CCS/CC1352P: 100% CPU CCS

Hi,

That is expected behaviour. EnergyTrace, just like any other trace functionality on CCS, offloads a great amount of data from the target and performs processing to properly display it on its view. This is particularly true for the EnergyTrace HDR present on CC13x2/26x2 launchpads, since it samples the power at considerable speeds and therefore generates a large amount of data in a short period of time. 

Regards,

Rafael

I don't think this is a feature. It will take up 200% (2 cores), never returns and appears hung. The only option I have is to kill the process otherwise it will stay stuck and hold up my PC's core forever. And typically its good programming practice not to take up 100% of CPU cycles for a long period of time.

Hi

Well, the amount of time it takes to complete this operation is directly proportional to the amount of time the EnergyTrace data was captured. However, one thing is not clear: do you see this increase in CPU usage while capturing/transferring data or after the data was finished capturing but CCS then becomes locked up? 

At any rate, you can try to see if things improve by deleting the EnergyTrace temporary files typically stored in the subdirectory /home/user/.TI-trace

I will check the functionality on my Ubuntu host. 

Regards,

Rafael

I didn't know there were temp files, that is a step in the right direction. I will try deleting  and test again.

Just so you know, this doesn't seem to happen on a fresh install. Keep in mind I'm running this tool perhaps 5-25 times a day when I'm working on this project.

I also need to power cycle the board to get an accurate measurement otherwise there is a ~1-2mA mean current bias. Put that's another issue...

thank you

It would appear that having Uniflash running in the background (and idle, doing nothing) causes problems with EnergyTrace if you try to run the "Start Trace Collection" more than once. I also eliminated using a USB hub, I don't know if that matters yet.

Please comment, thanks.

Hi,

I did some tests here on my Ubuntu 18.04 with both CCSv10.0.0 and 10.1.0 (yet unreleased), but couldn't reproduce the scenario you mentioned. 

The entire body of CPUs and threads is kept at capacity while capturing, displaying the EnergyTrace data AND running the target. However, as soon as the data capture is finished, the CPU load is scaled back to a level that is used by the target running by the Debug Server. As soon as I halt the target, the CPU becomes free. 

A similar scenario happens with EnergyTrace++ and its more comprehensive data capture and display. 

With this, I am unsure what else to test. My setup is a Dell Latitude 6520 (Core i7-2620M 2.70GHz quad core, 8GiB memory, Ubuntu 18.04.3 LTS) and a LAUNCHXL-CC26X2R1. 

I hope the removal of temporary files may have worked around this issue. 

Regards,

Rafael

Did you see my previous post?

Hi,

Uniflash should not cause any interference on EnergyTrace as long as all the targets are disconnected. 

Regards,

Rafael