-
Notifications
You must be signed in to change notification settings - Fork 25
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
Added AMD GPU Support to Zeus #57
Conversation
…amd_support rebase with master
Thanks for your work. Please resolve merge conflicts and make CI pass, and then request review. |
@jaywonchung fixed all merge conflicts and passing all tests, ready for review. |
@jaywonchung I went through and retested each method, and fixed any issues. It should be all correct now. |
info = amdsmi.amdsmi_get_power_cap_info(self.handle) # Returns in W | ||
amdsmi.amdsmi_set_power_cap( | ||
self.handle, 0, cap=int(info["default_power_cap"] * 1e6) | ||
) # expects value in microwatts |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is what I see, I was going off the source code for 6.0. So it should be correct for ROCM 6.0.
As for pytorch, it looks like it just got full support for ROCM 6.0 with the release of Pytorch 2.3 a week ago.
Should we stick to ROCM 6.0 then?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yep, if everything is consistent with 6.0, let's keep it that way!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, thank you for your wonderful work!
Added AMD GPU support to Zeus. Involved adding method implementations to Zeus.device.gpu for the AMDGPU class.