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.

AM62X-PET-CALC: How to use the AM62x Power Estimation Tool

Part Number: AM62X-PET-CALC

Could you tell me how to use the AM62x Power Estimation Tool?

For Operating Performance Point (OPP):

Q1. Set the operating frequency of each of the compute cores and clusters to 0 MHz to disable it?

For example, "GPU Frequency (MHz)" will be set to 0 for devices without GPU.

But "PRU-SS Frequency (MHz)" cannot be set to 0 since 0 cannot be selected from the list.

Q2. Cannot PRU-SS be disabled?

For Processor Core Utilization (%):

The application note "SPRAD31" is described as follows:

"For guidance, 0% is an "off" or "unused" state. 1% is "idle". 100% is maximum utilization (i.e. Dhrystone)."

Q3. Is the maximum utilization (100%) when processing that is forced to load the cores, such as using the dhrystone benchmark test?

Q4. Is the maximum utilization (100%) when CPU utilization keeps 100% on an OS such as Linux?

Q5. What is the worst case utilization (%) when used for control via various IO and interfaces, and for HMI display and control?

System firmware for security management and resource and power management is provided by TI. TI Foundational Security (TIFS) services will run on HSM, and Resource management (RM) and Power management (PM) services will run on R5F.

Q6. What is the HSM utilization (%) when TIFS is running?

Q7. What is the R5F utilization (%) when RM/PM is running?

The utilization of "Security Accelerator" is a static field.

Q8. Does "Security Accelerator" mean "Crypto Accelerators"?

Q9. Why is the utilization of "Security Accelerator" a static field?

For LVCMOS IO and Peripherals:

"off" and "unused" can be selected from the list in "Mode".

Q10. What is the difference between "off" and "unused"?

The application note "SPRAD31" is described as follows:

"Utilization (%): Specifies the utilization as a percentage of activity relative to a full load condition."

Q11. Does the full load condition mean that data is constantly being sent and received?

Q12. For IOs and Peripherals that support full-duplex, is the utilization less than 50% when only sending or receiving?

Q13. When VOUT is used, is the utilization 100%?

Q14. Is "DDR WR %" the percentage of write accesses out of total accesses?

MMC/SD supports UHS–I DDR50, but "ddr_50mbs" cannot be selected from the list.

Q15. Is it an error that "ddr_50mbs" cannot be selected from the list?

Attach the configured PET sheet here: AM62x_Power_Estimation_Tool_Public_1v1_AM6232ASGGHAALWonMCBD.xlsm

Best regards,


  • Hi Daisuke, 

    I am looking into your questions and will follow up after the weekend.


  • Hi Daisuke,

    Q1) We recommend users set the CLK freq to zero, and if that is not available then to set the Util% of the core to 0%.

    Q2) We recommend users set the PRU Util% to 0%.

    Q3) Your understanding is correct.

    Q4) A better example of core util% equal to 100% is dhrystone, instead of OS Boot. OS Idle is a heavy load procedure but not all IPs are 100%, and are customer application specific.

    Q5) These IO events are 'bursty', and not a continuous event. We recommend using the PET inputs and observing how the estimated power changes as a function of IO modulation.

    Q6) This is an application specific question; we do not have a nominal TIFS power profile.

    Q7) This is an application specific question; different applications have different PM software governing on the R5

    Q8) The Security Accelerator refers to the HSM in the TRM/Datasheet

    Q9) The power profile of the HSM is non trivial and we are still determining the best way to represent it in the PET. This is why we left it static in this version of the PET

    Q10) These modes were more unique during the early phases of the tool development. Over the iterations we kept both modes, but my personal observation is that the power profiles of each mode is the same.

    Q11) This is correct

    Q12) This is a reasonable understanding and model for PET inputs

    Q13) When VOUT is used, we recommend users chose the appropriate output resolution. Since the PET is suggested to be used for thermal modeling, it could be a good idea to set the Util% to 100% to understand the expected power profile; this is up to the Design Engineer.

    Q14) Your understanding is correct. 

    Q15) We recommend users choose SDR100 to model DDR50 power profile in the PET.



  • Hi Colin-san,

    Thank you for your reply.

    Our customer refer to the "AM62x Power Consumption Summary" application note in conjunction with PET. The application note shows actual measurements under various conditions, but the room temperature is always 25°C and the junction temperature (Tj) does not exceed 38°C under temperature controlled forced air.

    The temperature at the top of package of the AM6232ASGGHAALW on our customer's board exceeds 50°C at a room temperature of 23°C.

    Was a heatsink used for the AM62x on the board used for measurements in the application note?

    What is the airflow (m/s) under temperature controlled forced air used for measurements in the application note?

    Best regards,


  • Hi Daisuke-san,

    Was a heatsink used for the AM62x on the board used for measurements in the application note?

    There was not one used. Instead a Thermal Flow System, TFS.

    What is the airflow (m/s) under temperature controlled forced air used for measurements in the application note?

    about 15-17 scfm.