An Operating System support to low-overhead communications in NOW clusters

Marenzoni, Paolo; Rimassa, Giovanni; Bertozzi, Massimo; Conte, Gianni; Rossi, P.

doi:10.1007/3-540-62573-9_10

Cited by 14 publications

(4 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fewer research projects focused on lower cost ensembles of PC-based computation nodes networked by o -the-shelf Local Area Networks LANs hardware, aiming a t producing l e s s a m bitious NOWs in terms of absolute performance, but providing b e t ter cost performance gures, such a s t he NASA's Beowulf 11 , Cornell's U-Net on Linux 15 , PARMA 2 6 , and Illinois Fast Messages on Linux 8 .…”

Section: Introductionmentioning

confidence: 99%

Optimal communication performance on fast ethernet with GAMMA

Ciaccio

1998

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. The current prototype of the Genoa Active M essage MAchine GAMMA is a low-overhead, Active M essages-based inter-process communication layer implemented mainly at k ernel level in the Linux Operating System. It run s o n a p o o l o f l o w-cost Pentium-based Personal Computers PCs networked by a l o w-cost 100base-TX Ethernet hub to form a low-cost message-passing parallel platform. In this paper we describe in detail how GAMMA could achieve u nprecedented communication performance less than 13 s o n e-way user-to-user latency time a n d up to 98 of the communication throughput o f t he raw interconnection hardware on such a kind o f l o w-cost parallel architecture.

show abstract

Section: Introductionmentioning

confidence: 99%

Optimal communication performance on fast ethernet with GAMMA

Ciaccio

1998

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…PARMA2 [9] was an attempt at improving network performance by rewriting from scratch an IP-based protocol stack similar to UDP. Indeed, PARMA2 was comparable to UDP in performance, that is, only marginally better compared to TCP.…”

Section: Related Workmentioning

confidence: 99%

Architecture and implementation of the socket interface on top of GAMMA

Schneidenbach

Schnor

Petri

28th Annual IEEE International Conference on Local Computer Networks, 2003. LCN '03. Proceedings.

View full text Add to dashboard Cite

GAMMA (the Genoa Active Message MAchine) is a lightweight messaging system for Fast and Gigabit Ethernet. It is based on an Active Message-like paradigm and provides a performing, cost-effective alternative to proprietary high-speed networks, e. g. Myrinet, with the combination of low end-to-end latency and high throughput.GAMMA supports the important class of MPI based parallel applications via the MPI/GAMMA interface [5], but up to now support for the also important class of socket based cluster applications is still missing. This paper describes two different approaches to how the socket interface can be adapted to GAMMA: The first is transparent for both application and GAMMA layer, the second is only transparent for the application. First performance results with the so-called GAMMAsockets are given. They show that the second approach performs almost as good as native GAMMA communication.

show abstract

“…u PARMA [62] -is designed as a socket compatibility layer with no mechanism associated to flow control and data recovery. This is because the authors believed that the percentage of packets incorrectly delivered by a homogeneous local area network is negligible.…”

Section: Ethernet-based Systemsmentioning

confidence: 99%

Performance studies of high-speed communication on commodity cluster

Tam¹,

譚達俊²

View full text Add to dashboard Cite

Parallel computing is becoming increasingly accessible through advancement in microprocessors and networking technologies. It is found that the performances of the clusters do not match with their promises, although they are built on the most advanced components. Much effort has been devoted to address the software overhead problem in the past, which is known as the major hindrance in achieving high performance. This thesis shows that having a low-latency communication system does not guarantee high performance, as there are other communication issues that have not been addressed by the use of low-latency communication, such as contention, communication patterns and scheduling of communication events. The development of an efficient parallel application depends upon a realistic prediction of application behavior and the ability to explain the performance characteristics of an application on a parallel system; this requires in-depth understanding of both the application and the architecture characteristics.This dissertation proposes the use of a realistic communication model to guide the performance understanding and the algorithm design processes, which are the keys to achieve high performance. The model includes a collection of performance parameters which correspond to essential features in the communication architecture, and a collection of benchmark methodologies that quantify these performance features. This model can be used as a framework for programmers to conduct performance studies on various communication issues. We use this framework to examine the performance characteristics of a lightweight messaging system, and show that the model can be effectively used as an evaluation tool as well as an emulating tool.This thesis explores the congestion problem with the support of the communication model. Through modeling studies and experimental evaluations, we examine how different buffering mechanisms interact with a Go-Back-N reliable protocol, and show that the buffering mechanism dominates the congestion behavior of the communication systems. In the performance studies, we find that the timeout and flow control settings are the prevailing factors that interact with the buffering mechanism to determine how the congestion evolves. As our model is derived on a resource-centric view of how data move across the communication system, we use it to show that with careful design of communication schedules, we can achieve efficient communication as well as prevent congestion. We have developed a complete exchange algorithm, the Synchronous Shuffle Exchange, which is an optimal algorithm on the non-blocking network. To avoid congestion loss caused by the non-deterministic delays in communication events, a global congestion control scheme is introduced. This scheme uses a global windowing concept to coordinate all participating nodes to monitor and regulate the traffic load, which effectively avoids congestion loss and maintains sufficient throughput to maximize the performance.In summary, this thesis shows that the ...

show abstract

An Operating System support to low-overhead communications in NOW clusters

Cited by 14 publications

References 11 publications

Optimal communication performance on fast ethernet with GAMMA

Optimal communication performance on fast ethernet with GAMMA

Architecture and implementation of the socket interface on top of GAMMA

Performance studies of high-speed communication on commodity cluster

Contact Info

Product

Resources

About