Napari-MCP
Napari-MCP connects the open source napari software to modern LLMs. It includes a socket server plugin for napari as well as the MCP tools.
SEARCH · Engineering Papers
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.
Napari-MCP connects the open source napari software to modern LLMs. It includes a socket server plugin for napari as well as the MCP tools.
Scientific applications require solvers that work on many small size problems that are independent from each other. At the same time, the high-end hardware evolves rapidly and becomes ever more throughput-oriented and thus there is an increasing need for an effective approach to develop energy-efficient, high-performance codes for these small matrix problems that we call batched factorizations. The many applications that need this functionality could especially benefit from the use of GPUs, which currently are four to five times more energy efficient than multicore CPUs on important scientific workloads. This study, consequently, describes the development of the most common, one-sided factorizations, Cholesky, LU, and QR, for a set of small dense matrices. The algorithms we present together with their implementations are, by design, inherently parallel. In particular, our approach is based on representing the process as a sequence of batched BLAS routines that are executed entirely on a GPU. Importantly, this is unlike the LAPACK and the hybrid MAGMA factorization algorithms that work under drastically different assumptions of hardware design and efficiency of execution of the various computational kernels involved in the implementation. Thus, our approach is more efficient than what works for a combination of multicore CPUs and GPUs for the problems sizes of interest of the application use cases. The paradigm where upon a single chip (a GPU or a CPU) factorizes a single problem at a time is not at all efficient in our applications’ context. We illustrate all of these claims through a detailed performance analysis. With the help of profiling and tracing tools, we guide our development of batched factorizations to achieve up to two-fold speedup and three-fold better energy efficiency as compared against our highly optimized batched CPU implementations based on MKL library. Finally, the tested system featured two sockets of Intel Sandy Bridge CPUs and we compared with a batched LU factorizations featured in the CUBLAS library for GPUs, we achieve as high as 2.5× speedup on the NVIDIA K40 GPU.
Jefferson Lab's Web Extensible Display Manager (WEDM) allows staff to access EDM control system screens from a web browser in remote offices and from mobile devices. Native browser technologies are leveraged to avoid installing and managing software on remote clients such as browser plugins, tunnel applications, or an EDM environment. Since standard network ports are used firewall exceptions are minimized. To avoid security concerns from remote users modifying a control system, WEDM exposes read-only access and basic web authentication can be used to further restrict access. Updates of monitored EPICS channels are delivered via a Web Socket using a web gateway. The software translates EDM description files (denoted with the edl suffix) to HTML with Scalable Vector Graphics (SVG) following the EDM's edl file vector drawing rules to create faithful screen renderings. The WEDM server parses edl files and creates the HTML equivalent in real-time allowing existing screens to work without modification. Alternatively, the familiar drag and drop EDM screen creation tool can be used to create optimized screens sized specifically for smart phones and then rendered by WEDM.
TimePix3 is a 65k hybrid pixel readout chip with simultaneous Time-of-Arrival (ToA) and Time-over-Threshold (ToT) recording in each pixel*. The chip operates without a trigger signal with a sparse readout where only pixels containing events are read out. The flexible architecture allows 40 MHits/s/cm² readout throughput, using simultaneous readout and acquisition by sharing readout logic with transport logic of superpixel matrix formed using 2x4 structure. The chip ToA records 1.5625 ns time resolution. The X-ray and charged particle events are counted directly. However, indirect neutron counts use 6Li fission in a scintillator matrix, such as ZnS(Ag). The fission space-charge region is limited to 5-9 um. A photon from scintillator material excites a photocathode electron, which is further multiplied in dual-stack MCP. The neutron count event is a cluster of electron events at the chip. We report on the EPICS areaDetector** ADTimePix3 driver that controls Serval*** using json commands. The driver directs data to storage and to a real-time processing pipeline and configures the chip. The time-stamped data are stored in raw .tpx3 file format and passed through a socket where the clustering software identifies individual neutron events. The conventional 2D images are available as images for each exposure frame, and a preview is useful for sample alignment. The areaDetector driver allows integration of time-enhanced capabilities of this detector into SNS beamlines controls and unprecedented time resolution.
