Search NASA⌕ Search

SEARCH · Search NASA

Results for “synchronization overhead”

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 19 records

Accelerating Parallel Applications in Cloud Platforms via Adaptive Time-Slice Control

Cloud platforms can provide flexible and cost-effective environments for parallel applications. However, the resource over-commitment issues, i.e., cloud providers often provide much more executable virtual CPUs than available physical CPUs, still impede the synchronization operations of parallel applications, causing severe performance degradation. Existing methods optimize parallel applications by promoting the priorities of involved VMs. They cannot fully explore the performance of parallel applications, because they ignore the time-slice requirements of different phases of parallel applications. Furthermore, non-parallel applications experience unsatisfied performance because of low scheduling priorities. Given empirical analysis on time-slices of virtual machines (VMs), we find that shortening time-slices can mitigate synchronization overhead which incurs during communication phases, while over-short time-slices cause frequent cache misses in computation phases. Accordingly, we propose an Adaptive Time-slice Control (ATC) mechanism. ATC first detects the phases of parallel applications based on lock latency or cache misses. Then, ATC shortens time-slices during communication phases and prolongs time-slices during computation phases for parallel applications, and sets a uniform time-slice for non-parallel applications. Finally, we evaluate ATC using seven well-known benchmarks with 25+ applications. Experiments show that ATC obtains 1.5-75x performance gain for running parallel applications than state-of-the-art solutions, with nearly unaffected impact on non-parallel applications.

97 MATHEMATICS AND COMPUTING↗

Optimizing the hypre solver for manycore and GPU architectures

The solution of large-scale combustion problems with codes such as Uintah on modern computer architectures requires the use of multithreading and GPUs to achieve performance. Uintah uses a low-Mach number approximation that requires iteratively solving a large system of linear equations. The Hypre iterative solver has solved such systems in a scalable way for Uintah, but the use of OpenMP with Hypre leads to at least slowdown due to OpenMP overheads. The proposed solution uses the MPI Endpoints within Hypre, where each team of threads acts as a different MPI rank. This approach minimizes OpenMP synchronization overhead and performs as fast or (up to 1.44) faster than Hypre's MPI-only version, and allows the rest of Uintah to be optimized using OpenMP. The profiling of the GPU version of Hypre shows the bottleneck to be the launch overhead of thousands of micro-kernels. The GPU performance was improved by fusing these micro-kernels and was further optimized by using Cuda-aware MPI, resulting in an overall speedup of 1.16—1.44 compared to the baseline GPU implementation. The above optimization strategies were published in the International Conference on Computational Science 2020 [1]. This work extends the previously published research by carrying out the second phase of communication-centered optimizations in Hypre to improve its scalability on large-scale supercomputers. Additionally, this includes an efficient non-blocking inter-thread communication scheme, communication-reducing patch assignment, and expression of logical communication parallelism to a new version of the MPICH library that utilizes the underlying network parallelism [2]. The above optimizations avoid communication bottlenecks previously observed during strong scaling and improve performance by up to 2 on 256 nodes of Intel Knight's Landing processor.

97 MATHEMATICS AND COMPUTING↗

LC-MEMENTO: A Memory Model for Accelerated Architectures

With the advent of heterogeneous architectures, in particular, with the ubiquity of multi-GPU systems, it is becoming increasingly important to manage device memory efficiently in order to reap the benefits of the additional core count. To date, such responsibility mainly falls on the programmer where device-to-host data communication (and vice versa), if not done properly, may incur costly memory transfer operations and synchronization. The problem may be compounded by additional requirement to maintain system-wide memory consistency that may involve expensive synchronization overhead. In this paper, we present Location Consistency Memory Model for Enhanced Transfer Operations (LC-MEMENTO). This framework considers incorporating runtime techniques for multi-GPU memory management to support relaxed synchronization semantics and memory transfer operations automatically. Specifically, we implement a relaxed form of a memory consistency model based on the Location Consistency (LC) in an Asynchronous Many-Task Runtime (ARTS) and demonstrate that, this memory model enables additional optimization opportunities for the three representative applications encompassing different computational patterns (scientific computation, graphs, data streaming, etc.).

