GLANCED-IO: Taming IO Optimization for Deep Learning at Scale
Explore the source record for details and available documents.
Computer science↗
SEARCH · Search NASA
Results for “io optimizations”
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.
Understanding the I/O Performance Gap Between Cori KNL and Haswell
The Cori system at NERSC has two compute partitions with different CPU architectures: a 2,004 node Haswell partition and a 9,688 node KNL partition, which ranked as the 5th most powerful and fastest supercomputer on the November 2016 Top 500 list. The compute partitions share a common storage configuration, and understanding the IO performance gap between them is important, impacting not only to NERSC/LBNL users and other national labs, but also to the relevant hardware vendors and software developers. In this paper, we have analyzed performance of single core and single node IO comprehensively on the Haswell and KNL partitions, and have discovered the major bottlenecks, which include CPU frequencies and memory copy performance. We have also extended our performance tests to multi-node IO and revealed the IO cost difference caused by network latency, buffer size, and communication cost. Overall, we have developed a strong understanding of the IO gap between Haswell and KNL nodes and the lessons learned from this exploration will guide us in designing optimal IO solutions in many-core era.
97 MATHEMATICS AND COMPUTING↗
GLANCED-IO: Taming I/O Optimization for Deep Learning at Scale
Explore the source record for details and available documents.
Computer science↗
Parallel IO Libraries for Managing HEP Experimental Data
The computing and storage requirements of the energy and intensity frontiers will grow significantly during the Run 4 & 5 and the HL-LHC era. Similarly, in the intensity frontier, with larger trig ger readouts during supernovae explosions, the Deep Underground Neutrino Experiment (DUNE) will have unique computing challenges that could be addressed by the use of parallel and accelerated dataprocessing capabilities. Most of the requirements of the energy and intensity frontier experiments rely on increasing the role of high performance computing (HPC) in the HEP community. In this presentation, we will describe our ongoing efforts that are focused on using HPC resources for the next generation HEP experiments. The HEPCCE (High Energy Physics-Center for Computational Excellence) IOS (Input/Output and Storage) group has been developing approaches to map HEP data to the HDF5 , an IO library optimized for the HPC platforms to store the intermediate HEP data. The complex HEP data products are serialized using ROOT to allow for experiment independent general mapping approaches of the HEP data to the HDF5 format. The mapping approaches can be optimized for high performance parallel IO. Similarly, simpler data can be directly mapped into the HDF5, which can also be suitable for offloading into the GPUs directly. We will present our works on both complex and simple data model models.
Bashyal, Amit↗
Distributionally-robust chance constrained and interval optimization for integrated electricity and natural gas systems optimal power flow with wind uncertainties
Not Available
Fang, Xin (ORCID:000000027979803X)↗
Application-specific I/O Optimizations on Petascale Supercomputers
Data-intensive science frontiers and challenges are emerg ing as computer technology has evolved substantially. Large-scale simu lations demand significant I/O workload, and as a result the I/O performance often becomes a bottleneck preventing high performance in scientific applications. In this paper we introduce a variety of I/O optimization techniques developed and imple mented when scaling a seismic application to petascale. These techniques include file system striping, data aggreg ation, reader/writer limit ing and less interleaving of data, collective MPI-IO, and data staging. The optimizations result in nearly perfect scalability of the target applicat ion on some of the most advanced petascale systems. The techniques introduced in this paper are applicab le to other scientific applications facing similar petascale I/O challenges.
