Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Adding support for AWS x86 ParallelCluster 3.7.2 w/EFA #57

Merged
merged 4 commits into from
Nov 28, 2023
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 29 additions & 0 deletions configs/AWS-x86-ParallelCluster-3.7.2/README.txt
Original file line number Diff line number Diff line change
@@ -0,0 +1,29 @@
AWS x86 ParallelCluster 3.7.2
-----------------------------

This config should work on any AWS x86 ParallelCluster 3.7.2 instance with the
following caveats:

1) All compute instances must be x86 and EFA enabled. Supported instance types
can be found here:

https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/efa.html#efa-instance-types

2) Any OS supported by ParallelCluster 3.7.2 should work, but only Amazon
Linux 2 has been tested

3) ParallelCluster does not install optimized versions of BLAS/LAPACK. This
config uses the generic versions installed via:

sudo yum install lapack

3) OpenMPI is the only supported MPI flavor. IntelMPI is not yet supported.

OpenMPI is running in verbose mode so the user than confirm that EFA is
being used when running experiments. A line similar to the following in
slurm-NNN.out confirms EFA is being used:

mtl_ofi_component.c:362: mtl:ofi:provider: rdmap0s6-rdm

This debugging output can be silenced by removing the env variable
'OMPI_MCA_mtl_base_verbose=100' from the srun line in variables.yaml.
Original file line number Diff line number Diff line change
@@ -0,0 +1,14 @@
compilers:
- compiler:
spec: [email protected]
paths:
cc: /usr/bin/gcc
cxx: /usr/bin/g++
f77: /usr/bin/gfortran
fc: /usr/bin/gfortran
flags: {}
operating_system: alinux2
target: x86_64
modules: []
environment: {}
extra_rpaths: []
Original file line number Diff line number Diff line change
@@ -0,0 +1,27 @@
packages:
tar:
externals:
- spec: [email protected]
prefix: /usr
buildable: false
gmake:
externals:
- spec: [email protected]
prefix: /usr
blas:
externals:
- spec: [email protected]
prefix: /usr
buildable: false
lapack:
externals:
- spec: [email protected]
prefix: /usr
buildable: false
mpi:
externals:
- spec: [email protected]
prefix: /opt/amazon/openmpi
buildable: false
extra_attributes:
ldflags: "-L/opt/amazon/openmpi/lib -lmpi"
12 changes: 12 additions & 0 deletions configs/AWS-x86-ParallelCluster-3.7.2/spack.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,12 @@
spack:
packages:
default-compiler:
spack_spec: [email protected]
default-mpi:
spack_spec: [email protected]
compiler-gcc:
spack_spec: [email protected]
lapack:
spack_spec: [email protected]
mpi-gcc:
spack_spec: [email protected]
7 changes: 7 additions & 0 deletions configs/AWS-x86-ParallelCluster-3.7.2/variables.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,7 @@
variables:
batch_time: '02:00'
mpi_command: 'srun -N {n_nodes} -n {n_ranks} --mpi=pmix --export=ALL,FI_EFA_USE_DEVICE_RDMA=1,FI_PROVIDER="efa",OMPI_MCA_mtl_base_verbose=100'
batch_submit: 'sbatch {execute_experiment}'
batch_nodes: '#SBATCH -N {n_nodes}'
batch_ranks: '#SBATCH -n {n_ranks}'
batch_timeout: '#SBATCH -t {batch_time}:00'