Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Bandwidth”

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 991 records · Page 55

Architecture and Performance of Perlmutter’s 35 PB ClusterStor E1000 All-Flash File System

NERSC's newest system, Perlmutter, features a 35 PB all-flash Lustre file system built on HPE Cray ClusterStor E1000. We present its architecture, early performance figures, and performance considerations unique to this architecture. We demonstrate the performance of E1000 OSSes through low-level Lustre tests that achieve over 90% of the theoretical bandwidth of the SSDs at the OST and LNet levels. We also show end-to-end performance for both traditional dimensions of I/O performance (peak bulk-synchronous bandwidth) and non-optimal workloads endemic to production computing (small, incoherent I/Os at random offsets) and compare them to NERSC's previous system, Cori, to illustrate that Perlmutter achieves the performance of a burst buffer and the resilience of a scratch file system. Finally, we discuss performance considerations unique to all-flash Lustre and present ways in which users and HPC facilities can adjust their I/O patterns and operations to make optimal use of such architectures.

Lockwood, Glenn K↗

Design Space Exploration of Emerging Memory Technologies for Machine Learning Applications

Memory design space exploration methods study memory systems’ performances and limitations before implementation. The computer memory design space has grown exponentially because of the enormous growth of memory types, memory controllers, and application software. Computer simulators are commonly used for memory design space exploration. However, complex memory simulations take an enormous amount of time. Hence, in this paper, we proposed a machine learning-based design space exploration method for dynamic random-access memory and non-volatile memory systems. We applied our method to the CosmoGAN and LeNet applications to predict the following six memory response parameters: (i) bandwidth, (ii) power, (iii) average latency, (iv) average total latency, (v) memory reads, and (vi) memory writes. Our experimental results show that machine learning models can predict memory response parameter values faster than simulations. We used support vector machine, random forest, and gradient boosting machine learning models. We observed that the support vector machine provides better performance for bandwidth, average latency, and average total latency. The random forest model works better for memory reads and writes. The gradient boosting model provides superior prediction performance for power. We provide a detailed discussion on learning curve characteristics, error analysis, and memory type recommendation.

Hasan, S M Shamimul↗

Performance Potential of Mixed Data Management Modes for Heterogeneous Memory Systems

Many high-performance systems now include different types of memory devices within the same compute platform to meet strict performance and cost constraints. Such heterogeneous memory systems often include an upper-level tier with better performance, but limited capacity, and lower-level tiers with higher capacity, but less bandwidth and longer latencies for reads and writes. To utilize the different memory layers efficiently, current systems rely on hardware-directed, memory -side caching or they provide facilities in the operating system (OS) that allow applications to make their own data-tier assignments. Since these data management options each come with their own set of trade-offs, many systems also include mixed data management configurations that allow applications to employ hardware- and software-directed management simultaneously, but for different portions of their address space. Despite the opportunity to address limitations of stand-alone data management options, such mixed management modes are under-utilized in practice, and have not been evaluated in prior studies of complex memory hardware. In this work, we develop custom program profiling, configurations, and policies to study the potential of mixed data management modes to outperform hardware- or software-based management schemes alone. Our experiments, conducted on an Intel ® Knights Landing platform with high-bandwidth memory, demonstrate that the mixed data management mode achieves the same or better performance than the best stand-alone option for five memory intensive benchmark applications (run separately and in isolation), resulting in an average speedup compared to the best stand-alone policy of over 10 %, on average.

Effler, Chad↗

CP-DSSS: An OFDM Compatible Variable Rate Modulation for 5G and Beyond

Cyclic prefix direct sequence spread spectrum (CPDSSS) is a recently proposed waveform that has been designed for coexistence with OFDM (orthogonal frequency division multiplexing) in the same or a pair of parallel networks. The first contribution of this paper is to present equations that reveal similarities and differences of CP-DSSS and OFDM. Furthermore, we show that the channel model of CP-DSSS reduces to that of a cyclic prefixed block-wise single carrier modulation (CP-SCM). However, unlike CP-SCM whose symbol rate is always equal to the transmission bandwidth, CP-DSSS can be adopted to any symbol rate equal to or smaller than the transmission bandwidth. This reduction in symbol rate allows CP-DSSS power spectral density to drop to an arbitrarily low level, hence, facilitates it coexistence as a secondary waveform in a network of primary users. The peak-to-average power ratio (PAPR) of CP-DSSS is also explored and a method for reducing it is proposed.

