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.

Performance/Stress Testing Tools



Hi,

Any recommendations on performance/stress testing tools that we can use for TIVA C microcontrollers?

Thanks,

Rashmi

  • Hi,

    Regarding stress test, usually the MCU is run in a oven for several hours at a recommended high temperature.

    - kel
  • Physical stress or program stress?

    I don't know of any sets of program stress tools. Test Driven Development can do some of that.

    Physical stress, take a look at HALT. Note that unlike earlier tests the goal of HALT is to break the tested unit. Find the weak spots and fix them. Generally the test plan under HALT is to increase the stress until something breaks, fix/reinforce the break. Repeat until nothing breaks or anymore or the whole product breaks at once.

    Robert
  • An oven test - by itself - seems restricted.   Far more useful would be, "Shake/Bake" which provides much more of a, "real world" stress exercise.   Often too - the power supplied to the MCU is varied to encompass both the "Max & Min" legal levels - this while, "Shake/Bake" is in play.

    Usually "interconnects" - not the MCU - are most prone to failure.  

    Along w/Robert's "HALT" comes it's sidekick "HASS."    Google well defines/outlines.   Definitions below.

    HALT - Highly Accelerated Life Test

    HASS - Highly Accelerated Stress Screen

    While not strictly "stress/performance" related - the susceptibility of your device/board to EMI/RFI deserves investigation...

  • Hello Rashmi

    Do you mean CoreMark or Dhrystone Performance? I am not sure what you meant by Stress Tool?

    Regards
    Amit
  • Hi Amit,

    Yours is a "good" get - missed by those earlier.

    We seized upon "stress" - you locked-on to "non-stress" MCU performance.
    Perhaps "both" are important - this is the first occurrence of "serious stress" (here) that I recall....

    I've defined HALT (and sidekick) HASS (my post - 2 up) - to avoid poster's (search) demand...

  • Hello cb1,

    Performance: Quite clear. Stress: Not so. I have seen a lot of presentations from EDA companies on Stress Infrastructure but the "fine print" on the bottom reveals none as good as the engineer working on it...

    Regards
    Amit
  • Hi Amit,

    Agreed that a good/experienced/"sweat the details" engineer proves a huge asset - yet the value & benefits of a full & proper "Stress Test Procedure" remain - (surely) worth the "Risk-Reward."   (i.e. even the "best" engineer is unlikely to catch devices subject to unexpected "infant mortality" - fully outside of the engineer's domain & control...yet well caught/captured by an effective "stress" protocol...)

    It appears that the two together, (good engineer & stress procedure) yield "best/brightest" results...

  • Hello cb1

    Lets wait for the poster to elaborate on their definition as well. May be that shall bring in new evidence to further the case.

    Regards
    Amit
  • Here we must agree to disagree.     Poster's "specifics" cannot dilute the (clear) value of "Stress Tests."

    Thread here has value far beyond poster's intent...

  • Hello cb1

    The poster's original question is on recommendation for Tools that can do Stress Testing and not the Stress Tests themselves. Auto-pilot during take off and landing (not prescribed).

    Regards
    Amit
  • Amit Ashara said:
    Auto-pilot during take off and landing (not prescribed).  

    Those (fortunate enough to "return to ground") - after ride in cb1-piloted twin - may not (fully) agree...

    Poster appears unimpressed w/effort expended - his behalf...  

  • Hi

    I am not aware of the all the different parameters which come under performance testing for MCUs. Some of them can be parameters like speed, power consumption in active mode, standby and low power modes etc

    Thank you for your replies

    Regards,

    Rashmi

  • Hello Rashmi,

    Don't expect a tool to be smart enough to figure that out for you. First decide on what is that you want to stress in the device, then design a test environment that can generate and check for both correct and incorrect response to stimuli. There are some vendors that provide equipment and infrastructure that can build such environment but none automated to do it for you.

    Regards
    Amit
  • Pardon - but by staff (and my) read - "attempting to split the atom" should not occur before "full/proper study of that atom" has been achieved...
  • Hello cb1

    I hope that was what I conveyed in my post as well.

    Regards
    Amit