Keep a shared work command buffer open across logical GPU operations and submit it at the existing synchronization boundary. Co-authored-by: Martin Penkava <mpenkava1337@gmail.com>
Matthew Stanley