Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallel optimization”

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.

At least 1,747 records · Page 97

Enabling Highly Efficient Capsule Networks Processing Through A PIM-Based Architecture Design

In recent years, the CNNs have achieved great successes in the image processing tasks, e.g., image recognition and object detection. Unfortunately, traditional CNN's classication is found to be easily misled by increasingly complex image features due to the usage of pooling operations, hence unable to preserve accurate position and pose information of the objects. To address this challenge, a novel neural network structure called Capsule Network has been proposed, which introduces equivariance through capsules to signicantly enhance the learning ability for image segmentation and object detection. Due to its requirement of performing a high volume of matrix operations, CapsNets have been generally accelerated on modern GPU platforms that provide highly optimized software library for common deep learning tasks. However, based on our performance characterization on modern GPUs, CapsNets exhibit low effciency due to the special program and execution features of their routing procedure, including massive unshareable intermediate variables and intensive syn- chronizations, which are very dicult to optimize at software level. To address these challenges, we propose a hybrid computing architecture design named PIM-CapsNet. It preserves GPU's on-chip computing capability for accelerating CNN types of layers in CapsNet, while pipelining with an off-chip in-memory acceleration solution that effectively tackles routing procedure's ineffciency by leveraging the processing-in-memory capability of today's 3D stacked memory. Using routing procedure's inherent parallellization feature, our design enables hierarchical improvements on CapsNet inference effciency through minimizing data movement and maximizing parallel processing in memory. Evaluation results demonstrate that our proposed design can achieve substantial improvement on both performance and energy savings for CapsNet inference, with almost zero accuracy loss. The results also suggest good performance scalability in optimizing the routing procedure with increasing network size.

Zhang, Xingyao↗

Comparing the Performance of Julia on CPUs versus GPUs and Julia-MPI versus Fortran-MPI: a case study with MPAS-Ocean (Version 7.1)

Abstract. Some programming languages are easy to develop at the cost of slow execution, while others are fast at runtime but much more difficult to write. Julia is a programming language that aims to be the best of both worlds – a development and production language at the same time. To test Julia's utility in scientific high-performance computing (HPC), we built an unstructured-mesh shallow water model in Julia and compared it against an established Fortran-MPI ocean model, the Model for Prediction Across Scales–Ocean (MPAS-Ocean), as well as a Python shallow water code. Three versions of the Julia shallow water code were created: for single-core CPU, graphics processing unit (GPU), and Message Passing Interface (MPI) CPU clusters. Comparing identical simulations revealed that our first version of the Julia model was 13 times faster than Python using NumPy, where both used an unthreaded single-core CPU. Further Julia optimizations, including static typing and removing implicit memory allocations, provided an additional 10–20× speed-up of the single-core CPU Julia model. The GPU-accelerated Julia code was almost identical in terms of performance to the MPI parallelized code on 64 processes, an unexpected result for such different architectures. Parallelized Julia-MPI performance was identical to Fortran-MPI MPAS-Ocean for low processor counts and ranges from 2× faster to 2× slower for higher processor counts. Our experience is that Julia development is fast and convenient for prototyping but that Julia requires further investment and expertise to be competitive with compiled codes. We provide advice on Julia code optimization for HPC systems.

54 ENVIRONMENTAL SCIENCES↗

A direct transcription-based multiple shooting formulation for dynamic optimization

The growing need for fast and efficient solution techniques for solving dynamic optimization problems is driven by a broad spectrum of applications in scheduling and control. We suggest a novel framework for dynamic optimization that utilizes a multiple shooting “backbone” with discrete rather than continuous subproblems, thereby eliminating need for repeated time-integration. A Lagrangian relaxation (LR)-based decomposition scheme is proposed, which dualizes the state continuity requirements between subproblems and enables parallel solution of the problem. We demonstrate the applicability of our method on two case studies: the Van der Pol oscillator and a batch reactor.

