Maximizing available hardware parallelism on CPU and GPU architectures for finite element applications
Explore the source record for details and available documents.
SEARCH · Engineering Papers
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.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Abstract not provided.
Methods for analyzing and improving a target computer application and corresponding systems and computer-readable mediums. A method includes receiving the target application. The method includes generating a parallel control flow graph (ParCFG) corresponding to the target application. The method includes analyzing the ParCFG by the computer system. The method includes generating and storing the modified ParCFG for the target application.
We implement and benchmark parallel I/O methods for the fully-manycore driven particle-in-cell code PIConGPU. Identifying throughput and overall I/O size as a major challenge for applications on today’s and future HPC systems, we present a scaling law characterizing performance bottlenecks in state-of-the-art approaches for data reduction. Consequently, we propose, implement and verify multi-threaded data-transformations for the I/O library ADIOS as a feasible way to trade underutilized host-side compute potential on heterogeneous systems for reduced I/O latency.
A highly efficient program of massively parallel calculations for electron dynamics has been developed in an effort to apply the method to optical response of nanostructures of more than ten-nanometers in size. The approach is based on time-dependent density functional theory calculations in real-time and real-space. The computational code is implemented by using simple algorithms with a finite-difference method in space derivative and Taylor expansion in time-propagation. Since the computational program is free from the algorithms of eigenvalue problems and fast-Fourier-transformation, which are usually implemented in conventional quantum chemistry or band structure calculations, it is highly suitable for massively parallel calculations. Benchmark calculations using the K computer at RIKEN demonstrate that the parallel efficiency of the program is very high on more than 60 000 CPU cores. The method is applied to optical response of arrays of C{sub 60} orderly nanostructures of more than 10 nm in size. The computed absorption spectrum is in good agreement with the experimental observation.
A system, method and computer program product for supporting system initiated checkpoints in high performance parallel computing systems and storing of checkpoint data to a non-volatile memory storage device. The system and method generates selective control signals to perform checkpointing of system related data in presence of messaging activity associated with a user application running at the node. The checkpointing is initiated by the system such that checkpoint data of a plurality of network nodes may be obtained even in the presence of user applications running on highly parallel computers that include ongoing user messaging activity. In one embodiment, the non-volatile memory is a pluggable flash memory card.
High-performance computing (HPC) applications have traditionally relied on parallel file systems and file transfer services to manage data movement and storage. Alternative approaches have been proposed that use direct communications between application components, trading persistence and fault tolerance for speed. Event-driven architectures, as popularized in enterprise contexts, present a compelling middle ground, avoiding the performance cost and API constraints of parallel file systems while retaining persistence and offering impedance matching between application components. However, adapting streaming frameworks to HPC workloads requires addressing challenges unique to HPC systems. This paper investigates the potential for a streaming framework designed for HPC infrastructures and use cases. We introduce Mofka, a persistent event-streaming framework designed specifically for HPC environments. Mofka combines the capabilities of a traditional streaming service with optimizations tailored to the HPC context, such as support for massively multicore nodes, efficient scaling for large producer-consumer workflows, RDMA-enabled high-performance network communications, specialized network fabrics with multiple links per node, and efficient handling of large scientific data payloads. Built using the Mochi suite of HPC data service components, Mofka provides a lightweight, modular, and high-performance solution for persistent streaming in HPC systems. We present the architecture of Mofka and evaluate its performance against Kafka and Redpanda using benchmarks on diverse platforms, including Argonne's Polaris and Oak Ridge's Frontier supercomputers, showing up to 8× improvement in throughput in some scenarios. We then demonstrate its utility in several real-world applications: a tomographic reconstruction pipeline, a workflow for the discovery of metal-organic frameworks for carbon capture, and the instrumentation of Dask workflows for provenance tracking and performance analysis.
In this paper, commercial clouds bring a great opportunity to the scientific computing area. Scientific applications usually require significant resources, however not all scientists have access to sufficient high-end computing systems, may of which can be found in the Top500 list. Cloud Computing has gained the attention of scientists as a competitive resource to run HPC applications at a potentially lower cost. But as a different infrastructure, it is unclear whether clouds are capable of running scientific applications with a reasonable performance per money spent. This work studies the performance of public clouds and places this performance in context to price. We evaluate the raw performance of different services of AWS cloud in terms of the basic resources, such as compute, memory, network and I/O. We also evaluate the performance of the scientific applications running in the cloud. This paper aims to assess the ability of the cloud to perform well, as well as to evaluate the cost of the cloud running scientific applications. We developed a full set of metrics and conducted a comprehensive performance evlauation over the Amazon cloud. We evaluated EC2, S3, EBS and DynamoDB among the many Amazon AWS services. We evaluated the memory sub-system performance with CacheBench, the network performance with iperf, processor and network performance with the HPL benchmark application, and shared storage with NFS and PVFS in addition to S3. We also evaluated a real scientific computing application through the Swift parallel scripting system at scale. Armed with both detailed benchmarks to gauge expected performance and a detailed monetary cost analysis, we expect this paper will be a recipe cookbook for scientists to help them decide where to deploy and run their scientific applications between public clouds, private clouds, or hybrid clouds.
