Search NASA⌕ Search

SEARCH · Search NASA

Results for “Checkpoint restart”

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 73 records · Page 4

New Parallel computing framework for radiation transport codes

A new parallel computing framework has been developed to use with general-purpose radiation transport codes. The framework was implemented as a C++ module that uses MPI for message passing. The module is significantly independent of radiation transport codes it can be used with, and is connected to the codes by means of a number of interface functions. The framework was integrated with the MARS15 code, and an effort is under way to deploy it in PHITS. Besides the parallel computing functionality, the framework offers a checkpoint facility that allows restarting calculations with a saved checkpoint file. The checkpoint facility can be used in single process calculations as well as in the parallel regime. Several checkpoint files can be merged into one thus combining results of several calculations. The framework also corrects some of the known problems with the scheduling and load balancing found in the original implementations of the parallel computing functionality in MARS15 and PHITS. The framework can be used efficiently on homogeneous systems and networks of workstations, where the interference from the other users is possible.

43 PARTICLE ACCELERATORS↗

Failure recovery for bulk synchronous applications with MPI stages

When an MPI program experiences a failure, the most common recovery approach is to restart all processes from a previous checkpoint and to re-queue the entire job. A disadvantage of this method is that, although the failure occurred within the main application loop, live processes must start again from the beginning of the program, along with new replacement processes—this incurs unnecessary overhead for live processes. To avoid such overheads and concomitant delays, we introduce the concept of “MPI Stages.” MPI Stages saves internal MPI state in a separate checkpoint in conjunction with application state. Upon failure, both MPI and application state are recovered, respectively, from their last synchronous checkpoints and continue without restarting the overall MPI job. Live processes roll back only a few iterations within the main loop instead of rolling back to the beginning of the program, while a replacement of failed process restarts and reintegrates, thereby achieving faster failure recovery. As a result, this approach integrates well with large-scale, bulk synchronous applications and checkpoint/restart.

97 MATHEMATICS AND COMPUTING↗

Error recovery in shared memory multiprocessors using private caches

The problem of recovering from processor transient faults in shared memory multiprocesses systems is examined. A user-transparent checkpointing and recovery scheme using private caches is presented. Processes can recover from errors due to faulty processors by restarting from the checkpointed computation state. Implementation techniques using checkpoint identifiers and recovery stacks are examined as a means of reducing performance degradation in processor utilization during normal execution. This cache-based checkpointing technique prevents rollback propagation, provides rapid recovery, and can be integrated into standard cache coherence protocols. An analytical model is used to estimate the relative performance of the scheme during normal execution. Extensions to take error latency into account are presented.

Wu, Kun-Lung↗

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)↗

Recoverable distributed shared virtual memory

The problem of rollback recovery in distributed shared virtual environments, in which the shared memory is implemented in software in a loosely coupled distributed multicomputer system, is examined. A user-transparent checkpointing recovery scheme and a new twin-page disk storage management technique are presented for implementing recoverable distributed shared virtual memory. The checkpointing scheme can be integrated with the memory coherence protocol for managing the shared virtual memory. The twin-page disk design allows checkpointing to proceed in an incremental fashion without an explicit undo at the time of recovery. The recoverable distributed shared virtual memory allows the system to restart computation from a checkpoint without a global restart.

Wu, Kun-Lung↗

Accelerating Flash-X Simulations with Asynchronous I/O

Most high-fidelity physics simulation codes, such as Flash-X, need to save intermediate results (checkpoint files) to restart or gain insights into the evolution of the simulation. These simulation codes save such intermediate files synchronously, where computation is stalled while the data is written to storage. Depending on the problem size and computational requirements, this file write time can be a substantial portion of the total simulation time. In order to hide the I/O latency of checkpointing, asynchronous I/O methods have been introduced. These methods use background threads for performing I/O while the main threads continue with the simulation. The usage of background threads can compete for resources on the node as well as with communication. In this paper, we evaluate the overheads and the overall benefit of asynchronous I/O in HDF5 to simulations. Results from real-world high-fidelity simulations on the Summit supercomputer show that I/O operation is overlapped with application communication or computation or both, effectively hiding some or all of the I/O latency. Our evaluation shows that while using asynchronous I/O adds overhead to the application, the I/O time reduction is more significant, resulting in overall up to 1.5X performance speedup.

