Efficient collective communication on heterogeneous networks of workstations

Banikazemi, Mohammad; Moorthy, Vijay; Panda, Dhabaleswar K.

doi:10.1109/icpp.1998.708518

Cited by 81 publications

(130 citation statements)

References 14 publications

Supporting

Mentioning

124

Contrasting

Order By: Relevance

“…An approximation algorithm for multicast is given, with a competitive ratio logk where k is the number of destination of the multicast [4]. Banikazemi et al [2] proposed a simple model in which the heterogeneity among processors is characterized by the speed of sending processors, and show that a broadcast technique called fastest-node-first works well in practice. We will refer to this model as the sender-only model.…”

Section: Communication Modelmentioning

confidence: 99%

“…It is difficult to find the optimal broadcast tree that minimizes the total broadcast time in a heterogeneous cluster, therefore a simple heuristic called fastest-nodefirst (FNF) is proposed in [2] to find a reasonably good broadcast schedule for the original sender-only heterogeneous model [2].…”

Section: Fastest-node-first Techniquementioning

confidence: 99%

“…al. [2], is very effective in designing broadcast protocols for heterogeneous cluster systems. The fastest-node-first technique schedules the processors to receive the broadcast in the order of their communication speed, that is, the faster node should be scheduled earlier.…”

Section: Introductionmentioning

confidence: 99%

“…Every workstation cluster, be it homogeneous or heterogeneous, requires efficient collective communication [2]. For example, a barrier synchronization is often placed between two successive phases of computation to make sure that all processors finish the first phase before any can go to the next phase.…”

Section: Introductionmentioning

confidence: 99%

“…However, a cluster may consist of different types of processors and this heterogeneity within a cluster complicates the design of efficient collective communication protocols. This paper shows that a simple heuristic called fastest-node-first (FNF) [2] is very effective in reducing broadcast time for heterogeneous cluster systems. Despite the fact that FNF heuristic does not guarantee an optimal broadcast time for general heterogeneous network of workstation, we prove that FNF always gives near optimal broadcast time in a special case of cluster, and this finding helps us show that FNF delivers guaranteed performance for general clusters.…”

mentioning

confidence: 99%

See 4 more Smart Citations

An Approximation Algorithm for Broadcast Scheduling in Heterogeneous Clusters

Liu

Wang

Guo

2004

Real-Time and Embedded Computing Systems and Applications

View full text Add to dashboard Cite

Abstract. Network of workstation (NOW) is a cost-effective alternative to massively parallel supercomputers. As commercially available off-theshelf processors become cheaper and faster, it is now possible to build a PC or workstation cluster that provides high computing power within a limited budget. However, a cluster may consist of different types of processors and this heterogeneity within a cluster complicates the design of efficient collective communication protocols. This paper shows that a simple heuristic called fastest-node-first (FNF) [2] is very effective in reducing broadcast time for heterogeneous cluster systems. Despite the fact that FNF heuristic does not guarantee an optimal broadcast time for general heterogeneous network of workstation, we prove that FNF always gives near optimal broadcast time in a special case of cluster, and this finding helps us show that FNF delivers guaranteed performance for general clusters. In a previous paper we showed a similar bound on the competitive ratio in a send-only communication model. This paper extends the result to a more realistic sender-receiver model. We show that FNF gives a total broadcast of 2T + β, where T is the optimum time and β is a constant. This improves over the previous bound on 2αT + β [17], where α is a theoretically unbounded ratio of the processor performance in the cluster.

show abstract

Section: Communication Modelmentioning

confidence: 99%

Section: Fastest-node-first Techniquementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

An Approximation Algorithm for Broadcast Scheduling in Heterogeneous Clusters

Liu

Wang

Guo

2004

Real-Time and Embedded Computing Systems and Applications

View full text Add to dashboard Cite

show abstract

Broadcast scheduling for wide-area parallel distributed systems

Tasaki

Tamura

Sengoku

et al. 2004

Electron. Comm. Jpn. Pt. III

View full text Add to dashboard Cite

SUMMARYParallel distributed systems in which multiple computers are connected through LAN and WAN are widely used at present. One of the important functions of parallel distributed systems is broadcasting to distribute data over an entire system. Broadcast processing greatly affects the performance of parallel distributed systems, and it is desirable to develop a distribution procedure that completes the processing in a short time. Consequently, there have been many approaches to the minimum broadcast time problem, aiming at the optimal distribution procedure. However, not many past studies of this problem consider a wide-area system in which multiple parallel distributed systems are connected. In this context, this paper considers the minimum broadcast time problem for a wide-area parallel distributed system whose topology can be represented by a certain kind of split graph. It is shown that the problem can be solved in polynomial time in the homogeneous parallel distributed system in which the time required for communication is uniform. In addition, a heuristic algorithm is proposed for a heterogeneous parallel distributed system in which the time required for communication is not uniform. Simulation results reveal that an efficient distribution procedure can be derived in a short time.

show abstract

Scheduling for broadcast operation in heterogeneous parallel computing environments

Ooshita

Matsumae

Masuzawa

et al. 2004

Systems & Computers in Japan

View full text Add to dashboard Cite

SUMMARYParallel computing utilizes heterogeneous parallel computing environments in which multiple processors of various kinds are connected by communication links. Efficient collective communication operations are needed to realize efficient parallel computing in parallel computing environments. Although methods for realizing efficient collective communication operations have previously been proposed for homogeneous parallel computing environments, these methods applied as they are to heterogeneous parallel computing environments cannot realize efficient collective communication operations. In this paper, we consider design of an efficient broadcast, which is one of the most important collective communication operations. We show that we can extend the known sufficient condition for Fastest-Node-First (FNF) to output an optimal schedule, where FNF is an efficient heuristic algorithm for the broadcast. We show that we can reduce the time cost of the branch-and-bound search for the optimal schedule.

show abstract

Efficient collective communication on heterogeneous networks of workstations

Abstract: Networks of Workstations (NOW)

Cited by 81 publications

References 14 publications

An Approximation Algorithm for Broadcast Scheduling in Heterogeneous Clusters

An Approximation Algorithm for Broadcast Scheduling in Heterogeneous Clusters

Broadcast scheduling for wide-area parallel distributed systems

Scheduling for broadcast operation in heterogeneous parallel computing environments

Contact Info

Product

Resources

About