Super PI

[BenLi]

There's a benchmark called Super PI, I ran it for 2M digits of pi to calculate, the result came back like 17 seconds. My computer runs at 667 mhz with 256 mb of ram... I looked at the scores of others (that are all machines over 1.5 Ghz) and none of them are under a minute. I'm just wondering whats going on?

Oh and i was going to upload the screen shot, but my upload quota ran out

[md]

Definitely seems wrong to me...

Here's what superpi gives me (P4 @ 2.8, Gentoo Linux)

superpi wrote:

[john@saxifrage superpi]$ sh super_pi 21
Version 2.0 of the super_pi for Linux OS
Fortran source program was translated into C program with version 19981204 of
f2c, then generated C source program was optimized manually.
pgcc 3.2-3 with compile option of "-fast -tp px -Mbuiltin -Minline=size:1000 -Mnoframe -Mnobounds -Mcache_align -Mdalign -Mnoreentrant" was used for the
compilation.
------ Started super_pi run : Mon Dec 4 17:26:31 EST 2006
Start of PI calculation up to 2097152 decimal digits
End of initialization. Time= 1.628 Sec.
I= 1 L= 0 Time= 4.908 Sec.
I= 2 L= 0 Time= 5.632 Sec.
I= 3 L= 1 Time= 5.646 Sec.
I= 4 L= 2 Time= 5.668 Sec.
I= 5 L= 5 Time= 5.711 Sec.
I= 6 L= 10 Time= 5.636 Sec.
I= 7 L= 21 Time= 5.658 Sec.
I= 8 L= 43 Time= 5.666 Sec.
I= 9 L= 87 Time= 5.664 Sec.
I=10 L= 174 Time= 5.654 Sec.
I=11 L= 349 Time= 5.632 Sec.
I=12 L= 698 Time= 5.714 Sec.
I=13 L= 1396 Time= 5.648 Sec.
I=14 L= 2794 Time= 5.675 Sec.
I=15 L= 5588 Time= 5.627 Sec.
I=16 L= 11176 Time= 5.671 Sec.
I=17 L= 22353 Time= 5.603 Sec.
I=18 L= 44707 Time= 5.523 Sec.
I=19 L= 89415 Time= 5.380 Sec.
I=20 L= 178831 Time= 4.950 Sec.
End of main loop
End of calculation. Time= 117.162 Sec.
End of data output. Time= 0.440 Sec.
Total calculation(I/O) time= 117.602( 6.921) Sec.
------ Ended super_pi run : Mon Dec 4 17:28:30 EST 2006

[TheFerret]

Something defiently does seem wrong since me duel core amd x2 @ 2.2 GHz takes 98.625s to complete it... So, I think you ran yours wrong or misintereped the results...

[md]

Something better for calculating pi that I found and modified a wee bit

get from http://nxor.org/media/calc_pi.c.bz2 (bzip2 compressed file)
extract with

code:

$ zbip2 -d calc_pi.c.bz2

compile with

code:

$ gcc calc_pi.c -o pi.exec -lgmp -lm

rum with

code:

$ ./pi.exec <number of digits to calculate> <optional output filename>

mine:

Quote:

[john@saxifrage superpi]$ ./pi.exec 2000000
[john@saxifrage superpi]$ ./pi.exec 8000000
digits=8000000, #terms=564109, depth=21
sieve ...................................................
total time = 120.380s

[timmytheturtle]

For anyone trying to use md's program, make sure you have the GNU Multiple Precision Arithmetic Library installed.

Mine:

Quote:

chris@melchior:~$ ./pi.exec 8000000
digits=8000000, #terms=564109, depth=21
sieve ...................................................
total time = 119.660s

[War_Caymore]

can someone make a link for the super PI benchmark program, i wish to try it out on my computr so see how it performs.

anyone wanna take a guess how long a P4 2.93Ghz 512 DDR RAM will perform?

[Clayton]

why don't you google it? This is what I got with "superPI Benchmark"...

click here

[md]

SuperPI itself is actually not a very good program at all; it's only popular because so many stupid people use it. But you people can use superpi all you want... my posted app can calculate pi faster; so my computer must be faster

[BenLi]

yeah i see, I only counted the "initial value found thing" i didn't wait around to see what does after. My computer ended up being around 9 minutes...haha. But to what someone said, it doesn't matter if you're dual core. The application isn't.

And md, it doesn't matter if its not efficient. The point is that its standarized, so you can conpare different configurations. However, calculating pi is probably not a comprehensive evaluation of your computer at all

Just curious, which benchmarks do you guys use (if any)?

[cool dude]

My laptop took 2 minutes and a few seconds to calculate 2M although i had a bunch of open applications running. Wow i wish i had TheFerret's computer lol. 1.36 seconds. nice!

Edit: Forgot to mention my laptop specs.
2.8 GHz, Pentium 4, 512 MB RAM.

[rdrake]

One minute, twenty-one seconds here.

Running on an Intel Centrino Duo T2500@2.00 GHz with 1 GB RAM.

[War_Caymore]

1m 55s running a 2.93Ghz P4 with 512 of DDR RAM.

not bad

[Andy]

BenLi wrote:

But to what someone said, it doesn't matter if you're dual core. The application isn't.

err, incase you havent noticed, your system runs more than 1 program at a time. dual core systems will give you a noticeable increase in performance by pipelining tasks queues to the two cpus.

[md]

Andy wrote:

BenLi wrote:

But to what someone said, it doesn't matter if you're dual core. The application isn't.

err, incase you havent noticed, your system runs more than 1 program at a time. dual core systems will give you a noticeable increase in performance by pipelining tasks queues to the two cpus.

Yes, that speeds up the execution of programs in general; but it cannot speed one particular app up in particular if it's not multi-threaded. The best it could do is run the app as if it were getting 100% CPU time. So, faster? Yes. As past as a multi-threaded version on a dual-core? Probably not.

[Silent Avenger]

I've ran all of them (16K to 32M) on my computer which is running a Pentium D (dual core) at 3.6 Ghz with 4 GB RAM. I was also running some programs at the same time so it's not completely accurate but I guess it's close enough. Oh yeah I was too lazy to cut out the window so I just took a screen shot of my whole desktop.