Search NASA⌕ Search

SEARCH · Search NASA

Results for “code reuse”

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 109 records · Page 6

Benchmarking and Evaluating Unified Memory for OpenMP GPU Offloading

Here, the latest OpenMP standard offers automatic device offloading capabilities which facilitate GPU programming. Despite this, there remain many challenges. One of these is the unified memory feature introduced in recent GPUs. GPUs in current and future HPC systems have enhanced support for unified memory space. In such systems, CPU and GPU can access each other's memory transparently, that is, the data movement is managed automatically by the underlying system software and hardware. Memory over subscription is also possible in these systems. However, there is a significant lack of knowledge about how this mechanism will perform, and how programmers should use it. We have modified several benchmarks codes, in the Rodinia benchmark suite, to study the behavior of OpenMP accelerator extensions and have used them to explore the impact of unified memory in an OpenMP context. We moreover modified the open source LLVM compiler to allow OpenMP programs to exploit unified memory. The results of our evaluation reveal that, while the performance of unified memory is comparable with that of normal GPU offloading for benchmarks with little data reuse, it suffers from significant overhead when GPU memory is over subcribed for benchmarks with large amount of data reuse. Based on these results, we provide several guidelines for programmers to achieve better performance with unified memory.

97 MATHEMATICS AND COMPUTING↗

Raw diffraction data preservation and reuse: Overview, update on practicalities and metadata requirements

A topical review is presented of the rapidly developing interest in and storage options for the preservation and reuse of raw data within the scientific domain of the IUCr and its Commissions, each of which operates within a great diversity of instrumentation. A résumé is included of the case for raw diffraction data deposition. An overall context is set by highlighting the initiatives of science policy makers towards an `Open Science' model within which crystallographers will increasingly work in the future; this will bring new funding opportunities but also new codes of procedure within open science frameworks. Skills education and training for crystallographers will need to be expanded. Overall, there are now the means and the organization for the preservation of raw crystallographic diffraction data via different types of archive, such as at universities, discipline-specific repositories (Integrated Resource for Reproducibility in Macromolecular Crystallography, Structural Biology Data Grid), general public data repositories (Zenodo, ResearchGate) and centralized neutron and X-ray facilities. Formulation of improved metadata descriptors for the raw data types of each of the IUCr Commissions is in progress; some detailed examples are provided. Lastly, a number of specific case studies are presented, including an example research thread that provides complete open access to raw data.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Field Demonstration of the Brick Ontology to Scale up the Deployment of ASHRAE Guideline 36 Control Sequences

Many commercial buildings have a vast network of sensors as part of their building automation systems (BAS) that allows opportunities for energy consumption and cost savings by deploying advanced control sequences. However, this resource is often underutilized since BAS are typically programmed with simple control sequences with limited potential to deliver on these opportunities. The recent availability of ASHRAE Guideline 36 (G36) with advanced HVAC control sequences supports control retrofits in existing buildings to unlock much of the savings potential. However, barriers such as the lack of standard naming convention of building assets and data points, proprietary equipment and BAS, and the inherent uniqueness of buildings and their systems prevent building stakeholders from adopting any “plug-and-play” implementation of G36. Instead, control vendors must often undertake the manual and labor-intensive point mapping process to identify a data stream’s functional and spatial relationship within the HVAC system along with other relevant contexts and map it to the new control sequences. The vendor must carry out the point mapping process in each individual building since the mapping is unlikely to port over to another building. Even for the same building, the point mapping process can occur multiple times if various control vendors implement different control retrofits and/or multiple control retrofits happen over the lifecycle of the building. Then, there is the likelihood that G36 control sequences are programmed uniquely to the building, preventing the same implementation from being reused in another. Therefore, this paper presents a field demonstration of how we leveraged the Brick ontology with BACnet, OpenBuildingControl’s Control Description Language (CDL), and open-source support tools to implement scalable and portable advanced building controls. These tools provide standardized semantic descriptions and relationships of the building’s assets and data points (Brick), standardized communication protocol to read from and write to the building’s BAS (BACnet), and standardized code implementations (CDL) of standardized advanced control strategies (G36). We implemented G36’s hot water supply temperature setpoint reset in a Berkeley, CA building for this field demonstration. This field demonstration aims to show how integrating these tools may streamline the deployment of advanced control sequences such as G36 in a consistent manner regardless of differences found across buildings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Dense LU Factorization on Multicore Supercomputer Nodes

