Implementation and performance study of a hardware-VIA-based network adapter on gigabit ethernet

作者: Sejin Park , Sang-Hwa Chung , Ben Lee

DOI: 10.1016/J.SYSARC.2005.02.002

关键词: Remote direct memory accessFiber Distributed Data InterfaceRDMA over Converged EthernetComputer networkCarrier EthernetSystem area networkComputer scienceTCP offload engineGigabit EthernetNetwork interface controllerComputer hardware

摘要: This paper presents the implementation and performance of a hardware-VIA-based network adapter on Gigabit Ethernet. VIA is useer-level communication interface for high PC clustering. The 64-bit/66 MHz PCI plug-in card containing an FPGA Protocol Engine Ethernet chip to construct system area network. performs virtual-to-physical address translation, doorbell, RDMA write, send/receive completion operations in hardware without kernel intervention. In particular, Address Translation Table (ATT) stored local memory adapter, efficiently controls translation process by directly accessing ATT. addition, Prefetch Buffer used reduce time receiver. As result, overhead during transactions greatly reduced. Our experimental results show minimum latency 8.2 µs, maximum bandwidth 112.1 MB/s. terms latency, 2.8 times 3.3 faster than M-VIA, which software VIA, TCP/IP, respectively, over 24% 55% higher M-VIA respectively. These that HVIA-GE far better ServerNet II, version developed Tandem/Compaq.

参考文章(20)
In-Su Yoon, Sang-Hwa Chung, Ben Lee, Hyuk-Chul Kwon, Implementation and Performance Evaluation of M-VIA on AceNIC Gigabit Ethernet Card european conference on parallel processing. pp. 995- 1000 ,(2003) , 10.1007/978-3-540-45209-6_136
Paul A. Farrell, Hong Ong, Performance comparison of LAM/MPI, MPICH, and MVICH on a linux cluster connected by a gigabit ethernet network ALS'00 Proceedings of the 4th annual Linux Showcase & Conference - Volume 4. pp. 31- 31 ,(2000)
Mohammad Banikazemi, Bulent Abali, Dhabaleswar K. Panda, Comparison and Evaluation of Design Choices for Implementing the Virtual Interface Architecture (VIA) parallel computing. pp. 145- 161 ,(2000) , 10.1007/10720115_11
Alessandro Rubini, Jonathan Corbet, Greg Kroah-Hartman, Linux Device Drivers, 3rd Edition ,(2005)
Scott Pakin, Mario Lauria, Andrew Chien, High Performance Messaging on Workstations: Illinois Fast Messages (FM) for Myrinet conference on high performance computing (supercomputing). pp. 55- ,(1995) , 10.1145/224170.224360
C.R. Eddington, InfiniBridge: an InfiniBand channel adapter with integrated switch IEEE Micro. ,vol. 22, pp. 48- 56 ,(2002) , 10.1109/MM.2002.997879
Mohammad Banikazemi, Bulent Abali, Lorraine Herger, Dhabaleswar K. Panda, Design Alternatives for Virtual Interface Architecture and an Implementation on IBM Netfinity NT Cluster Journal of Parallel and Distributed Computing. ,vol. 61, pp. 1512- 1545 ,(2001) , 10.1006/JPDC.2001.1745
Hermann Hellwagner, Matthias Ohlenroth, VI architecture communication features and performance on the Giganet cluster LAN Future Generation Computer Systems. ,vol. 18, pp. 421- 433 ,(2002) , 10.1016/S0167-739X(01)00060-7
N.J. Boden, D. Cohen, R.E. Felderman, A.E. Kulawik, C.L. Seitz, J.N. Seizovic, Wen-King Su, Myrinet: a gigabit-per-second local area network IEEE Micro. ,vol. 15, pp. 29- 36 ,(1995) , 10.1109/40.342015
Mohammad Banikazemi, Jiuxing Liu, S Kutlug, A Ramakrishnan, P Sadayappan, H Shah, DK Panda, None, VIBe: a micro-benchmark suite for evaluating virtual interface architecture (VIA) implementations international parallel and distributed processing symposium. pp. 24- ,(2001) , 10.1109/IPDPS.2001.924960