42 ENGINEERING↗

Optimizing the FEDVR-TDCC Code for Exploring the Quantum Dynamics of Two-Electron Systems in Intense Laser Pulses

To efficiently solve the three-dimensional (3D) time-dependent linear and nonlinear Schrödinger equation, we have developed a large-scale parallel code RSP-FEDVR [B. I. Schneider, L. A. Collins, and S. X. Hu, Phys. Rev. E 73, 036708 (2006)], which combines the finite-element discrete variable representation (FEDVR) with the real-space product algorithm. Using the similar algorithm, we have derived an accurate approach to solve the time-dependent close-coupling (TDCC) equation for exploring two-electron dynamics in linearly polarized intense laser pulses. However, when the number N of partial waves used for the TDCC expansion increases, the FEDVR-TDCC code unfortunately slows down, because the potential-matrix operation scales as ~O(N^2). In this paper, we show that the full potential-matrix operation can be decomposed into a series of small-matrix operations utilizing the sparse property of the N N potential matrix. Such optimization speeds up the FEDVR-TDCC code by an order of magnitude for N=256. This may facilitate the ultimate solution to the 3D two-electron quantum dynamics in ultrashort intense optical laser pulses, where a large number of partial waves are required.

Hu, S X↗

Parallel-in-Time Multigrid Methods for Hyperbolic Problems, with a Focus on the Shallow Water Equations

The coming massive parallelism of exascale computing presents a pressing challenge for the many DOE simulations of time-dependent partial differential equations, which typically use traditional sequential time stepping methods. Since this traditional approach is inherently serial, it presents a sequential bottleneck when moving to exascale computing, because future performance gains will come through greater concurrency, not faster clock speeds. Thus, the goal of this work is to research parallelism in time, i.e., methods that compute multiple time values simultaneously, not sequentially. The focus will be on hyperbolic problems of programmatic interest to DOE, with the goal of enabling scalable simulations of time-dependent hyperbolic problems on future architectures. The difficulty lies in the fact that hyperbolic problems are well-known to be difficult for parallel-in-time methods, with the most common parallel-in-time method, Parareal, diverging in many cases. Here, the chosen methodology for scalably solving these hyperbolic space-time equations is multigrid-reductionin-time (MGRIT), because multigrid (when it works) is a powerful, optimal, and scalable solver for discretized PDEs. To further research in this area, this project will explore a model hyperbolic problem (the shallow water equations) in the context of recent advances (e.g., by Wingate and Haut) in constructing improved coarse time-grid time-propagators by using the slow asymptotic structure of the equations. In particular, we will investigate if such coarse time-propagators offer significant advantages in an MGRIT setting, and explore any broader insights gained into parallel-in-time for hyperbolic problems.

97 MATHEMATICS AND COMPUTING↗

Explicit spatial scattering for load balancing in conservatively synchronized parallel discrete-event simulations

We re-examine the problem of load balancing in conservatively synchronized parallel, discrete-event simulations executed on high-performance computing clusters, focusing on simulations where computational and messaging load tend to be spatially clustered. Such domains are frequently characterized by the presence of geographic 'hot-spots' - regions that generate significantly more simulation events than others. Examples of such domains include simulation of urban regions, transportation networks and networks where interaction between entities is often constrained by physical proximity. Noting that in conservatively synchronized parallel simulations, the speed of execution of the simulation is determined by the slowest (i.e most heavily loaded) simulation process, we study different partitioning strategies in achieving equitable processor-load distribution in domains with spatially clustered load. In particular, we study the effectiveness of partitioning via spatial scattering to achieve optimal load balance. In this partitioning technique, nearby entities are explicitly assigned to different processors, thereby scattering the load across the cluster. This is motivated by two observations, namely, (i) since load is spatially clustered, spatial scattering should, intuitively, spread the load across the compute cluster, and (ii) in parallel simulations, equitable distribution of CPU load is a greater determinant of execution speed than message passing overhead. Through large-scale simulation experiments - both of abstracted and real simulation models - we observe that scatter partitioning, even with its greatly increased messaging overhead, significantly outperforms more conventional spatial partitioning techniques that seek to reduce messaging overhead. Further, even if hot-spots change over the course of the simulation, if the underlying feature of spatial clustering is retained, load continues to be balanced with spatial scattering leading us to the observation that spatial scattering can often obviate the need for dynamic load balancing.

