Engineering Papers⌕ Search

Engineering topics

Morris, K.

Publications and source records attributed to Morris, K..

8 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2017

Exploring the interplay of resilience and energy consumption for a task-based partial differential equations preconditioner

Here, we discuss algorithm-based resilience to silent data corruptions (SDCs) in a task-based domain-decomposition preconditioner for partial differential equations (PDEs). The algorithm exploits a reformulation of the PDE as a sampling problem, followed by a solution update through data manipulation that is resilient to SDCs. The implementation is based on a server-client model where all state information is held by the servers, while clients are designed solely as computational units. Scalability tests run up to ~51K cores show a parallel efficiency greater than 90%. We use a 2D elliptic PDE and a fault model based on random single and double bit-flip to demonstrate the resilience of the application to synthetically injected SDC. We discuss two fault scenarios: one based on the corruption of all data of a target task, and the other involving the corruption of a single data point. We show that for our application, given the test problem considered, a four-fold increase in the number of faults only yields a 2% change in the overhead to overcome their presence, from 7% to 9%. We then discuss potential savings in energy consumption via dynamic voltage/frequency scaling, and its interplay with fault-rates, and application overhead.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2017

Partial differential equations preconditioner resilient to soft and hard faults

We present a domain-decomposition-based preconditioner for the solution of partial differential equations (PDEs) that is resilient to both soft and hard faults. The algorithm reformulates the PDE as a sampling problem, followed by a solution update through data manipulation that is resilient to both soft and hard faults. This reformulation allows us to recast the problem as a set of independent tasks, and exploit data locality to reduce global communication. We discuss two different parallel implementations: (a) a single program multiple data (SPMD) version based on a one-to-one mapping between subdomain and MPI processes responsible for both state and computation; and (b) an asynchronous server–client implementation where all state information is held by the servers and clients are designed solely as computational units. We present a scalability comparison of both implementations under nominal conditions, showing efficiency within ~80% for up to 12,000 cores. We present a resilience analysis under different fault scenarios based on the server–client implementation. This framework provides resiliency to hard faults such that if a client crashes, it stops asking for work, and the servers simply distribute the work among all of the other clients alive. Erroneous subdomain solves (e.g. due to soft faults) appear as corrupted data, which is either rejected if that causes a task to fail, or is seamlessly filtered out during the regression stage through a suitable noise model. Three different types of faults are modeled: hard faults modeling nodes (or clients) crashing; soft faults occurring during the communication of the tasks between server and clients; and soft faults occurring during task execution. We demonstrate the resiliency of the approach for a 2D elliptic PDE, and explore the effect of the faults at various failure rates.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2014

Wide-angle ITER-prototype tangential infrared and visible viewing system for DIII-D

An imaging system with a wide-angle tangential view of the full poloidal cross-section of the tokamak in simultaneous infrared and visible light has been installed on DIII-D. The optical train includes three polished stainless steel mirrors in vacuum, which view the tokamak through an aperture in the first mirror, similar to the design concept proposed for ITER. A dichroic beam splitter outside the vacuum separates visible and infrared (IR) light. Spatial calibration is accomplished by warping a CAD-rendered image to align with landmarks in a data image. The IR camera provides scrape-off layer heat flux profile deposition features in diverted and inner-wall-limited plasmas, such as heat flux reduction in pumped radiative divertor shots. Demonstration of the system to date includes observation of fast-ion losses to the outer wall during neutral beam injection, and shows reduced peak wall heat loading with disruption mitigation by injection of a massive gas puff.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
NASA NTRS2010

RDBE Development and Progress

A digital backend based on the ROACH board has been developed jointly by the National Radio Astronomy Observatory and MIT Haystack Observatory. The RDBE will have both Polyphase Filterbank and Digital Downconverter personalities. The initial configuration outputs sixteen 32-MHz channels, comprised of half the channels from the PFB processing of the two IF inputs, for use in the VLBI2010 geodetic system and in the VLBA sensitivity upgrade project. The output rate is 2x109 bits/second (1x10(exp 9) bits/sec = 1 Gbps) over a 10 GigE connection to the Mark 5C with the data written in Mark 5B format on disk.

