High Performance Linear Algebra Operations on Reconfigurable Systems
SC '05 Proceedings of the 2005 ACM/IEEE conference on Supercomputing
UPC: Distributed Shared Memory Programming (Wiley Series on Parallel and Distributed Computing)
UPC: Distributed Shared Memory Programming (Wiley Series on Parallel and Distributed Computing)
The AMD Opteron Northbridge Architecture
IEEE Micro
FPGA acceleration of a quantum Monte Carlo application
Parallel Computing
An open-source HyperTransport core
ACM Transactions on Reconfigurable Technology and Systems (TRETS)
VELO: A Novel Communication Engine for Ultra-Low Latency Message Transfers
ICPP '08 Proceedings of the 2008 37th International Conference on Parallel Processing
Hi-index | 0.00 |
This paper presents the very first implementation of a HyperTransport 3 physical layer interface for Field Programmable Gate Arrays. HyperTransport is a low latency, high bandwidth point-to-point interconnect technology that can be used to directly connect hardware accelerators to AMD's Opteron CPUs. Providing support for HyperTransport 3 on FPGAs is highly relevant for increasing the performance of accelerators based on reconfigurable logic. This paper shows the challenges of such an implementation and novel ideas to solve them successfully. A new architecture is presented that uses Fast Serializer Logic to keep up with the increasing speeds of current host interface protocols. A solid evaluation is provided using a specially developed FPGA board as a verification platform.