Article

Can User-Level Protocols Take Advantage of Multi-CPU NICs?

Authors:

Piyush Shivam,

Pete Wyckoff,

Dhabaleswar K. PandaAuthors Info & Claims

IPDPS '02: Proceedings of the 16th International Parallel and Distributed Processing Symposium

Page 88

Published: 15 April 2002 Publication History

Abstract

User-level protocols and their implementations on programmable network interface cards (NICs) have been alleviating the communication bottleneck for high speed interconnects. Most of the user-level protocols developed so far have been based on single-CPU NICs. One of the more popular current generation Gigabit Ethernet NICs includes two CPUs, though. This raises an open challenge whether performance of user-level protocols can be improved by taking advantage of a multi-CPU NIC. In this paper, we analyze the intrinsic issues associated with such a challenge and explore different parallelization and pipelining schemes to enhance the performance of our earlier developed EMP protocol for single-CPU Alteon NICs. Performance evaluation results indicate that parallelizing the receive path of the protocol can deliver 964 Mbps of bandwidth, close to the maximum achievable on Gigabit Ethernet. This scheme also delivers up to 8% improvement in latency for a range of message sizes. Parallelizing the send path leads to 17% improvement in bidirectional bandwidth.

References

[1]

N. Boden, D. Cohen, and R. Felderman. Myrinet: a gigabit per second local-area network. IEEE Micro, 15(1):29, 1995.

Digital Library

Google Scholar

[2]

Infiniband. http://www.infinibandta.org.

Google Scholar

[3]

E. Nahum, D. Yates, and J. Kurose. Performance issues in parallelized network protocols. In USENIX OSDI, 1994.

Digital Library

Google Scholar

[4]

S. Pakin, M. Lauria, and A. Chien. High performance messaging on workstations: Fast Messages, 1995.

Digital Library

Google Scholar

[5]

T. Porta and M. Schwartz. A high-speed protocol parallel implementation: design and analysis. In ICHPN, 1992.

Digital Library

Google Scholar

[6]

P. Shivam, P. Wyckoff, and D. Panda. Can user level protocols take advantage of multi-CPU NICs? Technical Report OSU-CISRC-08/01-TR09, The Ohio State University, 2001.

Google Scholar

[7]

P. Shivam, P. Wyckoff, and D. Panda. EMP: Zero-copy OSbypass NIC-driven gigabit ethernet message passing. In Proceedings of SC01, November 2001.

Digital Library

Google Scholar

[8]

VI. http://www.viarch.org, 1998.

Google Scholar

[9]

T. von Eicken, A. Basu, and V. Buch. U-Net: A user-level network interface for parallel computing. In ASPLOS, 1995.

Google Scholar

[10]

M. Zitterbart. High-speed protocol implementations on a multiproc. arch. In Protocols for Highspeed Networks, 1999.

Google Scholar

Cited By

View all

Shafer JRixner S(2007)RiceNICProceedings of the 2007 workshop on Experimental computer science10.1145/1281700.1281721(21-es)Online publication date: 13-Jun-2007
https://dl.acm.org/doi/10.1145/1281700.1281721
Gavrilovska ASchwan KBerenbaum ALi KTurner J(2005)Addressing data compatibility on programmable network platformsProceedings of the 2005 ACM symposium on Architecture for networking and communications systems10.1145/1095890.1095908(125-134)Online publication date: 26-Oct-2005
https://dl.acm.org/doi/10.1145/1095890.1095908
Gavrilovska AKumar SSundaragopalan SSchwan KFeng WMayer-Patel K(2005)Platform OverlaysProceedings of the international workshop on Network and operating systems support for digital audio and video10.1145/1065983.1066022(171-176)Online publication date: 13-Jun-2005
https://dl.acm.org/doi/10.1145/1065983.1066022
Show More Cited By

Index Terms

Can User-Level Protocols Take Advantage of Multi-CPU NICs?
1. General and reference
  1. Cross-computing tools and techniques
    1. Performance
2. Networks
  1. Network protocols

Index terms have been assigned to the content through auto-classification.

