Super PI Ubuntu - How fast is your cpu?

[mr clark25]

sorry for pasting so many times in a row.

i made a script that does the same thing as super pi (although it doesn't time itself). but, you must have pie installed to run it.

code:
sudo apt-get install pi

install pi, then download the script, and run it. i think it runs faster than super pi, because i got 11.5 seconds on a dual core running at 1.8ghz. you may need to make it executable. there is no need to open a terminal to run it. you can do it graphicly. (double click, select option "run in terminal") at the end it may display a time, but its not the time it took to complete. so, you just need to time it yourself...

[dbbolton]

sorry for pasting so many times in a row.

i made a script that does the same thing as super pi (although it doesn't time itself). but, you must have pie installed to run it.


install pi, then download the script, and run it. i think it runs faster than super pi, because i got 11.5 seconds on a dual core running at 1.8ghz. you may need to make it executable. there is no need to open a terminal to run it. you can do it graphicly. (double click, select option "run in terminal") at the end it may display a time, but its not the time it took to complete. so, you just need to time it yourself...

Your script seems to be calling 'pi' more than once. Still, if it is faster than Super Pi, then its results are totally irrelevant to all others in this thread.

If you really want to use the 'pi' program in the repositories to calculate pi to 2^20 digits and time the process, all you need to run is this command:

time pi 1048576 | egrep '.*total$'

And for me, running a T7600, this only took about 4 seconds. So I am pretty sure that this program isn't at all comparable to Super Pi.

The 'time' command is usually accurate to a few milliseconds:

time sleep 1
sleep 1  0.00s user 0.00s system 0% cpu 1.003 total

Considering human reaction time, I would guess that using a stop watch is going to be at least 1/3 of a second off, and probably way more for really short periods.

[markp1989]

Your script seems to be calling 'pi' more than once. Still, if it is faster than Super Pi, then its results are totally irrelevant to all others in this thread.

If you really want to use the 'pi' program in the repositories to calculate pi to 2^20 digits and time the process, all you need to run is this command:

time pi 1048576 | egrep '.*total$'

And for me, running a T7600, this only took about 4 seconds. So I am pretty sure that this program isn't at all comparable to Super Pi.

The 'time' command is usually accurate to a few milliseconds:

time sleep 1
sleep 1  0.00s user 0.00s system 0% cpu 1.003 total

Considering human reaction time, I would guess that using a stop watch is going to be at least 1/3 of a second off, and probably way more for really short periods.

here is mine, e8400@4ghz i dont know if it uses both cores?

mark@mark-desktop:~/superpi$ time pi 1048576 | egrep '.*total$'

real	0m2.391s
user	0m2.350s
sys	0m0.050s
mark@mark-desktop:~/superpi$

this is my e5200 @ stock (2.5ghz)

mark@torrentslave:~/scripts$ time pi 1048576 | egrep '.*total$'

real	0m4.227s
user	0m4.100s
sys	0m0.120s
mark@torrentslave:~/scripts$

time is def accurate ...

mark@mark-desktop:~$ time sleep 1 

real	0m1.001s
user	0m0.000s
sys	0m0.000s
mark@mark-desktop:~$

[dbbolton]

here is mine, e8400@4ghz i dont know if it uses both cores?

mark@mark-desktop:~/superpi$ time pi 1048576 | egrep '.*total$'

real    0m2.391s
user    0m2.350s
sys    0m0.050s
mark@mark-desktop:~/superpi$

this is my e5200 @ stock (2.5ghz)

mark@torrentslave:~/scripts$ time pi 1048576 | egrep '.*total$'

real    0m4.227s
user    0m4.100s
sys    0m0.120s
mark@torrentslave:~/scripts$

time is def accurate ...

mark@mark-desktop:~$ time sleep 1 

real    0m1.001s
user    0m0.000s
sys    0m0.000s
mark@mark-desktop:~$

If you are using bash, there is no need to grep for 'total' because bash will print it as 'real'. Of course it does suppress the number output of pi.

Then again, 'time' is pretty greedy. It will time whatever comes after it, including the pipe.

