Are you asking for a minimum CUDA SDK version or a restriction based
on the hardware of the machine?
We have no pre-built logic to enforce a minimum hardware component, as
we want to support building CUDA on a machine without a GPU.
As far as minimum CUDA SDK is concerned if you are doing it through
find_package(CUDA) you would check the version after and error out if
not high enough
As farm as CUDA as a first class language, you can check the
CMAKE_CUDA_COMPILER_VERSION and error out.
Thanks, I am interested in enforcing minimum computing capacity (which translates to hardware restraint). If there is no such feature, would you be interested for such a feature?
On August 16, 2019 7:41:33 AM PDT, Robert Maynard <[hidden email]> wrote:
>Are you asking for a minimum CUDA SDK version or a restriction based
>on the hardware of the machine?
>We have no pre-built logic to enforce a minimum hardware component, as
>we want to support building CUDA on a machine without a GPU.
>As far as minimum CUDA SDK is concerned if you are doing it through
>find_package(CUDA) you would check the version after and error out if
>not high enough
>As farm as CUDA as a first class language, you can check the
>CMAKE_CUDA_COMPILER_VERSION and error out.
>On Thu, Aug 15, 2019 at 9:34 PM Hong Xu <[hidden email]> wrote:
>> Is there a way to enforce a minimum CUDA arch version when finding
>> Powered by www.kitware.com
>> Please keep messages on-topic and check the CMake FAQ at:
>> Kitware offers various services to support the CMake community. For
>more information on each offering, please visit:
>> CMake Support: http://cmake.org/cmake/help/support.html >> CMake Consulting: http://cmake.org/cmake/help/consulting.html >> CMake Training Courses: http://cmake.org/cmake/help/training.html >>
>> Visit other Kitware open-source projects at
>> Follow this link to subscribe/unsubscribe:
else my code would not compile once I started using some newer directives.
I believe that more archs would be needed in the string for better
support of newer architectures, but just to get the starting point.
I would have preferred if there was a better directive, similar to
that can enable either C++14 or C++17 or later (depending on other
components of the project) without actually harcoding c++14 and ending
up with "-std=c++14 -std=c++20" in flags. For example something along
the lines of