Problem 1 A cloud gaming server (think Stadia, but that people actually play) allows players to offload the actual omputation necessary to play a video-game. A single interaction between a user and the systems consists of he following steps: (1) A user submits control inputs over a network interface (NI); (2) the CPU interprets he inputs; (3) the GPU renders the next video frame; and (4) finally the NI transmits back the next rendered video frame to the players. Assume that each interaction is a self-contained process. Further assume that very time a process is completed by the system, another one is ready to be executed in the dispatch queue. Each process is identical in terms of resource demand and follows this precise timeline. First, the NI takes 3 ms to obtain the user inputs (during this time, the NI resource is busy); next, the CPU takes 5 ms to nterpret the user inputs and launch the GPU kernel (during this time, the CPU resource is busy); next, he GPU takes 7 ms to render the current frame (during this time, the GPU resource is busy); finally the NI takes another 6 ms to transmit the rendered frame back to the user. The latter step concludes the ingle-interaction process. nitially, the system is designed such that it can handle at most one process at a time. In other words, no processing over the next set of user inputs occurs until the result for the previous set of inputs has been sent o the user. Also, your server processes requests always in the same order, i.e., if a resource is available, it oes to the request that has started processing earlier. Answer the following. a) What are the steady-state utilizations of the NI, CPU, and GPU? b) What is the overall throughput of the system? Show your work. c) What is the bottleneck of the system? Motivate your answer.