[powel212]

i7 950 @ 4ghz

------ Started super_pi run : Sun Jan 17 00:01:53 CST 2010
Start of PI calculation up to 1048576 decimal digits
End of initialization. Time= 0.128 Sec.
I= 1 L= 0 Time= 0.300 Sec.
I= 2 L= 0 Time= 0.336 Sec.
I= 3 L= 1 Time= 0.252 Sec.
I= 4 L= 2 Time= 0.268 Sec.
I= 5 L= 5 Time= 0.244 Sec.
I= 6 L= 10 Time= 0.240 Sec.
I= 7 L= 21 Time= 0.260 Sec.
I= 8 L= 43 Time= 0.264 Sec.
I= 9 L= 87 Time= 0.264 Sec.
I=10 L= 174 Time= 0.272 Sec.
I=11 L= 349 Time= 0.256 Sec.
I=12 L= 698 Time= 0.260 Sec.
I=13 L= 1396 Time= 0.272 Sec.
I=14 L= 2794 Time= 0.244 Sec.
I=15 L= 5588 Time= 0.268 Sec.
I=16 L= 11176 Time= 0.244 Sec.
I=17 L= 22353 Time= 0.248 Sec.
I=18 L= 44707 Time= 0.236 Sec.
I=19 L= 89415 Time= 0.256 Sec.
End of main loop
End of calculation. Time= 5.320 Sec.
End of data output. Time= 0.048 Sec.
Total calculation(I/O) time= 5.368( 0.224) Sec.
------ Ended super_pi run : Sun Jan 17 00:02:03 CST 2010

[drphilngood]

i7 950 @ 4ghz

------ Started super_pi run : Sun Jan 17 00:01:53 CST 2010
Start of PI calculation up to 1048576 decimal digits
End of initialization. Time= 0.128 Sec.
I= 1 L= 0 Time= 0.300 Sec.
I= 2 L= 0 Time= 0.336 Sec.
I= 3 L= 1 Time= 0.252 Sec.
I= 4 L= 2 Time= 0.268 Sec.
I= 5 L= 5 Time= 0.244 Sec.
I= 6 L= 10 Time= 0.240 Sec.
I= 7 L= 21 Time= 0.260 Sec.
I= 8 L= 43 Time= 0.264 Sec.
I= 9 L= 87 Time= 0.264 Sec.
I=10 L= 174 Time= 0.272 Sec.
I=11 L= 349 Time= 0.256 Sec.
I=12 L= 698 Time= 0.260 Sec.
I=13 L= 1396 Time= 0.272 Sec.
I=14 L= 2794 Time= 0.244 Sec.
I=15 L= 5588 Time= 0.268 Sec.
I=16 L= 11176 Time= 0.244 Sec.
I=17 L= 22353 Time= 0.248 Sec.
I=18 L= 44707 Time= 0.236 Sec.
I=19 L= 89415 Time= 0.256 Sec.
End of main loop
End of calculation. Time= 5.320 Sec.
End of data output. Time= 0.048 Sec.
Total calculation(I/O) time= 5.368( 0.224) Sec.
------ Ended super_pi run : Sun Jan 17 00:02:03 CST 2010

Strange, you're showing a 5.368 second run but it took 10 seconds.

[powel212]

Strange, you're showing a 5.368 second run but it took 10 seconds.

http://www.youtube.com/watch?v=Qcagk5mzoyA

I just ran it. I didn't try to do anything funny with it. Just wanted to demonstrate what is going on with my i7.

This is Jaunty running on my mac in a Vbox guest. I gave Jaunty 4 processors and 2 gigs of ram.

No funny business. Check the attached screencast.

Powel

[drphilngood]

http://www.youtube.com/watch?v=Qcagk5mzoyA

I just ran it. I didn't try to do anything funny with it. Just wanted to demonstrate what is going on with my i7.

This is Jaunty running on my mac in a Vbox guest. I gave Jaunty 4 processors and 2 gigs of ram.

No funny business. Check the attached screencast.

Powel

My apologies if my post seemed accusatory; I was just making conversation and it was not meant that way. I was only curious about the difference between the score and test time because your score is so much higher than any I can remember seeing for a 4Ghz Bloomfield.
Congrats and kudos, my friend, and, once again, my apologies. =)

[powel212]

My apologies if my post seemed accusatory

No worries. I actually didn't notice the difference in the number total until you mentioned it. It is a bit odd.

I can tell you this though. If you are wondering about switching to i7. I can tell you, I love it. The one drawback though is I can no longer use anyone else's computer. Can't deal with the lag.

If you are happy with what you've got enjoy it but i7 is definitely worth it.

P.

[drphilngood]

...i7 is definitely worth it.

P.

They sure are. I built my new Lynnfield rig about four months ago, after running my last build - an AMD socket 939, x2 4800+ - for about five years, and I am almost as impressed with the i7 as I was with the x2(my first dually). =)