Dense LU factorization is a prominent benchmark used to rank the performance of supercomputers. Many implementations, including the reference code HPL, use block-cyclic distributions of matrix blocks onto a two-dimensional process grid. The process grid dimensions drive a trade-off between communication and computation and are architecture- and implementation-sensitive. We show how the critical panel factorization steps can be made less communication-bound by overlapping asynchronous collectives for pivot identification and exchange with the computation of rank-k updates. By shifting this trade-off, a modified block-cyclic distribution can beneficially exploit more available parallelism on the critical path, and reduce panel factorization's memory hierarchy contention on now-ubiquitous multi-core architectures. The missed parallelism in traditional block-cyclic distributions arises because active panel factorization, triangular solves, and subsequent broadcasts are spread over single process columns or rows (respectively) of the process grid. Increasing one dimension of the process grid decreases the number of distinct processes in the other dimension. To increase parallelism in both dimensions, periodic 'rotation' is applied to the process grid to recover the row-parallelism lost by a tall process grid. During active panel factorization, rank-1 updates stream through memory with minimal reuse. In a column-major process grid, the performance of this access pattern degrades as too many streaming processors contend for access to memory. A block-cyclic mapping in the more popular row-major order does not encounter this problem, but consequently sacrifices node and network locality in the critical pivoting steps. We introduce 'striding' to vary between the two extremes of row- and column-major process grids. As a test-bed for further mapping experiments, we describe a dense LU implementation that allows a block distribution to be defined as a general function of block to processor. Other mappings can be tested with only small, local changes to the code.

99 GENERAL AND MISCELLANEOUS↗

Tool Support for Software Lookup Table Optimization

A number of scientific applications are performance-limited by expressions that repeatedly call costly elementary functions. Lookup table (LUT) optimization accelerates the evaluation of such functions by reusing previously computed results. LUT methods can speed up applications that tolerate an approximation of function results, thereby achieving a high level of fuzzy reuse. One problem with LUT optimization is the difficulty of controlling the tradeoff between performance and accuracy. The current practice of manual LUT optimization adds programming effort by requiring extensive experimentation to make this tradeoff, and such hand tuning can obfuscate algorithms. In this paper we describe a methodology and tool implementation to improve the application of software LUT optimization. Our Mesa tool implements source-to-source transformations for C or C++ code to automate the tedious and error-prone aspects of LUT generation such as domain profiling, error analysis, and code generation. We evaluate Mesa with five scientific applications. Our results show a performance improvement of 3.0× and 6.9× for two molecular biology algorithms, 1.4× for a molecular dynamics program, 2.1× to 2.8× for a neural network application, and 4.6× for a hydrology calculation. We find that Mesa enables LUT optimization with more control over accuracy and less effort than manual approaches.

Wilcox, Chris↗

Data on Cu- and Ni-Si-Mn-rich solute clustering in a neutron irradiated austenitic stainless steel

The data presented in this article is supplementary to the research article “Phase instabilities in austenitic steels during particle bombardment at high and low dose rates” (Levine et al.). Needle-shaped samples were prepared with focused ion beam milling from a 304L stainless steel that was irradiated with fast neutrons (E 0.1 MeV) in the BOR-60 reactor at 318 °C to 47.5 dpa. Atom probe tomography (APT) experiments in voltage mode were then conducted on a Cameca LEAP 5000X HR. Atom position, range, and mass spectrum files after reconstruction with Cameca’s IVAS software are included. Cu- and Ni-Si-Mn-rich solute nanoclusters were identified and analyzed using the Open Source Characterization of APT Reconstructions (OSCAR) program. Python code for OSCAR, information on the program’s underlying algorithm, and sample output files are provided. A proximity histogram of a Ni-Si-Mn-rich cluster and a 1D density/solute concentration profile of a Cu-rich cluster are given to demonstrate OSCAR’s analytical functionalities. The provided APT dataset is valuable for benchmarking phase instabilities in neutron-irradiated austenitic stainless steels that occur at high doses. The OSCAR program can be reused to process other APT data sets where solute nanoclustering is of interest.