Memory Models, Accelerators, Adaptive Optimization↗

Fast synchronization-free algorithms for parallel sparse triangular solves with multiple right-hand sides

The sparse triangular solve kernels, SpTRSV and SpTRSM, are important building blocks for a number of numerical linear algebra routines. Parallelizing SpTRSV and SpTRSM on today's many-core platforms, such as GPUs, is not an easy task since computing a component of the solution may depend on previously computed components, enforcing a degree of sequential processing. As a consequence, most existing work introduces a preprocessing stage to partition the components into a group of level-sets or colour-sets so that components within a set are independent and can be processed simultaneously during the subsequent solution stage. However, this class of methods requires a long preprocessing time as well as significant runtime synchronization over-heads between the sets. To address this, we

HPC↗

Reversible Discrete Event Formulation and Optimistic Parallel Execution of Vehicular Traffic Models

Vehicular traffic simulations are useful in applications such as emergency planning and traffic management. High speed of traffic simulations translates to speed of response and level of resilience in those applications. Discrete event formulation of traffic flow at the level of individual vehicles affords both the flexibility of simulating complex scenarios of vehicular flow behavior as well as rapid simulation time advances. However, efficient parallel/distributed execution of the models becomes challenging due to synchronization overheads. Here, a parallel traffic simulation approach is presented that is aimed at reducing the time for simulating emergency vehicular traffic scenarios. Our approach resolves the challenges that arise in parallel execution of microscopic, vehicular-level models of traffic. We apply a reverse computation-based optimistic execution approach to address the parallel synchronization problem. This is achieved by formulating a reversible version of a discrete event model of vehicular traffic, and by utilizing this reversible model in an optimistic execution setting. Three unique aspects of this effort are: (1) exploration of optimistic simulation applied to vehicular traffic simulation (2) addressing reverse computation challenges specific to optimistic vehicular traffic simulation (3) achieving absolute (as opposed to self-relative) speedup with a sequential speed close to that of a fast, de facto standard sequential simulator for emergency traffic. The design and development of the parallel simulation system is presented, along with a performance study that demonstrates excellent sequential performance as well as parallel performance. The benefits of optimistic execution are demonstrated, including a speed up of nearly 20 on 32 processors observed on a vehicular network of over 65,000 intersections and over 13 million vehicles.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Towards Flexible Exascale Stream Processing System Simulation

Stream processing is an important emerging computational model for performing complex operations on and across multi-source, high-volume, unpredictable dataflows. We present Flow, a platform for parallel and distributed stream processing system simulation that provides a flexible modeling environment for analyzing stream processing applications. The Flow stream processing system simulator is a high-performance, scalable simulator that automatically parallelizes chunks of the model space and incurs near-zero synchronization overhead for acyclic stream application graphs. We show promising parallel and distributed event rates exceeding 149 million events per second on a cluster with 512 processor cores.

99 GENERAL AND MISCELLANEOUS↗

Highly Efficient Compensation-Based Parallelism for Wavefront Loops on GPUs

Wavefront loops are widely used in many scientific applications, e.g., partial differential equation (PDE) solvers and sequence alignment tools. However, due to the data dependencies in wavefront loops, it is challenging to fully utilize the abundant compute units of GPUs and to reuse data through their memory hierarchy. Existing solutions can only optimize for these factors to a limited extent. For example, tiling-based methods optimize memory access but may result in load imbalance; while compensation-based methods, which change the original order of computation to expose more parallelism and then compensate for it, suffer from both global synchronization overhead and limited generality. In this paper, we first prove under which circumstances that breaking data dependencies and properly changing the sequence of computation operators in our compensation-based method does not affect the correctness of results. Based on this analysis, we design a highly efficient compensation-based parallelism on GPUs. Our method provides weighted scan-based GPU kernels to optimize the computation and combines with the tiling method to optimize memory access and synchronization. The performance results on the NVIDIA K80 and P100 GPU platforms demonstrate that our method can achieve significant improvements for four types of real-world application kernels over the state-of-the-art research.

