ORNL / HydraGNN

Distributed PyTorch implementation of multi-headed graph convolutional neural networks
BSD 3-Clause "New" or "Revised" License
68 stars 29 forks source link

ADIOS ImportError is being ignored #205

Closed kshitij-v-mehta closed 4 months ago

kshitij-v-mehta commented 12 months ago

https://github.com/ORNL/HydraGNN/blob/main/hydragnn/utils/adiosdataset.py#L10C1-L13C9

An ImportError caused by a missing ADIOS installation is not handled. This leads to errors with the ADIOS datasets further in the code. https://github.com/ORNL/HydraGNN/blob/main/hydragnn/utils/adiosdataset.py#L300

NameError: name 'ad2' is not defined

jychoi-hpc commented 12 months ago

We put try ... except block in the code to make HydraGNN work even without adios2 python module. When there is no adios2 python module and someone is trying to use adios related work, it will make such error.

On Frontier, I recommended people to use the following command:

module swap PrgEnv-${PE_ENV,,} PrgEnv-gnu
module load rocm/5.3.0
module load cmake
module load craype-accel-amd-gfx90a
module unload darshan-runtime

module use -a /lustre/orion/world-shared/phy122/jyc/frontier/sw/modulefiles
module load anaconda3/2022.10
module load adios2/devel

Now I recommend to use the following:

module load adios2/v2.8.3

or

module load adios2/v2.9.0
allaffa commented 4 months ago

@kshitij-v-mehta @jychoi-hpc Is this issue still relevant or can we close it?

kshitij-v-mehta commented 4 months ago

It is still open. We need to add a clear error message when someone tries to use ADIOS when it is not installed.