man introu (Commandes) - introduction to LAM user interface commands
NAME
introu - introduction to LAM user interface commands
LIST OF COMMANDS
Name Appears Description on Page bfctl bfctl.1 Control message buffers (optional). bfstate bfstate.1 Get message buffer status (optional). doom doom.1 Deliver a signal (optional). fctl fctl.1 Control remote file access (optional). fstate fstate.1 Get remote file status (optional). hboot hboot.1 Start LAM on a local node. mpicc mpicc.1 Compile LAM C programs. mpiCC mpicc.1 Compile LAM C++ programs. mpif77 mpif77.1 Compile LAM F77 programs. lamboot lamboot.1 Start LAM. lamclean lamclean.1 Clean all nodes. lamexec lamexec.1 Run non-MPI programs. lamgrow lamgrow.1 Add a node. lamhalt lamhalt.1 Terminate LAM. lamshrink lamshrink.1 Remove a node. lamtrace lamtrace.1 Unload trace data. loadgo loadgo.1 Execute program (optional). mpimsg mpimsg.1 Display MPI buffered messages. mpirun mpirun.1 Run an MPI application. mpitask mpitask.1 Display MPI processes. recon recon.1 Verify LAM setup. state state.1 Get process status (optional). sweep bfctl.1 Clean out message buffers (optional). tkill tkill.1 Terminate local node LAM session. tping tping.1 Echo messages to a node. lamwipe lamwipe.1 Terminate LAM (deprecated; use lamhalt).
Node Specification
Target nodes are given on the command line. Nodes are generically identified as n<list>, where <list> can be a single node identifier or a list of node identifiers. For example:
n1 n1,3,5-10
Additionally, c<list> can be used to list specific CPUs on a given machine (lamboot(1) is used to specify how many CPUs LAM may use on each node).
Node/CPU identifiers are established in the boot schema (see bhost(5)), and can be written in decimal or hexadecimal notation. In addition to explicit node identification, LAM has special mnemonics that refer to special nodes or a group of nodes.
Process Specification
LAM processes can be specified in two ways: by process identifier from the underlying operating system or by LAM process index. PIDs are written as p<list>, where <list> can be a single PID or a list of PIDs. Process indices are written as i<list>, where <list> can be a single index or a list of indices.