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.

TMDSCNCD263: Clarification required for one NOP action taken 3.5ns

Part Number: TMDSCNCD263

Hi Team,

Note:

  1. we are integrated in vector OS.
  2. Gel file core clock value

Method to verify core clock frequency in app (CycleCounterP_getCount32):

  1. The below observation shall get the time stamp between 1000, 2000 and 3000 “NOP” operations.

  2. The result for the delta value 3500 CPU cycle counter value as number for each 1000 NOP. Then for one NOP It would take around 3.5ns.

  3. The core clock was assuming as per the gel file configured for 400Mhz, then one cycle operation (Single “NOP “) will take 2.5ns theoretically.

  4. Hence practical value is not matched with the theoretical value.

Could you please clarify mentioned below points guide us? 

  1. My understanding the above delta value is not a tick value, CycleCounterP_getCount32 return value in ns.

  2. If the CycleCounterP_getCount32 return the tick value means, the 1000 NOP match with 1000 ticks.

  3. how to get almost match value with practical

Thanks

Vivek Arivazhagan

  • Hi Vivek,

    1. The CycleCounterP_getCount32() returns the CPU cycles (2.5ns/cycle for AM263x) not ns

    2. It also depends on how you do you 1000 NOPs. If you use the while or for loop, then it is not accurate. If you put the code in on-chip memory, then it is not accurate. The best way to do it is to put the code in TCM and measure the difference between 11 NOPs and 10 NOPs. It should be 1 CPU cycle.

    Best regards,

    Ming