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 goal of this report is to provide insight to the development of vt-tv, a C++ HPC visualization tool designed for insightful analysis of load-balancing metrics in the DARMA toolkit. In particular, it delves into its modular data model and diverse usage scenarios, emphasizing adaptability and efficiency.
Abstract not provided.
The computational capacity of High-Performance Computing (HPC) systems increases continuously with the rapid development of central processing units (CPUs) and graphic processing units (GPUs), while the in-/output (IO) subsystem develops relatively slowly and storage capacity is also limited. Data-intensive applications, which are designed to leverage the high computational capacity of HPC resources, typically generate a considerable amount of data for post-processing visualizations and data analytics. The limited IO speed and storage space could lead to constraints in the actual performance of these applications and, therefore, scientific discovery. In-situ techniques, where data is visualized/analysed while still in memory rather than through disk, can contribute to alleviating these problems as they can reduce or even fully avoid data writing/reading through the IO subsystem to/from storage. However, the overall efficiency of insitu techniques crucially depends on the characteristics of both the in-situ tasks and the applications, and the resource distribution among them. Therefore, choosing the right in-situ approach (synchronous, asynchronous, or hybrid) and resource allocation is essential to minimize overhead and maximize the benefits of concurrent execution. In this paper, we present a performance model of in-situ techniques to find the most beneficial in-situ approach and the preferred resource configuration. We verify the high accuracy of our approach with over 6800 measurements and provide use cases with different applications.
Hyperparameters employed by deep learning (DL) methods play a substantial role in the performance and reliability of these methods in practice. Unfortunately, finding performance-optimizing hyperparameter settings is a notoriously difficult task. Hyperparameter search methods typically have limited production-strength implementations or do not target scalability within a highly parallel machine, portability across different machines, experimental comparison between different methods, and tighter integration with workflow systems. In this paper, we present DeepHyper, a Python package that provides a common interface for the implementation and study of scalable hyperparameter search methods. It adopts the Balsam workflow system to hide the complexities of running large numbers of hyperparameter configurations in parallel on high-performance computing (HPC) systems. We implement and study asynchronous model-based search methods that consist of sampling a small number of input hyperparameter configurations and progressively fitting surrogate models over the input-output space until exhausting a user-defined budget of evaluations. We evaluate the efficacy of these methods relative to approaches such as random search, genetic algorithms, Bayesian optimization, and hyperband on DL benchmarks on CPU- and GPU-based HPC systems.
Most parallel scientific programs contain compiler directives (pragmas) such as those from OpenMP (Hermanns in Parallel programming in Fortran 95 using openMP, 2002. School of Aeronautical Engineering, Universidad Politécnica de Madrid, España, 2011), explicit calls to runtime library procedures such as those implementing the Message Passing Interface (MPI) (in A message-passing interface standard version 4.0, 2021. https://www.mpi-forum.org/docs/mpi-4.0/mpi40-report.pdf), or compiler-specific language extensions such as those provided by CUDA (Ruetsch and Fatica in CUDA Fortran for scientists and engineers: best practices for efficient CUDA Fortran programming, Elsevier, 2013). By contrast, the recent Fortran standards empower developers to express parallel algorithms without directly referencing lower-level parallel programming models (Numrich in Parallel programming with co-arrays, CRC Press, 2018, and Curcic in Modern Fortran: building efficient parallel applications, Manning Publications, 2020). Fortran’s parallel features place the language within the Partitioned Global Address Space (PGAS) class of programming models. When writing programs that exploit data parallelism, application developers often find it straightforward to develop custom parallel algorithms. Problems involving complex, heterogeneous, staged calculations, however, pose much greater challenges. Such applications require careful coordination of tasks in a manner that respects dependencies prescribed by a directed acyclic graph. When rolling one’s own solution proves difficult, extending a customizable framework becomes attractive. Further, the paper presents the design, implementation, and use of the Framework for Extensible Asynchronous Task Scheduling (FEATS), which we believe to be the first task scheduling tool written in modern Fortran. We describe the benefits and compromises associated with choosing Fortran as the implementation language, and we propose ways in which future Fortran standards can best support the use case in this paper.