Niell, A.↗
NASA NTRS1995

Ice Processes and Growth History on Arctic and Sub-Arctic Lakes Using ERS-1 SAR Data

A survey of ice growth and decay processes on a selection of shallow and deep sub-Arctic and Arctic lakes was conducted using radiometrically calibrated ERS-1 SAR images. Time series of radar backscatter data were compiled for selected sites on the lakes during the period ot ice cover (September to June) for the years 1991-1992 and 1992-1993. A variety of lake-ice processes could be observed, and significant changes in backscatter occurred from the time of initial ice formation in autumn until the onset of the spring thaw. Backscatter also varied according to the location and depth of the lakes. The spatial and temporal changes in backscatter were most constant and predictable at the shallow lakes on the North Slope of Alaska. As a consequence, they represent the most promising sites for long-term monitoring and the detection of changes related to global warming and its effects on the polar regions.

Morris, K.↗
NASA NTRS1994

Structural and stratigraphic features and ERS 1 synthetic aperture radar backscatter characteristics of ice growing on shallow lakes in NW Alaska, winter 1991-1992

Changes in Earth Remote-Sensing Satellite (ERS) 1 C band synthetic aperture radar (SAR) backscatter intensity (sigma(exp 0)) from ice growing on shallow tundra lakes at three locations in NW Alaska are described. Ice core analysis shows that all lakes on the coast at Barrow the ice, whether floating or frozen to the bottom, includes an inclusion-free layer overlying a layer of ice with tubular bubbles oriented parallel to the direction of growth. The clear ice may also be overlain by a discontinuous layer of bubbly snow ice. Backscatter is low (-16 to -22 dB) at the time of initial ice formation, probably due to the specular nature of the upper and lower ice surfaces causing the radar pulse to be reflected away from the radar. As the ice thickens during the autumn, backscatter rises steadily. Once the ice freezes to the lake bottom, regardless of the presence of foward scattering tubular bubbles, low backscatter values of -17 to -18 dB are caused by absorption of the radar signal in the lake bed. For ice that remains afloat all winter the ice-water interface and the tubular bubbles combine, presumably via an incoherent double-bounce mechanism, to cause maximum backscatter values of the order of -6 to -7 dB. The sigma(exp 0) saturates at -6 to -7 dB before maximum ice thickness and tubular bubble content are attained. A simple ice growth model suggests that the layer of ice with tubular bubbles need be only a few centimeters thick midway through the growth season to cause maximum backscatter from floating ice. During the spring thaw a previously unreported backscatter reversal is observed on the floating and grounded portions of the coastal lakes but not on the lakes farther inland. This reversal may be related to the ice surface topography and wetness plus the effects of a longer, cooler melt period by the coast. Time series of backscatter variations from shallow tundra lakes are a record of (1) the development of tubular bubbles in the ice and, by association, changes in the gas content of the underlying water and (2) the freezing of ice to the bottoms of the lakes and therefore lake bathymetry and water availability. SAR is also able to detect the onset of lake ice growth in autumn and the initiation of the spring thaw and thus has potential for monitoring high-altitude lake ice growth and decay processes in relation to climate variability.

Jeffries, M. O.↗
NASA NTRS1993

Backscatter from ice growing on shallow tundra lakes near Barrow, Alaska, winter 1991-1992

The timing of freeze-up and break-up of Arctic lake ice is a potentially useful environmental indicator that could be monitored using SAR. In order to do this, it is important to understand how the properties and structure of the ice during its growth and decay affect radar backscatter and thus lake ice SAR signatures. The availability of radiometrically and geometrically calibrated digital SAR data time series from the Alaska SAR Facility has made it possible for the first time to quantify lake ice backscatter intensity (sigma(sup o)) variations. This has been done for ice growing on shallow tundra lakes near Barrow, NW Alaska, from initial growth in September 1991 until thawing and decay in June 1992. Field and laboratory observations and measurements of the lake ice were made in late April 1992. The field investigations of the coastal lakes near Barrow confirmed previous findings that, (1) ice frozen to the lake bottom had a dark signature in SAR images, indicating weak backscatter, while, (2) ice that was floating had a bright signature, indicating strong backscatter. At all sites, regardless of whether the ice was grounded or floating, there was a layer of clear, inclusion-free ice overlaying a layer of ice with dense concentrations of vertically oriented tubular bubbles. At some sites, there was a third layer of porous, snow-ice overlaying the clear ice.

Jeffries, M. O.↗