SEARCH · Search NASA
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.
The Kokkos library implements thread-parallel execution policies and shared-memory multidimensional array data structures that enable applications and domain libraries to develop computational kernels that are performance portable across multicore-CPU and manycore-accelerator (e.g. GPU) computing architectures.
Abstract Background Pattern matching is a key step in a variety of biological sequence analysis pipelines. The FM-index is a compressed data structure for pattern matching, with search run time that is independent of the length of the database text. Implementation of the FM-index is reasonably complicated, so that increased adoption will be aided by the availability of a fast and flexible FM-index library. Results We present AvxWindowedFMindex (AWFM-index), a lightweight, open-source, thread-parallel FM-index library written in C that is optimized for indexing nucleotide and amino acid sequences. AWFM-index introduces a new approach to storing FM-index data in a strided bit-vector format that enables extremely efficient computation of the FM-index occurrence function via AVX2 bitwise instructions, and combines this with optional on-disk storage of the index’s suffix array and a cache-efficient lookup table for partial k-mer searches. The AWFM-index performs exact match count and locate queries faster than SeqAn3’s FM-index implementation across a range of comparable memory footprints. When optimized for speed, AWFM-index is $$\sim $$ ∼ 2–4x faster than SeqAn3 for nucleotide search, and $$\sim $$ ∼ 2–6x faster for amino acid search; it is also $$\sim $$ ∼ 4x faster with similar memory footprint when storing the suffix array in on-disk SSD storage. Conclusions AWFM-index is easy to incorporate into bioinformatics software, offers run-time performance parameterization, and provides clients with FM-index functionality at both a high-level (count or locate all instances of a query string) and low-level (step-wise control of the FM-index backward-search process). The open-source library is available for download at https://github.com/TravisWheelerLab/AvxWindowFmIndex.
family of schemes has been devised for organizing the output of an algorithm for predictive data compression of hyperspectral imagery so as to allow efficient parallelization in both the compressor and decompressor. In these schemes, the compressor performs a number of iterations, during each of which a portion of the data is compressed via parallel threads operating on independent portions of the data. The general idea is that for each iteration it is predetermined how much compressed data will be produced from each thread.
Background Oriented Schlieren (BOS) is a widely used technique that provides density gradient information in flow fields of interest, without imposing stringent optical quality requirements on the facility/experiment windows and/or optics used in the BOS setup. Typically, the BOS reference image is acquired before the test begins (flow off) and then the "live" image data are acquired during the actual testing/experiment (flow on). The raw BOS image data, while displayed in real-time as they are acquired from the camera, unfortunately provide little if any visual indication of the density gradients in the flow. Generally, the "live" images must be processed off-line after the testing is completed, providing no indication of the success of the BOS setup and no feedback on the operational success of the test. Advances in computer processing hardware enables the implementation of real-time processing and display of the BOS image data. Two different approaches to implementing the real-time BOS (RT-BOS) processing capability are described herein. First, a traditional multi-core Central Processing Unit (CPU) based approach using scheduled parallel threads is used to build a RT-BOS processing engine. In the second approach, a Graphical Processing Unit (GPU) approach is used to costruct a RT-BOS processing engine. Generally, high core count CPU processors can provide a useful processing rate for RT-BOS. However, the GPU based approach exceeds the processing capability of the CPU approach, at a fraction of the cost. The GPU approach places no restrictions on the Host PC processing capability, except that it be capable of acquiring the BOS image data from the camera in real-time.
We review our work done to optimize the staggered conjugate gradient (CG) algorithm in the MILC code for use with the Intel Knights Landing (KNL) architecture. KNL is the second gener- ation Intel Xeon Phi processor. It is capable of massive thread parallelism, data parallelism, and high on-board memory bandwidth and is being adopted in supercomputing centers for scientific research. The CG solver consumes the majority of time in production running, so we have spent most of our effort on it. We compare performance of an MPI+OpenMP baseline version of the MILC code with a version incorporating the QPhiX staggered CG solver, for both one-node and multi-node runs.