Geophysical models of the atmosphere and ocean invariably involve parameterizations. These represent two distinct areas: a) Subgrid processes which the model cannot (yet) resolve, due to its discrete resolution, and b) sources in the equation, due to radiation for example. Hence coupling between these physics parameterizations and the resolved fluid dynamics and also between the dynamics of the different fluids in the system (air and water) is necessary. This coupling is an important aspect of geophysical models. However, often model development is strictly segregated into either physics or dynamics. Hence, this area has many more unanswered questions than in-depth understanding. Furthermore, recent developments in the design of dynamical cores (significant increase of resolution, move to non-hydrostatic equation sets etc), extended process physics (prognostic micro physics, 3D turbulence, non-vertical radiation etc) and predicted future changes of the computational infrastructure (Exascale with its need for task parallelism, data locality and asynchronous time stepping for example) is adding even more complexity and new questions.
This report is an outcome of the ASC ATDM Level 2 Milestone 6015: Asynchronous Many-Task Software Stack Demonstration. It comprises a summary and in depth analysis of DARMA and a DARMA-compliant Asynchronous Many-Task (AMT) runtime software stack. Herein performance and productivity of the over- all approach are assessed on benchmarks and proxy applications representative of the Sandia ATDM applications. As part of the effort to assess the perceived strengths and weaknesses of AMT models compared to more traditional methods, experiments were performed on ATS-1 (Advanced Technology Systems) test bed machines and Trinity. In addition to productivity and performance assessments, this report includes findings on the generality of DARMAs backend API as well as findings on interoperability with node- level and network-level system libraries. Together, this information provides a clear understanding of the strengths and limitations of the DARMA approach in the context of Sandias ATDM codes, to guide our future research and development in this area.
We begin by presenting an overview of the general philosophy that is guiding the novel DARMA developments, followed by a brief reminder about the background of this project. We finally present the FY19 design requirements. As the Exascale era arises, DARMA is uniquely positioned at the forefront of asychronous many-task (AMT) research and development (R&D) to explore emerging programming model paradigms for next-generation HPC applications at Sandia, across NNSA labs, and beyond. The DARMA project explores how to fundamentally shift the expression(PM) and execution(EM)of massively concurrent HPC scientific algorithms to be more asynchronous, resilient to executional aberrations in heterogeneous/unpredictable environments, and data-dependency conscious—thereby enabling an intelligent, dynamic, and self-aware runtime to guide execution.
We propose IMPACC, an MPI+OpenACC framework for heterogeneous accelerator clusters. IMPACC tightly integrates MPI and OpenACC, while exploiting the shared memory parallelism in the target system. IMPACC dynamically adapts the input MPI+OpenACC applications on the target heterogeneous accelerator clusters to fully exploit target system-specific features. IMPACC provides the programmers with the unified virtual address space, automatic NUMA-friendly task-device mapping, efficient integrated communication routines, seamless streamlining of asynchronous executions, and transparent memory sharing. We have implemented IMPACC and evaluated its performance using three heterogeneous accelerator systems, including Titan supercomputer. Results show that IMPACC can achieve easier programming, higher performance, and better scalability than the current MPI+OpenACC model.
Finding frequent subgraph patterns in a big graph is an important problem with many applications such as classifying chemical compounds and building indexes to speed up graph queries. Since this problem is NP-hard, some recent parallel and distributed systems have been developed to accelerate the mining. However, they often have a huge memory cost, very long running time, suboptimal load balancing, poor scale-out capability, and possibly inaccurate results. In this article, we propose an efficient system called T-FSM for parallel mining of frequent subgraph patterns in a big graph. T-FSM supports a new anti-monotonic frequentness measure called Fraction-Score, which is more accurate than the widely used MNI measure. The execution engine of T-FSM supports both intra-machine parallelism and inter-machine parallelism. For intra-machine parallelism, T-FSM adopts a novel task-based execution model to ensure high multithreading concurrency, bounded memory consumption, and effective load balancing. For inter-machine parallelism, T-FSM ensures good scale-out performance with a lightweight pattern rebalancing approach that reduces workload skewness of pattern evaluations among machines. To avoid recomputing the contexts for migrated patterns, we design a novel context cache table to support concurrent and asynchronous requesting and caching of remote context data, which can timely evict and garbage collect used pattern contexts that are no longer needed to keep memory consumption bounded. Extensive experiments show that T-FSM is orders of magnitude faster than existing state-of-the-art parallel systems (more than 10×, 51×, 131×, 55× speedup over ScaleMine, DistGraph, Pangolin and Peregrine, respectively) and distributed systems (more than 42× and 88× over ScaleMine and DistGraph, respectively) for frequent subgraph pattern mining, and it scales out satisfactorily to 512 CPU cores on the Polaris supercomputer at Argonne National Laboratory.
