Guided Self-Scheduling: A Practical Scheduling Scheme for Parallel Supercomputers

Polychronopoulos, Constantine D.; Kuck, David J.

doi:10.1109/tc.1987.5009495

Cited by 518 publications

(301 citation statements)

References 10 publications

Supporting

Mentioning

292

Contrasting

Unclassified

Order By: Relevance

“…tion boundaries; the applications examine and adjust to the number of available processors each time they begin an iteration, but do not do so while executing any one iteration. It is clearly possible to do much more dynamic scheduling, e.g., [20,1,6, 14]; we did not do so because of the very large incremental implementation cost relative to our more restrictive change, and because we expect that ST-EQUI would perform even better when jobs are more responsive to changes in their allocations. (Of the three policies, ST-EQUI reallocates processors most frequently, and is therefore most sensitive to the latency with which applications can respond to changing allocations.…”

Section: Methodsmentioning

confidence: 99%

Using runtime measured workload characteristics in parallel processor scheduling

Nguyen

Vaswani

Zahorjan

1996

Job Scheduling Strategies for Parallel Processing

View full text Add to dashboard Cite

We consider the use of runtime measured workload characteristics in parallel processor scheduling. Although many researchers have considered the use of application characteristics in this domain, most of this work has assumed that such information is available a priori. In contrast, we propose and evaluate experimentally dynamic processor allocation policies that rely on determining job characteristics at runtime; in particular, we focus on measuring and using job efficiency and speedup.Our work is intended to be a first step towards the eventual development of production schedulers that use runtime measured workload characteristics in making their decisions. We consider two distinct scheduling scenarios: interactive systems, where minimizing response time is the goal, and batch systems, where maximizing useful instruction throughput is the goal. In both these environments, our experimental results validate the following observations:• Despite the inherent inaccuracies of runtime measurements and the added overhead of more frequent reallocations, schedulers that use runtime measurements of workload characteristics can significantly outperform schedulers that are oblivious to these characteristics.• Runtime measurements are sufficient for schedulers to achieve performance surprisingly close to that possible when efficiency and speedup information is available a priori.• The primary performance loss, relative to the use of a priori information, is due to the transient decisions of the schedulers as they acquire information on the running applications, rather than to measurement and reallocation overheads.Our experiments are performed using prototype implementations running on a 50-node KSR-2 shared memory multiprocessor.

show abstract

Section: Methodsmentioning

confidence: 99%

Using runtime measured workload characteristics in parallel processor scheduling

Nguyen

Vaswani

Zahorjan

1996

Job Scheduling Strategies for Parallel Processing

View full text Add to dashboard Cite

show abstract

“…IMSAME uses a modified Guided Self Scheduling (GSS) [19] to handle workload assignment. In this line, the query metagenome (the reference metagenome is only processed to generate the hash table) is separated into M partitions each of which will contain m i subpartitions, with i ranging from 1 to M .…”

Section: Dynamic Workload Partitioning and Distribution To Threadsmentioning

confidence: 99%

Accelerating Exhaustive Pairwise Metagenomic Comparisons

Pérez-Wohlfeil

Torreño

Trelles

2017

Algorithms and Architectures for Parallel Processing

View full text Add to dashboard Cite

Abstract. In this manuscript, we present an optimized and parallel version of our previous work IMSAME, an exhaustive gapped aligner for the pairwise and accurate comparison of metagenomes. Parallelization strategies are applied to take advantage of modern multiprocessor architectures. In addition, sequential optimizations in CPU time and memory consumption are provided. These algorithmic and computational enhancements enable IMSAME to calculate near optimal alignments which are used to directly assess similarity between metagenomes without requiring reference databases. We show that the overall efficiency of the parallel implementation is superior to 80% while retaining scalability as the number of parallel cores used increases. Moreover, we also show that sequential optimizations yield up to 8x speedup for scenarios with larger data. BackgroundA metagenome is defined as a collection of genetic material directly recovered from the environment. In particular, a metagenome is composed of a large number of reads (DNA strings) drawn from the species present in the original population. To this day, the field of comparative metagenomics has become big-data driven [1] due to new technological improvements in high-throughput sequencing. However, the analysis of large metagenomic datasets represents a computational challenge and poses several processing bottlenecks, specially to sequence comparison algorithms.Traditional metagenomics comparison involve intermediate pairwise (and individual) comparisons against a reference database. This procedure allows to extract a mapping distribution between reads and species, and thus enables to later on compare these distributions. A similarity measure can then be computed from the two distributions. However, due to the unknown and complex composition Corresponding author.

show abstract

“…To account for a good workload balancing, the different loop indexes are associated to the concurrent threads using the guided self-scheduling algorithm [41] available in OpenMP. For the class Complex, see Figure 1, we use the C++ standard complex class.…”

Section: Begin_algorithm // Tridiagonalization Of the N X N A Matrix mentioning

confidence: 99%

A Concurrent Object-Oriented Approach to the Eigenproblem Treatment in Shared Memory Multicore Environments

Niño

Muñoz‐Caro

Reyes

2011

Computational Science and Its Applications - ICCSA 2011

View full text Add to dashboard Cite

Abstract. This work presents an object-oriented approach to the concurrent computation of eigenvalues and eigenvectors in real symmetric and Hermitian matrices on present memory shared multicore systems. This can be considered the lower level step in a general framework for dealing with large size eigenproblems, where the matrices are factorized to a small enough size. The results show that the proposed parallelization achieves a good speedup in actual systems with up to four cores. Also, it is observed that the limiting performance factor is the number of threads rather than the size of the matrix. We also find that a reasonable upper limit for a "small" dense matrix to be treated in actual processors is in the interval 10000-30000.

show abstract

Guided Self-Scheduling: A Practical Scheduling Scheme for Parallel Supercomputers

Cited by 518 publications

References 10 publications

Using runtime measured workload characteristics in parallel processor scheduling

Using runtime measured workload characteristics in parallel processor scheduling

Accelerating Exhaustive Pairwise Metagenomic Comparisons

A Concurrent Object-Oriented Approach to the Eigenproblem Treatment in Shared Memory Multicore Environments

Contact Info

Product

Resources

About