97 MATHEMATICS AND COMPUTING↗
Measurements of file transfer rates over dedicated long-haul connections
Wide-area file transfers are an integral part of several High-Performance Computing (HPC) scenarios. Dedicated network connections with high capacity, low loss rate and low competing traffic, are increasingly being provisioned over current HPC infrastructures to support such transfers. To gain insights into these file transfers, we collected transfer rate measurements for Lustre and xfs file systems between dedicated multi-core servers over emulated 10 Gbps connections with round trip times (rtt) in 0-366 ms range. Memory transfer throughput over these connections is measured using iperf, and file IO throughput on host systems is measured using xddprof. We consider two file system configurations: Lustre over IB network and xfs over SSD connected to PCI bus. Files are transferred using xdd across these connections, and the transfer rates are measured, which indicate the need to jointly optimize the connection and host file IO parameters to achieve peak transfer rates. In particular, these measurements indicate that (i) peak file transfer rate is lower than peak connection and host IO throughput, in some cases by as much as 50% or lower, (ii) xdd request sizes that achieve peak throughput for host file IO do not necessarily lead to peak file transfer rates, and (iii) parallelism in host IO and TCP transport does not always improve the file transfer rates.
log-haul connections↗
Measurements of file transfer rates over dedicated long-haul connections
Wide-area file transfers are an integral part of several High-Performance Computing (HPC) scenarios. Dedicated network connections with high capacity, low loss rate and low competing traffic, are increasingly being provisioned over current HPC infrastructures to support such transfers. To gain insights into these file transfers, we collected transfer rate measurements for Lustre and xfs file systems between dedicated multi-core servers over emulated 10 Gbps connections with round trip times (rtt) in 0-366 ms range. Memory transfer throughput over these connections is measured using iperf, and file IO throughput on host systems is measured using xddprof. We consider two file system configurations: Lustre over IB network and xfs over SSD connected to PCI bus. Files are transferred using xdd across these connections, and the transfer rates are measured, which indicate the need to jointly optimize the connection and host file IO parameters to achieve peak transfer rates. In particular, these measurements indicate that (i) peak file transfer rate is lower than peak connection and host IO throughput, in some cases by as much as 50% or lower, (ii) xdd request sizes that achieve peak throughput for host file IO do not necessarily lead to peak file transfer rates, and (iii) parallelism in host IO and TCP transport does not always improve the file transfer rates.
Rao, Nageswara S↗
Low energy, high power hydrogen neutral beam for plasma heating
A high power, relatively low energy neutral beam injector was developed to upgrade of the neutral beam system of the gas dynamic trap device and C2-U experiment. The ion source of the injector produces a proton beam with the particle energy of 15 keV, current of up to 175 A, and pulse duration of a few milliseconds. The plasma emitter of the ion source is produced by superimposing highly ionized plasma jets from an array of four arc-discharge plasma generators. A multipole magnetic field produced with permanent magnets at the periphery of the plasma box is used to increase the efficiency and improve the uniformity of the plasma emitter. Multi-slit grids with 48% transparency are fabricated from bronze plates, which are spherically shaped to provide geometrical beam focusing. The focal length of the Ion Optical System (IOS) is 3.5 m and the initial beam diameter is 34 cm. The IOS geometry and grid potentials were optimized numerically to ensure accurate beam formation. The measured angular divergences of the beam are ±0.01 rad parallel to the slits and ±0.03 rad in the transverse direction.
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
Improving I/O Performance for Exascale Applications through Online Data Layout Reorganization
The applications being developed within the U.S. Exascale Computing Project (ECP) to run on imminent Exascale computers will generate scientific results with unprecedented fidelity and record turn-around time. Many of these codes are based on particle-mesh methods and use advanced algorithms, especially dynamic load-balancing and mesh-refinement, to achieve high performance on Exascale machines. Yet, as such algorithms improve parallel application efficiency, they raise new challenges for I/O logic due to their irregular and dynamic data distributions. Thus, while the enormous data rates of Exascale simulations already challenge existing file system write strategies, the need for efficient read and processing of generated data introduces additional constraints on the data layout strategies that can be used when writing data to secondary storage. We review these I/O challenges and introduce two online data layout reorganization approaches for achieving good tradeoffs between read and write performance. We demonstrate the benefits of using these two approaches for the ECP particle-in-cell simulation WarpX, which serves as a motif for a large class of important Exascale applications. Here, we show that by understanding application I/O patterns and carefully designing data layouts we can increase read performance by more than 80 percent.