As electric grids experience high penetration levels of renewable generation, fundamental changes are required to address real-time situational awareness. Here, we utilize unique traits of tensors to devise a model-free situational awareness and energy forecasting framework for distribution networks. This work formulates the state of the network at multiple time instants as a three-way tensor; hence, recovering full state information of the network is tantamount to estimating all the values of the tensor. Given measurements received from µphasor measurement units and/or smart meters, the recovery of unobserved quantities is carried out using the low-rank canonical polyadic decomposition of the state tensor—that is, the state estimation task is posed as a tensor imputation problem utilizing observed patterns in measured (sampled) quantities. Two structured sampling schemes are considered, namely, asynchronous slab and fiber sampling. For both schemes, we present sufficient conditions on the number of sampled slabs and fibers that guarantee identifiability of the factors of the state tensor. Numerical results demonstrate the ability of the proposed framework to achieve high estimation accuracy in multiple sampling scenarios.
FleCSI is a compile-time configurable C++ framework designed to support multi-physics application development. As such, FleCSI attempts to provide a very general set of infrastructure design patterns that can be specialized and extended to suit the needs of a broad variety of solver and data requirements. This means that FleCSI is potentially useful to many different ECP projects. Current support includes multidimensional mesh topology, mesh geometry, and mesh adjacency information, n-dimensional hashed-tree data structures, graph partitioning interfaces, and dependency closures (to identify data dependencies between distributed-memory address spaces). FleCSI introduces a functional programming model with control, execution, and data abstractions that are consistent with state-of-the-art task-based runtimes such as Legion and Charm++. The model also provides support for fine-grained, data-parallel execution with backend support for runtimes such as OpenMP and C++17. The FleCSI abstraction layer provides the developer with insulation from the underlying runtimes, while allowing support for multiple runtime systems, including conventional models like asynchronous MPI. The intent is to give developers a concrete set of user-friendly programming tools that can be used now, while allowing flexibility in choosing runtime implementations and optimizations that can be applied to architectures and runtimes that arise in the future. This project is essential to the ECP Ristra Next-Generation Code project, part of ASC ATDM, because it provides a hierarchically parallel programming model that is consistent with the design of modern system architectures, but which allows for the straightforward expression of algorithmic parallelism in a portably performant manner.
DARMA (Distributed Asynchronous Resilient Models for Applications) is a runtime library supporting the Sandia ATDM (Advanced Technology Development and Mitigation) program. The main application drivers fall within the ECP milestone 2.2.5.03 ADNN03-ASC ATDM SNL Application, which includes applications that require load balancing and asynchronous communication for high performance. The DARMA runtime infrastructure has been modified to be compatible with Kokkos/OpenMP parallelization within tasks, which is a critical requirement for high performance for the Sandia ATDM apps. DARMA development has occurred in parallel with a verification milestone for ATDM in FY18. For FY19, DARMA should impact ATDM by enabling dynamic load balancing and communication through only incremental changes to the existing verified MPI codes. DARMA can now support the intra-kernel thread parallelization in the parent MPI apps, allowing DARMA to be easily added without rewriting individual math kernels. The results presented here demonstrate the DARMA results for an MPI mini-app.
The Hartree-Fock (HF) method is the fundamental first step for incorporating quantum mechanics into many-electron simulations of atoms and molecules, and it is an important component of computational chemistry toolkits like NWChem. The GTFock code is an HF implementation that, while it does not have all the features in NWChem, represents crucial algorithmic advances that reduce communication and improve load balance by doing an up-front static partitioning of tasks, followed by work stealing whenever necessary. To enable innovations in algorithms and exploit next generation exascale systems, it is crucial to support quantum chemistry codes using expressive and convenient programming models and runtime systems that are also efficient and scalable. Here, this paper presents an HF implementation similar to GTFock using UPC++, a partitioned global address space model that includes flexible communication, asynchronous remote computation, and a powerful multidimensional array library. UPC++ offers runtime features that are useful for HF such as active messages, a rich calculus for array operations, hardware-supported fetch-and-add, and functions for ensuring asynchronous runtime progress. We present a new distributed array abstraction, DArray, that is convenient for the kinds of random-access array updates and linear algebra operations on block-distributed arrays with irregular data ownership. Finally, we analyze the performance of atomic fetch-and-add operations (relevant for load balancing) and runtime attentiveness, then compare various techniques and optimizations for each. Our optimized implementation of HF using UPC++ and the DArrays library shows up to 20% improvement over GTFock with Global Arrays at scales up to 24,000 cores.