CAASCADE — Compiler-Assisted Application Source Code Analysis and DatabasE—is a tool that summarizes the use of parallel programming language features in application source code using compiler technology. This paper discusses the library detection capability within CAASCADE to find information about the usage of scientific libraries within the source code. The information that CAASCADE collects provides insights into the usage of library calls in an applications. CAASCADE can classify the APIs by scientific libraries (e.g. LAPACK, BLAS, FFTW, etc). It can also detect the context in which a library API is being invoked, for example within a serial or multi-threaded region. To collect this information, CAASCADE uses compiler plugins that summarize procedural information and uses Apache Spark to do inter-procedural analysis to reconstruct call chains. In addition to this, we also integrated CAASCADE to work with XALT to collect library information based on linkage and modules installed on a system.
This software comprises a portable, open-source implementation of a runtime library to support applications written in the Unified Parallel C (UPC) language. This library implements the UPC-specific functionality, including shared-memory allocation and distributed locks. The network-dependent functionality is implemented as a thin wrapper around the separate GASNet (Global-Address Space Networking) library, which is required for operation. This software implements the "Berkeley UPC Runtime Specification", which is targeted by at least four separate UPC compiler front-ends. The implementation is highly portable and runs over hardware ranging from laptops to custom supercomputers. An external library is required for certain local memory allocation operations; a well-defined interface allows for multiple implementations of this library, but at present the "umalloc" library from LBNL is the only compatible implementation.
The POSIX interface does not lend itself well to enabling good performance for high-end applications. Extensions are needed in the POSIX I/O interface so that high-concurrency HPC applications running on top of parallel file systems perform well. This paper presents the rationale, design, and evaluation of a reference implementation of a subset of the POSIX I/O interfaces on a widely used parallel file system (PVFS) on clusters. Experimental results on a set of micro-benchmarks confirm that the extensions to the POSIX interface greatly improve scalability and performance.
The increased need for efficient ways to implement domain-specific accelerators is driving design methodologies towards the use of abstractions higher than the Register Transfer Level (RTL). In this scenario, High Level Synthesis (HLS) plays a significant role by enabling the automatic generation of custom hardware accelerators starting from high level descriptions (e.g., C code). Conventional HLS tools exploit parallelism mostly at the Instruction Level (ILP). They statically schedule the input specifications, and build centralized Finite State Machine (FSM) controllers. However, aggressive exploitation of ILP in many applications has diminishing returns and, usually, centralized approaches do not efficiently exploit coarser parallelism because FSMs are inherently serial. In this paper we present a HLS framework able to synthesize applications that, beside ILP, also expose Task Level Parallelism (TLP). An application can expose TLP through annotations that identify the parallel functions (i.e., tasks). To generate accelerators that efficiently execute concur- rent tasks, we need to solve several issues: devise a mechanism to support concurrent execution flows, exploit memory parallelism, and manage synchronization. To support concurrent execution flows, we introduce a novel adaptive controller. The adaptive controller is composed of a set of interacting control elements that independently manage the execution of a single operation or function call. These control elements check dependencies and resource constraints at runtime, enabling as soon as possible execution. To support parallel access to shared memories and synchronization, we introduce a novel Hierarchical Memory Interface (HMI). With respect to previous solutions, the proposed interface supports multi-ported memories and atomic memory operations, which commonly occur in parallel programming. Our framework can generate the hardware implementation of C functions by employing two different approaches, depending on its characteristics. If a function exposes TLP, then the framework generates hardware implementations based on the adaptive controller. Otherwise, the framework implements the function by exploiting a more conventional FSM approach, which is optimized for ILP exploitation. We evaluate our framework on a set of parallel applications, and show substantial performance improvements (average speedup of 4.7) with limited area over- heads (average area increase of 5.48 times).