42 ENGINEERING↗

BioJS: an open source standard for biological visualisation – its status in 2014

BioJS is a community-based standard and repository of functional components to represent biological information on the web. The development of BioJS has been prompted by the growing need for bioinformatics visualisation tools to be easily shared, reused and discovered. Its modular architecture makes it easy for users to find a specific functionality without needing to know how it has been built, while components can be extended or created for implementing new functionality. The BioJS community of developers currently provides a range of functionality that is open access and freely available. A registry has been set up that categorises and provides installation instructions and testing facilities at http://www.ebi.ac.uk/tools/biojs/. The source code for all components is available for ready use at https://github.com/biojs/biojs.

59 BASIC BIOLOGICAL SCIENCES↗

VOLTTRON/volttron-GS

PNNL grid related applications Object oriented design intended to facilitate reuse and extensibility of system features. The template enforces that needed structure and behaviors are available. Users may inherit and adapt these structures and behaviors to meet their specific requirements. Recognizes separation of the transactive system calculations into (1) scheduling of assets, (2) balancing of supply and demand, and (3) coordination among independent agents. This separation facilitates useful computation boundaries in which computations can be completed without requiring end-to-end communication for each update. The code provides fundamental market solutions for balancing. The code provides mechanisms for inter-agent coordination. Allows event-driven agent interactions; not limited to prescribed time intervals. Aggregates value and electricity at nodes and conveys opportunities up and down through the network.

Hammerstrom, Don↗

Optimizing a Multiple Right-Hand Side Dslash Kernel for Intel Knights Corner

Lattice Quantum Chromo-dynamics (LQCD) stands unique as the only computationally tractable, non-perturbative, and model-independent quantum field theory of the strong nuclear force. LQCD achieves tractability by discretizing our familiar 4-dimensional space-time onto a hypercubic lattice. The core component of the linear solver used in LQCD is the Wilson-Dirac operator whose finite difference component is called the Dslash operator. Dslash is a nearest neighbor stencil operator summing matrix-vector multiplications over lattice points. The arithmetic intensity of a regular Dslash operator acting on a single vector in single precision is between 0.88-2.29 (depending on the tricks employed) making it memory bandwidth bound on most current architectures, including Xeon Phi. Maximizing the performance of Dslash on Xeon Phi requires deliberate exploitation of its architectural features. One key feature of Xeon Phi is its powerful vector processing unit with lengthy vector registers. There is a class of LQCD computations which aims to solve the Dirac equation for the same lattice gauge fields multiple times. In this case one can gain additional reuse of the gauge field, and potentially simplified vectorization, through applying Dslash to several vectors at the same time; often referred to as Multiple Right Hand Side (MRHS) Dslash. Previous work in this area focused on applying this approach to the Highly Improved Staggered operator, and in cases to the multi-Grid algorithm for Wilson clover fermions. Here, we present two approaches to MRHS Dslash: a vector register blocking approach and one using the software package QPhiX with a custom Code Generator for low-level intrinsics. We explore the Knights Corner instruction set, memory hierarchy, and threading capabilities. For standard lattice sizes used in testing, we observed significant speedups, with sustained performance of over 700 GFLOPS (single precision) in one instance. We achieved up to 27% of theoretical peak performance compared to a maximum of 12% obtained by the previous single RHS method in QPhiX.

Joó, Bálint↗

Influence of construction and demolition waste management on the environmental impact of buildings

Highlights: Black-Right-Pointing-Pointer Environmental impacts of different demolition practices. Black-Right-Pointing-Pointer 'Top-down' approach to the Life Cycle Analysis methodology. Black-Right-Pointing-Pointer Results based on real buildings measurements and demolition contractor activities. Black-Right-Pointing-Pointer Not every type of selective demolition brings about environmental benefits. - Abstract: The purpose of this study is to quantify comparable environmental impacts within a Life Cycle Analysis (LCA) perspective, for buildings in which the first (Materials) and last (End of Life) life cycle stages are adjusted to several waste/material management options. Unlike most LCAs, the approach is 'top-down' rather than 'bottom-up', which usually involves large amounts of data and the use of specific software applications. This approach is considered appropriate for a limited but expedient LCA designed to compare the environmental impacts of different life cycle options. Present results, based on real buildings measurements and demolition contractor activities, show that shallow, superficial, selective demolition may not result in reduced environmental impacts. Calculations actually show an increase (generally less than 5%) in most impact categories for the Materials and End of Life stages because of extra transportation needs. However, core material separation in demolition operations and its recycling and/or reuse does bring environmental benefits. A reduction of around 77% has been estimated in the climate change impact category, 57% in acidification potential and 81% in the summer smog impact (for the life cycle stages referred).