Jain, Rajeev↗

Resilience and fault tolerance in high-performance computing for numerical weather and climate prediction

Progress in numerical weather and climate prediction accuracy greatly depends on the growth of the available computing power. As the number of cores in top computing facilities pushes into the millions, increased average frequency of hardware and software failures forces users to review their algorithms and systems in order to protect simulations from breakdown. This report surveys hardware, application-level and algorithm-level resilience approaches of particular relevance to time-critical numerical weather and climate prediction systems. A selection of applicable existing strategies is analysed, featuring interpolation-restart and compressed checkpointing for the numerical schemes, in-memory checkpointing, user-level failure mitigation and backup-based methods for the systems. Numerical examples showcase the performance of the techniques in addressing faults, with particular emphasis on iterative solvers for linear systems, a staple of atmospheric fluid flow solvers. The potential impact of these strategies is discussed in relation to current development of numerical weather prediction algorithms and systems towards the exascale. Trade-offs between performance, efficiency and effectiveness of resiliency strategies are analysed and some recommendations outlined for future developments.

54 ENVIRONMENTAL SCIENCES↗

Report on local data recovery approaches suitable for weather and climate prediction (Deliverable 1.3) (V.1.0)

Numerical weather and climate prediction rates as one of the scientific applications whose accuracy improvements greatly depend on the growth of the available computing power. As the number of cores in top computing facilities pushes into the millions, increasing average frequency of hardware and software failures forces users to review their algorithms and systems in order to protect simulations from breakdown. This report surveys approaches for fault-tolerance in numerical algorithms and system resilience in parallel simulations from the perspective of numerical weather and climate prediction systems. A selection of existing strategies is analyzed, featuring interpolation-restart and compressed checkpointing for the numerics, in-memory checkpointing, user-level failure mitigation-based and backup-based methods for the systems. Numerical examples showcase the performance of the techniques in addressing faults, with particular emphasis on iterative solvers for linear systems, a staple of atmospheric fluid flow solvers. The potential impact of these strategies is discussed in relation to current development of numerical weather prediction algorithms and systems towards the exascale. Trade-offs between performance, efficiency and effectiveness of resiliency strategies are analyzed and some recommendations outlined for future developments.

97 MATHEMATICS AND COMPUTING↗

Stability-preserving Lossy Compression for Large-scale Partial Differential Equations

Checkpoint/Restart (C/R) strategies are vital for fault tolerance in PDE-based scientific simulations, yet traditional checkpointing incurs significant I/O overhead. Lossy compression offers a scalable solution by reducing checkpoint data size, but conventional methods often lack control over physical invariants (e.g., energy), leading to instability such as oscillations or divergence in Partial Differential Equations (PDE) systems. This paper introduces a stability-preserving compression approach tailored for PDE simulations by explicitly controlling kinetic and potential energy perturbations to ensure stable restarts. Extensive experiments conducted across diverse PDE configurations demonstrate that our method maintains numerical stability with minimal error magnification—even across multiple checkpoint-restart cycles—outperforming state-of-the-art lossy compressors. Parallel evaluations on the Frontier supercomputer show up to 8.4× improvement in checkpoint write performance and 6.3× in read performance, while maintaining relative L2 errors ∼ 2e-6 throughout continued simulation. These results provide practical guidance for balancing compression accuracy, stability, and computational efficiency in large-scale PDE applications.

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

MPI Stages: Checkpointing MPI State for Bulk Synchronous Applications

When an MPI program experiences a failure, the most common recovery approach is to restart all processes from a previous checkpoint and to re-queue the entire job. A disadvantage of this method is that, although the failure occurred within the main application loop, live processes must start again from the beginning of the program, along with new replacement processes---this incurs unnecessary overhead for live processes. To avoid such overheads and concomitant delays, we introduce the concept of "MPI Stages." MPI Stages saves internal MPI state in a separate checkpoint in conjunction with application state. Upon failure, both MPI and application state are recovered, respectively, from their last synchronous checkpoints and continue without restarting the overall MPI job. Live processes roll back only a few iterations within the main loop instead of rolling back to the beginning of the program, while a replacement of failed process restarts and reintegrates, thereby achieving faster failure recovery. This approach integrates well with large-scale, bulk synchronous applications and checkpoint/restart. In this paper, we identify requirements for production MPI implementations to support state checkpointing with MPI Stages, which includes capturing and managing internal MPI state and serializing and deserializing user handles to MPI objects. We evaluate our fault tolerance approach with a proof-of-concept prototype MPI implementation that includes MPI Stages. We demonstrate its functionality and performance using LULESH and microbenchmarks. Our results show that MPI Stages reduces the recovery time by 13× for LULESH in comparison to checkpoint/restart.

