This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

'Save Memory' Word Length Parameter Misleading

Other Parts Discussed in Thread: TMDSCNCD263

Hello,

I'm using CCS 12.3 with TMDSCNCD263.

I want to upload the memory contents from the processor using 'Save Memory' and dump to a text file.

  • The 'File Type' is TI Data.
  • The Format is 32-Bit Hex - TI Style
  • The origin is 0x7002E000 and length is 0x001D0FF0 (1,904,624 bytes or 952,312 words).

It takes less than 10 seconds to reach 50% then crawls along to 100% for the next 7 minutes.

On investigating why it takes so long, I used a small test file. In this example I would expect 4 words to be uploaded and presented as 2 x 32 bit records:

But the actual result is double the expected upload which goes some way to explain the extended upload time.

It seems that Length specified as "number of memory words" does not mean number of 16 bit units to upload, it means the number of records specified in the Format field. The Help on this feature does not elaborate on this.

  • Suggest to be able to specify the upload length in Minimum Addressable Units, one unit being 8 bits (or 16bits for C2000)? 
  • Failing that, can the the mention of "word" be removed and replaced by "record" or "units of Format" or similar? 

Thank you.

  • Hi Kier,

    On investigating why it takes so long, I used a small test file. In this example I would expect 4 words to be uploaded and presented as 2 x 32 bit records:

    Given that the size of a word on the Cortex-R is 32 bits, wouldn't you expect 4 words to be 4 32-bit records?

    Thanks

    ki

  • t seems that Length specified as "number of memory words" does not mean number of 16 bit units to upload, it means the number of records specified in the Format field.

    However, this is true. If the device was say a F28x device with a 16-bit word, the result would have been the same. Hence I agree that "records" makes more sense.

    I filed a bug for this. Tracking link: https://sir.ext.ti.com/jira/browse/EXT_EP-11420

    Thanks

    ki

  • Given that the size of a word on the Cortex-R is 32 bits, wouldn't you expect 4 words to be 4 32-bit records?

    Hi Ki,

    Well, yes and no. If I knew that the dialogue box meant architecture word then, yes, you got me! A quick note in the Help file might have helped here.

    The word "word" is open to interpretation and that is my point. "word" has a variety of bit widths depending on your background. A Windows programmer will say a word is 16 bits for example. My mistake was assuming the same.

    The use of "octet" would have eliminated all ambiguity but it seems that we can't specify the length in octets unfortunately.

  • I filed a bug for this.

    Awesome! Thank you.