Android AlarmManager - RTC_WAKEUP vs ELAPSED_REALTIME_WAKEUP

Can someone explain to me the difference between AlarmManager.RTC_WAKEUP and AlarmManager.ELAPSED_REALTIME_WAKEUP? I have read the documentation but still don't really understand the implication of using one over the other.

Example code:

    alarmManager.set(AlarmManager.ELAPSED_REALTIME_WAKEUP, 
                     scheduledAlarmTime, 
                     pendingIntent);

    alarmManager.set(AlarmManager.RTC_WAKEUP, 
                     scheduledAlarmTime, 
                     pendingIntent);

How different will the two lines of code execute? When will those two lines of code execute relative to each other?

I appreciate your help.


Solution 1:

AlarmManager.ELAPSED_REALTIME_WAKEUP type is used to trigger the alarm since boot time:

alarmManager.set(AlarmManager.ELAPSED_REALTIME_WAKEUP, 600000, pendingIntent);

will actually make the alarm go off 10 min after the device boots.

There is a timer that starts running when the device boots up to measure the uptime of the device and this is the type that triggers your alarm according to the uptime of the device.

Whereas, AlarmManager.RTC_WAKEUP will trigger the alarm according to the time of the clock. For example if you do:

long thirtySecondsFromNow = System.currentTimeMillis() + 30 * 1000;
alarmManager.set(AlarmManager.RTC_WAKEUP, thirtySecondsFromNow , pendingIntent);

this, on the other hand, will trigger the alarm 30 seconds from now.

AlarmManager.ELAPSED_REALTIME_WAKEUP type is rarely used compared to AlarmManager.RTC_WAKEUP.

Solution 2:

Despite the currently accepted and up-voted answer, AlarmManager.ELAPSED_REALTIME* types along with SystemClock.elapsedRealtime() has always been more reliable than the RTC clocks for alarms and timing.

Using ELAPSED_REALTIME_WAKEUP with AlarmManager will rely on a monotonic clock starting from boot time "and continues to tick even when the CPU is in power saving modes, so is the recommend basis for general purpose interval timing". So,

alarmManager.set(AlarmManager.ELAPSED_REALTIME_WAKEUP, SystemClock.elapsedRealtime()
                 + 60*1000, pendingIntent);

will make your PendingIntent fire in 1 min (60*1000 milliseconds).

Whereas, AlarmManager.RTC_WAKEUP is for the the standard "wall" time in milliseconds since the epoch. So,

alarmManager.set(AlarmManager.RTC_WAKEUP, System.currentTimeMillis()
                 + 60*10000, pendingIntent);

may also trigger the alarm 60 seconds from now, but not reliably, because as noted in the SystemClock documentation:

The wall clock can be set by the user or the phone network (see setCurrentTimeMillis(long)), so the time may jump backwards or forwards unpredictably. This clock should only be used when correspondence with real-world dates and times is important, such as in a calendar or alarm clock application. Interval or elapsed time measurements should use a different clock. If you are using System.currentTimeMillis(), consider listening to the ACTION_TIME_TICK, ACTION_TIME_CHANGED and ACTION_TIMEZONE_CHANGED Intent broadcasts to find out when the time changes.

Also, the question only referenced only the *_WAKEUP alarms but see also the AlarmManager documentation on that to make sure you understand what the wakeup vs non-wakeup alarms provide.

Solution 3:

Just a note. You can get the uptime millis calling:

long uptimeMillis =  SystemClock.elapsedRealtime();

So if you want to fire the alarm 30 seconds from now, and you want to use the uptime clock instead of the normal clock, you can do:

long thirtySecondsFromNow =  SystemClock.elapsedRealtime() + 30 * 1000;
alarmManager.set(AlarmManager.ELAPSED_REALTIME_WAKEUP, thirtySecondsFromNow, pendingIntent);

Whenever you want to check for some elapsed time instead of a specific date/time, it's best to use the uptime. That's because the current time set by the user in the device can change if the user changes it using the settings.