Prime95 Vs SuperPI

saying that though i found prime passed for over 12 hours before then fell over on 32m SuperPi, not sure what that says exactly but i would/do use both
 
tbh Prime95 for seeing how stable it is, SuperPi for showing off how fast it is.
Although personally If my overclock is stable enough to last SuperPi 4MB I'm happy :)
 
wdEvA said:
hmm so will it stil be considered stable, if u can pass 1 and fail the other?

Everyone has their own idea of stable. Some people think a few hours in Prime is enough, some people - such as me - wont settle for anything but rock solid stability which is why I do over 24 hours of Prime testing and then move onto Folding@Home.

F@H runs at 100% all the time and is very very sensitive to errors. Basically if you can go a week or more of running F@H with no issues then I would say your system is stable. I've had F@H running 24/7 for the last 2 months or so - no issues yet :D

SiriusB
 
Prime95 for cpu and to some extent RAM - 24hrs at least for me
Memtest86+ - RAM - 24hrs for me
3Dmark01 - 3D and general stability - looped 24hrs


No rig has given me problems hardware related that passed that.
 
Justintime said:
Prime95 for cpu and to some extent RAM - 24hrs at least for me
Memtest86+ - RAM - 24hrs for me
3Dmark01 - 3D and general stability - looped 24hrs


No rig has given me problems hardware related that passed that.

3 days of testing? what a waste of time... You could at least run F@H or something... ;)
 
I've got time to waste when clients and money are involved, saves me from unnecessary callouts, 99% of the time its their fault and a software error, not the hardware. Also i rather 3 days of torture and a troublefree rig forever after.
 
hmm okie, i would like prime for 24hours too.. sidetrack abit..
if my ram has errors in memtest.. does it mean it has reach its max? or could there be other issues?
 
Could be many things, too much FSB for it to handle, too tight timings, not enough VDIMM, Incompatiable with board, Bad RAM, Bad motherboard etc.. :D But if you overclock and its fine up to a certain point/speed/voltage/latency then starts erroring with anything higher, then yes, you might've reached the peak.
 
Why do people run say, 12 hours of prime to see if somethings stable? Surely just playing say... CoD2 for 3 hours or something will show to see if it crashes or not?
 
UKTopGun said:
Why do people run say, 12 hours of prime to see if somethings stable? Surely just playing say... CoD2 for 3 hours or something will show to see if it crashes or not?

Your cpu may run that game for days but do something else and fall over, prime95 basically computes numbers that are known already and sets your cpu computing them and compares your computed data to the known value, any variations usually show somethings not right. So while you may think your pc is stable you might do something to trigger that instability, also funny things may happen like lost data, corrupt OS, random reboots, lockups etc.. that will manifest itself over time in a unstable rig even though on the face of it it might seem stable. Prime does a few things that overclocking needs: checks accuracy and heats things up, stresses components and drains a bit of juice since its more or less a full load on the system, so weak points may show up. Superpi is a rather short test, so instability caused by heat or undervoltage etc.. may not manifest itself in the time taken to complete it.
 
Last edited:
Justintime said:
Your cpu may run that game for days but do something else and fall over, prime95 basically computes numbers that are known already and sets your cpu computing them and compares your computed data to the known value, any variations usually show somethings not right. So while you may think your pc is stable you might do something to trigger that instability, also funny things may happen like lost data, corrupt OS, random reboots, lockups etc.. that will manifest itself over time in a unstable rig even though on the face of it it might seem stable. Prime does a few things that overclocking needs: checks accuracy and heats things up, stresses components and drains a bit of juice since its more or less a full load on the system, so weak points may show up. Superpi is a rather short test, so instability caused by heat or undervoltage etc.. may not manifest itself in the time taken to complete it.

thanks for the detailed explanation =)

how about like what Úlfhednar had mentioned? SP2004? how's it compared to prime?
 
wdEvA said:
SP2004? how's it compared to prime?
SP2004 "is" Prime95 v24.14 iirc. With a front end GUI, SP2004-orthos is the multicore version.

One of the main problems with a cpu that looks stable and isn't is that as files are written to disk, errors get written, these build up and sooner or latest something important like the MFT gets fubared.
 
Last edited:
UKTopGun said:
Why do people run say, 12 hours of prime to see if somethings stable? Surely just playing say... CoD2 for 3 hours or something will show to see if it crashes or not?
No, I just ran Quake 4 for a laugh and the cpu didn't even get warm, only a little over the idle temperature. Its not working that hard, try compressing a huge zip file, or encoding. That's the sort of thing that pushes it, excluding the torture programs. S&M a good one for a quick test.
 
Last edited:
Also, if you have a multicore CPU there is a good chance your game will be running on just the one core. From experience I can tell you that more often than not one core is generally weaker than the other, meaning under stress it will be the first to fail.

Now, if your game is running on the stronger of the two cores, an instability or error may not show itself. No good when the very next program you run could bug out on the weak core and take something out with it!

SiriusB
 
wdEvA said:
how about like what Úlfhednar had mentioned? SP2004? how's it compared to prime?
SP2004 is Prime95 but with a full GUI with lots of handy things on it, as said there is a dual-core version of SP2004 but you can also just run two instances of SP2004.
 
Back
Top Bottom