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

Add Compute Capabillity 90 (Hopper) to the default cuda architectures #954

Merged
merged 1 commit into from
Nov 2, 2022

Conversation

ptheywood
Copy link
Member

Adds SM 90 (Hopper) to our default CUDA Architecture values, so Hopper is targetted by default with CUDA >= 11.8.

In the future we might want to add an 11.8 wheel to the matrix for Hopper users, but in general anyone with hopper could build it for themselves (i.e. HPC only).

Ada (40 series) is SM 89 (i.e. it's closer to Ampere than Hopper), so we already target that major arch by default.


Configuring with CUDA 11.8 using the new default:

-- The CUDA compiler identification is NVIDIA 11.8.89
... 
-- Generating Compute Capabilities: 35;50;60;70;80;90

Configuring with CUDA 11.7 using the new default

-- The CUDA compiler identification is NVIDIA 11.7.99
...
-- Generating Compute Capabilities: 35;50;60;70;80

Building for 89,90 is fine and emits no new warnings. Unable to test execution though as we don't have the hardware.

@mondus mondus merged commit 6c50337 into master Nov 2, 2022
@mondus mondus deleted the sm_90 branch November 2, 2022 16:10
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

Successfully merging this pull request may close these issues.

3 participants