WDT and System.getTimer() Behavior on Hardware

Would anyone be curious about trying the example below on hardware using CIQ_LOG.txt? Running SDK 1.1.0 on my laptop, it fails with a WDT timeout for values of SPIN_MILLIS > 7, but I'm wondering how it fares on a real device. I'm curious when the WDT would trip and also whether getTimer() will really provide updates every millisecond.

function onShow() {

        var currentTimer = Sys.getTimer();

        var lastTimerMillis = currentTimer;

        var SPIN_MILLIS = 10;



        for(var i = 1; i <= SPIN_MILLIS; i++) {

            while(currentTimer == lastTimerMillis) {

                currentTimer = Sys.getTimer();

            }

            lastTimerMillis = currentTimer;

            Sys.println(i);

        }

    }

travis.vitek over 10 years ago

I'll test it, but I'm fairly certain that we'll find that getTimer() returns a virtual timer. i.e., you can call it many times in a loop and the value won't change.

Sforzando over 10 years ago

Thanks! Appreciate your help.

travis.vitek over 10 years ago

I'm wrong... Big surprise.

Code runs fine on my 920XT. I modified your code slightly to display the timer value in addition to the outer loop count. This code

var currentTimer = Sys.getTimer();

        var lastTimerMillis = currentTimer;

        var SPIN_MILLIS = 10;



        for(var i = 1; i <= SPIN_MILLIS; i++) {

            while(currentTimer == lastTimerMillis) {

                currentTimer = Sys.getTimer();

            }

            lastTimerMillis = currentTimer;

            Sys.println(i + ": " + currentTimer);

        }

Produces this output in Garmin/APPS/LOGS/Test.txt...

1: 380355846

2: 380356231

3: 380356496

4: 380356715

5: 380356934

6: 380357152

7: 380357365

8: 380357586

9: 380357806

10: 380358031

Sforzando over 10 years ago

That is interesting; thanks for trying. Now I'm wondering what would happen if we moved Sys.println() out of the loop and stored the values in an array instead since the I/O may be expensive: I was surprised to see 200-300 mS between each timer change. Still, it looks like the real HW will run for at least a couple of seconds without a WDT trip.

Brandon.ConnectIQ over 10 years ago

I hacked this together for you:

for( var i = 0; i < SPIN_MILLIS; i++ ) {
    while(currentTimer == lastTimerMillis) {
        currentTimer = Sys.getTimer();
    }
    lastTimerMillis = currentTimer;
    millisArray= currentTimer;
}

for( var i = 0; i < SPIN_MILLIS; i++ ) {
      Sys.println( i + ": " + millisArray);
}
[/CODE]

Looks like the timing is much tighter without the i/o:

0: 2164097
1: 2164098
2: 2164100
3: 2164102
4: 2164103
5: 2164105
6: 2164108
7: 2164109
8: 2164111
9: 2164113

0: 2166253
1: 2166255
2: 2166257
3: 2166259
4: 2166260
5: 2166262
6: 2166264
7: 2166265
8: 2166267
9: 2166269

0: 2168404
1: 2168406
2: 2168408
3: 2168410
4: 2168411
5: 2168414
6: 2168416
7: 2168417
8: 2168419
9: 2168421

Sforzando over 10 years ago

Thanks, Brandon. I was going to try this now that I have hardware to work with, but you beat me to it.