GigaE PM: a high performance communication facility using a Gigabit Ethernet

A high performance communication facility, called the GigaE PM, has been de signed and implemented for parallel applications on clusters of computers u sing a Gigabit Ethernet. The GigaE PM provides not only a reliable high ban dwidth and low latency communication, but also supports existing network pr otocols such as TCP/IP. A reliable communication mechanism for a parallel a pplication is implemented on the firmware on a NIC while existing network p rotocols are handled by an operating system kernel. A prototype system has been implemented using an Essential Communications Gigabit Ethernet card. T he performance results show that a 58.3 mu s round trip time for a four byt e user message, Emd 56.7 MBytes/sec bandwidth for a 1,468 byte message have been achieved on Intel Pentium II 400 MHz PCs. We have implemented MPICH-P M on top of the GigaE PM, and evaluated the NAS parallel benchmark performa nce. The results show that the IS class S performance on the GigaE PM is 1. 8 times faster than that on TCP/IP.