SEARCH · Search NASA
Results for “map task”
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.
Locality-aware and load-balanced static task scheduling for MapReduce
Task scheduling for MapReduce jobs has been an active area of research with the objective of decreasing the amount of data transferred during the shuffle phase via exploiting data locality. In the literature, generally only the scheduling of reduce tasks is considered with the assumption that scheduling of map tasks is already determined by the input data placement. However, in cloud or HPC deployments of MapReduce, the input data is located in a remote storage and scheduling map tasks gains importance. Here, we propose models for simultaneous scheduling of map and reduce tasks in order to improve data locality and balance the processors’ loads in both map and reduce phases. Our approach is based on graph and hypergraph models which correctly encode the interactions between map and reduce tasks. Partitions produced by these models are decoded to schedule map and reduce tasks. A two-constraint formulation utilized in these models enables balancing processors’ loads in both map and reduce phases. The partitioning objective in the hypergraph models correctly encapsulates the minimization of data transfer when a local combine step is performed prior to shuffle, whereas the partitioning objective in the graph models achieve the same feat when a local combine is not performed. Here, we show the validity of our scheduling on the MapReduce parallelizations of two important kernel operations – sparse matrix–vector multiplication (SpMV) and generalized sparse matrix–matrix multiplication (SpGEMM) – that are widely encountered in big data analytics and scientific computations. Compared to random scheduling, our models lead to tremendous savings in data transfer by reducing data traffic from several hundreds of megabytes to just a few megabytes in the shuffle phase and consequently leading up to 2.6x and 4.2x speedup for SpMV and SpGEMM, respectively.
Geometric Mapping of Tasks to Processors on Parallel Computers with Mesh or Torus Networks
We present a new method for reducing parallel applications' communication time by mapping their MPI tasks to processors in a way that lowers the distance messages travel and the amount of congestion in the network. Assuming geometric proximity among the tasks is a good approximation of their communication interdependence, we use a geometric partitioning algorithm to order both the tasks and the processors, assigning task parts to the corresponding processor parts. In this way, interdependent tasks are assigned to “nearby” cores in the network. We also present a number of algorithmic optimizations that exploit specific features of the network or application to further improve the quality of the mapping. We specifically address the case of sparse node allocation, where the nodes assigned to a job are not necessarily located in a contiguous block nor within close proximity to each other in the network. Furthermore, our methods generalize to contiguous allocations as well, and results are shown for both contiguous and non-contiguous allocations. We show that, for the structured finite difference mini-application MiniGhost, our mapping methods reduced communication time up to 75% relative to MiniGhost's default mapping on 128K cores of a Cray XK7 with sparse allocation. For the atmospheric modeling code E3SM/HOMME, our methods reduced communication time up to 31% on 16K cores of an IBM BlueGene/Q with contiguous allocation.
Progress in the scene-to-map registration task
An assessment of geometric accuracy of scene to map registration of LANDSAT MSS and TM sensor products is made. An outline of improved procedures for the registration and rectification of LANDSAT data is presented.
New Geologic Map of the Argyre Region of Mars: Deciphering the Geologic History Through Mars Global Surveyor, Mars Odyssey, and Mars Express Data
The primary objective of the mapping effort is to produce a geologic map of the Argyre basin and surrounding region at 1:5,000,000 scale in both digital and print formats that will detail the stratigraphic and crosscutting relations among rock materials and landforms (30 deg. S to 65 deg. S, 290 deg. E to 340 deg E). There has not been a detailed geologic map produced of the Argyre region since the Viking-era mapping investigation. The mapping tasks include stratigraphic mapping, crater counting, feature mapping, quantitative landform analysis, and spectroscopic/ stratigraphic investigation feature mapping. The regional geologic mapping investigation includes the Argyre basin floor and rim materials, the transition zone that straddles the Thaumasia plateau, which includes Argyre impactrelated modification, and the southeast margin of the Thaumasia plateau using important new data sets from the Mars Global Surveyor, Mars Odyssey, Mars Express, and Mars Reconnaissance Orbiter. The geologic information unfolded by this new mapping project will be useful to the community for constraining the regional geology, paleohydrology, and paleoclimate, which includes but is not limited to the assessment of: (1) whether the Argyre basin contained lakes, (2) the extent of reported flooding and glaciation, (3) existing interpretations of the origin of the narrow ridges located in the southeast part of the basin floor, and (4) the extent of Argyre-related tectonism and its influence on the surrounding regions.
