MVAPICH/MVAPICH2 Project
MVAPICH: MPI over InfiniBand, 10GigE/iWARP and RDMAoE
Network-Based Computing Laboratory
Department of Computer Science and Engineering
Ohio State University



EM64T | InterNode | Performance | Network-Based Computing Laboratory

Performance Numbers of MVAPICH2 on Nehalem Architecture (03/12/10)

  • 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 8x PCI Express interfaces and are equipped with InfiniPath-QLE7280 HCAs with PCI Express interfaces. A 24-port DDR switch was used to connect the two nodes. The operating system used was RedHat Linux AS4. InfiniPath Software Release 2.1 was used to build MVAPICH2.
  • MVAPICH2 currently delivers one-sided put latency of 7.65 microseconds , one-sided get latency of 12.91 microseconds for 4 bytes, one-sided put bandwidth up to 1950 Million Bytes/sec, one-sided get bandwidth up to 1333 Million Bytes/sec and put bidirectional bandwidth upto 3286 Million Bytes/sec on the above testbed. (1 Mega Byte = 1,048,576 Bytes; 1 Million Byte = 1,000,000 Bytes)