SPEC OMP2012 — An Application Benchmark Suite for Parallel Systems Using OpenMP

Müller, Matthias S.; Baron, John; Brantley, William C.; Feng, Huiyu; Hackenberg, Daniel; Henschel, Robert; Jost, Gabriele; Molka, Daniel; Parrott, Chris; Robichaux, Joe; Shelepugin, Pavel; Waveren, Matthijs van; Whitney, Brian; Kumaran, Kalyan

doi:10.1007/978-3-642-30961-8_17

Cited by 67 publications

(30 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SPEC has several different testing components that can be utilized to benchmark a system. For our benchmarking comparison we will use the SPEC OMP2012 [16] because it appears to represent a vast array of new and emerging parallel applications while simultaneously providing a comparison to other SPEC benchmarks. SPEC OMP continues the SPEC tradition of giving HPC users the most objective and representative benchmark suite for measuring the performance of SMP (shared memory multi-processor) systems.…”

Section: A Spec Openmpmentioning

confidence: 99%

“…This shows there is a very large difference between the two systems that can occur. Looking at these 4 benchmarks in more detail from the SPEC OMP2012 publication [16], we see each application has a high memory footprint, indicating there is a substantial amount of memory transfers occurring between each SMP system's nodes. These RDMA memory accesses occur across the QDR InfiniBand network for the vSMP at a bandwidth of up to 40Gbs, whereas the SGI UV uses the NUMAlink5 interconnect with only a 15Gbs peak bandwidth through 2 unidirectional 7.5Gbs links.…”

Section: A Openmpmentioning

confidence: 99%

“…It is also worth stressing that we are only characterizing two deployed systems, which may not accurately represent other current or future ScaleMP vSMP or SGI UV deployments, as there exists many differences between each deployment. This discrepancy between deployments can be seen in the reported improved performance of the SGI UV results provided by SGI Corporation on the SPEC OMP2012 website [16].…”

Section: A Openmpmentioning

confidence: 99%

See 2 more Smart Citations

Evaluation of SMP Shared Memory Machines for Use with In-Memory and OpenMP Big Data Applications

Younge

Reidy

Henschel

et al. 2016

2016 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)

Self Cite

View full text Add to dashboard Cite

Abstract-While distributed memory systems have shaped the field of distributed systems for decades, the demand for many-core shared memory resources is increasing. Symmetric Multiprocessor Systems (SMPs) have become increasingly important recently among a wide array of disciplines, ranging from Bioinformatics to astrophysics, and beyond. With the increase in big data computing, the size and scope of traditional commodity server systems is often outpaced. While some big data applications can be mapped to distributed memory systems found through many cluster and cloud technologies today, this effort represents a large barrier of entry that some projects cannot cross. Shared memory SMP systems look to effectively and efficiently fill this niche within distributed systems by providing high throughput and performance with minimized development effort, as the computing environment often represents what many researchers are already familiar with.In this paper we look at the use of two common shared memory systems, the ScaleMP vSMP virtualized SMP deployment at Indiana University, and the SGI UV architecture deployed at University of Arizona. While both systems are notably different in their design, their potential impact on computing are remarkably similar. As such, we look to compare each system first under a set of OpenMP threaded benchmarks via the SPEC group, and follow up with our experience using each machine for Trinity de-novo assembly. We find both SMP systems are well suited to support various big data applications, with the newer vSMP deployment often slightly faster, however certain caveats and performance considerations are necessary when considering such SMP systems.

show abstract

Section: A Spec Openmpmentioning

confidence: 99%

Section: A Openmpmentioning

confidence: 99%

Section: A Openmpmentioning

confidence: 99%

See 1 more Smart Citation

Evaluation of SMP Shared Memory Machines for Use with In-Memory and OpenMP Big Data Applications

Younge

Reidy

Henschel

et al. 2016

2016 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Table 1 shows some profile statistics related to several applications taken from the SPEC CPU2006 suite [11], the Princeton Application Repository for Shared-Memory Computers (PARSEC 2.1) [2], and the SPEC OMP2012 suite [15]. The profiles have been obtained using the profiler aprof-0.2.1 [1] while running these applications on their reference workloads.…”

Section: Visual Mining Of Input-sensitive Profilesmentioning

confidence: 99%

An interactive visualization framework for performance analysis

Coppa¹

2015

Proceedings of the 8th International Conference on Performance Evaluation Methodologies and Tools

View full text Add to dashboard Cite

Input-sensitive profiling is a recent methodology for analyzing how the performance of a routine scales as a function of the workload size. As increasingly more detailed profiles are collected by an input-sensitive profiler, the information conveyed to a user can quickly become overwhelming. In this paper, we present an interactive graphical tool called aprof-plot for visualizing performance profiles. Exploiting curve fitting techniques, aprof-plot can estimate the asymptotic complexity of each routine, pointing the attention of the programmer to the most critical routines of an application. A variety of routine-based charts can be automatically generated by our tool, allowing the developer to analyze the performance scalability of a routine. Several examples based on real-world applications are discussed, showing how to conduct an effective performance investigation using aprof-plot.

show abstract

“…Though it was developed for server workloads, the SPEC power measurement methodology and tools can be applied to others, as Hackenberg [10] demonstrates with his analysis of SPECMPI workloads run on a small cluster of nodes. The SPEC High Performance Group [19] has recently included the option and specification for power submissions as part of the SPEC OMP2012 benchmark suite [20] as an optional add-on and has since been analyzed by Muller [17]. While SPEC's methodology is precise and widely applied in multiple domains, it is not designed with supercomputer-scale systems in mind.…”

Section: Related Workmentioning

confidence: 99%