... | ... | @@ -2,7 +2,7 @@ Getting the full output from Intel MPI showing all the MPI calls you |
|
|
need to use something like:
|
|
|
```
|
|
|
# Trace MPI calls.
|
|
|
source/cosma/local/intel/Parallel_Studio_XE_2016/parallel_studio_xe_2016.0.047/sxevars.sh
|
|
|
source /cosma/local/intel/Parallel_Studio_XE_2016-update1/parallel_studio_xe_2016.1.05/psxevars.sh
|
|
|
export VT_VERBOSE=3
|
|
|
export VT_FLUSH_PREFIX=/gpfs/data/pdraper
|
|
|
export VT_CHECK="GLOBAL:MSG:DATA_TRANSMISSION_CORRUPTED OFF LOCAL:MEMORY:ILLEGAL_MODIFICATION OFF LOCAL:REQUEST:NOT_FREED OFF CHECK LOCAL:MEMORY:OVERLAP OFF"
|
... | ... | @@ -11,3 +11,5 @@ and then run using: |
|
|
```
|
|
|
mpirun -check_mpi ...
|
|
|
```
|
|
|
|
|
|
You will need to adapt this for compiler updates. |
|
|
\ No newline at end of file |