In addition to standard timestamps we can also ask postgres to return the number of seconds from the epoch.  To the uninitiated the epoch is usually the number of seconds past a fixed point in time (usually, but not always, midnight on 01/01/1970).  If you would like to know when the epoch is on your system then from a psql shell enter the following command:

postgres=# SELECT TO_TIMESTAMP(0);

my system returned 01/01/1970 as expected:

      to_timestamp
------------------------
 1970-01-01 00:00:00+00
(1 row)

If we want to return a timestamp using seconds from the epoch then we use the EXTRACT function

postgres=# SELECT EXTRACT(EPOCH FROM now());

will return something like

    date_part
------------------
 1506028038.21499
(1 row)

As with standard timestamps this is a very high resolution however thankfully we can use the same functionality to tame our return value and get the resolution that we require.

DATE_TRUNC and AT TIME ZONE work the same for epoch arguments as the do for standard timestamp examples so I am not going to go into a great deal of detail about them here, but here are a couple of examples to help you with the syntax:

postgres=# SELECT EXTRACT(EPOCH FROM DATE_TRUNC('minutes' , now()));

will return something like

 date_part
------------
 1506028320
(1 row)

Unlike a standard timestamp, when returning seconds past the epoc rather than appending the string with +XX to denote the time zone, the function will add 3600 seconds for every hour so if you are in a timezone that is GMT +6 it will add 21600 to the value when you use AT TIME ZONE

postgres=# SELECT EXTRACT(EPOCH FROM DATE_TRUNC('minutes', now() AT TIME ZONE 'Europe/London'));

if the value of now() is 1506028320 without a time zone then now() AT TIME ZONE ‘Europe/London’ will return i.e. 1506028320 + 3600

 date_part 
------------ 
1506031920
 (1 row)

One of the big advantages of using seconds from the epoch is when performing time calculations.