Hi, I need to analyse the performances of my program, running on the target platform. Actually I need the performance analyzer, but not through simulator. Indeed, using the simulator is not enougth, because in the real use cases there are a lot of interactions with the outside world that I can not simulate. I don't need very precise timing measurement, but only having an idea on where the program spend most of its time. Can you give me a hint on this? I work with µVision3 V3.72, and Ulink2 probe, and the target is a STM32 chip (Cortex-M3 core). Thanks.
use a scope and a pin to see indication of how long the program spends handling interrupts and in user mode. you can also use something like a trace buffer.
Thanks, Indeed I found the Realtime Trace Capabilities and the logic analizer. I should even be able to avoid stealing an oscilloscope from the hardware guy. With this capability I can measure the time spent on the functions I choose, but I still need a way to spot time consuming function that I did not think about. For this purpose the performance analyser was the perfect tool.