The software SAFIRE implements a fault injection tool to evaluate the impact of soft errors on parallel, multi-threaded applications. It implements instruction-level fault models that emulate soft errors by flipping bits in the register operands of executed machine instructions. SAFIRE is based on the LLVM compiler and utilizes novel techniques to optimize instruction instrumentation and fault injection. By detailed measurements, SAFIRE is multiple times faster than other state-of-the-art fault injection techniques, while having the same accuracy.
Abstract not provided.
A canonical quantization procedure is applied to the interaction of elastic waves—phonons—with infinitely long dislocations that can oscillate about an equilibrium, straight line, configuration. The interaction is implemented through the well-known Peach–Koehler force. For small dislocation excursions away from the equilibrium position, the quantum theory can be solved to all orders in the coupling constant. We study in detail the quantum excitations of the dislocation line and its interactions with phonons. The consequences for the drag on a dislocation caused by the phonon wind are pointed out. We compute the cross-section for phonons incident on the dislocation lines for an arbitrary angle of incidence. The consequences for thermal transport are explored, and we compare our results, involving a dynamic dislocation, with those of Klemens and Carruthers, involving a static dislocation. In our case, the relaxation time is inversely proportional to frequency, rather than directly proportional to frequency. As a consequence, the thermal transport anisotropy generated on a material by the presence of a highly-oriented array of dislocations is considerably more sensitive to the frequency of each propagating mode, and, therefore, to the temperature of the material.
Development of the Program to Optimize Simulated Trajectories (POST) began in the 1970s. Since then, it has become widely utilized across NASA, industry, and academia to solve a variety of atmospheric ascent and entry problems. Its successor, POST2, has undergone many upgrades since its release in the 1990s. Recently, there has been an increasing desire to take advantage of the advances in parallel computing for both offline and online systems. Thus, modifications were made to allow POST2 to simulate multiple trajectories simultaneously without adversely affecting results. This capability is leveraged to calculate optimization solutions in parallel as opposed to sequentially. A demonstration of the benefits is presented using a small set of POST2 regression tests, as well as a project simulating a human-scale Lunar lander.
Development of the Program to Optimize Simulated Trajectories (POST) began in the 1970s. Since then, it has become widely utilized across NASA, industry, and academia to solve a variety of atmospheric ascent and entry problems. Its successor, POST2, has undergone many upgrades since its release in the 1990s. Recently, there has been an increasing desire to take advantage of the advances in parallel computing for both offline and online systems. Thus, modifications were made to allow POST2 to simulate multiple trajectories simultaneously without adversely affecting results. This capability is leveraged to calculate optimization solutions in parallel as opposed to sequentially. A demonstration of the benefits is presented using a small set of POST2 regression tests, as well as a project simulating a human-scale Lunar lander.
High-level parallel programming models (PMs) are becoming crucial in order to extract the computational power of current on-node multi-threaded parallelism. The most popular PMs, such as OpenMP or OmpSs, are directive-based: the complexity of the hardware is hidden by the underlying runtime system, improving coding productivity. The implementations of OpenMP usually rely on POSIX threads (pthreads), offering excellent performance for coarse-grained parallelism and a perfect match with the current hardware. OmpSs is a task oriented PM based on an ad hoc runtime solution called Nanos++; it is the precursor of the tasking parallelism in the OpenMP tasking specification. A recent trend in runtimes and applications points to leveraging massive on-node parallelism in conjunction with fine-grained and dynamic scheduling paradigms. In this paper we analyze the behavior of the OpenMP and OmpSs PMs on top of the recently emerged Generic Lightweight Threads (GLT) API. GLT exposses a common API for lightweight thread (LWT) libraries that offers the possibility of running the same application over different native LWT solutions. We describe the design details of those high-level PMs implemented on top of GLT and analyze different scenarios in order to assess where the use of LWTs may benefit application performance. Our work reveals those scenarios where LWTs overperform pthread-based solutions and compares the performance between an ad hoc solution and a generic implementation.
