CS-476: [Lab3] How much can we change the experiments?

Hello,

I was wondering how much we can change the experimentation to get the measurements that we want ? For example to measure the mutex timings I was thinking it would be way quicker to simply start the system, make the cpu0 claim the mutex, wait 3 seconds and then release it right after turning a LED on and the second cpu would wait 1 sec, try to claim the mutex until it is released and then turn on LEDs. The measurement would be on the signal tap analyzer and we would measure the bit set of the LED corresponding to the mutex release and the time the other LED bit is set to know that the mutex has been claimed again.

And for all experiments that include timing measurements we could proceed like that. This means that we could get rid of the multiple timing units in the design (except for one that we could use to compare the different timing measurements).

I don't really understand why we have to implement SDRAM controllers aswell as no experiment uses the SDRAM. And I don't understand why we have to add two mutexes in the design. Isn't one enough ?

Thank you for your response !

Re: [Lab3] How much can we change the experiments?

by René Beuchat - Wednesday, 13 May 2020, 10:31 PM

Hello,

the design is a simple basic general multiprocessor system. If we need more memory it is available. And you should know how to add SDRAM on your system after the first laboratories.

In fact, it was another page of questions for this lab and I remove it as it was too time consuming for the students, but I keep the original design.

If you just want to do the minimum hardware, no problem.

For the timing measurement,

you start your specific timer or GPIO to activate a pin, or you use the signal tap as you know to do timing measurment
you activate the mutex and wait until the proc. have it
to the activation of the selected GPIO pins
release the mutex
measure the timing
do some statistics on this timing

Best.