Performance Numbers of MVAPICH2 on Nehalem Architecture (10/29/09)

• Experimental Testbed: Each node of our testbed has eight cores (2.40 GHz dual quad-core) and 12 GB main memory. The CPUs based on Nehalem architecture and run in 64 bit mode. The nodes support 16x PCI Express Gen2 interfaces and are equipped with Mellanox ConnectX QDR HCAs with PCI Express interfaces. The nodes are connected using a 36 port Mellanox QDR InfiniBand switch with QSFP ports. The operating system used was RedHat Enterprise Linux Server 5.

• MVAPICH2 currently delivers one-sided put latency of 2.82 microseconds , one-sided get latency of 5.68 microseconds for 4 bytes, one-sided put bandwidth up to 3030 Million Bytes/sec, one-sided get bandwidth up to 3032 Million Bytes/sec and put bidirectional bandwidth upto 5831 Million Bytes/sec on the above testbed. (1 Mega Byte = 1,048,576 Bytes; 1 Million Byte = 1,000,000 Bytes)