CODECOMPOSER: Hardware Trace Analyzer : Meaning Of Cycles

Hi Kier,

Do you get this message at the bottom of the view?

I assume you are using the ETB?

Also, any reason your are using Hardware Trace Analyzer? Your device supports TVT. Hardware Trace Analyzer was deprecated in support long ago, even in CCS 12.

Thanks

ki

Hi Ki,

Ki said:

Do you get this message at the bottom of the view?

Yes, I do get that message. I figured this wasn't an issue as I was happy to have the 't' axis in cycles....as long as 1 cycle was a known time quantum.

Ki said:

I assume you are using the ETB?

Yes, that's correct.

Ki said:

Also, any reason your are using Hardware Trace Analyzer?

Only because I thought that was my only option. I wondered why the manual was so old! What is TVT please?

We'll look into that. Many thanks!

Hi Ki,

Thanks for this documentation. I have tried replicating the results of Hardware Trace Analyzer with the function call sequence graph etc. in the latest version of CCS Theia and the only thing I can see being available is the Core Trace/Discontinuity trace.

Is there a way to enable EnergyTrace++ for the AM263P4 through XDS110?

Is there a plan to add support for AM263Px devices too?

I assume that if it was available in the old Eclipse based platform, which is now deprecated as you mention above, it would also be supported in the successor of CCS Eclipse.

I have captured a trace with your suggested method above:

Do Cycles in this view mean CPU Cycles?

Manos Giatrakis said:

Is there a plan to add support for AM263Px devices too?

No.

Manos Giatrakis said:

I assume that if it was available in the old Eclipse based platform, which is now deprecated as you mention above, it would also be supported in the successor of CCS Eclipse.

Ideally this would be the case. However, we will evaluate features based on resources and popularity and remove features accordingly. 

Manos Giatrakis said:

Do Cycles in this view mean CPU Cycles?

Cycles typically refer to cpu cycles unless noted otherwise.

Manos Giatrakis I wonder if the use of WFI in both FreeRTOS and non-RTOS cores is causing this phenomenon?

I am still waiting for confirmation here (+) AM263P4-Q1: CPU Load Calculation In No RTOS - Arm-based microcontrollers forum - Arm-based microcontrollers - TI E2E support forums that WFI() does or does not cause the CPU cycle counter to stop.

Granted, Eclipse may not be able to calculate time since it doesn't know the clock frequency but I see no reason why the ETM would not record the number of cycles correctly.