Up to CUDA 11.7 one coud use something like:
#include “thrust/system/cuda/experimental/pinned_allocator.h”
…
thrust::host_vector<float,thrust::cuda::experimental::pinned_allocator> a(100);
In the thrust release for CUDA 12 the include file is not present. I suspect pinned memory is not experimental any more, but I can’t figure out what to do instead. Could you please provide and example. I am using Windows 10 x64. Thanks