The OpenCHK model is a pragma-based checkpointing model developed by the Programming Models group at the Barcelona Supercomputing Center.
The aim of this model is to provide a generic and portable way to checkpoint and recover data in C/C++ and Fortran High Performance Computing applications.
The prototype implementation of the model is based on two software components:
The OpenCHK model is a pragma-based checkpointing model. The sentinel used to recognized this kind of pragmas is chk.
The init construct defines the initialization of a checkpoint context.
#pragma chk init [clauses] // C/C++ syntax
!$chk init [clauses] // Fortran syntax
where clauses may be:
com(comm-expr)
The comm clause is used to specify the MPI communicator that should be used in the checkpoint context that is being created.
- At most one checkpoint context may be alive at the same time.
- It is mandatory to specify a MPI communicator through the comm clause.
The shutdown construct defines the finalization of a checkpoint context.
#pragma chk shutdown // C/C++ syntax
!$chk shutdown // Fortran syntax
The load construct specifies that some variables and memory regions are going to be updated with the stored values that we have for them in a previous checkpoint (if any).
#pragma chk load(data-expr-list) [clauses] // C/C++ syntax
!$chk load(data-expr-list) [clauses] // Fortran syntax
where clauses may be:
if(bool-expr)
The if clause is used as a switch off mechanism: the contruct will be ignored if the expression of the if clause evaluates to false.
- It must be used inside a checkpoint context.
The store construct specifies that some variables and memory regions are going to be saved in a new checkpoint.
#pragma chk store(data-expr-list) [clauses] // C/C++ syntax
!$chk store(data-expr-list) [clauses] // Fortran syntax
where clauses may be:
if(bool-expr)
id(integer-expr)
level(integer-expr)
kind(kind-expr)
The if clause is used as a switch off mechanism: the contruct will be ignored if the expression of the if clause evaluates to false.
The id clause is used to give a name to the new checkpoint.
The level clause is used to specify in which level the checkpoint should be done.
The kind clause is used to specify which kind of checkpoint you want to do. We support two different kinds of checkpoint right now: CHK_FULL and CHK_DIFF. If the kind clause is not present, the default kind will be CHK_FULL.
- It must be used inside a checkpoint context.
- It is mandatory to specify an identifier through the id clause.
- It is mandatory to specify the checkpoint level through the level clause.