[BUG]: thrust::host_vector with universal_host_pinned_memory_resource uses a kernel to initialize memory

[fkallen]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this bug and that I agree to the Code of Conduct

Type of Bug

Something else

Component

Thrust

Describe the bug

The profiler shows that the following code executes a kernel.

[5/7] Executing 'cuda_gpu_kern_sum' stats report

 Time (%)  Total Time (ns)  Instances  Avg (ns)   Med (ns)   Min (ns)  Max (ns)  StdDev (ns)                                                  Name                                                
 --------  ---------------  ---------  ---------  ---------  --------  --------  -----------  ----------------------------------------------------------------------------------------------------
    100,0          320.130          1  320.130,0  320.130,0   320.130   320.130          0,0  void thrust::cuda_cub::core::_kernel_agent<thrust::cuda_cub::__parallel_for::ParallelForAgent<thrus

How to Reproduce

//nvcc 12.3 , nvcc test.cu -o test

#include <thrust/mr/universal_memory_resource.h>
#include <thrust/mr/allocator.h>
#include <thrust/host_vector.h>

int main(){
    using Allocator = thrust::mr::stateless_resource_allocator<
        int, thrust::universal_host_pinned_memory_resource>;

    thrust::host_vector<int, Allocator> vec(1024*1024);
}

Expected behavior

I expect the host to initialize the memory to avoid pcie transfer.

Reproduction link

No response

Operating System

No response

nvidia-smi output

No response

NVCC version

No response