The Django Remote Submission (DRS) is a Django (Django, n.d.) application to manage long running job submission, including starting the job, saving logs, and storing results. It is an independent project available as a standalone pypi package (PyPi, n.d.). It can be easily integrated in any Django project. The source code is freely available as a GitHub repository (django-remote-submission, n.d.). To run the jobs in background, DRS takes advantage of Celery (Celery, n.d.), a powerful asynchronous job queue used for running tasks in the background, and the Redis Server (Redis, n.d.), an in-memory data structure store. Celery uses brokers to pass messages between a Django Project and the Celery workers. Redis is the message broker of DRS. In addition DRS provides real time monitoring of the progress of Jobs and associated logs. Through the Django Channels project (Channels, n.d.), and the usage of Web Sockets, it is possible to asynchronously display the Job Status and the live Job output (standard output and standard error) on a web page.
Aggressive engine downsizing, variable compression ratio and use of the Atkinson cycle are being combined to improve fuel economy by up to 40 percent relative to port fuel injected gasoline engines, while maintaining full engine power. Approach Engine downsizing is viewed by US and foreign automobile manufacturers as one of the best options for improving fuel economy. While this strategy has already demonstrated a degree of success, downsizing and fuel economy gains are currently limited. With new variable compression ratio technology however, the degree of engine downsizing and fuel economy improvement can be greatly increased. A small variable compression ratio (VCR) engine has the potential to return significantly higher vehicle fuel economy while also providing high power. Affordability and potential for near term commercialization are key attributes of the Envera VCR engine. VCR Technology To meet torque and power requirements, a smaller engine needs to do more work per stroke. This is typically accomplished by boosting the incoming charge with either a turbo or supercharger so that more energy is present in the cylinder per stroke to do the work. With current production engines the degree of engine boosting (which correlates to downsizing) is limited by detonation (combustion knock) at high boost levels. Additionally, the turbo or supercharger needs to be responsive and efficient while providing the needed boost. VCR technology eliminates the limitation of engine knock at high load levels by reducing compression ratio to {approx}9:1 (or whatever level is appropriate) when high boost pressures are needed. By reducing the compression ratio during high load demand periods there is increased volume in the cylinder at top dead center (TDC) which allows more charge (or energy) to be present in the cylinder without increasing the peak pressure. Cylinder pressure is thus kept below the level at which the engine would begin to knock. When loads on the engine are low the compression ratio can be raised (to as much as 18:1) providing high engine efficiency. It is important to recognize that for a well designed VCR engine cylinder pressure does not need to be higher than found in current production turbocharged engines. As such, there is no need for a stronger crankcase, bearings and other load bearing parts within the VCR engine. The Envera VCR mechanism uses an eccentric carrier approach to adjust engine compression ratio. The crankshaft main bearings are mounted in this eccentric carrier or 'crankshaft cradle' and pivoting the eccentric carrier 30 degrees adjusts compression ratio from 9:1 to 18:1. The eccentric carrier is made up of a casting that provides rigid support for the main bearings, and removable upper bearing caps. Oil feed to the main bearings transits through the bearing cap fastener sockets. The eccentric carrier design was chosen for its low cost and rigid support of the main bearings. A control shaft and connecting links are used to pivot the eccentric carrier. The control shaft mechanism features compression ratio lock-up at minimum and maximum compression ratio settings. The control shaft method of pivoting the eccentric carrier was selected due to its lock-up capability. The control shaft can be rotated by a hydraulic actuator or an electric motor. The engine shown in Figures 3 and 4 has a hydraulic actuator that was developed under the current program. In-line 4-cylinder engines are significantly less expensive than V engines because an entire cylinder head can be eliminated. The cost savings from eliminating cylinders and an entire cylinder head will notably offset the added cost of the VCR and supercharging. Replacing V6 and V8 engines with in-line VCR 4-cylinder engines will provide high fuel economy at low cost. Numerous enabling technologies exist which have the potential to increase engine efficiency. The greatest efficiency gains are realized when the right combination of advanced and new technologies are packaged together to provide the greatest gains at the least cost. Aggressive engine downsizing with variable compression ratio and use of the extended Atkinson cycle can provide large fuel economy gains that are exceptionally cost effective. Analysis indicates that a 2.2L supercharged Envera VCR engine can match the torque of a larger V8 engine at 2000 rpm. The VCR engine's high torque value at low engine speed is beneficial for maintaining the driving feel and responsiveness of the larger V8 engine. The Envera VCR engine will attain high efficiency at {approx}100 Nm primarily due to the combination of engine down-sizing and use of the Atkinson cycle. Qualitatively the fuel economy gain realized from down-sizing from a V8 to an Atkinson-cycle I-4 is about twice as large as the benefits from down-sizing from a V8 to a Turbo V6 when evaluated at 100 Nm 2000 rpm.