This paper describes an approach that seeks to parallelize the spatial search associated with computational contact mechanics. In contact mechanics, the purpose of the spatial search is to find “nearest neighbors,” which is the prelude to an imprinting search that resolves the interactions between the external surfaces of contacting bodies. In particular, we are interested in the contact global search portion of the spatial search associated with this operation on domain-decomposition-based meshes. Specifically, we describe an implementation that combines standard domain-decomposition-based MPI-parallel spatial search with thread-level parallelism (MPI-X) available on advanced computer architectures (those with GPU coprocessors). Our goal is to demonstrate the efficacy of the MPI-X paradigm in the overall contact search. Standard MPI-parallel implementations typically use a domain decomposition of the external surfaces of bodies within the domain in an attempt to efficiently distribute computational work. This decomposition may or may not be the same as the volume decomposition associated with the host physics. The parallel contact global search phase is then employed to find and distribute surface entities (nodes and faces) that are needed to compute contact constraints between entities owned by different MPI ranks without further inter-rank communication. Key steps of the contact global search include computing bounding boxes, building surface entity (node and face) search trees and finding and distributing entities required to complete on-rank (local) spatial searches. To enable source-code portability and performance across a variety of different computer architectures, we implemented the algorithm using the Kokkos hardware abstraction library. While we targeted development towards machines with a GPU accelerator per MPI rank, we also report performance results for OpenMP with a conventional multi-core compute node per rank. Results here demonstrate a 47 % decrease in the time spent within the global search algorithm, comparing the reference ACME algorithm with the GPU implementation, on an 18M face problem using four MPI ranks. As a result, while further work remains to maximize performance on the GPU, this result illustrates the potential of the proposed implementation.
An ionization front (IF) surrounding an H II region is a sharp interface where a cold neutral gas makes the transition to a warm ionized phase by absorbing UV photons from central stars. We investigate the instability of a plane-parallel D-type IF threaded by parallel magnetic fields, by neglecting the effects of recombination within the ionized gas. We find that weak D-type IFs always have the post-IF magnetosonic Mach number M{sub M2}≤1. For such fronts, magnetic fields increase the maximum propagation speed of the IFs, while reducing the expansion factor α by a factor of 1 + 1/(2β{sub 1}) compared to the unmagnetized case, with β{sub 1} denoting the plasma beta in the pre-IF region. IFs become unstable to distortional perturbations owing to gas expansion across the fronts, exactly analogous to the Darrieus-Landau instability of ablation fronts in terrestrial flames. The growth rate of the IF instability is proportional linearly to the perturbation wavenumber, as well as the upstream flow speed, and approximately to α{sup 1/2}. The IF instability is stabilized by gas compressibility and becomes completely quenched when the front is D-critical. The instability is also stabilized by magnetic pressure when the perturbations propagate in the direction perpendicular to the fields. When the perturbations propagate in the direction parallel to the fields, on the other hand, it is magnetic tension that reduces the growth rate, completely suppressing the instability when M{sub M2}{sup 2}<2/(2β{sub 1}−1). When the front experiences an acceleration, the IF instability cooperates with the Rayleigh-Taylor instability to make the front more unstable.
A current thread in parallel computation is the use of cluster computers created by networking a few to thousands of commodity general-purpose workstation-level commuters using the Linux operating system. For example on the Medusa cluster at NASA/GSFC, this provides for super computing performance, 130 G(sub flops) (Linpack Benchmark) at moderate cost, $370K. However, to be useful for scientific computing in the area of Earth science, issues of ease of programming, access to existing scientific libraries, and portability of existing code need to be considered. In this paper, I address these issues in the context of tools for rendering earth science remote sensing data into useful products. In particular, I focus on a problem that can be decomposed into a set of independent tasks, which on a serial computer would be performed sequentially, but with a cluster computer can be performed in parallel, giving an obvious speedup. To make the ideas concrete, I consider the problem of classifying hyperspectral imagery where some ground truth is available to train the classifier. In particular I will use the Support Vector Machine (SVM) approach as applied to hyperspectral imagery. The approach will be to introduce notions about parallel computation and then to restrict the development to the SVM problem. Pseudocode (an outline of the computation) will be described and then details specific to the implementation will be given. Then timing results will be reported to show what speedups are possible using parallel computation. The paper will close with a discussion of the results.