The Effects of Long-Duration Spaceflight on Training Retention and Transfer
Training our crew members for long duration, exploration-class missions will have to maximize long-term retention and transfer of the trained skills. The expected duration of the missions, our inability to predict all the possible tasks the crew will be called upon to perform, and the low training-to-mission time ratio require that the training be maximally effective such that the skills acquired during training will be retained and will be transferrable across a wide range of specific tasks that are different from the particular tasks used during training. However, to be able to design training that can achieve these ambitious goals, we must first understand the ways in which long-duration spaceflight affects training retention and transfer. Current theories of training retention and transfer are largely based on experimental studies conducted at university laboratories using undergraduate students as participants. Furthermore, all such studies have been conducted on Earth. We do not know how well the results of these studies predict the performance of crew members. More specifically, we do not know how well the results of these studies predict the performance of crew members in space and especially during long-duration missions. To address this gap in our knowledge, the current on-going study seeks to test the null hypothesis that performance of university undergraduate students on Earth on training retention and transfer tests do in fact predict accurately the performance of crew members during long-duration spaceflights. To test this hypothesis, the study employs a single 16-month long experimental protocol with 3 different participant groups: undergraduate university students, crew members on the ground, and crew members in space. Results from this study will be presented upon its completion. This poster presents results of study trials of the two tasks used in this study: a data entry task and a mapping task. By researching established training principles, by examining future needs, and by using current practices in spaceflight training as test beds, this research project is mitigating program risks and generating templates and requirements to meet future training needs.
The Effects of Long-Duration Spaceflight on Training Retention and Transfer
Training our crew members for long duration, exploration-class missions will have to maximize long-term retention and transfer of the trained skills. The expected duration of the missions, our inability to predict all the possible tasks the crew will be called upon to perform, and the low training-to-mission time ratio require that the training be maximally effective such that the skills acquired during training will be retained and will be transferrable across a wide range of specific tasks that are different from the particular tasks used during training. However, to be able to design training that can achieve these ambitious goals, we must first understand the ways in which long-duration spaceflight affects training retention and transfer. Current theories of training retention and transfer are largely based on experimental studies conducted at university laboratories using undergraduate students as participants. Furthermore, all such studies have been conducted on Earth. We do not know how well the results of these studies predict the performance of crew members. More specifically, we do not know how well the results of these studies predict the performance of crew members in space and especially during long-duration missions. To address this gap in our knowledge, the current on-going study seeks to test the null hypothesis that performance of university undergraduate students on Earth on training retention and transfer tests do in fact predict accurately the performance of crew members during long-duration spaceflights. To test this hypothesis, the study employs a single 16-month long experimental protocol with 3 different participant groups: undergraduate university students, crew members on the ground, and crew members in space. Results from this study will be presented upon its completion. This poster presents results of study trials of the two tasks used in this study: a data entry task and a mapping task. By researching established training principles, by examining future needs, and by using current practices in spaceflight training as test beds, this research project is mitigating program risks and generating templates and requirements to meet future training needs.
Scalable multiscale modeling of platelets with 100 million particles
Here, we developed the core components of the AI-aided multiple time stepping algorithm for multiscale modeling of cell dynamics. This algorithm was implemented and analyzed on two supercomputer architectures with an application of simulating the aggregation of 250 platelets, or 102 million particles. To scale on these computers with complex memory and network architectures with GPUs, we devised a biomechanics-informed task mapping scheme to optimize load imbalance, communications, and memory utilization. Our simulations, scaling well up to 192 nodes on a Summit-like supercomputer with a peak speed of 11 petaflops, achieved a rate of 423 μs/day which is 500 times faster than the conventional algorithm using static time step and this has enabled studies of record size blood clots at record spatial–temporal resolutions. Additionally, we discovered the sensitive dependence of the scalability and execution time on the methods of decomposition, CPU–GPU coupling, and task mapping.
