Computation of cyclic redundancy checks via table look-up
Communications of the ACM
Parallel view-dependent isosurface extraction using multi-pass occlusion culling
PVG '01 Proceedings of the IEEE 2001 symposium on parallel and large-data visualization and graphics
Distributed processing of very large datasets with DataCutter
Parallel Computing - Clusters and computational grids for scientific computing
Fault-Tolerant Wormhole Routing Algorithms for Mesh Networks
IEEE Transactions on Computers
Design of a Framework for Data-Intensive Wide-Area Applications
HCW '00 Proceedings of the 9th Heterogeneous Computing Workshop
The Quadrics Network (QsNet): High-Performance Clustering Technology
HOTI '01 Proceedings of the The Ninth Symposium on High Performance Interconnects
Sockets Direct Protocol over InfiniBand in clusters: is it beneficial?
ISPASS '04 Proceedings of the 2004 IEEE International Symposium on Performance Analysis of Systems and Software
On the performance of TCP splicing for URL-aware redirection
USITS'99 Proceedings of the 2nd conference on USENIX Symposium on Internet Technologies and Systems - Volume 2
Hot-Spot Avoidance With Multi-Pathing Over InfiniBand: An MPI Perspective
CCGRID '07 Proceedings of the Seventh IEEE International Symposium on Cluster Computing and the Grid
Advanced Flow-control Mechanisms for the Sockets Direct Protocol over InfiniBand
ICPP '07 Proceedings of the 2007 International Conference on Parallel Processing
IEEE Transactions on Information Technology in Biomedicine
Hi-index | 0.00 |
As high-end computing systems continue to grow, the needfor advanced networking capabilities, such as hot-spot avoidance andfault tolerance, is becoming important. While the traditional approachof utilizing intelligent network hardware has worked well to achieve highperformance, adding more and more features makes the hardware complexand expensive. Consequently, protocol stacks such as iWARP andMX for 10-Gigabit Ethernet and QLogic InfiniBand, utilize hybridhardware-software designs that take advantage of the processing powerof multi-core processors together with network hardware accelerators.However, upper-layer stacks on these networks, such as the Sockets DirectProtocol (SDP), have not kept pace with such shift in paradigm,and have continued to assume complete hardware offload, leading to redundantfeatures and performance loss. In this paper, we propose anenhanced design for SDP that allows network stacks to specify componentsimplemented in hardware and software, and uses this informationto optimize its execution.