Hou, Kaixi↗

A Fast Parallel Algorithm for Selected Inversion of Structured Sparse Matrices with Application to 2D Electronic Structure Calculations

We present an efficient parallel algorithm and its implementation for computing the diagonal of $H^-1$ where $H$ is a 2D Kohn-Sham Hamiltonian discretized on a rectangular domain using a standard second order finite difference scheme. This type of calculation can be used to obtain an accurate approximation to the diagonal of a Fermi-Dirac function of $H$ through a recently developed pole-expansion technique \cite{LinLuYingE2009}. The diagonal elements are needed in electronic structure calculations for quantum mechanical systems \citeHohenbergKohn1964, KohnSham 1965,DreizlerGross1990. We show how elimination tree is used to organize the parallel computation and how synchronization overhead is reduced by passing data level by level along this tree using the technique of local buffers and relative indices. We analyze the performance of our implementation by examining its load balance and communication overhead. We show that our implementation exhibits an excellent weak scaling on a large-scale high performance distributed parallel machine. When compared with standard approach for evaluating the diagonal a Fermi-Dirac function of a Kohn-Sham Hamiltonian associated a 2D electron quantum dot, the new pole-expansion technique that uses our algorithm to compute the diagonal of $(H-z_i I)^-1$ for a small number of poles $z_i$ is much faster, especially when the quantum dot contains many electrons.

74 ATOMIC AND MOLECULAR PHYSICS↗

Physics-aware adaptive checkpointing with shadow systems for nonlinear PDE simulations

Large-scale simulations of nonlinear partial differential equations (PDEs) that exhibit strongly transient behavior and pattern-forming dynamics produce enormous amounts of data, which, even with modern storage systems, cannot be stored for later curation. Current I/O strategies either write dense time series of snapshots, which is often prohibitive in I/O and storage, or store a few checkpoints that enable restart but incur expensive recomputation cost and provide no control over post-restart error growth, especially when lossy compression is used. Moreover, most, if not all, existing strategies take no account of the actual physical state of the system. Here, we present a simple physics-aware I/O framework in which a low-cost shadow system adaptively triggers lossy checkpoints when the shadow system deviates from the fine-scale simulation. The shadow system can be a coarsened replica of the fine-scale simulation that evolves concurrently. This means that checkpoints are taken based on the physical state of the system: fewer checkpoints are triggered when the system is quiescent while more are taken when the system undergoes a rapid change. This type of behavior is observed in many systems such as Brusselator and FitzHugh–Nagumo. We illustrate that our framework maintains stable restarts, keeps fine-scale restart errors bounded by shadow errors, and reconstructs the time history with significantly lower error and storage than interpolating fixed-interval snapshots, with low-cost shadow replay and modest online synchronization overhead.

Gong, Qian [ORNL] (ORCID:0000000235704142)↗

An Efficient Wait-Free Vector

The vector is a fundamental data structure, which provides constant-time access to a dynamically-resizable range of elements. Currently, there exist no wait-free vectors. The only non-blocking version supports only a subset of the sequential vector API and exhibits significant synchronization overhead caused by supporting opposing operations. Since many applications operate in phases of execution, wherein each phase only a subset of operations are used, this overhead is unnecessary for the majority of the application. To address the limitations of the non-blocking version, we present a new design that is wait-free, supports more of the operations provided by the sequential vector, and provides alternative implementations of key operations. These alternatives allow the developer to balance the performance and functionality of the vector as requirements change throughout execution. Compared to the known non-blocking version and the concurrent vector found in Intel’s TBB library, our design outperforms or provides comparable performance in the majority of tested scenarios. Over all tested scenarios, the presented design performs an average of 4.97 times more operations per second than the non-blocking vector and 1.54 more than the TBB vector. In a scenario designed to simulate the filling of a vector, performance improvement increases to 13.38 and 1.16 times. This work presents the first ABA-free non-blocking vector. Finally, unlike the other non-blocking approach, all operations are wait-free and bounds-checked and elements are stored contiguously in memory.