This report outlines the research, development, and support requirements for the Advanced Simulation and Computing (ASC ) Advanced Technology, Development, and Mitigation (ATDM) Performance Portability (a.k.a., Kokkos) project for 2015 - 2019 . The research and development (R&D) goal for Kokkos (v2) has been to create and demonstrate a thread - parallel programming model a nd standard C++ library - based implementation that enables performance portability across diverse manycore architectures such as multicore CPU, Intel Xeon Phi, and NVIDIA Kepler GPU. This R&D goal has been achieved for algorithms that use data parallel pat terns including parallel - for, parallel - reduce, and parallel - scan. Current R&D is focusing on hierarchical parallel patterns such as a directed acyclic graph (DAG) of asynchronous tasks where each task contain s nested data parallel algorithms. This five y ear plan includes R&D required to f ully and performance portably exploit thread parallelism across current and anticipated next generation platforms (NGP). The Kokkos library is being evaluated by many projects exploring algorithm s and code design for NGP. Some production libraries and applications such as Trilinos and LAMMPS have already committed to Kokkos as their foundation for manycore parallelism an d performance portability. These five year requirements includes support required for current and antic ipated ASC projects to be effective and productive in their use of Kokkos on NGP. The greatest risk to the success of Kokkos and ASC projects relying upon Kokkos is a lack of staffing resources to support Kokkos to the degree needed by these ASC projects. This support includes up - to - date tutorials, documentation, multi - platform (hardware and software stack) testing, minor feature enhancements, thread - scalable algorithm consulting, and managing collaborative R&D.
The new challenges posed by Exascale system architectures have resulted in difficulty achieving a desired scalability using traditional distributed memory runtimes. Task-based programming models show promise in addressing these challenges, providing application developers with a productive and performant approach to programming on next generation systems. Empirical studies show that task-based models can overcome load balancing issues that are inherent to traditional distributed memory runtimes, and that task-based runtimes perform comparably to those systems when balanced. This panel is designed to explore the advantages of task-based programming models on modern and future HPC systems from an industry, university, and national lab perspective. It aims at gathering application experts and proponents of these models to present concrete and practical examples of using task-based runtimes to overcome the challenges posed by Exascale system architectures. This report describes the objectives, activities, and outcomes of the panel TBAA: Task-Based Algorithms and Applications which was held at the International Conference for High Performance Computing, Networking, Storage, and Analysis (SC 20) on November 18, 2020.
Because they were developed for optimal sequential complexity, classical graph algorithms as found in textbooks have strictly-defined orders of operations. Enforcing a prescribed order of operations, or even an approximate order, in a distributed memory setting requires significant amounts of synchronization, which in turn can severely limit scalability. As a result, new algorithms are typically required to achieve scalable performance, even for solving well-known graph problems. Yet, even in these cases, parallel graph algorithms are written according to parallel programming models that evolved for, e.g., scientific computing, and that still have inherent, and scalability-limiting, amounts of synchronization. In this paper we present a new approach to parallel graph algorithms: synchronization-avoiding algorithms. To eliminate synchronization and its associated overhead, synchronization-avoiding algorithms perform work in an unordered and fully asynchronous fashion in such a way that the result is constantly refined toward its final state. “Wasted” work is minimized by locally prioritizing tasks using problem-dependent task utility metrics. We classify algorithms for graph applications into two broad categories: algorithms with monotonic updates (which evince global synchronization) and algorithms with nonmonotonic updates (which evince vertex-centric synchronization). We apply our approach to both classes and develop novel, synchronization-avoiding algorithms for solving exemplar problems: SSSP and connected components for the former, graph coloring for the latter. We demonstrate that eliminating synchronization in conjunction with effective scheduling policies and optimizations in the runtime results in improved scalability for both classes of algorithms.
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.