SEARCH · Engineering Papers
Results for “memory”
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.
Evaluating Online Global Recovery with Fenix Using Application-Aware In-Memory Checkpointing Techniques
Exascale systems promise the potential for computation at unprecedented scales and resolutions, but achieving exascale by the end of this decade presents significant challenges. A key challenge is due to the very large number of cores and components and the resulting mean time between failures (MTBF) in the order of hours or minutes. Since the typical run times of target scientific applications are longer than this MTBF, fault tolerance techniques will be essential. An important class of failures that must be addressed is process or node failures. While checkpoint/restart (C/R) is currently the most widely accepted technique for addressing processor failures, coordinated, stable-storage-based global C/R might be unfeasible at exascale when the time to checkpoint exceeds the expected MTBF. This paper explores transparent recovery via implicitly coordinated, diskless, application-driven checkpointing as a way to tolerate process failures in MPI applications at exascale. The discussed approach leverages User Level Failure Mitigation (ULFM), which is being proposed as an MPI extension to allow applications to create policies for tolerating process failures. Specifically, this paper demonstrates how different implementations of application-driven in-memory checkpoint storage and recovery compare in terms of performance and scalability. We also experimentally evaluate the effectiveness and scalability of the Fenix online global recovery framework on a production system -the Titan Cray XK7 at ORNL- and demonstrate the ability of Fenix to tolerate dynamically injected failures using the execution of four benchmarks and mini-applications with different behaviors.
Impact of Linearity and Write Noise of Analog Resistive Memory Devices in a Neural Algorithm Accelerator
Resistive memory (ReRAM) shows promise for use as an analog synapse element in energy-efficient neural network algorithm accelerators. A particularly important application is the training of neural networks, as this is the most computationally-intensive procedure in using a neural algorithm. However, training a network with analog ReRAM synapses can significantly reduce the accuracy at the algorithm level. In order to assess this degradation, analog properties of ReRAM devices were measured and hand-written digit recognition accuracy was modeled for the training using backpropagation. Bipolar filamentary devices utilizing three material systems were measured and compared: one oxygen vacancy system, Ta-TaO x , and two conducting metallization systems, Cu-SiO 2 , and Ag/chalcogenide. Analog properties and conductance ranges of the devices are optimized by measuring the response to varying voltage pulse characteristics. Key analog device properties which degrade the accuracy are update linearity and write noise. Write noise may improve as a function of device manufacturing maturity, but write nonlinearity appears relatively consistent among the different device material systems and is found to be the most significant factor affecting accuracy. As a result, this suggests that new materials and/or fundamentally different resistive switching mechanisms may be required to improve device linearity and achieve higher algorithm training accuracy.
MO-D-BRD-01: Memorial to Bengt Bjarngard - Memorial Lecture
We lost a legendary medical physicist, Dr. Bengt Erik Bjarngard, to angiosarcoma an aggressive type of cancer. He devoted his life to providing improved methods of radiation treatment for this devastating disease over the last 36 years. Bengt was born in a rural village of Bjarnum in southern Sweden, located near forest and is known for its furniture making. He migrated to USA at the age of 35 and was recruited by Dr. Samuel Hellman to lead a group of physicists that became the “mecca of medical physics” known as the Joint Center of Radiation Therapy (JCRT) at Harvard Medical School in Boston. Bengt mentored some of the best physicists in the country, and many of our modern treatments go back to the early days of research at the JCRT. These accomplishments, dating from 1969–1989, include: dose optimization using computer control; soft wedges; stereotactic radiosurgery (SRS); total-body irradiation (TBI); CT-planning; and radiation dosimetry. Bengt worked at Brown University in Rhode Island and at the University of Pennsylvania in Philadelphia, where he provided major contributions in radiation dosimetry, specifically with the head scatter model. He advocated superior calculation algorithm through the Helax treatment planning system that was on par from most commercial systems. Bengt served as AAPM president in 1979 and was a recipient of the Coolidge Award in 1998. He had a lifelong love of nature, retiring in 2000 from the University of Pennsylvania to take care of his 200 acres of homestead forest in Maine. His legacy continues through his contributions to radiation dosimetry. This session, on small field dosimetry, is a small tribute to his memory. Further details can be found in his obituary in Med Phy, 41(4), 040801, 2014.