97 MATHEMATICS AND COMPUTING↗

An efficient mixed-precision, hybrid CPU-GPU implementation of a nonlinearly implicit one-dimensional particle-in-cell algorithm

Recently, a fully implicit, energy- and charge-conserving particle-in-cell method has been developed for multi-scale, full-f kinetic simulations [G. Chen, et al., J. Comput. Phys. 230, 18 (2011)]. The method employs a Jacobian-free Newton-Krylov (JFNK) solver and is capable of using very large timesteps without loss of numerical stability or accuracy. A fundamental feature of the method is the segregation of particle orbit integrations from the field solver, while remaining fully self-consistent. This provides great flexibility, and dramatically improves the solver efficiency by reducing the degrees of freedom of the associated nonlinear system. However, it requires a particle push per nonlinear residual evaluation, which makes the particle push the most time-consuming operation in the algorithm. This paper describes a very efficient mixed-precision, hybrid CPU-GPU implementation of the implicit PIC algorithm. The JFNK solver is kept on the CPU (in double precision), while the inherent data parallelism of the particle mover is exploited by implementing it in single-precision on a graphics processing unit (GPU) using CUDA. Performance-oriented optimizations, with the aid of an analytical performance model, the roofline model, are employed. Despite being highly dynamic, the adaptive, charge-conserving particle mover algorithm achieves up to 300 400 GOp/s (including single-precision floating-point, integer, and logic operations) on a Nvidia GeForce GTX580, corresponding to 20 25% absolute GPU efficiency (against the peak theoretical performance) and 50-70% intrinsic efficiency (against the algorithm s maximum operational throughput, which neglects all latencies). This is about 200-300 times faster than an equivalent serial CPU implementation. When the single-precision GPU particle mover is combined with a double-precision CPU JFNK field solver, overall performance gains 100 vs. the double-precision CPU-only serial version are obtained, with no apparent loss of robustness or accuracy when applied to a challenging long-time scale ion acoustic wave simulation.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Implementation of High-Order Multireference Coupled-Cluster Methods on Intel Many Integrated Core Architecture

Here we discuss the implementation of multireference coupled-cluster formalism with singles, doubles, and noniterative triples (MRCCSD(T)), which is capable of taking advantage of the processing power of the Intel Xeon Phi coprocessor. We discuss the integration of two levels of parallelism underlying the MRCCSD(T) implementation with computational kernels designed to offload the computationally intensive parts of the MRCCSD(T) formalism to Intel Xeon Phi coprocessors. Special attention is given to the enhancement of the parallel performance by task reordering that has improved load balancing in the noniterative part of the MRCCSD(T) calculations. We also discuss aspects regarding efficient optimization and vectorization strategies.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Design and Interpretation of Colliding Pulse Injected Laser-Plasma Acceleration Experiments

The use of colliding laser pulses to control the injection of plasma electrons into the plasma wake of a laser-plasma accelerator is a promising approach to obtaining GeV scale electron bunches with reduced emittance and energy spread. Colliding Pulse Injection (CPI) experiments are being performed by groups around the world. We present recent particle-in-cell simulations, using the parallel VORPAL framework, of CPI for physical parameters relevant to ongoing experiments of the LOASIS program at LBNL. We perform parameter scans in order to optimize the quality of the bunch, and compare the results with experimental data. Effect of non-ideal gaussian pulses and laser self-focusing in the plasma channel on the trapped bunch are evaluated. For optimized parameters accessible in the experiment, a 20 pC electron beam can be accelerated to 300 MeV with percent level energy spread.

43 PARTICLE ACCELERATORS↗

ePix: a class of architectures for second generation LCLS cameras

ePix is a novel class of ASIC architectures, based on a common platform, optimized to build modular scalable detectors for LCLS. The platform architecture is composed of a random access analog matrix of pixel with global shutter, fast parallel column readout, and dedicated sigma-delta analog-to-digital converters per column. It also implements a dedicated control interface and all the required support electronics to perform configuration, calibration and readout of the matrix. Based on this platform a class of front-end ASICs and several camera modules, meeting different requirements, can be developed by designing specific pixel architectures. This approach reduces development time and expands the possibility of integration of detector modules with different size, shape or functionality in the same camera. The ePix platform is currently under development together with the first two integrating pixel architectures: ePix100 dedicated to ultra low noise applications and ePix10k for high dynamic range applications.