97 MATHEMATICS AND COMPUTING↗
Parameterization retrieval of trace gas volume mixing ratios from Airborne MAX-DOAS
We present a parameterization retrieval of volume mixing ratios (VMRs) from differential slant column density (dSCD) measurements by Airborne Multi-AXis Differential Optical Absorption Spectroscopy (AMAX-DOAS). The method makes use of the fact that horizontally recorded limb spectra (elevation angle 0°) are strongly sensitive to the atmospheric layer at instrument altitude. These limb spectra are analyzed using reference spectra that largely cancel out column contributions from above and below the instrument, so that the resulting limb dSCDs, i.e., the column integrated concentration with respect to a reference spectrum, are almost exclusively sensitive to the atmospheric layers around instrument altitude. The conversion of limb dSCDs into VMRs is then realized by calculating box air mass factors (Box-AMFs) for a Rayleigh atmosphere and applying a scaling factor constrained by O 4 dSCDs to account for aerosol extinction. An iterative VMR retrieval scheme corrects for trace gas profile shape effects. Benefits of this method are (1) a fast conversion that only requires the computation of Box-AMFs in a Rayleigh atmosphere; (2) neither local aerosol extinction nor the slant column density in the DOAS reference (SCD ref ) needs to be known; and (3) VMRs can be retrieved for every measurement point along a flight track, thus increasing statistics and adding flexibility to capture concentration gradients. Sensitivity studies are performed for bromine monoxide (BrO), iodine monoxide (IO) and nitrogen dioxide (NO 2 ), using (1) simulated dSCD data for different trace gas and aerosol profiles and (2) field measurements from the Tropical Ocean tRoposphere Exchange of Reactive halogen species and Oxygenated VOC (TORERO) field experiment. For simulated data in a Rayleigh atmosphere, the agreement between the VMR from the parameterization method (VMR para ) and the true VMR (VMR true ) is excellent for all trace gases. Offsets, slopes and R 2 values for the linear fit of VMR para over VMR true are, respectively (0.008 ± 0.001) pptv, 0.988 ± 0.001, 0.987 for BrO; (-0.0066 ± 0.0001) pptv, 1.0021 ± 0.0003, 0.9979 for IO; (-0.17 ± 0.03) pptv, 1.0036 ± 0.0001, 0.9997 for NO 2 . The agreement for atmospheres with aerosol shows comparable R 2 values to the Rayleigh case, but slopes deviate a bit more from one: (0.093 ± 0.002) pptv, 0.933 ± 0.002, 0.907 for BrO; (0.0021 ± 0.0004) pptv, 0.887 ± 0.001, 0.973 for IO; (8.5 ± 0.1) pptv, 0.8302 ± 0.0006, 0.9923 for NO 2 . VMR para from field data are further compared with optimal estimation retrievals (VMR OE ). Least orthogonal distance fit of the data give the following equations: BrO para = (0.1 ± 0.2) pptv + (0.95 ± 0.14) × BrO OE ; IO para = (0.01 ± 0.02) pptv + (1.00 ± 0.12) × IO OE ; NO 2para = (3.9 ± 2.5) pptv + (0.87 ± 0.15) × NO 2OE . Overall, we conclude that the parameterization retrieval is accurate with an uncertainty of 20 % for IO, 30 % for BrO and NO 2 , but not better than 0.05 pptv IO, 0.5 pptv BrO and 10 pptv NO 2 . Finally, the retrieval is applicable over a wide range of atmospheric conditions and measurement geometries and not limited to the interpretation of vertical profile measurements in the remote troposphere.
54 ENVIRONMENTAL SCIENCES↗
Using Formal Grammars to Predict I/O Behaviors in HPC: The Omnisc'IO Approach
The increasing gap between the computation performance of post-petascale machines and the performance of their I/O subsystem has motivated many I/O optimizations including prefetching, caching, and scheduling. In order to further improve these techniques, modeling and predicting spatial and temporal I/O patterns of HPC applications as they run has become crucial. In this paper we present Omnisc'IO, an approach that builds a grammar-based model of the I/O behavior of HPC applications and uses it to predict when future I/O operations will occur, and where and how much data will be accessed. To infer grammars, Omnisc'IO is based on StarSequitur, a novel algorithm extending Nevill-Manning's Sequitur algorithm. Omnisc'IO is transparently integrated into the POSIX and MPI I/O stacks and does not require any modification in applications or higher-level I/O libraries. It works without any prior knowledge of the application and converges to accurate predictions of any N future I/O operations within a couple of iterations. Its implementation is efficient in both computation time and memory footprint.