The manipulators used to support and position lattice elements are critical components of all particle accelerators. The increased use of large superconducting magnets and accelerator modules places even greater demands on these manipulators. However, the performance of these support systems has not kept pace with the advances made in other areas of accelerator technology. This results in accelerators that are difficult to align and may not be capable of achieving target luminosities. An innovative new type of positioning mechanism tailored to the requirements of the International Linear Collider is proposed. The Tri-Sphere System provides secure support for large lattice elements and precision adjustment in six degrees of freedom. Integrated target sockets allow the support system to be rapidly pre-aligned. The system's kinematic design passively guides lattice elements into their correct location during installation. A complimentary Portable Actuation Unit provides the advantages of automated adjustment and allows these adjustments to be completely decoupled from surveying.
New generator technology project is driven by the need to be able to remotely deploy generator technology where it is needed, when it is needed. Both the military and aid programs that provide assistance after disasters could use the ability to deploy energy generation that fits the needs of the situation. Currently, pre-specified generators are deployed, sometime more than half way around the world to provide electricity. Through our Phase-I to Phase III DARPA grant, we will provide a mechanism where a 3d print station and raw materials could be shipped to a deployment site and remotely deployed personnel. These remote personnel can collaborate with engineers at a home location where 3d print plans can be optimized for the remote purpose. The plans can then be sent electronically to the remote location for printing, much like NASA sent the plans for a socket wrench to the International Space Station for printing in . If multiple generators need to be deployed at different remote locations, within miles of each other the printer rig can be moved to print the generators where they are needed. 3d printing is growing in the field of manufacturing. 3d printing has matured to the point where many types of materials are now available for many types of manufacturing. Both magnetic and electrically conductive material materials have recently been developed which can now lead to 3d printing of engines and generators. Our project will provide a successful printer rig that can be remotely deployed, to print a generator design in the field as well as provide a process for deploying the printed generator as well. This Systems Engineering Management Plan(SEMP) will provide the planning required for a Phase I DARPA grant that may also include goals for Phase II and Phase II grants. The SEMP provides a proposed project schedule, references, system engineering processes, specialty engineering system deployment and product support sections. Each section will state how our company will provide the necessary services to make this project succeed.
This report focuses on the bare lamp performance of 31 linear LED lamps intended as an alternative to T8 fluorescent lamps. Data obtained in accordance with IES LM-79-08 indicated that the mean efficacy is similar to that of fluorescent lamps, but that lumen output is often much lower. This presents a situation where something must change in order for energy savings and equivalent illumination levels to be achieved simultaneously. In this case, the luminous intensity distribution of all the tested lamps was directional or semi-directional, rather than omnidirectional. Also discussed in this report are several issues related to the electrical configuration of the lamps, such as the required socket types and power feed location. While no configuration is necessarily better, the multitude of options can make specifying and installing linear LED lamps more difficult, with the potential for safety issues. Similarly, the variety of color and power quality attributes adds a layer of complexity to the specification process. Many products offered good or excellent quality attributes, but some did not and thus could be perceived as inferior to fluorescent lamps in some installations.