We present preliminary results of the Roofline Toolkit for multicore, many core, and accelerated architectures. This paper focuses on the processor architecture characterization engine, a collection of portable instrumented micro benchmarks implemented with Message Passing Interface (MPI), and OpenMP used to express thread-level parallelism. These benchmarks are specialized to quantify the behavior of different architectural features. Compared to previous work on performance characterization, these microbenchmarks focus on capturing the performance of each level of the memory hierarchy, along with thread-level parallelism, instruction-level parallelism and explicit SIMD parallelism, measured in the context of the compilers and run-time environments. We also measure sustained PCIe throughput with four GPU memory managed mechanisms. By combining results from the architecture characterization with the Roofline model based solely on architectural specifications, this work offers insights for performance prediction of current and future architectures and their software systems. To that end, we instrument three applications and plot their resultant performance on the corresponding Roofline model when run on a Blue Gene/Q architecture.
Optimal scheduling of air traffic over the entire National Airspace System is a computationally difficult task. To speed computation, Dantzig-Wolfe decomposition is applied to a known linear integer programming approach for assigning delays to flights. The optimization model is proven to have the block-angular structure necessary for Dantzig-Wolfe decomposition. The subproblems for this decomposition are solved in parallel via independent computation threads. Experimental evidence suggests that as the number of subproblems/threads increases (and their respective sizes decrease), the solution quality, convergence, and runtime improve. A demonstration of this is provided by using one flight per subproblem, which is the finest possible decomposition. This results in thousands of subproblems and associated computation threads. This massively parallel approach is compared to one with few threads and to standard (non-decomposed) approaches in terms of solution quality and runtime. Since this method generally provides a non-integral (relaxed) solution to the original optimization problem, two heuristics are developed to generate an integral solution. Dantzig-Wolfe followed by these heuristics can provide a near-optimal (sometimes optimal) solution to the original problem hundreds of times faster than standard (non-decomposed) approaches. In addition, when massive decomposition is employed, the solution is shown to be more likely integral, which obviates the need for an integerization step. These results indicate that nationwide, real-time, high fidelity, optimal traffic flow scheduling is achievable for (at least) 3 hour planning horizons.
Supercomputing applications are increasingly adopting the MPI+threads programming model over the traditional "MPI everywhere" approach to better handle the disproportionate increase in the number of cores compared with other on-node resources. In practice, however, most applications observe a slower performance with MPI+threads primarily because of poor communication performance. Recent research efforts on MPI libraries address this bottleneck by mapping logically parallel communication, that is, operations that are not subject to MPI's ordering constraints to the underlying network parallelism. Domain scientists, however, typically do not expose such communication independence information because the existing MPI-3.1 standard's semantics can be limiting. Researchers had initially proposed user-visible endpoints to combat this issue, but such a solution requires intrusive changes to the standard (new APIs). The upcoming MPI-4.0 standard, on the other hand, allows applications to relax unneeded semantics and provides them with many opportunities to express logical communication parallelism. Here in this article, we show how MPI+threads applications can achieve high performance with logically parallel communication. Through application case studies, we compare the capabilities of the new MPI-4.0 standard with those of the existing one and user-visible endpoints (upper bound). Logical communication parallelism can boost the overall performance of an application by over 2x.
Massive multi-threading in GPU imposes tremendous pressure on memory subsystems. Due to rapid growth in thread-level parallelism of GPU and slowly improved peak memory bandwidth, memory becomes a bottleneck of GPU’s performance and energy efficiency. In this article, we propose an integrated architectural scheme to optimize the memory accesses and therefore boost the performance and energy efficiency of GPU. First, we propose a thread batch enabled memory partitioning (TEMP) to improve GPU memory access parallelism. In particular, TEMP groups multiple thread blocks that share the same set of pages into a thread batch and applies a page coloring mechanism to bound each stream multiprocessor (SM) to the dedicated memory banks. After that, TEMP dispatches the thread batch to an SM to ensure high-parallel memory-access streaming from the different thread blocks. Second, a thread batch-aware scheduling (TBAS) scheme is introduced to improve the GPU memory access locality and to reduce the contention on memory controllers and interconnection networks. Experimental results show that the integration of TEMP and TBAS can achieve up to 10.3% performance improvement and 11.3% DRAM energy reduction across diverse GPU applications. We also evaluate the performance interference of the mixed CPU+GPU workloads when they are run on a heterogeneous system that employs our proposed schemes. Our results show that a simple solution can effectively ensure the efficient execution of both GPU and CPU applications.