TY - JOUR
T1 - JTLB
T2 - A Mechanism for Address Translation on Network Interfaces
AU - Chen, Yuqun
AU - Bilas, Angelos
AU - Damianakis, Stefanos N.
AU - Dubnicki, Cezary
AU - Li, Kai
PY - 1998/11
Y1 - 1998/11
N2 - An important aspect of a high-speed network system is the ability to transfer data directly between the network interface and application buffers. Such a direct data path requires the network interface to "know" the virtual-to-physical address translation of a user buffer, i.e., the physical memory location of the buffer. This paper presents an efficient address translation architecture, User-managed TLB (UTLB), which eliminates system calls and device interrupts from the common communication path. UTLB also supports application-specific policies to pin and unpin application memory. We report micro-benchmark results for an implementation on Myrinet PC clusters, A trace-driven analysis is used to compare the UTLB approach with the interrupt-based approach. It is also used to study the effects of UTLB cache size, associativity, and prefetching. Our results show that the UTLB approach delivers robust performance with relatively small translation cache sizes.
AB - An important aspect of a high-speed network system is the ability to transfer data directly between the network interface and application buffers. Such a direct data path requires the network interface to "know" the virtual-to-physical address translation of a user buffer, i.e., the physical memory location of the buffer. This paper presents an efficient address translation architecture, User-managed TLB (UTLB), which eliminates system calls and device interrupts from the common communication path. UTLB also supports application-specific policies to pin and unpin application memory. We report micro-benchmark results for an implementation on Myrinet PC clusters, A trace-driven analysis is used to compare the UTLB approach with the interrupt-based approach. It is also used to study the effects of UTLB cache size, associativity, and prefetching. Our results show that the UTLB approach delivers robust performance with relatively small translation cache sizes.
UR - http://www.scopus.com/inward/record.url?scp=0346339907&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0346339907&partnerID=8YFLogxK
U2 - 10.1145/291006.291046
DO - 10.1145/291006.291046
M3 - Article
AN - SCOPUS:0346339907
SN - 0362-1340
VL - 33
SP - 193
EP - 204
JO - SIGPLAN Notices (ACM Special Interest Group on Programming Languages)
JF - SIGPLAN Notices (ACM Special Interest Group on Programming Languages)
IS - 11
ER -