Deadlock-Free Message Routing in Multiprocessor Interconnection Networks
IEEE Transactions on Computers
In search of clusters (2nd ed.)
In search of clusters (2nd ed.)
IEEE Transactions on Parallel and Distributed Systems
Deadlock-Free Routing in InfiniBand through Destination Renaming
ICPP '02 Proceedings of the 2001 International Conference on Parallel Processing
Effective Strategy to Compute Forwarding Tables for InfiniBand Networks
ICPP '02 Proceedings of the 2001 International Conference on Parallel Processing
IPPS '99/SPDP '99 Proceedings of the 13th International Symposium on Parallel Processing and the 10th Symposium on Parallel and Distributed Processing
Improving the Up*/Down* Routing Scheme for Networks of Workstations
Euro-Par '00 Proceedings from the 6th International Euro-Par Conference on Parallel Processing
A New Methodology to Computer Deadlock-Free Routing Tables for Irregular Networks
CANPC '00 Proceedings of the 4th International Workshop on Network-Based Parallel Computing: Communication, Architecture, and Applications
IEEE Communications Magazine
Supporting adaptive routing in IBA switches
Journal of Systems Architecture: the EUROMICRO Journal - Special issue: Evolutions in parallel distributed and network-based processing
An efficient buffer allocation technique for virtual lanes in InfiniBand networks
HSI'03 Proceedings of the 2nd international conference on Human.society@internet
| Hi-index | 0.00 |
InfiniBand is very likely to become the de facto standard for communication between processing nodes and I/O devices as well as for interprocessor communication. The InfiniBand Architecture (IBA) supports up to 15 data virtual lanes per physical link, primarily intended for traffic prioritization, deadlock avoidance, and quality of service. However, virtual lanes may also contribute to improve performance by reducing the influence of the head-of-line blocking effect on input physical ports. On the other hand, when virtual lanes are used, crossbar complexity may be increased.The main goal of this paper is to show at what extent the use of virtual lanes may contribute to improve network performance on an InfiniBand environment, obtaining the trade-off between the number of virtual lanes and performance improvement. Different configurations (crossbar organization, crossbar bandwidth, and link bandwidth) are used. Evaluation results using up*/down* routing show that two virtual lanes are often enough to achieve the most of the improvement on performance, allowing the use of the remaining virtual lanes for other purposes. Additionally, by increasing the crossbar bandwidth, a lower complexity crossbar configuration can be used.