Sultana, Nawrin↗

Recoverable distributed shared virtual memory - Memory coherence and storage structures

This paper examines the problem of implementing rollback recovery in multicomputer distributed shared virtual memory environments, in which the shared memory is implemented in software and exists only virtually. A user-transparent checkpointing recovery scheme and new twin-page disk storage management are presented to implement a recoverable distributed shared virtual memory. The checkpointing scheme is integrated with the shared virtual memory management. The twin-page disk approach allows incremental checkpointing without an explicit undo at the time of recovery. A single consistent checkpoint state is maintained on stable disk storage. The recoverable distributed shared virtual memory allows the system to restart computation from a previous checkpoint due to a processor failure without a global restart.

Wu, Kun-Lung↗

Fault-tolerance for exascale systems.

Periodic, coordinated, checkpointing to disk is the most prevalent fault tolerance method used in modern large-scale, capability class, high-performance computing (HPC) systems. Previous work has shown that as the system grows in size, the inherent synchronization of coordinated checkpoint/restart (CR) limits application scalability; at large node counts the application spends most of its time checkpointing instead of executing useful work. Furthermore, a single component failure forces an application restart from the last correct checkpoint. Suggested alternatives to coordinated CR include uncoordinated CR with message logging, redundant computation, and RAID-inspired, in-memory distributed checkpointing schemes. Each of these alternatives have differing overheads that are dependent on both the scale and communication characteristics of the application. In this work, using the Structural Simulation Toolkit (SST) simulator, we compare the performance characteristics of each of these resilience methods for a number of HPC application patterns on a number of proposed exascale machines. The result of this work provides valuable guidance on the most efficient resilience methods for exascale systems.

99 GENERAL AND MISCELLANEOUS↗

Checkpointing for a hybrid computing node

According to an aspect, a method for checkpointing in a hybrid computing node includes executing a task in a processing accelerator of the hybrid computing node. A checkpoint is created in a local memory of the processing accelerator. The checkpoint includes state data to restart execution of the task in the processing accelerator upon a restart operation. Execution of the task is resumed in the processing accelerator after creating the checkpoint. The state data of the checkpoint are transferred from the processing accelerator to a main processor of the hybrid computing node while the processing accelerator is executing the task.

97 MATHEMATICS AND COMPUTING↗

Checkpoint triggering in a computer system

According to an aspect, a method for triggering creation of a checkpoint in a computer system includes executing a task in a processing node of the computer system. A monitoring block size is determined for the checkpoint. A checkpoint interval is determined based on the monitoring block size, a checkpoint bandwidth, and a failure rate of the computer system. Based on determining that the checkpoint interval has elapsed, the checkpoint including state data of the task is created to enable restarting execution of the task upon a restart operation. The state data of the checkpoint is restored from a memory responsive to detecting an error condition at the processing node. Execution of the task is restarted in the processing node based on the state data restored from the memory.

97 MATHEMATICS AND COMPUTING↗

Hybrid Checkpointing for MPI Jobs in HPC Environments

As the core count in high-performance computing systems keeps increasing, faults are becoming common place. Check pointing addresses such faults but captures full process images even though only a subset of the process image changes between checkpoints. We have designed a hybrid check pointing technique for MPI tasks of high-performance applications. This technique alternates between full and incremental checkpoints: At incremental checkpoints, only data changed since the last checkpoint is captured. Our implementation integrates new BLCR and LAM/MPI features that complement traditional full checkpoints. This results in significantly reduced checkpoint sizes and overheads with only moderate increases in restart overhead. After accounting for cost and savings, benefits due to incremental checkpoints are an order of magnitude larger than overheads on restarts. We further derive qualitative results indicating an optimal balance between full/incremental checkpoints of our novel approach at a ratio of 1:9, which outperforms both always-full and always-incremental check pointing.