In his 2011 State of the Union address, President Obama called for one million electric vehicles on the road by 2015 [1]. With large-scale Electric Vehicle (EV) or Plug-in Electric Vehicle (PEV or EV for short) or Plug-in Hybrid Electric Vehicle (PHEV) penetration into the US market, there will be drastic reduction in fossil fuel consumption, thus significantly reducing our dependency on foreign oil [2-6]. There will also be significant reduction on Green House Gas (GHG) emissions and smog in the major US cities [3, 7, 8]. Similar studies have also been done other industrial counties [9]. For the fuel cost, with the home electricity rate around $0.13 per kWh, it would cost about $0.05 per mile for DC operation and $0.03 cents per mile for AC operation. But, assuming 25 miles per gallon for a typical vehicle and $4 per gallon, fossil fuel will cost $0.16 per mile [10]. The overall lifecycle cost of PEVs will be several folds lower than the existing fossil fueled vehicles. Despite the above advantages of the EVs, the current cost of EVSE is not affordable for the average consumer. Presently, the cost of installing state-of-the-art residential EVSE ranges from $1500 to $2500 [11]. Low priced EVSE technology, which is easy to install, and affordable to operate and maintain by an average consumer, is essential for the large-scale market penetration of EVs. In addition, the long-term success of this technology is contingent on the PEVs having minimal excessive load and shift impact on the grid, especially at peak times. In a report [2] published by the Pacific Northwest National Laboratory (PNNL), the exiting electric power generation infrastructure, if used at its full capacity 24 hours a day, would support up to 84% of the nation’s cars, pickup trucks and SUVs for an average daily drive of 33 miles. This mileage estimate is certainly much below what an average driver would drive his/her vehicle per day. Another report [3] by the National Renewable Energy Laboratory (NREL) shows that an increased PEV penetration would significantly increase pressure on the peak generation, if no controlled charging strategy was put in place. Investigations from Oak Ridge National Laboratory (ORNL) show that in many regions, additional power generation facilities must be put in place and operate in evening times to recharge the EVs [12]. By all accounts, large PEV penetration will bring to the power grid enormous challenges due to the excessive and stochastic demand, and can entirely change the peak time distribution and behavior, perhaps, into a bi-modal distribution capable of exhausting primary, secondary and even reserves (spinning or non-spinning). To minimize the infrastructure upgrade costs and risks to the grid, and to ensure that power quality and reliability remain within the set standards, the demand for EV plug-ins must then be controlled and coordinated locally and at regional levels. Novel control techniques must be devised to allow for close collaboration between neighboring plug-in requestors, between neighboring communities, and between these and more central power authorities. The concept of electric drive vehicle is not new. The development of electric vehicle has been around since 19th century [13]. But due to a number of reasons and practical limitations at the time, including lower cost of gasoline compared to electricity, excessive refueling times, and abundance of gasoline, the automobile industry embraced gasoline-powered vehicles worldwide [13]. With the global warming, ever reducing reservoirs of fossil oil around the world and increasing political pressure to reduce the national dependency on foreign oil, the last decade of the 20th century witnessed major technological breakthroughs in Alternative Fueled Vehicle (AFV) technologies, including electric vehicles. With GHG emissions and carbon footprint in the minds of many more consumers and politicians, the first decade of the 21stCentury witnessed more breakthroughs with some real life experimentation and sporadic deployment of these technologies [14]. By many accounts, the second decade of the 21st Century is expected to be the time when mass volume production and popular usage of these AFV technologies, especially EV, will materialize. The current DOE request for proposals recognizes the need for major technological changes to ensure that the above national goal is realizable. Two major challenges have been identified: (1) major reduction in the cost of ownership of EVSEs, and (2) managing additional EV loads in the power grid while maintaining power quality, reliability, and affordability. We note that the two challenges are closely linked – A holistic approach to true lifecycle cost of EVSE ownership will certainly include any taxes and surcharges that can be put in place for major potential investments in the grid, and higher electricity charges in case of more frequent and longer peak periods. From a societal perspective, this cost could also include the lost GDP (computed on a local basis) and revenue for businesses at local and regional levels when the grid is no longer capable of meeting the demand and unexpected outages occur. A typical end-point electrical distribution system delivers power to a residential EVSE from the neighborhood distribution pole, as shown in Fig.1. This pole has a transformer (neighboring step-down transformer) that steps down the utility medium voltage to dual 120VAC single phase (also called 240VAC split phase). This voltage is fed through a meter into the residential load control center. The load control center consists of branch circuit breakers and distributes the power supply within various areas of the residential unit. One of the branch circuits from the load control center feeds EV charging station for the unit. An electric vehicle charger is plugged into the socket of the EV charging station and other end of this charger is connected to the vehicle during charging. Figure 1 illustrates a typical configuration of the power grid. The left side of the figure shows the power grid from the power generation to the neighboring step-down transformer, while the right side of the figure shows multiple EVs with the respective charging stations. The typical step-down transformer has an upper limit representing the maximum load that can be requested from these neighboring houses altogether (typically 24 kW). In case the total load increases beyond the supported limit, the protection system (e.g. a circuit breaker) attached to the step-down transformer gets activated automatically.