Recommendations

Can User-Level Protocols Take Advantage of Multi-CPU NICs?
IPDPS '02: Proceedings of the 16th International Symposium on Parallel and Distributed Processing

User-level protocols and their implementations on programmable network interface cards (NICs) have been al-leviating the communication bottleneck for high speed interconnects. Most of the user-level protocols developed so far have been based on single-...
Optimized HPL for AMD GPU and multi-core CPU usage

The installation of the LOEWE-CSC ( http://csc.uni-frankfurt.de/csc/__ __51 ) supercomputer at the Goethe University in Frankfurt lead to the development of a Linpack which can fully utilize the installed AMD Cypress GPUs. At its core, a fast DGEMM for ...
Transparently bridging semantic gap in CPU management for virtualized environments

Consolidated environments are progressively accommodating diverse and unpredictable workloads in conjunction with virtual desktop infrastructure and cloud computing. Unpredictable workloads, however, aggravate the semantic gap between the virtual ...

Comments

Information & Contributors

Information

Published In

IPDPS '02: Proceedings of the 16th International Parallel and Distributed Processing Symposium

April 2002

ISBN:0769515738

Publisher

IEEE Computer Society

United States

Publication History

Published: 15 April 2002

Qualifiers

Article

Contributors

Other Metrics

View Article Metrics

Bibliometrics & Citations

Bibliometrics

Article Metrics

7
Total Citations
View Citations
0
Total Downloads

Downloads (Last 12 months)0
Downloads (Last 6 weeks)0

Reflects downloads up to 10 Aug 2024

Other Metrics

View Author Metrics

Citations

Cited By

View all

Shafer JRixner S(2007)RiceNICProceedings of the 2007 workshop on Experimental computer science10.1145/1281700.1281721(21-es)Online publication date: 13-Jun-2007
https://dl.acm.org/doi/10.1145/1281700.1281721
Gavrilovska ASchwan KBerenbaum ALi KTurner J(2005)Addressing data compatibility on programmable network platformsProceedings of the 2005 ACM symposium on Architecture for networking and communications systems10.1145/1095890.1095908(125-134)Online publication date: 26-Oct-2005
https://dl.acm.org/doi/10.1145/1095890.1095908
Gavrilovska AKumar SSundaragopalan SSchwan KFeng WMayer-Patel K(2005)Platform OverlaysProceedings of the international workshop on Network and operating systems support for digital audio and video10.1145/1065983.1066022(171-176)Online publication date: 13-Jun-2005
https://dl.acm.org/doi/10.1145/1065983.1066022
Majumder SRixner SCox ASubhlok J(2004)Comparing Ethernet and Myrinet for MPI communicationProceedings of the 7th workshop on Workshop on languages, compilers, and run-time support for scalable systems10.1145/1066650.1066659(1-7)Online publication date: 22-Oct-2004
https://dl.acm.org/doi/10.1145/1066650.1066659
Kim HPai VRixner S(2003)Exploiting task-level concurrency in a programmable network interfaceACM SIGPLAN Notices10.1145/966049.78150638:10(61-72)Online publication date: 11-Jun-2003
https://dl.acm.org/doi/10.1145/966049.781506
Kim HPai VRixner SEigenmann RRinard M(2003)Exploiting task-level concurrency in a programmable network interfaceProceedings of the ninth ACM SIGPLAN symposium on Principles and practice of parallel programming10.1145/781498.781506(61-72)Online publication date: 11-Jun-2003
https://dl.acm.org/doi/10.1145/781498.781506
Jaganathan RUnderwood KSass RMcGraw J(2003)A Configurable Network Protocol for Cluster Based Communications using Modular Hardware Primitives on an Intelligent NICProceedings of the 2003 ACM/IEEE conference on Supercomputing10.1145/1048935.1050173Online publication date: 15-Nov-2003
https://dl.acm.org/doi/10.1145/1048935.1050173

View Options

View options

Get Access

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Abstract

References

Cited By

Index Terms

Recommendations