5G↗

Leaky Buddies: Cross-Component Covert Channels on Integrated CPU-GPU Systems

Graphics Processing Units (GPUs) are ubiquitous components used across the range of today’s computing plat- forms, from phones and tablets, through personal computers, to high-end server class platforms. With the increasing importance of graphics and video workloads, recent processors are shipped with GPU devices that are integrated on the same chip. Integrated GPUs share some resources with the CPU and as a result, there is a potential for microarchitectural attacks from the GPU to the CPU or vice versa. We consider the potential for covert channel attacks that arise either from shared microarchitectural components (such as caches) or through shared contention domains (e.g., shared buses). We illustrate these two types of channels by developing two reliable covert channel attacks. The first covert channel uses the shared LLC cache in Intel’s integrated GPU architectures. The second is a contention based channel targeting the ring bus connecting the CPU and GPU to the LLC. This is the first demonstrated microarchitectural attack crossing the component boundary (GPU to CPU or vice versa). Cross-component channels introduce a number of new challenges that we had to overcome since they occur across heterogeneous components that use different computation models and are interconnected using asymmetric memory hierarchies. We also exploit GPU parallelism to increase the bandwidth of the communication, even without relying on a common clock. The LLC based channel achieves a bandwidth of 120 kbps with a low error rate of 2%, while the contention based channel delivers up to 400 kbps with a 0.8% error rate. We also demonstrate a proof- of-concept prime-and-probe side channel attack that probes the full LLC from the GPU.

Cyber secruity, high performance comptuing, Hetero↗

On the Phase Grouping Mechanism for a Magnetron Coherent Oscillation

CW magnetrons, developed for industrial heaters, but driven by an injection-locking signal were suggested to power Superconducting RF (SRF) cavities due to higher efficiency and lower cost of generated RF power per Watt than traditionally used RF sources (klystrons, IOTs, solid-state amplifiers). When the magnetrons are intended to feed Room Temperature (RT) cavities, the injected phase or frequency locking signal may provide required phase or frequency stability of the accelerating field. However, when the magnetron RF sources are intended to feed high Q-factor SRF cavities, the sources must be controlled in phase and power in a wide bandwidth to compensate parasitic phase and amplitude modulations caused by microphonics. In dependence on parameters of magnetron and the injection-locking signal one can choose regime most suitable for feeding SRF cavities, enabling magnetron almost coherent oscillation at the wide bandwidth of control. A novel approach considering magnetrons as quasi-coherent or coherent RF generators enables choosing the tube parameters and operation most suitable for various SRF accelerators.

43 PARTICLE ACCELERATORS↗

A Modern Ethernet Data Acquisition Architecture for Fermilab Beam Instrumentation

The Fermilab Accelerator Division, Instrumentation Department is adopting an open-source framework to replace our embedded VME-based data acquisition systems. Utilizing an iterative methodology, we first moved to embedded Linux, removing the need for VxWorks. Next, we adopted Ethernet on each data acquisition module eliminating the need for the VME backplane in addition to communicating with a rack mount server. Development of DDCP (Distributed Data Communications Protocol), allowed for an abstraction between the firmware and software layers. Each data acquisition module was adapted to read out using 1 GbE and aggregated at a switch which up linked to a 10 GbE network. Current development includes scaling the system to aggregate more modules, to increase bandwidth to support multiple systems and to adopt MicroTCA as a crate technology. The architecture was utilized on various beamlines around the Fermilab complex including PIP2IT, FAST/IOTA and the Muon Delivery Ring. In summary, we were able to develop a data acquisition framework which incrementally replaces VxWorks & VME hardware as well as increases our total bandwidth to 10 Gbit/s using off the shelf Ethernet technology.

43 PARTICLE ACCELERATORS↗

Systems, methods and computer program products for collaborative agent control

Systems, methods and unmanned agents for collaboratively controlling agents in a collaborative network by one or more agents continuously simulating numeric models of one or more other agents in the network to dramatically reduce the computational bandwidth required between agents, and improve the quality of shared estimates of the agent locations as well as the locations and characteristics of other objects of interest, e.g. targets. Bandwidth is reduced by using the models to intelligently filter data before communicating.

