Hi

In those project euler postings I find "runs in 0.xx s" or so.

Wonder how they can say so.

Did this with gforth on an empty FOR NEXT loop with k=1000000:

5 runtimetests

usertime: 80531 ms <0>
cputime: 41995 ms <0>
systime: 450 ms <0>

usertime: 43113 ms <0>
cputime: 43510 ms <0>
systime: 426 ms <0>

usertime: 75839 ms <0>
cputime: 43210 ms <0>
systime: 509 ms <0>

usertime: 61606 ms <0>
cputime: 42545 ms <0>
systime: 568 ms <0>

On Jun 9, 6:46 am, Michael wrote:

This is interesting. You've said you're running a powerbook, which is
a few iterations behind my model (a penryn macbook) but did the tests
using usertime and cpu time actually take 80 secs and 41 secs etc?

I ran your code unmodified on a gforth 0.6.9-20070604 (no fancy build
options, just linked to ffcall) and I got really weird results too.

ie - and 20 iterations went by in less than a sec.

usertime: 14688 ms <0>
cputime: 4001 ms <0>
systime: 0 ms <0>

Hi Michael, there are *micro*-seconds, not milliseconds!

best regards,
humpty-dumpty

Michael writes:

Try the following:
first do all measurements, then do all i/o.

At second, you measure sys/cpu/usr time for *different* runs.
Why don't you measure them for the same run?

On 11 Jun., 09:07, m_l_g3 yahoo.com> wrote:

Ok, lets try this:

\ test: Keeping track of Time with gforth - Mac OSX (PowerBook G4)

vocabulary test
test definitions

0 [if]

user: 14586256 44859 45282 43424 <0>
0.958968243452144
sys : 1383017 839 2644 493 <0>
0.186459909228442
ut : 1213225092442357 58390 383728 49172 <0>
0.128142851186257
time t2-t1 max min .s ratio

Hm, utimes reads the real time clock of the computer, am I right?

So, in those euler calculations utime will give the total time from
start of code until a result apears on screen, while you do email,
YouTube, ... ;-)

On Jun 12, 3:59 pm, Michael wrote:

On 13 Jun., 04:30, brian....@rogers.com wrote:
...

So you use utime too.

This also answers my question from the "Ultimate Prime Sieve

Michael writes:

It may, on some OSs. On Unix variants, typically the OS reads the
time on startup from the real-time clock, and then maintains it
through software; many systems also use NTP to keep the correct clock.

I any case UTIME gives you the real (aka "wall-clock") time.

Yes, if you use UTIME for timings, you will see the time you would
have also measured with a stop-watch.

On Unix variants, it is.