Communication in iWarp systems

Groß, Thomas

doi:10.1145/76263.76311

Cited by 14 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…if (si d i) does not con ict (6) with any connection in Ck (7) then (8) Ck = Ck S f (si d i) g (9) R = R ; f (si d i) g (10) end if (11) end for (12) MC = MC S f Ck g (13) k = k + 1 (14) until R = In this case, (0, 2) will be in time slot 1, (1, 3) in time slot 2, (3, 4) in time slot 1 and (2, 4) in time slot 3. Therefore, a multiplexing degree of 3 is needed to establish the paths for the four connections.…”

Section: Greedy Algorithmmentioning

confidence: 99%

“…The metric used to compare the algorithms is the multiplexing degree needed to establish the connections. The performance of the algorithms is evaluated using randomly (1) PhaseRank *] = 0 (2) for(si d i) 2 R do (3) let (si d i ) 2 Ak (4) PhaseRank k] = PhaseRank k] + length((si d i)) (5) end for (6) sort phase according to PhaseRank (7) Reorder R according the sorted phases. (8) call greedy algorithm Figure 5: Ordered AAPC scheduling algorithm.…”

Section: Performance Of the Scheduling Algorithmsmentioning

confidence: 99%

See 1 more Smart Citation

Compiled communication for all-optical TDM networks

Yuan

Melhem

Gupta

1996

Proceedings of the 1996 ACM/IEEE Conference on Supercomputing

View full text Add to dashboard Cite

While all{optical net works o er large bandwidth for transferring data, the control mechanisms to dynamically establish all{optical paths incur large overhead. In this paper, we consider adapting all{optical multiplexed networks in multiprocessor or multicomputer environment b y using compiled c ommunication as an alternative to dynamic network control. Compiled comm unication eliminates the runtime overhead by managing network resources statically. T h us, it can employ complex o {line algorithms to impro ve resource utilization. W e studied several o {line connection scheduling algorithms for minimizing the multiplexing degree required to satisfy communication requests. The performance of compiled communication is evaluated and compared with that of dynamically controlled communication for static communications in a number of application programs. Our results show t h a t compiled comm unication out{performs dynamic comm unication to a large degree. Since most of the comm unication patterns in scienti c applications are static, we conclude that compiled comm unication is an e ective m e c hanism for all{optical networks in multiprocessor environments.

show abstract

Section: Greedy Algorithmmentioning

confidence: 99%

Section: Performance Of the Scheduling Algorithmsmentioning

confidence: 99%

Compiled communication for all-optical TDM networks

Yuan

Melhem

Gupta

1996

Proceedings of the 1996 ACM/IEEE Conference on Supercomputing

View full text Add to dashboard Cite

show abstract

“…It has been used in combination with message passing in the iWarp system [17,18]. In [19], the compiler inserts the commands needed to establish the needed connections in the network before the communication takes place.…”

Section: Compile-time and Load-time Prediction Of Working Setsmentioning

confidence: 99%

Switch Design to Enable Predictive Multiplexed Switching in Multiprocessor Networks

Ding

Hoare

Jones

et al.

19th IEEE International Parallel and Distributed Processing Symposium

View full text Add to dashboard Cite

show abstract

“…The multiplexed channels and buffer pools provided by the iWarp hardware serve better to decouple general message-passing and specialized systolic communication. The iWarp VLSI component allows both alternatives, since it supports bidirectional links as well as logical channels [Gro89]. …”

Section: Deadlocks Through Torus Wrap Around Linksmentioning

confidence: 99%

Message Routing On Irregular 2d-meshes And Tori

Stricker

The Sixth Distributed Memory Computing Conference, 1991. Proceedings

View full text Add to dashboard Cite

Wormhole message routing is supported by the communication hardware of several distributed memory machines. This particular method of message routing has numerous advantages but creates the problem of a routing deadlock. When long messages compete for the same channels in the network, some messages will be blocked until the the first message is fully consumed by the processor at the destination of the message. A deadlock occurs if a set of messages mutually blocks, and no message can progress towards its destination. Most deadlock free routing schemes previously known are designed to work on regular binary hypercubes, a very special case of multicomputer interconnection networks. However, these routing schemes do not provide enough flexibility to deal with the irregular 2-D-tori and attached auxiliary cells found on many newer parallel systems.To handle irregular topologies elegantly, a simple proof is necessary to verify the router code. The new proof given in this report is carried out directly on the network graph. It is constructive in the sense that it reveals the design optiors to deal with irregularities and shows how additional flexibility can be used to achieve better load balancing.Based on the modified routing model, a set of deadlock free router functions relevant to the iWarp system configurations are described and proven to be correct.

show abstract

Communication in iWarp systems

Cited by 14 publications

References 6 publications

Compiled communication for all-optical TDM networks

Compiled communication for all-optical TDM networks

Switch Design to Enable Predictive Multiplexed Switching in Multiprocessor Networks

Message Routing On Irregular 2d-meshes And Tori

Contact Info

Product

Resources

About