The discovery of a new allotropic form of elemental carbon – the fullerenes – and subsequently other novel forms of elemental carbon with pyramidalized surfaces, most notably single-walled and multi-walled carbon nanotubes, introduced a novel structural motif to the polycyclic aromatic hydrocarbons (PAHs) with nonplanar surfaces. Our research program supported by BES DOE grant DE-FG02-04ER15514 has dealt with the synthesis, structural studies, and chemistry of the novel curved-surface PAHs with carbon frameworks structurally related to fullerenes. They are referred to as “buckybowls”. We prepared several new buckybowls and, even more importantly, developed the efficient, gram-scale synthetic methodologies for the preparation of small buckybowls, most notably corannulene (C20H10) and its derivatives. In addition, the employment of the corannulene-based synthons previously developed in our laboratory led to a number of highly nonplanar molecular architectures with two or more corannulene subunits with a potential for the applications as novel materials in separation sciences, nanoelectronics, photovoltaics and catalysis. In collaboration with Professor Angelici (Iowa State) we prepared and characterized several transition metal complexes of corannulene, providing the first structural characterization of η6 metal complexes of buckybowls by a single crystal X-ray diffraction. In addition to the definitive structural characterization of the complexes we demonstrated that the (η6-C6Me6)Ru2+ unit in some relatively stable complexes activate the corannulene ligand to react with proper nucleophiles suggesting that such complexex may be used in catalysis. (Section C). We have explored the efficiency of the dispersion-based interactions of curved-surface conjugated carbon networks by high-level computational models. We showed that the curvature of such networks does not reduce the van der Waals attractions as compared to the planar systems of similar size. We than concentrated on the design, synthesis and testing of the previously unknown molecular receptors with corannulene pincers which are capable of forming the strong ball-and-socket inclusion complexes with fullerenes in solution and in the solid state. Buckycatcher I (a molecular receptor with two corannulene pincers preorganized on a tetrabensocyclooctatetraene tether) inclusion complexes with fullerenes provided the first experimental evidence for the importance of concave-convex π – π interactions in the supramolecular chemistry of fullerene carbon cages with buckybowls. C60@Buckycatcher I has become the prototypical supramolecular system formed by the relatively weak (on the atom-to-atom bases) and has been extensively used by the computational community to test the quality of the theoretical models. In addition, Buckycatcher I shows an exceptional ability to adopt other guest molecules and several its inclusion complexes and/or solvates have been structurally characterized. These unusual structures show the potential of the buckycatchers to be tested as novel organic materials. “Intelligent design” of molecular receptors with corannulene receptors using computational approach allowing for a priori prediction of their binding potentials led to the preparation of other buckycatchers. Both bi- and tridentate corannulene pincers were tested and two of the bidentate receptors (Buckycatchers II and III) exhibited the outstanding affinity toward fullerenes, significantly exciding Buckycatcher I performance. Careful tailoring of the tethers resulted in the optimization of binding energies of the receptors with guest carbon cages and in the reduction of the entropy/solvation penalties. Practical preparation of dicorannulenopentacene opens a new avenue for the synthesis of a pool of bis-corannulene receptors possessing polar groups on their dibenzobarrelene tether. The polar “anchors” can be used to attach these efficient fullerene receptors to solid supports in order to modify their surfaces with the potential applications in separation sciences, catalysis and photovoltaic materials. additionally, the presence of such groups will improve the solubility of the receptors in polar solvents, expanding the scope of their supramolecular chemistry in solutions. Rigorous studies of thermodynamics of host-guest association of molecular clips and tweezers in organic solvents with C60/C70 inclusion complexes with Buckycatcher I as a model system have provided a full set of thermodynamical data including ΔG, ΔH, and –TΔS, for the formation of the inclusion complexes over a range of temperatures and in several solvents. Both 1H NMR and Isothermal Calorimetry (ITC) titrations provide virtually identical association constants, confirming the reliability of the results. WE have shown that thermodynamics of the host-guest complex formation is more complicated than anticipated and proposed by the existing computational models. These data are of premium importance for the testing of the computational solvation models and for a deeper understanding of the supramolecular interactions in organic solvents.
