Impact of heterogeneity on DSM performance

Figueiredo, Renato; Fortes, José A. B.

doi:10.1109/hpca.2000.824336

Cited by 14 publications

(13 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, these algorithms have limited use in practice due to simplified and often unrealistic assumptions, such as a priori knowledge of task runtime and dependencies. Figueiredo and Fortes [10] studied heterogeneous distributed shared-memory (DSM) multiprocessors and proposed a static scheduling policy that uses processor performance ratios. This policy is similar to our asymmetry-aware load balancing algorithm, but different in that it performs only static assignments, whereas our algorithm performs dynamic load balancing.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Efficient operating system scheduling for performance-asymmetric multi-core architectures

Baumberger

Koufaty

et al. 2007

Proceedings of the 2007 ACM/IEEE Conference on Supercomputing

176

119

View full text Add to dashboard Cite

Recent research advocates asymmetric multi-core architectures, where cores in the same processor can have different performance. These architectures support single-threaded performance and multithreaded throughput at lower costs (e.g., die size and power). However, they also pose unique challenges to operating systems, which traditionally assume homogeneous hardware. This paper presents AMPS, an operating system scheduler that efficiently supports both SMPand NUMA-style performance-asymmetric architectures. AMPS contains three components: asymmetry-aware load balancing, fastercore-first scheduling, and NUMA-aware migration. We have implemented AMPS in Linux kernel 2.6.16 and used CPU clock modulation to emulate performance asymmetry on an SMP and NUMA system. For various workloads, we show that AMPS achieves a median speedup of 1.16 with a maximum of 1.44 over stock Linux on the SMP, and a median of 1.07 with a maximum of 2.61 on the NUMA system. Our results also show that AMPS improves fairness and repeatability of application performance measurements.

show abstract

Section: Related Workmentioning

confidence: 99%

“…To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. SC07 November [10][11][12][13][14][15][16]2007, Reno, Nevada, USA Copyright 2007 ACM 978-1-59593-764-3/07/0011 ...$5.00. making scheduling decisions.…”

Section: Introductionmentioning

confidence: 99%

Efficient operating system scheduling for performance-asymmetric multi-core architectures

Baumberger

Koufaty

et al. 2007

Proceedings of the 2007 ACM/IEEE Conference on Supercomputing

176

119

View full text Add to dashboard Cite

show abstract

“…Menasce and Almeida [20] proposed the use of heterogeneous processors in supercomputers, with two different types of processors used to speed up the parallel and sequential portions of a computation. Figueiredo and Fortes [8] explored heterogeneous distributed-shared-memory multiprocessors with a few nodes with large caches designed for single-thread performance, and a larger number of nodes with smaller caches designed for multi-threaded parallelism. Morad et al [21] presented an analysis of why an asymmetric chip multiprocessor can achieve higher performance for a given area and power budget.…”

Section: Related Workmentioning

confidence: 99%

Mitigating Amdahl's Law through EPI Throttling

Annavaram¹,

Grochowski²,

Shen³

2005

SIGARCH Comput. Archit. News

136

164

View full text Add to dashboard Cite

show abstract

“…Traditional graph-based algorithms [4,27,28] are impractical due to assumptions such as a priori knowledge of task runtime and dependencies. Figueiredo and Fortes [10] studied heterogeneous distributed shared-memory multiprocessors and proposed an algorithm using processor performance ratios, similar to our CPU ratings. Their algorithm, however, is static and not suitable for OS scheduling.…”

Section: Related Workmentioning

confidence: 99%