97 MATHEMATICS AND COMPUTING↗
Design, performance, and grounding aspects of the International Thermonuclear Experimental Reactor ion cyclotron range of frequencies antenna
ITER's Ion Cyclotron Range of Frequencies (ICRF) system [Lamalle et al., Fusion Eng. Des. 88, 517–520 (2013)] comprises two antenna launchers designed by CYCLE (a consortium of European associations listed in the author affiliations above) on behalf of ITER Organisation (IO), each inserted as a Port Plug (PP) into one of ITER's Vacuum Vessel (VV) ports. Each launcher is an array of 4 toroidal by 6 poloidal RF current straps specified to couple up to 20 MW in total to the plasma in the frequency range of 40 to 55 MHz but limited to a maximum system voltage of 45 kV and limits on RF electric fields depending on their location and direction with respect to, respectively, the torus vacuum and the toroidal magnetic field. A crucial aspect of coupling ICRF power to plasmas is the knowledge of the plasma density profiles in the Scrape-Off Layer (SOL) and the location of the RF current straps with respect to the SOL. The launcher layout and details were optimized and its performance estimated for a worst case SOL provided by the IO. The paper summarizes the estimated performance obtained within the operational parameter space specified by IO. Aspects of the RF grounding of the whole antenna PP to the VV port and the effect of the voids between the PP and the Blanket Shielding Modules (BSM) surrounding the antenna front are discussed. These blanket modules, whose dimensions are of the order of the ICRF wavelengths, together with the clearance gaps between them will constitute a corrugated structure which will interact with the electromagnetic waves launched by ICRF antennas. The conditions in which the grooves constituted by the clearance gaps between the blanket modules can become resonant are studied. Simple analytical models and numerical simulations show that mushroom type structures (with larger gaps at the back than at the front) can bring down the resonance frequencies, which could lead to large voltages in the gaps between the blanket modules and perturb the RF properties of the antenna if they are in the ICRF operating range. The effect on the wave propagation along the wall structure, which is acting as a spatially periodic (toroidally and poloidally) corrugated structure, and hence constitutes a slow wave structure modifying the wall boundary condition, is examined.
70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
Integrated operation scenarios: Chapter 6 of the special issue: on the path to tokamak burning plasma operation
Here we report the progress of the development and optimization of operational scenarios for ITER and beyond, focusing upon baseline, hybrid, and steady-state scenarios since 2007. This includes advancements made by the integrated operation scenarios (IOS) topical group of the international tokamak physical activity as well as contributions from the broader tokamak community. The key area of research involves developing IOSs that encompass tokamak physics, operation, and technology by utilizing integrated modeling and control strategies. This requires leveraging available actuators to simultaneously control plasma position and shape, MHD activities that could lead to disruptions, transport, plasma-wall interaction and power exhaust, fuel cycle, fusion burn, and tritium breeding. The control extends from the plasma initiation phase, through the current ramp-up, flattop, start and end of the fusion burn, and current ramp-down, to the plasma termination phase. A review of the currently developed scenarios and modeling is provided in terms of (i) optimizing plasma initiation in ITER, (ii) preparing for the low activation phase to fully commission all tokamak systems and establish and validate physics and scenario conditions in preparation for deuterim-tritium (DT) operation, (iii) developing and preparing baseline and hybrid scenarios to demonstrate the feasibility of achieving these regimes within device constraints, (iv) exploring steady-state scenarios to meet ITER’s steady-state goals, (v) evaluating and preparing actuators for ITER, (vi) developing integrated control solutions using shared actuators. The most notable achievements include; (i) the development of ITER demonstration discharges by matching various dimensionless parameters, (ii) the development of scenarios in an ITER-like tungsten environment and DT operation, and (iii) the development of scenarios in superconducting tokamaks, enabling long-pulse operations with similar coil constraints to ITER. Along with these significant achievements, outstanding issues and recommendations for further research and development are provided. Importantly, this study goes beyond simply updating the ITER Physics Basis; it carries profound implications for the broader field of burning plasma research, offering valuable insights and guidance for the next generation of fusion experiments and devices.
