Your goal is simple to tell, but hard to implement. Draw the text MAIL (as it is one of the few words that can be read on a 1D graph) on the CPU graph.
It should look something like this:
Elaborating a bit more on what qualifies:
- The code does not need to be cross-platform (so you won't need unknown APIs to deal with).
- You may capture it in any general CPU usage utility that you have.
- The graph looks a bit worse on a different machine: I trust you this time.
- The base CPU usage % must be continuous, so if you generate a random wave and highlight something that looks like the word MAIL, it's obviously cheating.
- You may choose the maximum load to use, but it has to be substantial enough to clearly see it.
- You must follow the linearity of the example. (For M it looks like this: base %, then sudden increase to the specified max, fall gradually to a lower %, rise back to max and sudden drop to the base % again.)
- If it's unreadable, voters will notice after all.
Standard loopholes apply. Post the images too!