We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
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
There is no mention of CMAKE_CUDA_ARCHITECTURES in the main documentation (https://koide3.github.io/glim/installation.html).
CMAKE_CUDA_ARCHITECTURES
If user cmake is recent enough (>= 3.24), then CMAKE_CUDA_ARCHITECTURES=native is used in gtsam_points, which definitely looks like a sensible default. Unfortunatelly, the current cmake on Ubuntu 22.04 is slightly older so gtsam_points falls back to compute capability >= 8.0 here: https://github.com/koide3/gtsam_points/blob/529d1d190067a0c3176275f4cbaeb6bd23575d9c/CMakeLists.txt#L72
CMAKE_CUDA_ARCHITECTURES=native
gtsam_points
Considering the popularity of old Jetson TX2/nano and cie (which support compute <= 5.3), maybe it could be worth to:
BUILD_WITH_CUDA_MULTIARCH
What do you think?
The text was updated successfully, but these errors were encountered:
OK, I think it would be good do add a few more options to the default CMAKE_CUDA_ARCHITECTURES (53 62 87) to better support Jetson series. I'll make a PR soon. Do you think if we need more architectures? https://github.com/koide3/gtsam_points/blob/529d1d190067a0c3176275f4cbaeb6bd23575d9c/CMakeLists.txt#L72
53 62 87
Sorry, something went wrong.
I think these are good defaults
I just added several architectures to the default koide3/gtsam_points#54. I'll soon merge it.
set(CMAKE_CUDA_ARCHITECTURES 53 62 80 87 89 90 CACHE STRING "CUDA architectures" FORCE)
No branches or pull requests
There is no mention of
CMAKE_CUDA_ARCHITECTURES
in the main documentation (https://koide3.github.io/glim/installation.html).If user cmake is recent enough (>= 3.24), then
CMAKE_CUDA_ARCHITECTURES=native
is used ingtsam_points
, which definitely looks like a sensible default. Unfortunatelly, the current cmake on Ubuntu 22.04 is slightly older sogtsam_points
falls back to compute capability >= 8.0 here: https://github.com/koide3/gtsam_points/blob/529d1d190067a0c3176275f4cbaeb6bd23575d9c/CMakeLists.txt#L72Considering the popularity of old Jetson TX2/nano and cie (which support compute <= 5.3), maybe it could be worth to:
CMAKE_CUDA_ARCHITECTURES
for old GPUs in the main documentationBUILD_WITH_CUDA_MULTIARCH
)CMAKE_CUDA_ARCHITECTURES
is unset?What do you think?
The text was updated successfully, but these errors were encountered: