Evaluation of Collective I/O Implementations on Parallel Architectures

Dickens, Phillip M.; Thakur, Rajeev

doi:10.1006/jpdc.2000.1733

Cited by 12 publications

(9 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This approach requires a tight coupling between MPI and the underlying file system [25]. Algorithms termed as "two-phase I/O" [9], [29] enable efficient collective I/O implementations by aggregating requests and by adapting the write pattern to the file layout across multiple data servers [7]. Collective I/O avoids metadata redundancy as opposed to the file-per-process approach.…”

Section: B Approaches To I/o Management In Hpc Simulationsmentioning

confidence: 99%

Damaris: How to Efficiently Leverage Multicore Parallelism to Achieve Scalable, Jitter-free I/O

Dorier

Antoniu

Cappello

et al. 2012

2012 IEEE International Conference on Cluster Computing

View full text Add to dashboard Cite

With exascale computing on the horizon, the performance variability of I/O systems represents a key challenge in sustaining high performance. In many HPC applications, I/O is concurrently performed by all processes, which leads to I/O bursts. This causes resource contention and substantial variability of I/O performance, which significantly impacts the overall application performance and, most importantly, its predictability over time. In this paper, we propose a new approach to I/O, called Damaris, which leverages dedicated I/O cores on each multicore SMP node, along with the use of sharedmemory, to efficiently perform asynchronous data processing and I/O in order to hide this variability. We evaluate our approach on three different platforms including the Kraken Cray XT5 supercomputer (ranked 11th in Top500), with the CM1 atmospheric model, one of the target HPC applications for the Blue Waters postpetascale supercomputer project. By overlapping I/O with computation and by gathering data into large files while avoiding synchronization between cores, our solution brings several benefits: 1) it fully hides jitter as well as all I/O-related costs, which makes simulation performance predictable; 2) it increases the sustained write throughput by a factor of 15 compared to standard approaches; 3) it allows almost perfect scalability of the simulation up to over 9,000 cores, as opposed to state-of-the-art approaches which fail to scale; 4) it enables a 600% compression ratio without any additional overhead, leading to a major reduction of storage requirements.

show abstract

Section: B Approaches To I/o Management In Hpc Simulationsmentioning

confidence: 99%

Damaris: How to Efficiently Leverage Multicore Parallelism to Achieve Scalable, Jitter-free I/O

Dorier

Antoniu

Cappello

et al. 2012

2012 IEEE International Conference on Cluster Computing

View full text Add to dashboard Cite

show abstract

“…Collective I/O plays a critical role in cooperating processes to generate aggregated I/O requests, instead of performing noncontiguous small I/Os independently [9,37]. As we discussed in the background section, a widely-used implementation of collective I/O is the two-phase I/O protocol [37].…”

Section: Motivationmentioning

confidence: 99%

“…5 shows the aggregator's sending time, which is the time this aggregator takes to send the data to the collective buffer. The lower dashed portion in other bars (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16) shows the waiting time of non-aggregators and the higher portion refers to their receiving time. We can see that the data exchange time between aggregator and non-aggregators is various, due to which, we can argue that the shuffle cost is determined by the maximum data exchange time (in this case, i.e., 2.3 ms).…”

Section: Analysis and Prediction Of Shuffle Costmentioning

confidence: 99%

Hierarchical Collective I/O Scheduling for High-Performance Computing

2015

View full text Add to dashboard Cite

The non-contiguous access pattern of many scientific applications results in a large number of I/O requests, which can seriously limit the data-access performance. Collective I/O has been widely used to address this issue. However, the performance of collective I/O could be dramatically degraded in today's high-performance computing systems due to the increasing shuffle cost caused by highly concurrent data accesses. This situation tends to be even worse as many applications become more and more data intensive. Previous research has primarily focused on optimizing I/O access cost in collective I/O but largely ignored the shuffle cost involved. Previous works assume that the lowest average response time leads to the best QoS and performance, while that is not always true for collective requests when considering the additional shuffle cost. In this study, we propose a new hierarchical I/O scheduling (HIO) algorithm to address the increasing shuffle cost in collective I/O. The fundamental idea is to schedule applications' I/O requests based on a shuffle cost analysis to achieve the optimal overall performance, instead of achieving optimal I/O accesses only. The algorithm is currently evaluated with the MPICH3 and PVFS2. Both theoretical analysis and experimental tests show that the proposed hierarchical I/O scheduling has a potential in addressing the degraded performance issue of collective I/O with highly concurrent accesses.

show abstract

“…[1,2,7,9,11,12,13,17]. Most work is about the implementation of the well-known, portable open-source ROMIO implementation of MPI-IO [12].…”

Section: Introductionmentioning

confidence: 99%

“…For collective I/O, data sieving is combined with techniques like two-phase I/O [11,13]. Orthogonal work has dealt with hiding file access time through active buffering and I/O threads [2,7], or optimizations for specific, low-level file systems [1,9]. To the best of our knowledge, no attention per se has been paid to the efficient handling of non-contiguous, typed MPI data buffers as are always involved in non-contiguous file access.…”

Section: Introductionmentioning

confidence: 99%

Fast Parallel Non-Contiguous File Access

Worringen

Träff

Ritzdorf

2003

Proceedings of the 2003 ACM/IEEE Conference on Supercomputing

View full text Add to dashboard Cite

Many applications of parallel I/O perform non-contiguous file accesses: instead of accessing a single (large) block of data in a file, a number of (smaller) blocks of data scattered throughout the file needs to be accessed in each logical I/O operation. However, only few file system interfaces directly support this kind of non-contiguous file access. In contrast, the most commonly used parallel programming interface, MPI, incorporates a flexible model of parallel I/O through its MPI-IO interface. With MPI-IO, arbitrary non-contiguous file accesses are supported in a uniform fashion by the use of derived MPI datatypes set up by the user to reflect the desired I/O pattern.Despite a considerable amount of recent work in this area, current MPI-IO implementations suffer from low performance of such non-contiguous accesses when compared to the performance of the storage system for contiguous accesses. In this paper we analyze an important bottleneck in the efficient handling of non-contiguous access patterns in current implementations of MPI-IO. We present a new technique, termed listless I/O, that can be incorporated into MPI-IO implementations like the well-known ROMIO implementation, and completely eliminates this bottleneck. We have implemented the technique in MPI/SX, the MPI implementation for the NEC SX-series of parallel vector computers. Results with a synthetic benchmark and an application kernel show that listless I/O is able to increase the bandwidth for non-contiguous file access by sometimes more than a factor of 500 when compared to the traditional approach.

show abstract

Evaluation of Collective I/O Implementations on Parallel Architectures

Cited by 12 publications

References 20 publications

Damaris: How to Efficiently Leverage Multicore Parallelism to Achieve Scalable, Jitter-free I/O

Damaris: How to Efficiently Leverage Multicore Parallelism to Achieve Scalable, Jitter-free I/O

Hierarchical Collective I/O Scheduling for High-Performance Computing

Fast Parallel Non-Contiguous File Access

Contact Info

Product

Resources

About