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
{{ message }}
This repository was archived by the owner on Mar 21, 2024. It is now read-only.
Reference CUDA back end functions only when CUDA back end is enabled
Add a #if block to temporary_allocator<T,S>::allocate in
thrust/detail/allocator/temporary_allocator.inl so that the CUDA back end
function thrust::system::cuda::detail::terminate_with_message is referenced
only when the CUDA back end is enabled. Putting the reference within the
device-specific branch of an NV_IF_TARGET isn't good enough; there are some
situations, e.g. 'nvc++ -cuda -stdpar=multicore', where that doesn't work
and an explicit check for the CUDA back end is necessary.
0 commit comments