measuring the execution time on multicore application

[moh_algoblan]

Hi
I want to test the performance of Xcore 200 when using more than one core, and comparing it with the case of using only one core. so I decided to to use the Time Analyzer with single core application first. it works well, but when I tried with multi-core app it shows me only the expected time of each task, not the whole program running with task in parallel. I though maybe because it's hard, not accurate, too, to estimate the execution time of a multi-core application, so I want to know if there is a way to get the execution time by running my program on the Xcore 200 micro-controller, not by using the Time Analyzer.

[Gothmag]

Typically it's simple math but the second cores interact and ports are involved all bets are off since they run independently. That's why you typically look at timing of functions. If you really want to time it running I'd just use 1 port per core and toggle high at begin, low at end, measure the +pulse time.

[moh_algoblan]

Typically it's simple math but the second cores interact and ports are involved all bets are off since they run independently. That's why you typically look at timing of functions. If you really want to time it running I'd just use 1 port per core and toggle high at begin, low at end, measure the +pulse time.

thanks, I will try that, but I prefer testing it using the simulator since it's easier, I think if I use a port that changes its value after the par function, then I will be able to get execution time of that part of the code by looking at the waveform of the port after running it on the simulator.

[MuellerNick]

> thanks, I will try that, but I prefer testing it using the simulator since it's easier,

You still can do that in the simulation. Really quite easy.

Nick