3D-ReG: A 3D ReRAM-based Heterogeneous Architecture for Training Deep Neural Networks
Deep neural network (DNN) models are being expanded to a broader range of applications. The computational capability of traditional hardware platforms cannot accommodate the growth of model complexity. Among recent technologies to accelerate DNN, resistive memory (ReRAM)-based processing-in-memory (PIM) emerged as a promising solution for DNN inference due to its high efficiency for matrix-based computation. We face two major technical challenges in extending the use of ReRAM-based accelerators for training: (1) full-precision data is essential in back-propagation; (2) the need to support both feed-forward and back-propagation aggravates the data-movement burden. We propose a heterogeneous architecture named as 3D-ReG, which leverages full-precision GPU to ensure training accuracy and low-overhead 3D integration to provide low-cost data movements. Moreover, we introduce conservative and aggressive task-mapping schemes, which partition the computation phases in different ways to balance execution efficiency and training accuracy. We evaluate 3D-ReG implemented with two 3D integration technologies, through-silicon vias (TSVs) and monolithic inter-tier vias (MIVs), and compare them with GPU-only and PIM-only counterparts. Various GPU-only platforms using two main-memory technologies (DRAM, ReRAM) and three interconnect technologies (2D, TSV, MIV) are evaluated as well. Experimental results show that 3D-ReG can achieve on average 5.64× training speedup and 3.56× higher energy efficiency compared with the GPU with DRAM as main memory, at the cost of 0.05%–3.39% accuracy drop. We define a new metric, gain-loss ratio (GLR), which quantitatively evaluates the capability of a DNN training hardware in terms of the model accuracy and hardware efficiency. The results of our comparison show that the aggressive task-mapping scheme on MIV-based 3D-ReG outperforms the other methods.
Machine processing of S-192 and supporting aircraft data: Studies of atmospheric effects, agricultural classifications, and land resource mapping
Two tasks of machine processing of S-192 multispectral scanner data are reviewed. In the first task, the effects of changing atmospheric and base altitude on the ability to machine-classify agricultural crops were investigated. A classifier and atmospheric effects simulation model was devised and its accuracy verified by comparison of its predicted results with S-192 processed results. In the second task, land resource maps of a mountainous area near Cripple Creek, Colorado were prepared from S-192 data collected on 4 August 1973.
Mission Adaptive UAS Platform for Earth Science Resource Assessment
NASA Ames Research Center has led a number of important Earth science remote sensing missions including several directed at the assessment of natural resources. A key asset for accessing high risk airspace has been the 180 kg class SIERRA UAS platform, providing mission durations of up to 8 hrs at altitudes up to 3 km. Recent improvements to this mission capability are embodied in the incipient SIERRA-B variant. Two resource mapping problems having unusual mission characteristics requiring a mission adaptive capability are explored here. One example involves the requirement for careful control over solar angle geometry for passive reflectance measurements. This challenges the management of resources in the coastal ocean where solar angle combines with sea state to produce surface glint that can obscure the ocean color signal. Furthermore, as for all scanning imager applications, the primary flight control priority to fly the UAS directly to the next waypoint should compromise with the requirement to minimize roll and crab effects in the imagery. A second example involves the mapping of natural resources in the Earth's crust using precision magnetometry. In this case the vehicle flight path must be oriented to optimize magnetic flux gradients over a spatial domain having continually emerging features, while optimizing the efficiency of the spatial mapping task. These requirements were highlighted in several recent Earth Science missions including the October 2013 OCEANIA mission directed at improving the capability for hyperspectral reflectance measurements in the coastal ocean, and the Surprise Valley Mission directed at mapping sub-surface mineral composition and faults, using high-sensitivity magentometry. This paper reports the development of specific aircraft control approaches to incorporate the unusual and demanding requirements to manage solar angle, aircraft attitude and flight path orientation, and efficient (directly geo-rectified) surface and sub-surface mapping, including the near-time optimization of these sometimes conflicting requirements. *
Automatic Halo Management for the Uintah GPU-Heterogeneous Asynchronous Many-Task Runtime
The Uintah computational framework is used for the parallel solution of partial differential equations on adaptive mesh refinement grids using modern supercomputers. Uintah is structured with an application layer and a separate runtime system. Uintah is based on a distributed directed acyclic graph of computational tasks, with a task scheduler that efficiently schedules and executes these tasks on both CPU cores and on-node accelerators. The runtime system identifies task dependencies, creates a task graph prior to the execution of these tasks, automatically generates MPI message tags, and automatically performs halo transfers for simulation variables. Automating halo transfers in a heterogeneous environment poses significant challenges when tasks compute within a few milliseconds, as runtime overhead affects wall time execution, or when simulation variables require large halos spanning most or all of the computational domain, as task dependencies become expensive to process. These challenges are magnified at production scale when application developers require each compute node perform thousands of different halo transfers among thousands simulation variables. The principal contribution of this work is to (1) identify and address inefficiencies that arise when mapping tasks onto the GPU in the presence of automated halo transfers, (2) implement new schemes to reduce runtime system overhead, (3) minimize application developer involvement with the runtime, and (4) show overhead reduction results from these improvements.
