TM4C129ENCPDT: Program download speed

Ok I found this https://e2e.ti.com/support/microcontrollers/arm-based-microcontrollers-group/arm-based-microcontrollers/f/arm-based-microcontrollers-forum/863792/tm4c129encpdt-tm4c129encpdt-download-and-simulation-tools

Can someone report about the speed differences: Is the XDS560v2 faster than the XDS110 or is the speed limit related to the TM4C129 capabilities?

Hello Michael,

Michael Schuster said:

Can someone report about the speed differences: Is the XDS560v2 faster than the XDS110 or is the speed limit related to the TM4C129 capabilities?

I don't think we ever really benchmarked these points but a lot of the bottleneck is simply the time it takes to write to Flash memory over JTAG. I am not sure how much faster an XDS560 or even a Seggar J-Link would be, but I've never heard of customers using them for speed, only extra debug capabilities.

Michael Schuster said:

Additionally both debuggers fail to load the program in about 30..50 % -  as the running program is sometimes not properly terminated.

That seems to be a CCS issue with debug sessions though I don't usually see that so probably some issue with the steps taken to program too - regardless in production you'd use CCS Uniflash and not have this issue at all.

Best Regards,

Ralph Jacobi

Referring the speed issue: I'll have to evaluate on my own, then [sigh].

With the Uniflash we donot face problems with downloads in production. But in that case, the controllers are in factory delivery state, so this is no wonder. When debugging with a "fresh" part, I also donot encounter any problem with CCS.

Hi Michael,

I can try and do a quick check with 1MB image on a DK-TM4C129X using ICDI vs XDS200 vs Seggar J-Link on Uniflash this afternoon to see if the latter offers any discernable advantage to XDS200. I don't have an XDS560 to test with.

Best Regards,

Ralph Jacobi

Hi Ralph,

that sounds great!

Regards

Micky

Debug Probe Performance Comparison has some benchmarks on a XDS560 (and other debug probes) on a MSP432E401Y is a sister device of a TM4C129, so should be comparable.

Hi Ralph,

Thanks for your help.

If Debug Probe Performance Comparison is also valid for the TM4C,  the XDS560v2 is not faster (if you can evaluate for this particular device, this is welcome, too).

I will buy a segger then. I will check also the different options there. Is there a spreadsheet for J-Link integration or is this just out of the box?

Regards

Micky

Hi Michael,

I wasn't able to find an XDS560 to test with unfortunately. That said, I would expect that what Chester shared would reflect very closely with what to expect with TM4C as the devices are very similar, so it looks like XDS560 doesn't have any advantages over XDS200 whereas J-Link will be faster than both.

Best Regards,

Ralph Jacobi

Hi Ralph,

That's ok so far. I checked segger's homepage and their products. I've seen, they also have a arm-flasher unit, which might be useful in our production. It is fast and can be controlled via e.g. simple TCP-commands. That is even easier than using uniflash. We will evaluate this. I will report, if you don't mind.

I will check the different J-Link Debuggers and Trace devices, and see, how I can use them. For now, I'm using CCS for debugging only, anyway. I compile with make and use emacs (why I do so, is another story to tell with a lot of pain...). Well, my dream: Work complete in emacs, download fast, debug robust, do all inside emacs. OMG, this would be a wonderful day...

Regards

Micky.

Hi

as I told to give feedback: With the Segger J-Link Ultra+ debugger I have a downloadtime of 9 seconds.

So for 500 kB I have here

+ ICD Debugger 29 sec

+ XDS110  20 sec.

+ JLink 9 sec.

EDIT: I have to correct myself, sorry. The next sentence is only true, if I change nothing in the code and do a simple reload.

and: The debugger is also checking, weather a flash block needs update at all, so if I change only in some functions, only these blocks are renewed, which is mostly the case during development. This speeds up to less than 1 second. And it seems to be much more robust, and more than 6 breakpoints ...

I'm really angry with myself that I didnot investigate this earlier. I will also check their gdbsever (and will report). I really can recommend this debugger, if you code more than smaller projects. So thumbs up!!!! We also bought the Segger Arm Flasher, which I think will be as fast. As we flash a lot of devices, this will save a lot of time.

Regards

Micky