 So in this example, our workload has three tasks. Task number one is going to require three instances of task two before it can run. And task number two is going to require ten instances of task number three before it can start running. So we'll want to start by doing one. So we'll want to start by calculating how many times we need to run each task to complete one instance of task number one. So obviously, we will run task number one one time. In order to run task one once, we need the outputs from three instances of task number two. And for each instance of task two that we've run, we're going to need to run task number three ten times. So task three will get run ten times for each instance of task number two. So we'll multiply our ten by three. Tell us that we need a total of 30 instances of task two in order to complete one instance of task one. Now we can look at how long it takes each of these machines to run this set of tasks. So machine A needs to run one instance of task one. And that's going to take ten microseconds. Ten needs to run three instances of task number two. And each of those is going to take 100 nanoseconds. And then we're going to run 30 instances of task three, each of which are taking five microseconds. So this will give me 10 microseconds. Three times 100 is 300 nanoseconds or 0.3 microseconds. So it's in the same terms as the other two. And then 30 times five will give me 150 microseconds. Add that together, I have 10 and 150 is 160 plus the 0.3. We'll do the same thing for machine B. Again, I am going to need to run task one one time. And that will take 15 microseconds this time. I need three instances of task two, each of which will take 250 nanoseconds. And then I need my 30 instances of task three, each of which take 30 microseconds. So I'll get 15 microseconds plus 750 nanoseconds or 0.75 microseconds. And then 30 times 30 will give me 900 microseconds. Added together, 915.75 microseconds. Now I'm going to want to put these into my relative performance equation. And I'm interested in the performance of machine A relative to machine B. So again, since I have execution times, I'm going to want machine A to be on the bottom of my fraction and machine B to be on the top. So I would have, and machine B is, and machine A is. Now this time I don't have either of my numbers in terms of seconds. They are in terms of microseconds. However, since I'm looking at a ratio here, and they are both in the same terms, I know that both of these are going to cancel. And I'm still going to end up with the ratio that I expect. Now I've just got a number of 915.75 divided by 160.3. And if I put this into my calculator, I'm going to come out with a number. About 5.71. So that means that machine A is about 5.71 times faster than machine B. Clearly, machine A is taking much less time than machine B is. So it's reasonable that we should conclude that machine A is several times faster than machine B.