hello maxdz8,
thanks for your reply.
I read your message and i know about the blocking nature of this calls. I don't understand why one of the calls should block another host program from enqueueing commands in the queue of the second gpu.
host program nr 1 only handles gpu nr 1 and host program nr 2 handles gpu nr 2.
If i reduce my host program to:
...
| /* Create a command queue */ |
queue = clCreateCommandQueue(context, device, 0, &err);
if (err < 0) {
| perror("Couldn't create a command queue"); | |
| client_cleanup(EXIT_FAILURE); |
}
/* Create a kernel */
kernel = clCreateKernel(program, KERNEL_FUNC, &err);
if (err < 0) {
| perror("Couldn't create a kernel"); | |
| client_cleanup(EXIT_FAILURE); |
}
/* Create kernel arguments */
err = clSetKernelArg(kernel, 0, sizeof(input_buffer), &input_buffer);
err |= clSetKernelArg(kernel, 1, sizeof(output_buffer), &output_buffer);
if (err < 0) {
| perror("Couldn't create a kernel argument"); | |
| client_cleanup(EXIT_FAILURE); |
}
debug("enqueueing kernel\n");
/* Enqueue kernel */
err = clEnqueueNDRangeKernel(queue, kernel, 1, NULL, &global_size,
| &local_size, 0, NULL, NULL); |
if (err < 0) {
| perror("Couldn't enqueue the kernel"); | |
| client_cleanup(EXIT_FAILURE); |
}
sleep(60);
| /* Deallocate resources */ |
clReleaseKernel(kernel);
clReleaseMemObject(output_buffer);
clReleaseMemObject(input_buffer);
clReleaseCommandQueue(queue);
clReleaseProgram(program);
clReleaseContext(context);
client_cleanup(EXIT_SUCCESS);
-----------------------------------------------------------------
Now i should be able to call a clXXX function (for example clGetPlatformIDs()) from host program nr. 2.
But every clXXX-call from host program nr 2. will block until the kernel from host program nr. 1 got finished.
At the moment i don't really know where the problem could be.
greets