assessing uncertainties in structural reliability analysis and composite micromechanics. The primary objectives which were achieved in performing the development were: (1) Utilization of the power of parallel processing and static/dynamic load balancing optimization to make the complex simulation of structure, material and processing of high temperature composite affordable; (2) Computational integration and synchronization of probabilistic mathematics, structural/material mechanics and parallel computing; (3) Implementation of an innovative multi-level domain decomposition technique to identify the inherent parallelism, and increasing convergence rates through high- and low-level processor assignment; (4) Creating the framework for Portable Paralleled architecture for the machine independent Multi Instruction Multi Data, (MIMD), Single Instruction Multi Data (SIMD), hybrid and distributed workstation type of computers; and (5) Market evaluation. The results of Phase-2 effort provides a good basis for continuation and warrants Phase-3 government, and industry partnership.

Abdi, Frank↗
DOE OSTI2021

Distribution of Infiltration in the 216-U-10, 216-B-3 Pond, and 216-T-4 Pond Systems 1944-1997

A modeling system to simulate fate and transport of contaminant plumes in groundwater beneath the Central Plateau to support the Cumulative Impact Evaluation (CIE) and Composite Analysis (CA) is being developed. The CIE will evaluate the cumulative effect of multiple Central Plateau radiological and chemical sources on groundwater quality. This will enable cleanup decisions to be evaluated in the context of all Central Plateau waste sites and existing groundwater conditions. The purpose of this Environmental Calculation File (ECF) is to partition the infiltration of wastewater and contaminant inventory between the main pond and each expansion lobe of the 216-B-3 Pond system, and to partition the infiltration of wastewater and contaminant inventory between the ditches and ponds of the 216-U-10 Pond and 216-T-4 Pond systems.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2023

AI-Based Faster-Than-Real-Time Stability Assessment of Large Power Systems with Applications on WECC System

Achieving clean energy goals will require significant advances in regard to addressing the computational needs for next-generation renewable-dominated power grids. One critical obstacle that lies in the way of transitioning today’s power grid to a renewable-dominated power grid is the lack of a faster-than-real-time stability assessment technology for operating a fast-changing power grid. This paper proposes an artificial intelligence (AI) -based method that predicts the system’s stability margin information (e.g., the frequency nadir in the frequency stability assessment and the critical clearing time (CCT) value in the transient stability assessment) directly from the system operating conditions without performing the conventional time-consuming time-domain simulations over detailed dynamic models. Since the AI method shifts the majority of the computational burden to offline training, the online evaluation is extremely fast. This paper has tested the AI-based stability assessment method using multiple dispatch cases that are converted and tuned from actual dispatch cases of the Western Electricity Coordinating Council (WECC) system model with more than 20,000 buses. The results show that the AI-based method could accurately predict the stability margin of such a large power system in less than 0.2 milliseconds using the offline-trained AI agent. Therefore, the proposed method has great potential to achieve faster-than-real-time stability assessment for practical large power systems while preserving sufficient accuracy.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2018

Advanced Photovoltaic Inverter Control Development and Validation in a Controller-Hardware-in-the-Loop Test Bed

Penetration of solar photovoltaic (PV) installations into the grid has increased in recent years. In the past, photovoltaic installations have not included grid-support functionality. Grid support functions including Volt-VAR (VVAR), Frequency-Watt (FW), Volt-Watt (VW), frequency/voltage-ride through, as well as other inverter functions such as anti-islanding are recommended by standards such as IEEE 1547 to allow PV inverters to contribute to acceptable operation and stability of distribution systems with high penetrations of PV and to ensure predictable response of the PV inverters. First, low-level and high-level grid support inverter control functions were developed and tested in simulation. The developed controller is able to integrate advanced controller functions in a coordinated approach, avoiding any conflict between the different control objectives. The algorithms were then programmed on an off-the-shelf embedded controller that included a dual-core CPU and FPGA. The details of the high-level grid support function and the lower level switching control will be presented. Second, this programmed controller is tested using a controller-hardware-in-the-loop (CHIL) test bed setup using an FPGA-based real-time simulator. The CHIL setup has the capability to run at a 500 nano-second time-step to accommodate the 20 kHz switching frequency of the developed controller. The advanced control function and CHIL test bed details provided here will aide future researchers to design, implement, and test advanced inverter functions for PV inverters.

14 SOLAR ENERGY↗
DOE OSTI2021

Optimal Demand Response Incorporating Distribution LMP with PV Generation Uncertainty

The utilization of aggregated demand-side flexibility via demand response (DR) has become a promising pathway for the integration of renewable energy resources in power systems. Nowadays, there are several management strategies for DR such as the price-based transactive control strategies. However, many of such existing price-based control strategies neglect the physics and operational constraints of the underlying distribution networks when computing the price, raising concerns regarding their theoretical and practical values. This paper studies this issue and investigates optimal DR (ODR) by incorporating the distribution locational marginal price (DLMP). In particular, we discuss DR in connection with DLMPs and propose a multiperiod bilevel optimization problem to find the ODR strategy. Here, the objective is to minimize the peak load, load fluctuation, and payments of load aggregators. In addition, a robust bilevel ODR model is formulated to provide a robust ODR strategy while minimizing operating costs under the worst-case realization of uncertainties; this mitigates the impact of forecasting errors on renewable energy resources. Then, we propose an efficient solution approach by employing the Karush-Kuhn-Tucker conditions and strong duality. Simulation results are presented to illustrate the mutual impacts of the interaction between DR and DLMP and the benefits of the robust ODR strategy.

