You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
rapids-cmake should provide a rapids_cpm_libcudacxx function that allows for a unified version of libcudacxx across all rapids-cmake projects.
This is important as more RAPIDS projects start using newer versions of libcudacxx compared to the one provided by the CUDA Toolkit.
Describe the solution you'd like
rapids_cpm_libcudacxx would provide a consistent API to find an existing install of libcudacxx or download it from source.
Since currently libcudacxx doesn't provide a way to install just the headers, or a libcudacxx-config.cmake to allow find_package(libcudacxx) to work rapids_cpm_libcudacxx will need to do the following:
Specify custom install rules for libcudacxx
Generate a build directory libcudacxx-config.cmake
Generate an install directory libcudacxx-config.cmake
Like thrust, libcudacxx is provided as part of the CUDA Toolkit and therefore the issues outlined in #95 also apply. Primarily the generated libcudacxx-config.cmake needs to ensure that the include paths are brought in as non-imported.
The text was updated successfully, but these errors were encountered:
Provides an unified version of libcudacxx for all rapids-cmake based projects.
Since the libcudacxx is focused on building the project, we can't rely
on its install rules. Instead we manually produce install rules and a
libcudacxx-config.cmake to allow find_package(libcudacxx).
Like Thrust we also make sure the includes are namespaced and
always marked as user include so that consumers use this version
of libcudacxx instead of the one provided in the system cuda toolkit.
Fixesrapidsai#97
Provides an unified version of libcudacxx for all rapids-cmake based projects.
Since the libcudacxx is focused on building the project, we can't rely
on its install rules. Instead we manually produce install rules and a
libcudacxx-config.cmake to allow find_package(libcudacxx).
Like Thrust we also make sure the includes are namespaced and
always marked as user include so that consumers use this version
of libcudacxx instead of the one provided in the system cuda toolkit.
Fixesrapidsai#97
Provides an unified version of libcudacxx for all rapids-cmake based projects.
Since the libcudacxx is focused on building the project, we can't rely
on its install rules. Instead we manually produce install rules and a
libcudacxx-config.cmake to allow find_package(libcudacxx).
Like Thrust we also make sure the includes are namespaced and
always marked as user include so that consumers use this version
of libcudacxx instead of the one provided in the system cuda toolkit.
Fixes#97
Authors:
- Robert Maynard (https://github.com/robertmaynard)
Approvers: None
URL: #111
Is your feature request related to a problem? Please describe.
rapids-cmake should provide a rapids_cpm_libcudacxx function that allows for a unified version of
libcudacxx
across all rapids-cmake projects.This is important as more RAPIDS projects start using newer versions of
libcudacxx
compared to the one provided by the CUDA Toolkit.Describe the solution you'd like
rapids_cpm_libcudacxx
would provide a consistent API to find an existing install oflibcudacxx
or download it from source.Since currently
libcudacxx
doesn't provide a way to install just the headers, or alibcudacxx-config.cmake
to allowfind_package(libcudacxx)
to workrapids_cpm_libcudacxx
will need to do the following:libcudacxx
libcudacxx-config.cmake
libcudacxx-config.cmake
Like
thrust
,libcudacxx
is provided as part of the CUDA Toolkit and therefore the issues outlined in #95 also apply. Primarily the generatedlibcudacxx-config.cmake
needs to ensure that the include paths are brought in as non-imported.The text was updated successfully, but these errors were encountered: