Superpi

Soldato
Joined
2 Oct 2004
Posts
4,362
Location
N.W London
Hi,

I just overclocked my amd 3500 newcastle and was wondering what type of test to run on superpi?

How does the prog work? Should I try and run 16k tests or 32mb tests?

Whats the setting to test a system in superpi once overclocked?
 
Superpi is a performance benchmark, though 32M is often used for testing stability as it takes the longest. All of the tests are equally valid, though 1M, 8M, and 32M are the normal ones that are run and compared with other systems.

Jokester
 
Run the 1m test and post you're time. There's a thread on Superpi times on here with a table of peoples time so you can compare yours to them.
 
is there a chart or something I can use for future tests?

I mean how did you work it out? Or does it come with experience :-)
 
Difficult, kind of comes when you see a lot of scores. I can't say it 100% accurate either as I'm currently out of the loop but a lot of peoples system running at about 2.7-2.8GHz were hitting 30s time, this might be down to a better RAM setup as you are running 4 x 256 sticks.
 
People sometimes use the 32M run as test of stability, which is ok as a rough guide to detect unstable system. 1Mb is used normally to bench other peoples scores and compare, usually as it's quite quick and easy, but the bigger ones are also used. You can also get versions of that show what fraction of a second results are at e.g. 29.029s just to show more accurate results.
 
I have my 4400+ X2 at 2.64 ghz and get a time of 33 seconds on 1M test, the CPU temp reaches 46c and chipset 41c but the PWM area goes to 60c, is this too hot? I have a fan directly over the PWM area trying to cool it also.
 
Its hard to judge superpi 1mb times as most people in the benchies have tweaked their system - clean installs, no background apps etc to lower their times. I dropped 2s from 29s to 27s on 1mb time by windows tweaking. Did same tweaks with dual core and it made hardly any difference so guess 2nd core was running background stuff
 
the best time iv had for a 1mb test is 36s, i think this program prefers faster memory than faster clock speeds?

if i close all the background tasks it makes it quicker :p
 
1mb test is an ok test for stabilty, Ive had a system overclock that would run BF2 for an hour but wouldnt complete the 1mb test so it must be pretty strict
 
Not sure geiger. A game like BF2 will stress your graphics a lot so the load on the CPU won't be as high, I don't think that is a good indication on CPU stability.

If you want to test a CPU for true stability it should be able to pass anything, like OCCT, Prime 24.14, SnM. I think SuperPi is good to get a quick estimate, like if it can't pass 8Mb then don't bother incrementing without upping the vcore a bit, or reduce the OC by a few hz and see if it passes.
 
Yea I wasnt saying BF2 was a good test just that even on a pc that appears to work fine, just a quick 40 second SuperPi test could show it wasnt stable really.

BF2 doesnt use the CPU much, its all about big textures and the GPU. Memory & disk speed would most likely help more then my overclock.

Prime doesnt show stability that well Ive realised, Ive had it prime for 24 hours then fall over in OCCT in a few minutes. S+M is a nightmare, only works at stock speed on mine
 
Back
Top Bottom