Buerger, Stephen↗

SRF Cavity Emulator for PIP-II LLRF Lab and Field Testing

There are many stages in the LLRF and RF system development process for any new accelerator that can take advantage of hardware emulation of the high-power RF system and RF cavities. LLRF development, bench testing, control system development and testing of installed systems must happen well before SRF cavities are available for test. The PIP-II Linac has three frequencies of SRF cavities, 162.5 MHz, 325 MHz and 650 MHz and a simple analog emulator design has been chosen that can meet the cavity bandwidth requirements, provide tuning errors to emulate Lorentz force detuning and microphonics for all cavity types. This emulator design utilizes a quartz crystal with a bandwidth of 65Hz at an IF of ~ 4 MHz, providing a Q of ~ 1.3 x 10^7 at 650MHz. This paper will discuss the design and test results of this emulator.

43 PARTICLE ACCELERATORS↗

Stability and Fidelity of Power-Hardware-in-the-Loop Using NREL's CGI

The presentation focuses on dynamic analysis of power hardware-in-the-loop (PHIL) experiments using National Renewable Energy Laboratory Flatirons Campus' grid simulator - 7MVA Controllable Grid Interface (CGI). A control diagram of typical PHIL experiment is described, and transfer functions of critical components are explained. An impedance-based definition of PHIL experiment bandwidth is defined using open loop transfer functions of the system. Further an example resistive divider circuit is introduced that is used to explain implication of PHIL on accuracy of simulation by comparing measured results against ideal expected signals. At the end the real-life PHIL simulation using Maui power system model is described showing various oscillation modes observed and discussing the meaning of these in context of dynamic bandwidth of simulation.

grid simulator↗

Multi-chip photonics transceiver

A photonics transceiver is described herein, wherein the photonics transceiver exhibits improved areal bandwidth density and improved energy per bit consumption relative to conventional photonics transceivers. The photonics transceiver achieves an areal bandwidth density of at least 5 Tbps/mm2 with an energy consumption of less than 500 fJ/bit (sum of energy consumed for both a transmitted bit and a received bit). The photonics transceiver is a multi-chip module, where chips in the multi-chip module are tightly integrated with one another. The multi-chip module includes light source, photodetector, photonics, and control/logic chips. The photonics chip includes transparent conducting oxide integrated optical modulators and multiplexers and demultiplexers based on MEMS-tunable optical ring resonators.

Wood, Michael↗

Evaluation and Parameter Analysis of Burn up Calculations for the Assessment of Radioactive Waste - 13187

Burn up calculations facilitate a determination of the composition and nuclear inventory of spent nuclear fuel, if operational history is known. In case this information is not available, the total nuclear inventory can be determined by means of destructive or, even on industrial scale, nondestructive measurement methods. For non-destructive measurements however only a few easy-to-measure, so-called key nuclides, are determined due to their characteristic gamma lines or neutron emission. From these measured activities the fuel burn up and cooling time are derived to facilitate the numerical inventory determination of spent fuel elements. Most regulatory bodies require an independent assessment of nuclear waste properties and their documentation. Prominent part of this assessment is a consistency check of inventory declaration. The waste packages often contain wastes from different types of spent fuels of different history and information about the secondary reactor parameters may not be available. In this case the so-called characteristic fuel burn up and cooling time are determined. These values are obtained from a correlations involving key-nuclides with a certain bandwidth, thus with upper and lower limits. The bandwidth is strongly dependent on secondary reactor parameter such as initial enrichment, temperature and density of the fuel and moderator, hence the reactor type, fuel element geometry and plant operation history. The purpose of our investigation is to look into the scaling and correlation limitations, to define and verify the range of validity and to scrutinize the dependencies and propagation of uncertainties that affect the waste inventory declarations and their independent verification. This is accomplished by numerical assessment and simulation of waste production using well accepted codes SCALE 6.0 and 6.1 to simulate the cooling time and burn up of a spent fuel element. The simulations are benchmarked against spent fuel from the real reactor Obrigheim in Germany for which sufficiently precise experimental reference data are available. (authors)

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Runtime extension for neural network training with heterogeneous memory

Systems, apparatuses, and methods for managing buffers in a neural network implementation with heterogeneous memory are disclosed. A system includes a neural network coupled to a first memory and a second memory. The first memory is a relatively low-capacity, high-bandwidth memory while the second memory is a relatively high-capacity, low-bandwidth memory. During a forward propagation pass of the neural network, a run-time manager monitors the usage of the buffers for the various layers of the neural network. During a backward propagation pass of the neural network, the run-time manager determines how to move the buffers between the first and second memories based on the monitored buffer usage during the forward propagation pass. As a result, the run-time manager is able to reduce memory access latency for the layers of the neural network during the backward propagation pass.