97 MATHEMATICS AND COMPUTING↗

Data-driven Fault Tolerance for Work Stealing Computations

Checkpoint-restart approaches to fault tolerance typically roll back all the processes to the previous checkpoint in the event of a failure. Work stealing is a promising technique to dynamically tolerate variations in the execution environment, including faults, system noise, and energy constraints. In this paper, we present fault tolerance mechanisms for task parallel computations, a popular computation idiom, employing work stealing. The computation is organized as a collection of tasks with data in a global address space. The completion of data operations, rather than the actual messages, is tracked to derive an idempotent data store. This information is used to accurately identify the tasks to be re-executed, therefore to recompute only the lost data, in the presence of random work stealing. We consider three recovery schemes that present distinct trade-offs -- lazy recovery with potentially increased re-execution cost, immediate collective recovery with associated synchronization overheads, and noncollective recovery enabled by additional communication. We employ distributed work stealing to dynamically rebalance the tasks on the live processes and evaluate the three schemes using candidate application benchmarks. We demonstrate that the overheads (space and time) of the fault tolerance mechanism are low, the cost incurred due to failures are small, and the overheads decrease with per-process work at scale.

fault tolerance, task parallelism, load balancing,↗

Fast and Scalable Sparse Triangular Solver for Multi-GPU Based HPC Architectures

Designing efficient and scalable sparse linear algebra kernels on modern multi-GPU based HPC systems is a daunting task due to significant irregular memory references and workload imbalance across the GPUs. This is particularly the case for \textit{Sparse Triangular Solver (SpTRSV)} which introduces additional two-dimensional computation dependencies among subsequent computation steps. Dependency information is exchanged and shared among GPUs, thus warrant for efficient memory allocation, data partitioning, and workload distribution as well as fine-grained communication and synchronization support. In this work, we demonstrate that directly adopting unified memory can adversely affect the performance of SpTRSV on multi-GPU architectures, despite linking via fast interconnect like NVLinks and NVSwitches. Alternatively, we employ the latest NVSHMEM technology based on Partitioned Global Address Space programming model to enable efficient fine-grained communication and drastic synchronization overhead reduction. Furthermore, to handle workload imbalance, we propose a malleable task-pool execution model which can further enhance the utilization of GPUs. By applying these techniques, our experiments on the NVIDIA multi-GPU supernode V100-DGX-1 and DGX-2 systems demonstrate that our design can achieve on average 3.53x (up to 9.86x) speedup on a DGX-1 system and 3.66x (up to 9.64x) speedup on a DGX-2 system with 4-GPUs over the Unified-Memory design. The comprehensive sensitivity and scalability studies also show that the proposed zero-copy SpTRSV is able to fully utilize the computing and communication resources of the multi-GPU system.

Xie, Chenhao↗

Topology-Dependent Performance of Free-Space Photonic Quantum Networks Under Noise

Photonic quantum communication enables secure and high-fidelity information transfer beyond classical limits, with direct relevance to emerging quantum networks operating in free-space environments. While physical-layer models of depolarizing noise, Gamma–Gamma turbulence statistics, entanglement swapping, and decoy-state QKD security bounds are individually well established, prior work typically treats these components in isolation or under fixed network assumptions. In this work, we develop a unified topology-aware analytical framework that simultaneously integrates free-space optical link budgets, turbulence-induced visibility degradation, depolarizing qubit noise, multi-hop entanglement cascade dynamics, teleportation fidelity thresholds, CHSH nonlocality certification, and asymptotic decoy-state secret key rate bounds across star, mesh, and ring graph structures. Rather than introducing new physical channel models, we demonstrate that identical physical links exhibit fundamentally different end-to-end performance once embedded within different network topologies. Mesh architectures minimize visibility cascade through hop-count reduction but incur quadratic hardware scaling. Star topologies minimize link count but concentrate noise and synchronization overhead at the hub. Ring configurations offer linear hardware scaling with multiplicative fidelity degradation. The results establish topology as a first-order design parameter in near-term free-space quantum networks operating without full quantum repeater infrastructures. While motivated by distributed multi-agent architectures, the framework applies broadly to terrestrial, airborne, and satellite-assisted photonic quantum communication systems.