Wang, Chao↗

...And Eat it Too: High Read Performance in Write-Optimized HPC I/O Middleware File Formats

As HPC applications run on increasingly high process counts on larger and larger machines, both the frequency of checkpoints needed for fault tolerance and the resolution and size of Data Analysis Dumps are expected to increase proportionally. In order to maintain an acceptable ratio of time spent performing useful computation work to time spent performing I/O, write bandwidth to the underlying storage system must increase proportionally to this increase in the checkpoint and computation size. Unfortunately, popular scientific self-describing file formats such as netCDF and HDF5 are designed with a focus on portability and flexibility. Extra care and careful crafting of the output structure and API calls is required to optimize for write performance using these APIs. To provide sufficient write bandwidth to continue to support the demands of scientific applications, the HPC community has developed a number of I/O middleware layers, that structure output into write-optimized file formats. However, the obvious concern with any write optimized file format would be a corresponding penalty on reads. In the log-structured filesystem, for example, a file generated by random writes could be written efficiently, but reading the file back sequentially later would result in very poor performance. Simulation results require efficient read-back for visualization and analytics, and though most checkpoint files are never used, the efficiency of a restart is very important in the face of inevitable failures. The utility of write speed improving middleware would be greatly diminished if it sacrificed acceptable read performance. In this paper we examine the read performance of two write-optimized middleware layers on large parallel machines and compare it to reading data natively in popular file formats. The two I/O middleware layer examined in this paper are the Adaptable IO System (ADIOS), a library-based approach developed at Oak Ridge National Laboratory to provide a high-level IO API that can be used in place of netCDF or HDF5 to do much more aggressive write-behind and efficient reordering of data locations within the file; and the Parallel Log-structured Filesystem (PLFS), a stackable FUSE filesystem approach developed at Los Alamos National Laboratory that decouples concurrent writes to improve the speed of checkpoints. Since ADIOS is an I/O componentization that affords selection of different I/O methods at or during runtime, through a single API, users can have access to MPI-IO, Posix-IO, HDF5, netCDF, and staging methods. The ADIOS BP file format is a new log-file format that has a superset of the features of both HDF5 and netCDF, but is designed to be portable and flexible while being optimized for writing. PLFS, on the other hand, is mounted as a stackable filesystem on top of an existing parallel filesystem. Reads or writes to the PLFS filesystem are transparently translated into operations on per-process log files stored in the underlying parallel filesystem. Since PLFS performs this translation without application modification, users can write in HDF5, netCDF, or app-specific file formats and PLFS will store the writes in a set of efficiently written log-formatted files, while presenting the user with a logical 'flat' file on reads. Despite their different approaches, the commonality behind both of these middleware systems is that they both write to a log file format. As shown in previous publications, writes are fully optimized in both systems, sometimes resulting in 100x improvements over writing data in popular file formats. But as mentioned above, writing is only one part of the story. In this paper we examine the read performance of our middleware layers on large parallel machines and compare these to reading data either natively or from other popular file formats. We compare the reading performance in two different scenarios: (1) Reading back restarts from the same number of processors as wrote the data and (2) Reading back restart data from a different number of processors. We observe that not only can write-optimized I/O middleware be built to not penalize read speeds, but for important workloads, techniques that improve write performance can, perhaps counterintuitively, improve read speeds over reading from a contiguously organized file format. In the remainder of this paper, we investigate this further through case studies of PLFS and ADIOS on simulation checkpoint restarts.

99 GENERAL AND MISCELLANEOUS↗

Software Resilience using Kokkos Ecosystem

Due to the cost of hardware failures within mission critical and scientific applications, it is necessary for software to provide a mechanism to prevent or recover from interruptions. The Kokkos ecosystem is a programming environment that provides performance and portability to many applications that run on DOE supercomputers as well as smaller scale systems. These applications require a higher level of service due to the cost associated with each simulation or the critical nature of the mission. Software resilience enables an application of manage hardware failures reducing the cost of an interruption. Two different resilience methodologies have been added to the Kokkos ecosystem: checkpointing has been added for restart capabilities and a resilient execution model has been added to account for failures in compute devices. The design and implementation of each of these additions are described, and appropriate examples are included for end users.

97 MATHEMATICS AND COMPUTING↗

X10: A FORTRAN direct access data management system

The XIO system is a set of subroutines that provide generalized data management capability for FORTRAN programs using a direct access file. Arrays of integer, real, double precision, and character data may be stored, each logical group of data identified by a unique matrix number. A matrix may be organized and stored as batches to reduce core requirements. Batches may be accessed randomly or sequentially. The file may be checkpointed and retained, allowing for restarts with stored values. The XIO subroutines operate on either IBM 360-370/OS/VS or DEC PDP-11/RSX computing systems.

Roland, D. P.↗