24 POWER TRANSMISSION AND DISTRIBUTION↗
NASA NTRS2001

Tropical Ocean Evaporation/SST Sensitivity and It's Link to Water and Energy Budget Variations During ENSO

The continuing debate over feedback mechanisms governing tropical sea surface temperatures (SSTs) and tropical climate in general has highlighted the diversity of potential checks and balances within the climate system. Competing feedbacks due to changes in surface evaporation, water vapor, and cloud long- and shortwave radiative properties each may serve critical roles in stabilizing or destabilizing the climate system. It is also intriguing that even those climate variations having origins internal to the climate system - changes in ocean heat transport for example, apparently require complementary equilibrating effects by changes in atmospheric energy fluxes. Perhaps the best observational evidence of this is the relatively invariant nature of tropically averaged net radiation exiting the top-of-atmosphere (TOA) as measured by broadband satellite sensors over the past two decades. Thus, analyzing how these feedback mechanisms are operating within the context of current interannual variability may offer considerable insight for anticipating future climate change. In this paper we focus primarily on interannual variations of ocean evaporative fluxes and their significance for coupled water and energy cycles within the tropical climate system. In particular, we use both the da Silva estimates of surface fluxes (based on the Comprehensive Ocean Atmosphere Data Set, COADS) and numerical simulations from several global climate models to examine evaporation sensitivity to perturbations in SST associated with warm and cold ENSO events. The specific questions we address are as follows: (1) What recurring patterns of surface wind and humidity anomalies are present during ENSO and how do they combine to yield systematic evaporation anomalies?, (2) What is the resulting tropical ocean mean evaporation-SST sensitivity associated with this climate perturbation?, and (3) What role does this evaporation play in tropical heat and water balance over tropical oceanic regions? We use the da Silva ocean flux data to identify composite structure of departures of latent heat flux from climatology. We also show how these patterns arise out of associated wind and humidity anomaly distributions. Our preliminary work shows that evaporation sensitivity estimates from the da Silva / COADS data, computed for the tropical oceans (30 degrees N/S) are in the neighborhood of 5 to 6 W/square m K. Model estimates are also quite close to this figure. This rate is only slightly less than a rate corresponding to constant relative humidity; however, substantial regional departures from constant relative humidity are present. These patterns are robust and we relate the associated wind and humidity fluctuations noted in previous investigations to the derived evaporation anomalies. Finally, these results are interpreted with other data from the Earth radiation Budget Experiment (ERBE), Global Precipitation Climatology Project (GPCP) and NASA's Surface Radiation Budget (SRB) data set to characterize the tropical energetics of ENSO-related climate variability.

Robertson, Franklin R.↗
NASA NTRS2012

The Gas Distribution in Galaxy Cluster Outer Regions

Aims. We present the analysis of a local (z = 0.04 - 0.2) sample of 31 galaxy clusters with the aim of measuring the density of the X-ray emitting gas in cluster outskirts. We compare our results with numerical simulations to set constraints on the azimuthal symmetry and gas clumping in the outer regions of galaxy clusters. Methods. We exploit the large field-of-view and low instrumental background of ROSAT/PSPC to trace the density of the intracluster gas out to the virial radius. We perform a stacking of the density profiles to detect a signal beyond r200 and measure the typical density and scatter in cluster outskirts. We also compute the azimuthal scatter of the profiles with respect to the mean value to look for deviations from spherical symmetry. Finally, we compare our average density and scatter profiles with the results of numerical simulations. Results. As opposed to some recent Suzaku results, and confirming previous evidence from ROSAT and Chandra, we observe a steepening of the density profiles beyond approximately r(sub 500). Comparing our density profiles with simulations, we find that non-radiative runs predict too steep density profiles, whereas runs including additional physics and/or treating gas clumping are in better agreement with the observed gas distribution. We report for the first time the high-confidence detection of a systematic difference between cool-core and non-cool core clusters beyond 0.3r(sub 200), which we explain by a different distribution of the gas in the two classes. Beyond r(sub 500), galaxy clusters deviate significantly from spherical symmetry, with only little differences between relaxed and disturbed systems. We find good agreement between the observed and predicted scatter profiles, but only when the 1% densest clumps are filtered out in the simulations. Conclusions. Comparing our results with numerical simulations, we find that non-radiative simulations fail to reproduce the gas distribution, even well outside cluster cores. Although their general behavior is in better agreement with the observations, simulations including cooling and star formation convert a large amount of gas into stars, which results in a low gas fraction with respect to the observations. Consequently, a detailed treatment of gas cooling, star formation, AGN feedback, and taking into account gas clumping is required to construct realistic models of cluster outer regions.

Eckert, D.↗
DOE OSTI2019

A microbenchmark characterization of the Emu chick

The Emu Chick is a prototype system designed around the concept of migratory memory-side processing. Rather than transferring large amounts of data across power-hungry, high-latency interconnects, the Emu Chick moves lightweight thread contexts to near-memory cores before the beginning of each memory read. The current prototype hardware uses FPGAs to implement cache-less “Gossamer” cores for computational work and rely on a typical stationary core (PowerPC) to run basic operating system functions and migrate threads between nodes. In this multi-node characterization of the Emu Chick, we extend an earlier single-node investigation of the the memory bandwidth characteristics of the system through benchmarks like STREAM, pointer chasing, and sparse matrix-vector multiplication. Here, we compare the Emu Chick hardware to architectural simulation and an Intel Xeon-based platform. Our results demonstrate that for many basic operations the Emu Chick can use available memory bandwidth more efficiently than a more traditional, cache-based architecture although bandwidth usage suffers for computationally intensive workloads like SpMV. Moreover, the Emu Chick provides stable, predictable performance with up to 65% of the peak bandwidth utilization on a random-access pointer chasing benchmark with weak locality.

97 MATHEMATICS AND COMPUTING↗