stress test basics

rowedogg · 26 Feb 2012 at 21:55

yo guys.

i recently built my comp this weekend, (2500k,8gig,p8z68-v) and its currently running on the stock settings just fine (i think).

Ive downloaded prime95, and IBT along with some monitors like realtemp and cpuid.

i could do with some basic how to's, so that when i overclock over the coming weeks, i can test its stability accurately. i searched google, but a lot of the info that comes up is from years ago(2003)!
any help appreciated.

jkkc313 · 26 Feb 2012 at 22:44

prime95 stress test is more than capable, just running it overnight, if no errors, then u pretty much have a stable system

rowedogg · 26 Feb 2012 at 22:48

just leave it on the blend test with changing settings?

panyan · 26 Feb 2012 at 23:05

id do the extreme intel burn test with maximum ram and see if it can handle it

Justintime · 26 Feb 2012 at 23:08

Quick and dirty - 8hrs prime and 50 ibt at max. Good enough for many, not me but it'll likely be stable.

rustynuts · 26 Feb 2012 at 23:09

As said Prime95 on blend leave it running over night and if its still going in the morning when you wake up, you know the system is pukka.

j.col · 27 Feb 2012 at 13:28

Justintime said:
Quick and dirty - 8hrs prime and 50 ibt at max. Good enough for many, not me but it'll likely be stable.

this, also i would do 50 passes of IBT, (not sure Justintime means 50 hours

)

rowedogg · 27 Feb 2012 at 14:56

50 passes on IBT sounds a lot? i could have thought that i remember reading somewhere about not going over 15? cant remember where i saw it tho

Justintime · 27 Feb 2012 at 17:45

j.col said:
this, also i would do 50 passes of IBT, (not sure Justintime means 50 hours )

lol 50 passes, though when i stress test (i do 150) it comes close.

rowedogg said:
50 passes on IBT sounds a lot? i could have thought that i remember reading somewhere about not going over 15? cant remember where i saw it tho

You likely read that from someone who dosen't have a clue. I don't have a clue either BUT i have experience with tens of cpus not being damaged whatesoever with 150+ runs of LINX/IBT etc..

rowedogg · 27 Feb 2012 at 18:51

what about the memory on IBT, its says standard, 1024mb available ram 6088mb. leave it on standard or increase?

j.col · 27 Feb 2012 at 21:16

maximum for a quick and dirty test, but the proper way is to use a custom amount of ram.
Have a look under the performance tab in task manager.
Then make a note of how much "Free" ram is listed.
enter the custom ram, just below this figure.
eg. if i have 2550mb of ram free, i will enter 2500mb.

This way you are using just free ram, and not also cached ram, so you should get higher gflops, and therfore a more stable test.

Thread here on how to run IBT properly and explanations.
http://forums.overclockers.co.uk/showthread.php?t=18297411

rowedogg · 29 Feb 2012 at 00:22

yo guys. so heres a screen of me after one hour prime95

[/URL] Uploaded with ImageShack.us[/IMG]

what you guys reckoning? the temp maxed out at 71* but only after 20 mins, then for the rest of the hour it was sitting around 60+.

i put the cpu voltage to 1.36, as that was the first one that allowed me to boot into windows, first i tryed 1.25, from clunk website but that was a no go, so prob could get it down a little.

the VID in coretemp says 1.39 should i be worried about this?

thanks

Justintime · 29 Feb 2012 at 17:59

rowedogg said:
y60+.

i put the cpu voltage to 1.36, as that was the first one that allowed me to boot into windows, first i tryed 1.25, from clunk website but that was a no go, so prob could get it down a little.

the VID in coretemp says 1.39 should i be worried about this?

thanks

If 1.36 was the first that allowed you into windows, it might not be stable as Windows loading is less demanding then when the cpu is under stress, VID is what the chip wants, but you're giving it 1.36 so no, no need to worry. Temps are ok.

rowedogg · 29 Feb 2012 at 19:37

it seems that it wouldnt let me in as i had turned off pll overvoltage as advised on a clunk.org. ive done some further testing today and seemed to have managed a stable 1.32vcore at 4.5gh. the temps were 73 max in IBT and 67max on prime95

1kiwi · 1 Mar 2012 at 20:39

Justintime said:
lol 50 passes, though when i stress test (i do 150) it comes close.

You likely read that from someone who dosen't have a clue. I don't have a clue either BUT i have experience with tens of cpus not being damaged whatesoever with 150+ runs of LINX/IBT etc..

Then you would know that IBT/Lynx is rubbish for Sandybridge. As above use Prime Blend for at least 4 hours.

Justintime · 2 Mar 2012 at 04:04

kiwi1 said:
Then you would know that IBT/Lynx is rubbish for Sandybridge. As above use Prime Blend for at least 4 hours.

Since when? IBT/LINX without AVX support and SP1 is pointless, when it has those things its as good as it ever was and gets things cooking.

gurusan · 2 Mar 2012 at 04:09

IBT is a great tool for SB. IMO essential, it's just that it needs to be used in conjunction with prime95 blend for an IMC test.

Justintime · 2 Mar 2012 at 04:53

Yea it seems people like to read one persons opinion and somehow take it for 'expert' if they post on xtreme etc..(lots of guys know their stuff there, not saying that they don't) then repost it as 'the rules' IBT is definetly a great tool for stability checking, prime blend as you say is also very good.
I've downclocked many overclocks that many would pass as 'stable' (i.e 8hrs prime 10 runs IBT) because they failed IBT at some point before my own 150 mark or failed prime within my 24hr min run. Leads me to believe that many of these 4.6/4.8 awesome overclocks ain't exactly 100% solid.