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

An Evidence-Based Critique of U.S. Nuclear Waste Policy Using Data from Reactor and Cyclotron Sites - 18568

Concrete at nuclear facilities becomes radioactive when it is exposed to a neutron flux, and when such facilities are vacated, decommissioning activities are enacted to make the site safe for reuse. These activities generate wastes that the public expects to be managed in their best interest. This paper provides side-by-side comparisons of concrete bio-shielding encountered at typical reactor and cyclotron decommissioning projects. Three sites are considered, and results from characterization samples are used to demonstrate that the wastes generated by such sites are similar in terms of content and risk. First, the dose potential represented by each waste is determined from mass-based radioactivity concentration with the RESRAD-BUILD computer code. The resulting doses are translated into population-risk estimates by applying dose-response predictions from the National Academy of Sciences' 2006 BEIR VII report. In terms of possible adverse health outcomes, the risk of solid cancer or leukemia associated with the wastes this paper examines is 8 to 50 extra cases per 100,000 persons. The regulatory framework applicable to each site is also reviewed, and the waste management strategies used at each site are contrasted. This paper demonstrates that the wastes it examines are virtually indistinguishable. Contrary to this finding, the wastes are regulated, managed, and disposed in very different manners according to current U.S. nuclear waste policy. The value of this paper is that the side-by-side comparisons it presents provides an opportunity to examine the risk-benefit of existing policies in terms of the benefits or disparities they represent. (author)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Resource-Adaptive Federated Text Generation with Differential Privacy

In cross-silo federated learning (FL), sensitive text datasets remain confined to local organizations due to privacy regulations, making repeated training for each downstream task both communication-intensive and privacy-demanding. A promising alternative is to generate differentially private (DP) synthetic datasets that approximate the global distribution and can be reused across tasks. However, pretrained large language models (LLMs) often fail under domain shift, and federated finetuning is hindered by computational heterogeneity: only resource-rich clients can update the model, while weaker clients are excluded, amplifying data skew and the adverse effects of DP noise. We propose a flexible participation framework that adapts to client capacities. Strong clients perform DP federated finetuning, while weak clients contribute through a lightweight DP voting mechanism that refines synthetic text. To ensure the synthetic data mirrors the global dataset, we apply control codes (e.g., labels, topics, metadata) that represent each client’s data proportions and constrain voting to semantically coherent subsets. This two-phase approach requires only a single round of communication for weak clients and integrates contributions from all participants. Experiments show that our framework improves distribution alignment and downstream robustness under DP and heterogeneity.

Wang, Jiayi [ORNL]↗

Profile Generation for GPU Targets

GPU accelerators are ubiquitous, but their ecosystem is far less evolved than the host one. Compiler heuristics are often tuned for CPUs and reused for GPU. Similarly, tooling and more evolved optimization techniques are historically not available on GPU targets. In this work, we address one of these shortcomings and enable profile generation and profile-guided optimizations (PGO) for GPU targets. While this is only a single step towards a CPU equivalent ecosystem for offload devices, it shows how old misconceptions on the limitations of GPUs are often not warranted anymore. Through our implementation in LLVM/Offload, we enable device-side PGO for full scientific applications and open up tooling opportunities, including code coverage analysis and compiler-built-in roofline analysis. Our evaluation highlights the performance implications of profile generation, the insights gained from these profiles, and the (missed) opportunities in utilizing the information for GPU compilation.

McDonough, Ethan Luis [Lawrence Livermore National↗

StreamJIT: a commensal compiler for high-performance stream programming

There are many domain libraries, but despite the performance benefits of compilation, domain-specific languages are comparatively rare due to the high cost of implementing an optimizing compiler. We propose commensal compilation, a new strategy for compiling embedded domain-specific languages by reusing the massive investment in modern language virtual machine platforms. Commensal compilers use the host language's front-end, use host platform APIs that enable back-end optimizations by the host platform JIT, and use an autotuner for optimization selection. The cost of implementing a commensal compiler is only the cost of implementing the domain-specific optimizations. We demonstrate the concept by implementing a commensal compiler for the stream programming language StreamJIT atop the Java platform. Our compiler achieves performance 2.8 times better than the StreamIt native code (via GCC) compiler with considerably less implementation effort.