The increased need for efficient ways to implement domain-specific accelerators is driving design methodologies towards the use of abstractions higher than the Register Transfer Level (RTL). In this scenario, High Level Synthesis (HLS) plays a significant role by enabling the automatic generation of custom hardware accelerators starting from high level descriptions (e.g., C code). Conventional HLS tools exploit parallelism mostly at the Instruction Level (ILP). They statically schedule the input specifications, and build centralized Finite State Machine (FSM) controllers. However, aggressive exploitation of ILP in many applications has diminishing returns and, usually, centralized approaches do not efficiently exploit coarser parallelism because FSMs are inherently serial. In this paper we present a HLS framework able to synthesize applications that, beside ILP, also expose Task Level Parallelism (TLP). An application can expose TLP through annotations that identify the parallel functions (i.e., tasks). To generate accelerators that efficiently execute concur- rent tasks, we need to solve several issues: devise a mechanism to support concurrent execution flows, exploit memory parallelism, and manage synchronization. To support concurrent execution flows, we introduce a novel adaptive controller. The adaptive controller is composed of a set of interacting control elements that independently manage the execution of a single operation or function call. These control elements check dependencies and resource constraints at runtime, enabling as soon as possible execution. To support parallel access to shared memories and synchronization, we introduce a novel Hierarchical Memory Interface (HMI). With respect to previous solutions, the proposed interface supports multi-ported memories and atomic memory operations, which commonly occur in parallel programming. Our framework can generate the hardware implementation of C functions by employing two different approaches, depending on its characteristics. If a function exposes TLP, then the framework generates hardware implementations based on the adaptive controller. Otherwise, the framework implements the function by exploiting a more conventional FSM approach, which is optimized for ILP exploitation. We evaluate our framework on a set of parallel applications, and show substantial performance improvements (average speedup of 4.7) with limited area over- heads (average area increase of 5.48 times).
As hardware advances continue to modify the supercomputing landscape, traditional scientific software development practices will become more outdated, ineffective, and inefficient. The process of rewriting/retooling existing software for new architectures is a Sisyphean task, and results in substantial hours of development time, effort, and money. Software libraries which provide an abstraction of the resources provided by such architectures are therefore essential if the computational engineering and science communities are to continue to flourish in this modern computing environment. The Multiphysics Object Oriented Simulation Environment (MOOSE) framework enables complex multiphysics analysis tools to be built rapidly by scientists, engineers, and domain specialists, while also allowing them to both take advantage of current HPC architectures, and efficiently prepare for future supercomputer designs. MOOSE employs a hybrid shared-memory and distributed-memory parallel model and provides a complete and consistent interface for creating multiphysics analysis tools. In this paper, a brief discussion of the mathematical algorithms underlying the framework and the internal object-oriented hybrid parallel design are given. Representative massively parallel results from several applications areas are presented, and a brief discussion of future areas of research for the framework are provided.
Improving data transfer throughput over high-speed long-distance networks has become increasingly difficult. Numerous factors such as nondeterministic congestion, dynamics of the transfer protocol, and multiuser and multitask source and destination endpoints, as well as interactions among these factors, contribute to this difficulty. A promising approach to improving throughput consists in using parallel streams at the application layer.We formulate and solve the problem of choosing the number of such streams from a mathematical optimization perspective. We propose the use of direct search methods, a class of easy-to-implement and light-weight mathematical optimization algorithms, to improve the performance of data transfers by dynamically adapting the number of parallel streams in a manner that does not require domain expertise, instrumentation, analytical models, or historic data. We apply our method to transfers performed with the GridFTP protocol, and illustrate the effectiveness of the proposed algorithm when used within Globus, a state-of-the-art data transfer tool, on productionWAN links and servers. We show that when compared to user default settings our direct search methods can achieve up to 10x performance improvement under certain conditions. We also show that our method can overcome performance degradation due to external compute and network load on source end points, a common scenario at high performance computing facilities.
The MOOSE framework is a foundational capability used by the NEAMS program to create over 15 different simulation tools for advanced nuclear reactors. Due to MOOSE’s broad use, improvements to the framework in support of modeling and simulation goals are critical to the program. Such improvements can take many forms, including optimization, improved user experience, streamlined application programming interfaces (APIs), parallelism, and new capabilities. The work described in this report was conducted in direct support of the simulation tools and has already been deployed. The capabilities were implemented in the same order as they are covered in this report: multiple time integrators in the same input file, initial design of framework Components, an input file Application block, extension of NetGen to 3D geometries in MOOSE, and deployment of executors in the multi-system paradigm. These five additions are fundamental capabilities that will be leveraged by many NEAMS applications.
As scalable parallel systems evolve towards more complex nodes with many-core architectures and larger trans-petascale & upcoming exascale deployments, there is a need to understand, characterize and quantify the underlying execution models being used on such systems. Execution models are a conceptual layer between applications & algorithms and the underlying parallel hardware and systems software on which those applications run. This paper presents the SCaLeM (Synchronization, Concurrency, Locality, Memory) framework for characterizing and execution models. SCaLeM consists of three basic elements: attributes, compositions and mapping of these compositions to abstract parallel systems. The fundamental Synchronization, Concurrency, Locality and Memory attributes are used to characterize each execution model, while the combinations of those attributes in the form of compositions are used to describe the primitive operations of the execution model. The mapping of the execution model’s primitive operations described by compositions, to an underlying abstract parallel system can be evaluated quantitatively to determine its effectiveness. Finally, SCaLeM also enables the representation and analysis of applications in terms of execution models, for the purpose of evaluating the effectiveness of such mapping.