developmentseed / developmentseed/ml-workstation
configurable workstation variables
- Dominant language
- Python
- Stars
- 0
- Forks
- 0
- PR merge metrics
- No merged PRs in 30d
Description
These are taken from our terraform variables for a gcp instance, and it would be helpful to have an analogue of these for an aws workstation.
location/region: "europe-west4" for gcp or the equivalent in aws
instance-type: determines number of cores and memory, linkage possibilities between the ssd and the gpu (NVLink? not super familiar with this part)
project: Not sure if there is an aws analogue, but gcp has project ids, with many projects to one gcp account
gpu-count: 1-4
gpu-type: type of gpu(s) to attach to instance. example would be nvidia-tesla-t4 vs much cheaper, much dinkier Nvidia K80s
### additional variables
EFS Mount s3 prefix path and mount path. I'm not sure what the performance of this would be. How fast is it compared to a local file system? In the past I used https://github.com/s3fs-fuse/s3fs-fuse on our terraform instance, but found it to be too slow. I switched to using the aws cli to sync data to the ssd or using fsspec in my python code to load files directly from the bucket prefixes.
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.