clEnqueueNDRangeKernel hangs on HD 630（i3-8100H）

[FateWong]

I have a problem when I'm doing tests with opencl, when I keep restarting the computer and then running the following code, the thread hangs Incidentally (very low probability, occurs once in a hundred times) on the event of clEnqueueNDRangeKernel，and does not return any error number. Getting the status code is always a commit.All initialisations return success. The CPU driver version is 30.0.101.1069； My code: init opencl... queue = clCreateCommandQueue(context, device, CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE, &err); if (err != CL_SUCCESS) { LOGI_(ErrorLog) << "Error: Failed to creat command queue!" << err; return false; } cl_event write_evt; status = clEnqueueWriteBuffer(queue, cl_input, 1, 0, data_size * sizeof(cl_float), input_data, 0, 0, &write_evt); if (status != CL_SUCCESS) { LOGI_(ErrorLog) << "Error: Failed to transport input_data to device!" << status; return false; } err = clWaitForEvents(1, &write_evt); err = clSetKernelArg(kernel, 0, sizeof(cl_mem), (void*)&cl_input); err = clSetKernelArg(kernel, 1, sizeof(cl_mem), (void*)&cl_output); err = clSetKernelArg(kernel, 2, sizeof(cl_int), (void*)&image_row); err = clSetKernelArg(kernel, 3, sizeof(cl_int), (void*)&image_col); err = clSetKernelArg(kernel, 4, sizeof(cl_int), (void*)&d); err = clSetKernelArg(kernel, 5, sizeof(cl_int), (void*)&d); err = clSetKernelArg(kernel, 6, sizeof(cl_float), (void*)&alpha1); err = clSetKernelArg(kernel, 7, sizeof(cl_float), (void*)&alpha2); if (err != CL_SUCCESS) { LOGI_(ErrorLog) << "Error: Failed to set kernel parameters!" << err; return false; } err = clEnqueueNDRangeKernel(queue, kernel, 2, NULL, globalThreads, localThreads, 0, NULL, &run_evt); if (err != CL_SUCCESS) { LOGI_(ErrorLog) << "Error: Failed to run range kernel!" << err; return false; } err = clWaitForEvents(1, &run_evt);-------------------------------------------------hangs here if (err != CL_SUCCESS) { LOGI_(ErrorLog) << "Error: Failed to wait event run_evt!" << err; return false; } cl_event read_evt; status = clEnqueueReadBuffer(queue, cl_output, 1, 0, data_size * sizeof(cl_float), output_data, 0, 0, &read_evt); if (status != CL_SUCCESS) { LOGI_(ErrorLog) << "Error: Failed to transport input_data to device!" << status; return false; } err = clWaitForEvents(1, &read_evt); release........ My kernel Code： __kernel void Test(__global float *input_data, __global float *output_data, int in_rows, int in_cols, int kernel_rows, int kernel_cols, float alpha1, float alpha2) { int x = get_global_id(0); int y = get_global_id(1); return;