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

GPU details #1

Open
banesullivan opened this issue Jun 25, 2019 · 5 comments
Open

GPU details #1

banesullivan opened this issue Jun 25, 2019 · 5 comments

Comments

@banesullivan
Copy link
Owner

How would we get info about the GPU?

@banesullivan
Copy link
Owner Author

There's an excellent description of the GPU/display hardware in pyvista/pyvista#301 (comment)

@banesullivan
Copy link
Owner Author

PyVista will need a way to programmatically see what brand of GPU is being used (ref pyvista/pyvista#277 (comment)) e.g. NVIDIA, Radeon, Intel, etc.

@prisae
Copy link
Collaborator

prisae commented Jul 10, 2019

IMHO, anything GPU related should be optional.
E.g., scooby.Report(usual_args, gpu_info=False/True).

@banesullivan
Copy link
Owner Author

Ah yes definitely - I think we'll have to use external libraries to get this info which would mean optional dependencies.

We talked about the structure for extras at some point... maybe on slack?

@prisae
Copy link
Collaborator

prisae commented Jul 11, 2019

I think you referred to the following, it was indeed on Slack; here a condensed version:

  • Bane [5:07 PM]
    my hesitation to drop to 2 files is that before long, I’m going to figure out how to report info about the GPU (need for PyVista) and I imagine that down the road, we may need other info like stats about Fortran, C++, etc. compilers (thinking about GeoBIPy and some geostats libs that interface to Fortran)

    Since we need to keep scooby as having no non-standard dependencies, maybe it would make since to have a mixins module (a directory like what I made for discretize to add support for vtk and omf) where we could have all the extras based on the dependencies they require (edited)

  • prisae [5:07 PM]
    That might be a good way forward

  • Bane [5:08 PM]
    Also, what about if we need to report on a cluster/distributed network?
    Maybe a simple mixin would work?

  • prisae [5:08 PM]
    Again, in that case, report.py would mostly stay the way it is

  • Bane [5:11 PM]
    I say let’s go forward with your reduction to two files
    it cleans it up well and then when I figure out the GPU stuff I’ll refactor the stuff that requires optional dependencies into a new mixins module

  • prisae [5:13 PM]
    Great. That is good. With time knowledge might only contain dictionaries, and other files will be for system, gpu, network, etc

  • Bane [5:14 PM]
    That’s exactly what I’m thinking - but for now it works well

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants