| 139 | | * Most installations should use either the ch3:ssm or ch3:sock methods. If you need multi-threaded MPI, you must use ch3:sock in this release. If you have a cluster of SMPs and do not need a multi-threaded MPI, then select ch3:ssm. If you are interested in trying out our highest-performance device, and need neither threads nor the dynamic process routines from MPI, then you can use ch3:nemesis. |
| | 139 | * ch3:sock This is the default communication method. It uses sockets for all communications between processes. |
| | 140 | * ch3:ssm This method uses sockets between nodes and shared memory within a node. |
| | 141 | * ch3:shm This method only uses shared memory and only works within a single SMP. It does not support the MPI dynamic process routines such as MPI Comm spawn. |
| | 142 | * ch3:nemesis This method is our new, high performance method. It supports sockets, shared memory, and Myrinet-GM at present. |
| | 143 | * Most installations should use either the '''ch3:ssm''' or '''ch3:sock''' methods. If you need __'''multi-threaded MPI'''__, you must __'''use ch3:sock'''__ in this release. If you have __'''a cluster of SMPs and do not need a multi-threaded MPI'''__, then select __'''ch3:ssm'''__. If you are interested in trying out our highest-performance device, and need neither threads nor the dynamic process routines from MPI, then you can use ch3:nemesis. |