Mappouras, Georgios↗

Accelerating Application Bulk Synchronous Writes in HPC Environments

High-bandwidth storage tiers are becoming more common for their capability to absorb high-rate, bursty I/Os. Notably, the designs of these fast storage tiers differ from system to system. The variation of these layers and non-uniform methods of access can pose chal- lenges for applications seeking to run at multiple HPC facilities. Therefore, in this work, we present Spectral, a rapid-output ab- straction library to accelerate application, bulk-synchronous writes on HPC systems. We design Spectral to enable applications to use high-bandwidth storage, such as node-local storage and dis- tributed, write-caches (e.g., burst buffers) transparently without requiring modifications to the application or file system source code. The key idea is to allow applications to spend most of the time performing productive work and to not require any source code changes for maximum portability on different HPC archi- tectures. Spectral internally re-routes write-only files through available, high-performance I/O resources before ultimately mi- grating them to the shared global parallel file system. For instance, on Summit, Spectral transparently places application outputs on node-local storage and then utilizes asynchronous migration to the center-wide GPFS file system. We evaluate Spectral on the Summit HPC system (1024 nodes) using the IOR benchmark and real scientific applications. Spectral shows linear performance scaling, improving application write performance by over an order of magnitude when compared to GPFS.

Khan, Awais↗

On the Speed-up of Wave-like Dark Matter Searches with Entangled Qubits

Qubit-based sensing platforms offer promising new directions for wave-like dark matter searches. Recent proposals demonstrate that entangled qubits can achieve quadratic scaling of the signal in the number of qubits. In this work we expand on these proposals to analyze the bandwidth and scan rate performance of entangled qubit protocols across different error regimes. We find that the phase-based readout of entangled protocols preserves the search bandwidth independent of qubit number, in contrast to power-based detection schemes, thereby achieving a genuine scan-rate advantage. We derive coherence time and error rate requirements for qubit systems to realize this advantage. Applying our analysis to dark photon searches, we find that entangled states of approximately 100 qubits can become competitive with benchmark photon-counting cavity experiments for masses $\gtrsim 30{-}40~μ{\rm eV}$, provided sufficiently low error rates are achieved. The advantage increases at higher masses where cavity volume scaling becomes less favorable.

Bodas, Arushi [Fermilab; Chicago U.] (ORCID:000000↗

Suppression of auger recombination in ""giant"" core/shell nanocrystals

Many potential applications of semiconductor nanocrystals are hindered by nonradiative Auger recombination wherein the electron-hole (exciton) recombination energy is transferred to a third charge carrier. This process severely limits the lifetime and bandwidth of optical gain, leads to large nonradiative losses in light emitting diodes and photovoltaic cells, and is believed to be responsible for intermittency ('blinking') of emission from single nanocrystals. The development of nanostructures in which Auger recombination is suppressed has been a longstanding goal in colloidal nanocrystal research. Here, we demonstrate that such suppression is possible using so-called 'giant' nanocrystals that consist of a small CdSe core and a thick CdS shell. These nanostructures exhibit a very long biexciton lifetime ({approx}10 ns) that is likely dominated by radiative decay instead of non-radiative Auger recombination. As a result of suppressed Auger recombination, even high-order multiexcitons exhibit high emission efficiencies, which allows us to demonstrate optical amplification with an extraordinarily large bandwidth (>500 me V) and record low excitation thresholds.

