CPU-bound vs. I/O bound

1.

Modify alarm so that instead of doing only CPU work, it alternates between bursts of CPU and I/O, where the I/O in this case will be printing things on the screen. In other words, it should loop for a little while, print something, loop for a little while, print something, etc.

2.

Run the experiment again and see if the behavior of the background process (alarm) affects the amount of CPU time allocated to the foreground process (loop). Modify alarm so that it does different amounts of CPU work between each print statement. How does the frequency of the background process's I/O affect the amount of CPU time the foreground process gets?

3.

Modify alarm so that instead of doing I/O to the screen it does I/O to the disk. You can do this internally in alarm, by using fprintf, or you can do it at the shell level by redirecting the output from alarm to a file. Does it change your results when the background process is doing disk I/O instead of screen I/O?