Dragone, A.↗

The Reverse Cuthill-McKee Algorithm in Distributed-Memory

Ordering vertices of a graph is key to minimize fill-in and data structure size in sparse direct solvers, maximize locality in iterative solvers, and improve performance in graph algorithms. Except for naturally parallelizable ordering methods such as nested dissection, many important ordering methods have not been efficiently mapped to distributed-memory architectures. In this paper, we present the first-ever distributed-memory implementation of the reverse Cuthill-McKee (RCM) algorithm for reducing the profile of a sparse matrix. Our parallelization uses a two-dimensional sparse matrix decomposition. We achieve high performance by decomposing the problem into a small number of primitives and utilizing optimized implementations of these primitives. Our implementation shows strong scaling up to 1024 cores for smaller matrices and up to 4096 cores for larger matrices.

97 MATHEMATICS AND COMPUTING↗

Grid Optimization Lawrence Livermore Nonlinear Programming

Modelling and solution library for security-constrained AC optimal power flow problems that uses nonlinear programming methods (interior-point methods) and solvers (Ipopt, HiOp). The code is written in C++ and provides parallel computing capabilities in the form of task parallelism (solving contingency problems in parallel). Code can run on a variety of commodity high-performance computing.

Petra, CosminG↗

Adorym: a multi-platform generic X-ray image reconstruction framework based on automatic differentiation

We describe and demonstrate an optimization-based X-ray image reconstruction framework called Adorym. Our framework provides a generic forward model, allowing one code framework to be used for a wide range of imaging methods ranging from near-field holography to fly-scan ptychographic tomography. By using automatic differentiation for optimization, Adorym has the flexibility to refine experimental parameters including probe positions, multiple hologram alignment, and object tilts. It is written with strong support for parallel processing, allowing large datasets to be processed on high-performance computing systems. We demonstrate its use on several experimental datasets to show improved image quality through parameter refinement.

36 MATERIALS SCIENCE↗

Shear horizontal surface acoustic wave microsensor for Class A viral and bacterial detection.

The rapid autonomous detection of pathogenic microorganisms and bioagents by field deployable platforms is critical to human health and safety. To achieve a high level of sensitivity for fluidic detection applications, we have developed a 330 MHz Love wave acoustic biosensor on 36{sup o} YX Lithium Tantalate (LTO). Each die has four delay-line detection channels, permitting simultaneous measurement of multiple analytes or for parallel detection of single analyte containing samples. Crucial to our biosensor was the development of a transducer that excites the shear horizontal (SH) mode, through optimization of the transducer, minimizing propagation losses and reducing undesirable modes. Detection was achieved by comparing the reference phase of an input signal to the phase shift from the biosensor using an integrated electronic multi-readout system connected to a laptop computer or PDA. The Love wave acoustic arrays were centered at 330 MHz, shifting to 325-328 MHz after application of the silicon dioxide waveguides. The insertion loss was -6 dB with an out-of-band rejection of 35 dB. The amplitude and phase ripple were 2.5 dB p-p and 2-3{sup o} p-p, respectively. Time-domain gating confirmed propagation of the SH mode while showing suppression of the triple transit. Antigen capture and mass detection experiments demonstrate a sensitivity of 7.19 {+-} 0.74{sup o} mm{sup 2}/ng with a detection limit of 6.7 {+-} 0.40 pg/mm{sup 2} for each channel.

60 APPLIED LIFE SCIENCES↗

Community Project for Accelerator Science and Simulation (ComPASS)

After concluding our initial exercise (solving a simplified statistical inverse problem with unknown parameter laser intensity) of coupling Vorpal and our parallel statistical library QUESO, we shifted the application focus to DLA. Our efforts focused on developing a Gaussian process (GP) emulator within QUESO for efficient optimization of power couplers within woodpiles. The smaller simulation size (compared with LPA) allows for sufficient “training runs” to develop a reasonable GP statistical emulator for a parameter space of moderate dimension.