QKD↗

Parallel Asynchronous Distributed-Memory Maximal Independent Set Algorithm with Work Ordering

The maximal independent set (MIS) graph problem arises in many applications such as computer vision, information theory, molecular biology, and process scheduling. The growing scale of graph data suggests the use of distributed memory hardware as a cost-effective approach to providing necessary compute and memory resources. Existing distributed memory parallel MIS algorithms rely on synchronous communication and use techniques such as subgraph computations. In this paper, we present an asynchronous distributed-memory parallel graph algorithm that relies on a virtual directed acyclic graph (DAG) that is created during the algorithm execution. We introduce two additional algorithms that save computations by ordering generated work. The first algorithm applies ordering globally to reduce computations, and the second algorithm applies ordering locally at the level of threads to minimize the synchronization overhead. We use two different implementations of Luby’s algorithm variants as baseline to compare the performance of the presented algorithms: (1) vertex-centric Luby A and Luby B implementations, and (2) the CombBLAS linear-algebra Luby A implementation. Results show that proposed algorithms outperform both implementations of Luby algorithms, especially in distributed execution. Furthermore, we show that for low-diameter graphs the algorithm that applies global ordering scales better than other algorithms and for high diameter graphs the original asynchronous algorithm and thread-level ordering algorithm show better performance.

maximal independent set (MIS), distributed memory ↗

Exploring Integer Sum Reduction using Atomics on Intel CPU

Atomic functions are useful in updating a shared variable by multiple threads, barrier synchronizations, constructing complex data structures, and building high-level frameworks. In this paper, we focus on the evaluation and analysis of integer sum reduction, a common data parallel primitive. We convert the sequential reduction into parallel OpenCL implementations on the CPU. We also develop three micro kernels, which allow us to understand the relationships between the kernel performance and the operations involved in reduction. The results of the micro kernels show that increasing the work-group size linearly can linearly improve the kernel performance. There is a sweet spot in the relationship between the work-group size and barrier synchronization overhead. The performance of the atomics over local memory are not sensitive to the work-group size. The sum reduction kernel with vectorized memory accesses can improve the performance of the baseline kernel for a wide range of work-group sizes. However, the vectorization efficiency shrinks with the growing work-group size. We also find that the vendor’s default OpenCL kernel optimization does not improve the kernel performance. On average, disabling the optimization can reduce the execution time of the kernel with vectorized memory accesses by 15%. We attribute the performance drop to the fact that the default kernel optimizations instantiate a large number of atomics over global memory when implicitly vectorizing the kernel computation.

atomics↗

Synchronization-Avoiding Graph Algorithms

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.

Firoz, Jesun S.↗

Synchronization Algorithms for Co-Simulation of Power Grid and Communication Networks

The ongoing modernization of power grids consists of integrating them with communication networks in order to achieve robust and resilient control of grid operations. To understand the operation of the new smart grid, one approach is to use simulation software. Unfortunately, current power grid simulators at best utilize inadequate approximations to simulate communication networks, if at all. Cooperative simulation of specialized power grid and communication network simulators promises to more accurately reproduce the interactions of real smart grid deployments. However, co-simulation is a challenging problem. A co-simulation must manage the exchange of informa- tion, including the synchronization of simulator clocks, between all simulators while maintaining adequate computational perfor- mance. This paper describes two new conservative algorithms for reducing the overhead of time synchronization, namely Active Set Conservative and Reactive Conservative. We provide a detailed analysis of their performance characteristics with respect to the current state of the art including both conservative and optimistic synchronization algorithms. In addition, we provide guidelines for selecting the appropriate synchronization algorithm based on the requirements of the co-simulation. The newly proposed algorithms are shown to achieve as much as 14% and 63% im- provement, respectively, over the existing conservative algorithm.

co-simulation↗

Synchronization for CXL Based Memory

Compute Express Link (CXL) is an important emerging standard for disaggregated memory. While this standard provisions coherency across numerous hosts and devices, implementing hardware support for type three devices is challenging. In this work, we look at the overhead of software synchronization and using software-based coherency. Moreover, we discuss the limits of software-based coherency in fully expressing modern synchronization techniques for a CXL-based disaggregate memory system. We demonstrate our approach using a CXL hardware prototype and running a version of the famous Peterson Lock (enhanced to run with more than two threads). We analyze its performance and share how more advanced synchronization techniques might interact with software-based coherence CXL hardware and program execution models.

High Performance Computing (HPC)↗