ITER↗
An Optimizing Compiler for Petascale I/O on Leadership-Class Architectures
In high-performance computing (HPC), parallel I/O architectures usually have very complex hierarchies with multiple layers that collectively constitute an I/O stack, including high-level I/O libraries such as PnetCDF and HDF5, I/O middleware such as MPI-IO, and parallel file systems such as PVFS and Lustre. Our DOE project explored automated instrumentation and compiler support for I/O intensive applications. Our project made significant progress towards understanding the complex I/O hierarchies of high-performance storage systems (including storage caches, HDDs, and SSDs), and designing and implementing state-of-the-art compiler/runtime system technology that targets I/O intensive HPC applications that target leadership class machine. This final report summarizes the major achievements of the project and also points out promising future directions Two new sections in this report compared to the previous report are IOGenie and SSD/NVM-specific optimizations.
97 MATHEMATICS AND COMPUTING↗
I/O Bottleneck Detection and Tuning: Connecting the Dots using Interactive Log Analysis
Using parallel file systems efficiently is a tricky problem due to inter-dependencies among multiple layers of I/O software, including high-level I/O libraries (HDF5, netCDF, etc.), MPI-IO, POSIX, and file systems (GPFS, Lustre, etc.). Profiling tools such as Darshan collect traces to help understand the I/O performance behavior. However, there are significant gaps in analyzing the collected traces and then applying tuning options offered by various layers of I/O software. Seeking to connect the dots between I/O bottleneck detection and tuning, we propose DXT Explorer, an interactive log analysis tool. In this paper, we present a case study using our interactive log analysis tool to identify and apply various I/O optimizations. We report an evaluation of performance improvement achieved for four I/O kernels extracted from science applications.
Bez, Jean Luca↗
A surface acoustic wave hydrogen sensor with tin doped indium oxide layers for intermediate temperatures
We report high temperature surface acoustic wave (SAW) gas sensors with conducting sensing layers require tuning of the sheet conductivity for optimal response. Conducting metal oxides are attractive sensing materials for their tunable electronic properties and high thermal stability, amongst others. Here, we have investigated the application of indium oxide (IO) and indium tin oxide (ITO) films on langasite (LGS)-based SAW reflective delay line sensor devices for monitoring hydrogen at 350 °C. Specifically, we modeled the effect of the IO and ITO sensing layer thickness on the wave velocity, attenuation, and effective electromechanical coefficient. This was followed by an experimental demonstration of tuning of the ITO film sheet conductivity by controlling the dopant concentration and yielding an improvement in the sensor sensitivity. The current study provides a pathway towards the development of conductivity-based sensing layers for high temperature SAW gas sensors with improved sensitivity.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
An Optimizing Compiler for Petascale I/O on Leadership Class Architectures
In high-performance computing systems, parallel I/O architectures usually have very complex hierarchies with multiple layers that collectively constitute an I/O stack, including high-level I/O libraries such as PnetCDF and HDF5, I/O middleware such as MPI-IO, and parallel file systems such as PVFS and Lustre. Our project explored automated instrumentation and compiler support for I/O intensive applications. Our project made significant progress towards understanding the complex I/O hierarchies of high-performance storage systems (including storage caches, HDDs, and SSDs), and designing and implementing state-of-the-art compiler/runtime system technology that targets I/O intensive HPC applications that target leadership class machine. This final report summarizes the major achievements of the project and also points out promising future directions.
97 MATHEMATICS AND COMPUTING↗