43 PARTICLE ACCELERATORS↗

Effective Vectorization with OpenMP 4.5

This paper describes how the Single Instruction Multiple Data (SIMD) model and its extensions in OpenMP work, and how these are implemented in different compilers. Modern processors are highly parallel computational machines which often include multiple processors capable of executing several instructions in parallel. Understanding SIMD and executing instructions in parallel allows the processor to achieve higher performance without increasing the power required to run it. SIMD instructions can significantly reduce the runtime of code by executing a single operation on large groups of data. The SIMD model is so integral to the processor s potential performance that, if SIMD is not utilized, less than half of the processor is ever actually used. Unfortunately, using SIMD instructions is a challenge in higher level languages because most programming languages do not have a way to describe them. Most compilers are capable of vectorizing code by using the SIMD instructions, but there are many code features important for SIMD vectorization that the compiler cannot determine at compile time. OpenMP attempts to solve this by extending the C++/C and Fortran programming languages with compiler directives that express SIMD parallelism. OpenMP is used to pass hints to the compiler about the code to be executed in SIMD. This is a key resource for making optimized code, but it does not change whether or not the code can use SIMD operations. However, in many cases critical functions are limited by a poor understanding of how SIMD instructions are actually implemented, as SIMD can be implemented through vector instructions or simultaneous multi-threading (SMT). We have found that it is often the case that code cannot be vectorized, or is vectorized poorly, because the programmer does not have sufficient knowledge of how SIMD instructions work.

97 MATHEMATICS AND COMPUTING↗

Rapid measurement of RH-dependent aerosol hygroscopic growth using a humidity-controlled fast integrated mobility spectrometer (HFIMS)

Abstract. The ability of aerosol particles to uptake water (hygroscopic growth) is an important determinant of aerosol optical properties and radiative effects. Aerosol hygroscopic growth is traditionally measured by humidified tandem differential mobility analyzers (HTDMA), in which size-selected dry particles are exposed to elevated relative humidity (RH), and the size distribution of humidified particles is subsequently measured using a scanning mobility particle sizer. As a scanning mobility particle sizer can measure only one particle size at a time, HTDMA measurements are time consuming, and ambient measurements are often limited to a single RH level. Pinterich et al. (2017b) showed that fast measurements of aerosol hygroscopic growth are possible using a humidity-controlled fast integrated mobility spectrometer (HFIMS). In HFIMS, the size distribution of humidified particles is rapidly captured by a water-based fast integrated mobility spectrometer (WFIMS), leading to a factor of ∼10 increase in measurement time resolution. In this study we present a prototype HFIMS that extends fast hygroscopic growth measurements to a wide range of atmospherically relevant RH values, allowing for more comprehensive characterizations of aerosol hygroscopic growth. A dual-channel humidifier consisting of two humidity conditioners in parallel is employed such that aerosol RH can be quickly stepped among different RH levels by sampling from alternating conditioners. The measurement sequence is also optimized to minimize the transition time between different particle sizes. The HFIMS is capable of measuring aerosol hygroscopic growth of six particle diameters under five RH levels ranging from 20 % to 85 % (30 separate measurements) every 25 min. The performance of this HFIMS is characterized and validated using laboratory-generated ammonium sulfate aerosol standards. Measurements of ambient aerosols are shown to demonstrate the capability of HFIMS to capture the rapid evolution of aerosol hygroscopic growth and its dependence on both size and RH.

54 ENVIRONMENTAL SCIENCES↗

A graphite crystal polarimeter for stellar X-ray astronomy.

The first crystal X-ray polarimeter to be used for X-ray astronomy is described. Polarization is measured by modulation of the X rays diffracted at an average 45 deg glancing angle from large, curved graphite crystal panels as these rotate about an axis parallel to the incident X-ray flux. Arrangement of the crystal panels, the design of the detector, and the signal-processing circuitry were optimized to minimize systematic effects produced by off-axis pointing of the rocket and cosmic ray induced events. The in-flight performance of the instrument in relation to the observed background signal is discussed.

Weisskopf, M. C.↗