36 MATERIALS SCIENCE↗

Evaluating the Computational Requirements of using SVM software to train Data-Intensive Problems

Support vector machine (SVM) technology is a growing class of computational algorithms for solving classification problems using supervised learning. There are many freely available SVM codes with implementations of varying granularity in the core optimization task. The performance of SVM implementations is related to four key elements; (1) the size and dimension of the data set on which they are operating, (2) the granularity of their core SVM optimization implementation, (3) the kernel transformation applied to the data, and the (4) underlying hardware on which the implementation is running. To assess the performance of different SVM implementations, several freely available codes representing the spectrum of optimization granularity were built on a variety of hardware—two linux clusters, and a shared memory machine. Binary classifiers were trained on datasets of varying size containing anywhere from a few hundred to over 50,000 vectors. Performance of the methods was measured in terms of wall-clock time for training, statistical quality of the resulting classifier, hardware performance (memory footprint, memory bandwidth, floating point unit usage, I/O bandwidth, and instruction stalls), robustness and portability.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

...And Eat it Too: High Read Performance in Write-Optimized HPC I/O Middleware File Formats

As HPC applications run on increasingly high process counts on larger and larger machines, both the frequency of checkpoints needed for fault tolerance and the resolution and size of Data Analysis Dumps are expected to increase proportionally. In order to maintain an acceptable ratio of time spent performing useful computation work to time spent performing I/O, write bandwidth to the underlying storage system must increase proportionally to this increase in the checkpoint and computation size. Unfortunately, popular scientific self-describing file formats such as netCDF and HDF5 are designed with a focus on portability and flexibility. Extra care and careful crafting of the output structure and API calls is required to optimize for write performance using these APIs. To provide sufficient write bandwidth to continue to support the demands of scientific applications, the HPC community has developed a number of I/O middleware layers, that structure output into write-optimized file formats. However, the obvious concern with any write optimized file format would be a corresponding penalty on reads. In the log-structured filesystem, for example, a file generated by random writes could be written efficiently, but reading the file back sequentially later would result in very poor performance. Simulation results require efficient read-back for visualization and analytics, and though most checkpoint files are never used, the efficiency of a restart is very important in the face of inevitable failures. The utility of write speed improving middleware would be greatly diminished if it sacrificed acceptable read performance. In this paper we examine the read performance of two write-optimized middleware layers on large parallel machines and compare it to reading data natively in popular file formats. The two I/O middleware layer examined in this paper are the Adaptable IO System (ADIOS), a library-based approach developed at Oak Ridge National Laboratory to provide a high-level IO API that can be used in place of netCDF or HDF5 to do much more aggressive write-behind and efficient reordering of data locations within the file; and the Parallel Log-structured Filesystem (PLFS), a stackable FUSE filesystem approach developed at Los Alamos National Laboratory that decouples concurrent writes to improve the speed of checkpoints. Since ADIOS is an I/O componentization that affords selection of different I/O methods at or during runtime, through a single API, users can have access to MPI-IO, Posix-IO, HDF5, netCDF, and staging methods. The ADIOS BP file format is a new log-file format that has a superset of the features of both HDF5 and netCDF, but is designed to be portable and flexible while being optimized for writing. PLFS, on the other hand, is mounted as a stackable filesystem on top of an existing parallel filesystem. Reads or writes to the PLFS filesystem are transparently translated into operations on per-process log files stored in the underlying parallel filesystem. Since PLFS performs this translation without application modification, users can write in HDF5, netCDF, or app-specific file formats and PLFS will store the writes in a set of efficiently written log-formatted files, while presenting the user with a logical 'flat' file on reads. Despite their different approaches, the commonality behind both of these middleware systems is that they both write to a log file format. As shown in previous publications, writes are fully optimized in both systems, sometimes resulting in 100x improvements over writing data in popular file formats. But as mentioned above, writing is only one part of the story. In this paper we examine the read performance of our middleware layers on large parallel machines and compare these to reading data either natively or from other popular file formats. We compare the reading performance in two different scenarios: (1) Reading back restarts from the same number of processors as wrote the data and (2) Reading back restart data from a different number of processors. We observe that not only can write-optimized I/O middleware be built to not penalize read speeds, but for important workloads, techniques that improve write performance can, perhaps counterintuitively, improve read speeds over reading from a contiguously organized file format. In the remainder of this paper, we investigate this further through case studies of PLFS and ADIOS on simulation checkpoint restarts.

99 GENERAL AND MISCELLANEOUS↗