Servers’ field energy use remains poorly understood, given heterogeneous computing loads, configurable hardware and software, and operation over a wide range of management practices. This paper explores various characteristics of 1- and 2-socket volume servers that affect energy consumption, and quantifies the difference in power demand between higher-performing SPEC and ENERGY STAR servers and our best understanding of a typical server operating today. We first establish general characteristics of the U.S. installed base of volume servers from existing IDC data and the literature, before presenting information on server hardware configurations from data collection events at a major online retail website. We then compare cumulative distribution functions of server idle power across three separate datasets and explain the differences between them via examination of the hardware characteristics to which power draw is most sensitive. We find that idle server power demand is significantly higher than ENERGY STAR benchmarks and the industry-released energy use documented in SPEC, and that SPEC server configurations—and likely the associated power-scaling trends—are atypical of volume servers. Next, we examine recent trends in server power draw among high-performing servers across their full load range to consider how representative these trends are of all volume servers before inputting weighted average idle power load values into a recently published model of national server energy use. Finally, we present results from two surveys of IT managers (n=216) and IT vendors (n=178) that illustrate the prevalence of more-efficient equipment and operational practices in server rooms and closets; these findings highlight opportunities to improve the energy efficiency of the U.S. server stock.
The primary objective of this task was to provide a recommendation for a replaceable paddle that could be incorporated into the Saltstone Production Facility (SPF) process. The replaceable paddle concept was proposed given the time it took to replace the existing paddle designs was approximately a week of downtime. Two different types of replaceable paddles were considered, a split paddle and a replaceable tip paddle. The split paddle design was developed; however, a review of the drawings by maintenance personnel determined that changeout of the new design would take about the same time/effort as replacing the existing paddles. Five different replaceable tip designs were considered: side attachment, V neck, straight notch, angled notch, and dovetail. The recommended replaceable tip design was the angled notch, which includes a central hub with two angled notches, where the replaceable tips are secured to the hub using two ½” – 13 Unified National Coarse Threads (UNC) socket head cap screws (SHCS). Design drawing for both the flat and helical paddle blades are complete and are included in Appendix A. A structural calculation (M-CLC-Z-00137) was performed on this design which determined the design is structurally adequate for normal or extreme loads that would be encountered in the SPF READCOTM mixer. This calculation is included in Appendix C. Based on the calculation, mechanical failure will not limit the life of the tip or tip attachment. The quantity of erosion will be the primary driver dictating tip replacement schedules.
The first step towards running PSLF in a high-performance computing (HPC) environment was porting PSLF to the Linux operating system from the current shipping target of MS Windows. The target machine was the LLNL Quartz system which is a Linux cluster configured for large parallel jobs. It has 2,688 batch nodes with Dual Socket Intel E5-2695 v4 (Broadwell) CPUs with 18 cores each, with a total of 344,064 GB of memory, each node having 128 GB of memory. Nodes are connected with an Intel Omni-Path 100 GB/s interconnect. Quartz runs the Tri-Laboratory Operating System Software (TOSS) that includes Red Hat Enterprise Linux (RHEL) distribution used on the LLNL HPC machines.
