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

Build: Set default CUDA_ARCHITECTURES for builds #648

Closed
wants to merge 3 commits into from

Conversation

tbirdso
Copy link
Contributor

@tbirdso tbirdso commented Dec 19, 2024

Set a default CMAKE_CUDA_ARCHITECTURES list for builds to inherit that covers most modern architectures.

The list here is at parity with Holoscan SDK build defaults. Subprojects in HoloHub may choose to override this value with their own default settings or set CUDA_ARCHITECTURES directly on targets.

Resolves an observed issue on IGX 1.1 where not setting this property resulted in a default value of "52" in the Multi-AI Ultrasound application, leading to a "PTX" failure message. With this fix the Multi-AI Ultrasound
application inherits the updated list of compute capabilities by default.

Note that existing builds may need to clear the CMakeCache.txt or the
specific architectures cache entry to apply the fix.

Reference: https://cmake.org/cmake/help/v3.24/variable/CMAKE_CUDA_ARCHITECTURES.html

Set a default `CMAKE_CUDA_ARCHITECTURES` list for builds to inherit that covers
most modern architectures.

The list here is at parity with Holoscan SDK build defaults. Subprojects
in HoloHub may choose to override this value with their own default settings
or set `CUDA_ARCHITECTURES` directly on targets.

Resolves an observed issue on IGX 1.1 where not setting this property
resulted in a default value of "52" in the Multi-AI Ultrasound application,
leading to a "PTX" failure message. With this fix the Multi-AI Ultrasound
application inherits the updated list of compute capabilities by default.

Note that existing builds may need to clear the `CMakeCache.txt` or the
specific architectures cache entry to apply the fix.

co-authored-by: Wendell Hom <whom@nvidia.com>
Signed-off-by: Tom Birdsong <tbirdsong@nvidia.com>
CMakeLists.txt Outdated Show resolved Hide resolved
@tbirdso tbirdso self-assigned this Dec 20, 2024
CMakeLists.txt Outdated Show resolved Hide resolved
CMakeLists.txt Outdated Show resolved Hide resolved
@jjomier jjomier self-requested a review January 2, 2025 16:23
@tbirdso
Copy link
Contributor Author

tbirdso commented Jan 2, 2025

Moved forward with the "native" stopgap in #651. Let's continue discussion on best practice for a long term fix in this PR.

AndreasHeumann and others added 2 commits January 7, 2025 07:18
Co-authored-by: Tom Birdsong <40648863+tbirdso@users.noreply.github.com>
Signed-off-by: Andreas Heumann <aheumann@nvidia.com>
@tbirdso
Copy link
Contributor Author

tbirdso commented Jan 7, 2025

Closing as duplicate of #651

@tbirdso tbirdso closed this Jan 7, 2025
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