Skip to content

MPI

The following MPIs can be used with the ABCI system.

To use one of these libraries, it is necessary to configure the user environment in advance using the module command. If you run the module command in an interactive node, environment variables for compilation are set automatically. If you run the module command in a compute node, environment variables both for compilation and execution are set automatically.

[username@es1 ~]$ module load openmpi/4.0.5
[username@es1 ~]$ module load cuda/11.0 mvapich/mvapich2-gdr/2.3.5
[username@es1 ~]$ module load mvapich/mvapich2/2.3.5
[username@es1 ~]$ module load intel-mpi/2019.9

The following is a list MPI versions installed in the ABCI system.

Open MPI

Compute Node (V):

openmpi/ Compiler version w/o CUDA cuda8.0 cuda9.0 cuda9.1 cuda9.2 cuda10.0 cuda10.1 cuda10.2 cuda11.0 cuda11.1 cuda11.2
2.1.6 gcc/4.8.5 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
2.1.6 gcc/7.4.0 Yes - - - Yes Yes Yes Yes Yes Yes Yes
2.1.6 gcc/9.3.0 Yes - - - - - Yes Yes Yes Yes Yes
2.1.6 pgi/20.4 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
2.1.6 nvhpc/20.11 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
2.1.6 nvhpc/21.2 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
3.1.6 gcc/4.8.5 Yes - - - Yes Yes Yes Yes Yes Yes Yes
3.1.6 gcc/7.4.0 Yes - - - Yes Yes Yes Yes Yes Yes Yes
3.1.6 gcc/9.3.0 Yes - - - Yes Yes Yes Yes Yes Yes Yes
3.1.6 pgi/20.4 Yes - - - Yes Yes Yes Yes Yes Yes Yes
3.1.6 nvhpc/20.11 Yes - - - Yes Yes Yes Yes Yes Yes Yes
3.1.6 nvhpc/21.2 Yes - - - Yes Yes Yes Yes Yes Yes Yes
4.0.5 gcc/4.8.5 Yes - - - Yes Yes Yes Yes Yes Yes Yes
4.0.5 gcc/7.4.0 Yes - - - Yes Yes Yes Yes Yes Yes Yes
4.0.5 gcc/9.3.0 Yes - - - Yes Yes Yes Yes Yes Yes Yes
4.0.5 pgi/20.4 Yes - - - Yes Yes Yes Yes Yes Yes Yes
4.0.5 nvhpc/20.11 Yes - - - Yes Yes Yes Yes Yes Yes Yes
4.0.5 nvhpc/21.2 Yes - - - Yes Yes Yes Yes Yes Yes Yes

Compute Node (A):

openmpi/ Compiler version w/o CUDA cuda10.01 cuda10.11 cuda10.21 cuda11.0 cuda11.1 cuda11.2
2.1.6 gcc/7.4.0 Yes Yes Yes Yes Yes Yes Yes
2.1.6 gcc/8.3.1 Yes Yes Yes Yes Yes Yes Yes
2.1.6 gcc/9.3.0 Yes - Yes Yes Yes Yes Yes
2.1.6 pgi/20.4 Yes - - - Yes Yes Yes
2.1.6 nvhpc/20.11 Yes Yes Yes Yes Yes Yes Yes
2.1.6 nvhpc/21.2 Yes Yes Yes Yes Yes Yes Yes
3.1.6 gcc/7.4.0 Yes Yes Yes Yes Yes Yes Yes
3.1.6 gcc/8.3.1 Yes Yes Yes Yes Yes Yes Yes
3.1.6 gcc/9.3.0 Yes Yes Yes Yes Yes Yes Yes
3.1.6 pgi/20.4 Yes - - - Yes Yes Yes
3.1.6 nvhpc/20.11 Yes Yes Yes Yes Yes Yes Yes
3.1.6 nvhpc/21.2 Yes Yes Yes Yes Yes Yes Yes
4.0.5 gcc/7.4.0 Yes Yes Yes Yes Yes Yes Yes
4.0.5 gcc/8.3.1 Yes Yes Yes Yes Yes Yes Yes
4.0.5 gcc/9.3.0 Yes Yes Yes Yes Yes Yes Yes
4.0.5 pgi/20.4 Yes - - - Yes Yes Yes
4.0.5 nvhpc/20.11 Yes Yes Yes Yes Yes Yes Yes
4.0.5 nvhpc/21.2 Yes Yes Yes Yes Yes Yes Yes

MVAPICH2

mvapich/mvapich2/ Compiler version Compute Node (V) Compute Node (A)
2.3.5 gcc/4.8.5 Yes -
2.3.5 gcc/7.4.0 Yes Yes
2.3.5 gcc/8.3.1 - Yes
2.3.5 gcc/9.3.0 Yes Yes
2.3.5 pgi/20.4 Yes Yes
2.3.5 nvhpc/20.11 Yes Yes
2.3.5 nvhpc/21.2 Yes Yes

MVAPICH2-GDR

mvapich/mvapich2-gdr/ Compiler version cuda10.0 cuda10.1 cuda10.2 cuda11.0
2.3.5 gcc/4.8.5 - - Yes Yes

Note

Compute Node (A) does not currently provide MVAPICH2-GDR.

Note

If you need MVAPICH2-GDR for PGI, please contact Customer Support.

Intel MPI

intel-mpi/ Compute Node (V) Compute Node (A)
2019.9 Yes Yes

  1. Provided only for experimental use. NVIDIA A100 is supported on CUDA 11+.