Ibis Networks is a full-stack cleantech company that provides plug-level energy monitoring and control to solve energy and asset management problems for the enterprise. During DATES – DATES, an NREL research team worked with the Ibis team to validate their product, the InteliSocket. The Ibis InteliSocket is a pass-through plug-load energy monitor and controller that is designed to reduce energy wasted by common 120 V plug-in devices in commercial office buildings, such as computer peripherals, conference room AV equipment, and break-room appliances. The system can shut off supply power to these end uses via remote control, manual switches, pre-set schedules, or automated control algorithms. The scope of this IN2 project was the development and refinement of “smart” learning behavior algorithms (LBAs), which could help installation processes and dramatically expand the sockets’ capabilities and energy-saving potential by suggesting suitable control schedules that are based on monitored use patterns. While Ibis has the analytical and software expertise for algorithm development, the lack of test data, both in a controlled laboratory setting and in real-life deployment scenarios, represented a key barrier toward commercialization of the product. Assistance through the IN2 program provided an opportunity to conduct the needed “trial and error” algorithm development. The project included baseline field-data collection, laboratory testing, and field validation components, all of which were conducted at the NREL campus between April 2017 and July 2019.
Triangle counting is a fundamental building block in graph algorithms. In this paper, we propose a block-based triangle counting algorithm to reduce data movement during both sequential and parallel execution. Our block-based formulation makes the algorithm naturally suitable for heterogeneous architectures. The problem of partitioning the adjacency matrix of a graph is well-studied. Our task decomposition goes one step further: it partitions the set of triangles in the graph. By streaming these small tasks to compute resources, we can solve problems that do not fit on a device. We demonstrate the effectiveness of our approach by providing an implementation on a compute node with multiple sockets, cores and GPUs. The current state-of-the-art in triangle enumeration processes the Friendster graph in 2.1 seconds, not including data copy time between CPU and GPU. Using that metric, our approach is 20 percent faster. When copy times are included, our algorithm takes 3.2 seconds. This is 5.6 times faster than the fastest published CPU-only time.
Power distribution upgrades that include: A new PMH-5, 12KV/208, Delta/WYE transformer. 600A Main Switchboard (MSB) with 600A LSI Main CB and push button capability. New feeders from transformer to MSB. New panels with NEMA 1 enclosures, door-in-door, with 4" gutter space on each side, New NEMA 4X CT Cabinet and meter socket. New 300A & 150A LSI circuit breakers, Trenching for grounding and an allowance of $12,000 for grounding system provided by SNL electrical Operations Engineer.
We describe an electronics system to operate SIPMs in the laboratory. The system is comprised of a single channel amplifier board (SIPM Amp) and a 4-channel control board to which up to 4 SIPM Amp boards can be connected. The control board supplies a common low voltage power for the amplifiers. All channels feature individually regulated bias and Peltier power voltages, along withPt10K RTD readout. A graphical user interface allows control of the bias voltages, amplifier voltage, Peltier voltages, and also provides for readback of leakage currents and other voltages. The GUI app has a feedback loop to measure the SIPM temperature (with the PT10K) and generate correction voltages to the Peltier to regulate the temperature. The temperatures, which can be achieved by Peltier cooling depend on the thermal design of the setup, the practical limit being about 50C below ambient. The SIPM Amp. board is a small (3x5cm) PCB that has a single SIPM readout channel. The SIPM scan be surface mounted on the pcb or connected to a 2-pin 0.100" socket. The area around the SIPM footprint is thermally isolated by perforations in the PCB. The RTD is located inside the thermal break. Onboard is BV and LV filtering. The 10X amplifier is based on MiniCircuits' GALI-S66+ integrated circuit, tweaked for a 6V power. The amplifier was selected for a combination of high bandwidth, low noise, and low power. An accoupled amplifier output is connected to an SMA connector. For best operation of this inverting amplifier the output signal should be in the range of 0 to -1V.The Amplifier board is connected to the Control board with a 10-conductor 0.025" pitch ribbon cable, carrying all the services. Peltier power for currents more than 1A should use a separate 2-contact connector.