Instructions for running HCP data with ndmg (functional and diffusion) on AWS with Clowdr/Boutiques
Links:
- https://github.com/gkiar/hcp-ndmg-connectomes
- To log in to the console: https://console.aws.amazon.com
- To check the usage: log in with email and check the billing dashboard
- To describe tools: https://boutiques.github.io
- To launch tools: https://clowdr.github.io
AWS dictionary:
- EC2/ECS: shows what computers are running/turned on
- Batch: manages your cluster
- S3: manages your storage
- IAM: manages your users
- Cloudwatch: monitors your logs
Using Clowdr:
... On Amazon:
clowdr cloud <tool-descriptor.json> \
<tool-inputs.json> \
s3://<path-to-clowdr-outputs> \
s3://<path-to-input-data> \
aws \
<path-to-aws-credentials.csv> \
-Vb \
--region us-east-1
... On Compute Canada (using specific files this time rather than generic things):
clowdr local experiment/ndmg-descriptor-singularity.json \
experiment/invocation-computecanada.json \
~/executions/clowdr-ndmg-hcp/ \
--cluster slurm \
-bV \
--simg ~/bids-ndmg.simg \
-a account:rpp-aevans-ab,time:5:00:00,mem=22G \
-v /project/6008063/gkiar/:/project/6008063/gkiar/
(-Vb
makes it verbose and BIDS-aware)