TY - GEN
T1 - On-the-fly sending
T2 - 5th International Symposium on Parallel Architectures, Algorithms, and Networks, I-SPAN 2000
AU - Tanabe, Noboru
AU - Yamamoto, Junji
AU - Nishi, Hiroaki
AU - Kudoh, Tomohiro
AU - Hamada, Yoshihiro
AU - Nakajo, Hironori
AU - Amano, Hideharu
N1 - Publisher Copyright:
© 2000 IEEE.
PY - 2000
Y1 - 2000
N2 - Low latency high bandwidth network interface architectures are described. This paper proposes two types of architectures, atomic on-the& (OTF)sending with a header TLB, and block on-the-$y sending with protection stampable window memory. These techniques work very effectively with MEMOnet which is a class of network interface card(NIC)plugged into a memory slot. Weare developing a network interface controller LSI called Martini. The Martini chip is used in two prototype network interface cards, DIMMnet-1 based on MEMOnet, and RHiNET-2INI based on PCI. On a DIMMnet-I, the software overhead needed to generate a message is only 1 CPU cycle and the estimated hardware delay is less than 1OOns using atomic OTF sending. The estimated achievable sending bandwidth of DIMMnet-1 using block OTF sending is 984MBls which was observed in our experiments. This bandwidth is 7.4 times higher than the maximum bandwidth of PCI. This excellent performance is available for cheap personal computers with DIMM slots. This paper also discribes the e$ fects of block OTF sending for a PCI-based NIC.
AB - Low latency high bandwidth network interface architectures are described. This paper proposes two types of architectures, atomic on-the& (OTF)sending with a header TLB, and block on-the-$y sending with protection stampable window memory. These techniques work very effectively with MEMOnet which is a class of network interface card(NIC)plugged into a memory slot. Weare developing a network interface controller LSI called Martini. The Martini chip is used in two prototype network interface cards, DIMMnet-1 based on MEMOnet, and RHiNET-2INI based on PCI. On a DIMMnet-I, the software overhead needed to generate a message is only 1 CPU cycle and the estimated hardware delay is less than 1OOns using atomic OTF sending. The estimated achievable sending bandwidth of DIMMnet-1 using block OTF sending is 984MBls which was observed in our experiments. This bandwidth is 7.4 times higher than the maximum bandwidth of PCI. This excellent performance is available for cheap personal computers with DIMM slots. This paper also discribes the e$ fects of block OTF sending for a PCI-based NIC.
UR - http://www.scopus.com/inward/record.url?scp=34548258548&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=34548258548&partnerID=8YFLogxK
U2 - 10.1109/ISPAN.2000.900284
DO - 10.1109/ISPAN.2000.900284
M3 - Conference contribution
AN - SCOPUS:34548258548
T3 - Proceedings - International Symposium on Parallel Architectures, Algorithms, and Networks, I-SPAN 2000
SP - 186
EP - 193
BT - Proceedings - International Symposium on Parallel Architectures, Algorithms, and Networks, I-SPAN 2000
A2 - Sudborough, Hal
A2 - Monien, Burkhard
A2 - Hsu, D. Frank
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 7 December 2000 through 9 December 2000
ER -