Nanometer scale nonvolatile memory device and method for storing binary and quantum memory states
Example implementations include an electronic memory device with a metallic layer having a first planar crystalline structure, a first encapsulating layer including an encapsulating material having a second planar crystalline structure, and disposed adjacent to a first planar surface of the metallic layer, and a second encapsulating layer including the encapsulating material, and disposed adjacent to a second planar surface of the metallic layer. Example implementations also include a method of depositing graphite crystals onto a substrate to form a gate bottom layer, depositing BN crystals onto the graphite bottom layer to form a BN bottom layer, depositing tungsten ditelluride (WTe2) crystals onto the BN bottom layer to form a metallic layer, depositing the BN crystals onto the BN bottom layer and the metallic layer to form a BN top layer, and depositing the graphite crystals onto the BN top layer to form a gate top layer.
Fluorogel Elastomers with Tunable Transparency, Elasticity, Shape-Memory, and Antifouling Properties**
Omniphobic fluorogel elastomers were prepared by photocuring perfluorinated acrylates and a perfluoropolyether crosslinker. By tuning either the chemical composition or the temperature that control the crystallinity of the resulting polymer chains, a broad range of optical and mechanical properties of the fluorogel can be achieved. After infusing with fluorinated lubricants, the fluorogels showed excellent resistance to wetting by various liquids and anti-biofouling behavior, while maintaining cytocompatiblity.
Magnetic and magneto-transport studies of substrate effect on the martensitic transformation in a NiMnIn shape memory alloy
The effect of substrates on the magnetic and transport properties of Ni{sub 2}Mn{sub 1.5}In{sub 0.5} ultra-thin films were studied theoretically and experimentally. High quality 8-nm films were grown by laser-assisted molecular beam epitaxy deposition. Magneto-transport measurements revealed that the films undergo electronic structure transformation similar to those of bulk materials at the martensitic transformation. The temperature of the transformation depends strongly on lattice parameters of the substrate. To explain this behavior, we performed DFT calculations on the system and found that different substrates change the relative stability of the ferromagnetic (FM) austenite and ferrimagnetic (FiM) martensite states. We conclude that the energy difference between the FM austenite and FiM martensite states in Ni{sub 2}Mn{sub 1.5}In{sub 0.5} films grown on MgO (001) substrates is ΔE = 0.20 eV per NiMnIn f.u, somewhat lower compared to ΔE = 0.24 eV in the bulk material with the same lattice parameters. When the lattice parameters of Ni{sub 2}Mn{sub 1.5}In{sub 0.5} film have values close to those of the MgO substrate, the energy difference becomes ΔE = 0.08 eV per NiMnIn f.u. These results suggest the possibility to control the martensitic transition in thin films through substrate engineering.
HANDS ON WITH OPENMP4.5 and UNIFIED MEMORY: DEVELOPING APPLICATIONS FOR IBM'S HYBRID CPU+GPU SYSTEMS (PART II)
Abstract not provided
Mechanisms for phase transformation induced slip in shape memory alloy micro-crystals
Not Available
Element-specific amorphization of vacancy-ordered GeSbTe for ternary-state phase change memory
Not Available
Strong electron-polarized atom chain in amorphous phase-change memory Ge Sb Te alloy
Not Available
Tunability of martensitic transformation in Mg-Sc shape memory alloys: A DFT study
Not provided.
Heterogeneity and inelasticity of deformation in a notched martensitic NiTi shape memory alloy specimen
Not provided.
H-phase precipitation and its effects on martensitic transformation in NiTi-Hf high-temperature shape memory alloys
Not provided.