97 MATHEMATICS AND COMPUTING↗

Composable optimization and control toolkit for scientific applications

Applications of Artificial Intelligence (AI) and Machine Learning (ML) can improve the computational efficiency and scientific research output. In order to improve interoperability and reuse of AI/ML software, a composable approach is required. This talk presents a composable approach for scientific workflow development that allows seamless integration of various modules developed by independent researchers. These practices will reduce redundant software development by allowing re-use of workflow modules across projects, teams, departments and facilities. We will present three use cases that follow the composable approach namely, Scientific Optimization and Control Toolkit (SOCT), SciDAC QuantOm workflow, and JLab Nuclear Physics experimental workflows. This talk will dive deeper into SOCT and present the details of the composable code development for optimization and control algorithms using reinforcement learning.

Rajput, Kishansingh↗

LibraryX: A Framework for Cross-Library-Call Optimization

Scientific applications utilize performance libraries as a software engineering concept: these libraries encapsulate important and well-understood (mathematical) operations, allow for reuse, and are implemented and tuned by experts. Domain scientists then implement complex algorithms based on these domainspecific libraries. While individual library calls are optimized, larger performance gains across sequences of calls—sometimes spanning multiple libraries—are often unrealized, forcing a trade-off between performance and implementation complexity.To overcome this issue, we propose LibraryX, an approach and a system that allows for cross-library-call optimization even when library calls stem from multiple performance libraries. LibraryX annotates library calls with semantic information and optimizes entire directed acyclic graphs (DAGs) of calls dynamically using the SPIRAL code generation system. We demonstrate its effectiveness across a range of memory bound workloads, achieving significant speedups on Nvidia, AMD, and Intel accelerators compared to code using native libraries without cross-call optimization.

Rao, Sanil [Carnegie Mellon University,Department ↗

ALE3D: An Arbitrary Lagrangian-Eulerian Multi-Physics Code

ALE3D is a multi-physics numerical simulation software tool utilizing arbitrary-Lagrangian- Eulerian (ALE) techniques. The code is written to address both two-dimensional (2D plane and axisymmetric) and three-dimensional (3D) physics and engineering problems using a hybrid finite element and finite volume formulation to model fluid and elastic-plastic response of materials on an unstructured grid. As shown in Figure 1, ALE3D is a single code that integrates many physical phenomena.

42 ENGINEERING↗

The coyote universe extended: Precision emulation of the matter power spectrum

Modern sky surveys are returning precision measurements of cosmological statistics such as weak lensing shear correlations, the distribution of galaxies, and cluster abundance. To fully exploit these observations, theorists must provide predictions that are at least as accurate as the measurements, as well as robust estimates of systematic errors that are inherent to the modeling process. In the nonlinear regime of structure formation, this challenge can only be overcome by developing a large-scale, multi-physics simulation capability covering a range of cosmological models and astrophysical processes. As a first step to achieving this goal, we have recently developed a prediction scheme for the matter power spectrum (a so-called emulator), accurate at the 1% level out to k ∼ 1 Mpc{sup –1} and z = 1 for wCDM cosmologies based on a set of high-accuracy N-body simulations. It is highly desirable to increase the range in both redshift and wavenumber and to extend the reach in cosmological parameter space. To make progress in this direction, while minimizing computational cost, we present a strategy that maximally reuses the original simulations. We demonstrate improvement over the original spatial dynamic range by an order of magnitude, reaching k ∼ 10 h Mpc{sup –1}, a four-fold increase in redshift coverage, to z = 4, and now include the Hubble parameter as a new independent variable. To further the range in k and z, a new set of nested simulations run at modest cost is added to the original set. The extension in h is performed by including perturbation theory results within a multi-scale procedure for building the emulator. This economical methodology still gives excellent error control, ∼5% near the edges of the domain of applicability of the emulator. A public domain code for the new emulator is released as part of the work presented in this paper.

79 ASTRONOMY AND ASTROPHYSICS↗