Optimized orbits for the constellation of Space Station polar platforms
The orbital configuration of the Eos polar-platform constellation is examined to determine if baseline orbital parameters are appropriately specified to meet the mission task of measuring global oceanic mesoscale variability. Using the Nyquist sampling theorem, spatial and temporal measurement requirements are first established for the constellation. Based upon these requirements, the optimum solution indicates that a total of five spacecraft equipped with nadir-pointing altimeters, or two multibeam-equipped platforms, would be required to meet the mesoscale mapping task. Additional analyses were performed on both one- and and two-platform configurations as well as the complete four-platform Eos constellation. The results of these calculations show that the preferred scenario would be to equip all four Eos polar platforms with multibeam altimeters.
On optimizing a constellation of altimetric satellites for measuring global oceanic mesoscale
The orbital configuration of the Eos polar-platform constellation is examined to determine if baseline orbital parameters are appropriately specified to meet the mission goal of measuring global oceanic mesoscale variability. Using the Nyquist sampling theorem, spatial and temporal measurement requirements are first established for the constellation. Based upon these requirements, the optimum solution indicates that a total of five spacecraft are required to meet the mesoscale mapping task. Additional analyses are performed on both a two-platform configuration as well as the complete four-platform Eos constellation. The results of these calculations show that an optimized configuration of four properly-phased Eos platforms, placed into identical 835-km sun-synchronous orbits with 13-day repeat periods, can very nearly meet statistically significant sampling criteria for mapping global oceanic mesoscale variability.
Frameshifting Stimulatory Sequence Induces Large Structural Change of Ribosomal Proteins When Bound to E. coli Ribosomes
Biological macromolecular machines occupy a continuum of structural conformations to perform cellular tasks. Mapping this conformational space provides an insight into its functionality. While the cryo-electron microscopy resolution revolution has expanded our ability to characterize the conformational continuums, there are obstacles in structurally characterizing regions of high flexibility. These technical barriers have impeded characterization of flexible ribosomal proteins when the ribosome is interacting with mRNA stem-loop structures such as a frameshifting stimulatory sequence (FSS). Small-angle neutron/X-ray scattering and electron microscopy were used to study ribosomal samples and compared structural differences between a ribosome that is bound to an FSS stem-loop compared to a ribosome bound to linear mRNA. This comparison shows that a large protein stalk elongates by 22% when the 70S interacts with an mRNA stem-loop. Finally, our results suggest that ribosomal proteins have extensive flexibility and may influence important ribosomal mechanisms, such as those that involve FSS.
IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library
This paper presents IRIS-BLAS, a novel heterogeneous and performance portable BLAS library. IRIS-BLAS is built on top of the IRIS runtime and multiple vendor and open-source BLAS libraries. It can transparently use all the architectures/devices available in a heterogeneous system, using the appropriate BLAS library based on the task mapping at run time. Thus, IRIS-BLAS is portable across a broad spectrum of architectures and BLAS libraries, alleviating the worry of application developers about modifying the application source code. Even though the emphasis is on portability, IRIS-BLAS provides competitive or even better performance than other state-of-the-art references. Moreover, IRIS-BLAS offers new features such as efficiently using extremely heterogeneous systems composed of multiple GPUs from different hardware vendors.
Towards Automated/Semiautomated Extraction of Faults from Lidar Data
The Pajarito fault system is a complex zone of deformation and a seismically active region nestled within the Rio Grande rift in north-central New Mexico. Numerous laterally discontinuous faults and associated folds and fractures interact in a manner that has important implications for seismic hazards and risk mitigation. Previous efforts have established a foundation for the location of lineaments and structures in the Pajarito fault system; however, ensuring the completeness of the current lineament mapping is required for identifying areas for field validation, evaluating the potential for future seismic activity, and better understanding fault interaction. Assistance with this fault-mapping task via automated or semiautomated techniques as applied to lidar data over a large area of interest is highly desirable. A proof-of-concept processing flow which transforms lidar point-cloud data into a raster of surficial fault candidates is described and illustrated herein. We report these initial results hold great promise toward achieving our ultimate goal.
Approaches to digital snow mapping with LANDSAT-1 data
Applying the same LANDSAT-1 data to three substantially different image processing systems, a snow mapping task was performed. LARSYS Ver.3, STANSORT-2, and General Electric Image-100 did all the jobs of detecting the snowline in forested mountainous terrain, and to determine the snowcovered area. While the control and accuracy achieved with LARSYS is remarkable, time and effort to perform the processing favor the systems STANSORT and Image-100. The experiences and results demonstrate the need for a fast interactive system for operational snowmapping with multispectral satellite data.