This is the sample checking kernel progress for Alea GPU

Program.cs

using Alea;
using Alea.CSharp;
using System;
using System.Threading;

namespace AleaCheckKernelProgressSample
{
    class Program
    {
        static void Main(string[] args)
        {
            unsafe
            {
                var gpu = Gpu.Default;  // to avoid the 'cudaErrorInitializationError = 3' error that indicates the CUDA driver and runtime could not be initialized

                const int Incs = 100;
                int* d_data, h_data;
                var rc = default(CUDAInterop.cudaError_enum);
                if ((rc = CUDAInterop.cuMemHostAlloc((IntPtr*)&h_data, (IntPtr)sizeof(int), 2)) != CUDAInterop.cudaError_enum.CUDA_SUCCESS)
                    throw new InvalidOperationException($"'cudaHostAlloc' is failed. CUDA Error Code = { (int)rc }({ rc })");
                if ((rc = CUDAInterop.cuMemHostGetDevicePointer((IntPtr*)&d_data, (IntPtr)h_data, 0)) != CUDAInterop.cudaError_enum.CUDA_SUCCESS)
                    throw new InvalidOperationException($"'cudaHostGetDevicePointer' is failed. CUDA Error Code = { (int)rc }({ rc })");

                *h_data = 0;
                var e = gpu.CreateEvent();

                Console.WriteLine("kernel starting");
                var d_dataPtr = new deviceptr<int>((IntPtr)d_data);
                gpu.Launch(d_dataPtr_ =>
                {
                    for (var i = 0; i < Incs; i++)
                    {
                        DeviceFunction.AtomicAdd(d_dataPtr_, 1);
                        DeviceFunction.ThreadFenceSystem();
                        var time = DeviceFunction.Clock64();
                        while ((DeviceFunction.Clock64() - time) < 100000000) { }
                    }
                }, new LaunchParam(1, 1), d_dataPtr);

                var value = 0;
                do
                {
                    e.Record();
                    CUDAInterop.cuEventQuery(e.Handle);
                    Thread.Sleep(100);  // if there are too many requests, value reflection stops
                    var value1 = *h_data;
                    if (value1 > value)
                    {
                        Console.WriteLine("h_data = {0}", value1);
                        value = value1;
                    }
                } while (value < (Incs - 1));
            }
        }
    }
}

To make this, I reference the following native CUDA codes:

• cuda - How can I check the progress of matrix multiplication? - Stack Overflow
  https://stackoverflow.com/questions/20345702/how-can-i-check-the-progress-of-matrix-multiplication
• CUDA mapped memory: device -> host writes are not visible on host - Stack Overflow
  https://stackoverflow.com/questions/33455396/cuda-mapped-memory-device-host-writes-are-not-visible-on-host