Realistic World Stress Test

Devrij · 10 Jan 2013 at 10:57

nkata said:
Agreed, prime95 and heaven benchmark concurrent is a test I often run.

Same here. I also find FC3 is a good real world test as it's quite cpu dependent and stresses the gpu very well too. It's also more fun than watching a benchmark run.

If your temps are too hot in prime, then your cooler sucks, or your case needs better cooling. End of. There are progs that run hotter than prime, so if it's too hot in prime then you want to keep your eye on those temps.

mmj_uk · 10 Jan 2013 at 21:27

It's not even a heatsink problem with Ivy Bridge, most of the heat from the core is getting trapped on the core due to their decision to use cheap thermal paste between it and heatspreader, this paste is hindering the heat dissipation process which is a major issue when overclocking and especially when overvolting.

As for Prime95 it's the type of application that computers were designed for and the sort of thing the scientific community still uses them for (heavy number crunching), if Intel shipped processors which weren't stable in applications such as that they'd lose the majority of their contracts.

It's like when Ferrari test their engine designs for reliability they do it at the top speed the engine is designed for not what they think their customers will drive the car at day to day, we consumers don't have Intel's complex tools for monitoring processor die performance so we have to rely on the most stressful applications available to us to do our overclock stability testing - Prime95 has always been among the best.

zepzeppelin · 11 Jan 2013 at 01:20

doyll said:
What did 8 Pack say that in context with? Could you post a link to original? There are many real world applications that work systems as hard as Prime does. Rendering, encoding and converting graphics for example.

http://forums.overclockers.co.uk/showthread.php?t=18476308 post #14

doyll · 11 Jan 2013 at 01:34

zepzeppelin said:
http://forums.overclockers.co.uk/showthread.php?t=18476308 post #14

Post #14

8 Pack said:
30 mins is enough prime. Then just use your computer as normal, thats the best way of stability testing.

Not the post you quoted below.

Originally Posted by 8 Pack View Post
Remember Prime is synthetic you wont hit these temps in everyday use anyway unless your rig folds or similar 24/7. Using your PC is a better guide than Prime to its ability to be stable for what you need it for.

Loxa · 11 Jan 2013 at 01:52

zepzeppelin said:
Recently Stress testing with prime95 has, (in not so many words) become a real pain in the ass for Ivy Bridge users, especially with the recent version 27.9 which kick the hell out of a 3570k and makes it near impossible to get stable with a x45 Ratio. I've had temps reaching 90+ in less than an hour, and unable to get an hour out of Prime unless the vcores 1.304~1.312 or even 1.320v. That's way to high, and the developers of Prime need to relax or it'll be totally impossible before too long.

So!, I bought myself a copy of AIDA64 Extreme Edition, and I can get x44 with little fuss, and presumably higher. I have read on the asus website that AIDA is their recommendation for Ivy Bridge, as prime95 is classed as being too synthetic in real world stress testing, and I agree. This is what my volts and temps were after just over 2hrs on AIDA64:

http://img803.imageshack.us/img803/706/45518501.jpg

The difference between HW monitor temps and Realtemp is kinda big

doyll · 11 Jan 2013 at 01:59

51-62-63-55c HW Monitor
57-61-63-55c Real Temp

Loxa · 11 Jan 2013 at 02:04

doyll said:
51-62-63-55c HW Monitor
57-61-63-55c Real Temp

57 68 68 64 HW
68 76 75 71 Realtemp

You go off max temps not there current value

JonJ678 · 11 Jan 2013 at 09:32

Well. Prime95 does a load of maths and tells you if your system is getting the maths wrong. So if you can't pass prime95, your computer gets maths wrong.

How much you care about that depends on how much you care about the computer getting calculations wrong. If you play games & don't give a **** about it periodically corrupting your data, then go for it. A pixel in the wrong place or a saved game nuked ultimately doesn't matter.

If you're using your computer for things where the answer matters - such as when doing maths - you want it to be stable. Properly stable, as in gets the numbers right. Not "stable" as in "doesn't fall over often if treated gently".

Consequently I agree with PCZ here. I care when my computer gets its sums wrong. Right now, my box is not stable - I think the RAM is failing. That's annoying, but I can still post in threads like this while it fails IBT. Until it's passing IBT, I won't trust it with FEA and probably wont trust it with CFD - even if it doesn't crash, it's apt to get the numbers wrong.

ubersonic · 11 Jan 2013 at 10:31

zepzeppelin said:
Recently Stress testing with prime95 has, (in not so many words) become a real pain in the ass for Ivy Bridge users, especially with the recent version 27.9 which kick the hell out of a 3570k and makes it near impossible to get stable with a x45 Ratio. I've had temps reaching 90+ in less than an hour, and unable to get an hour out of Prime unless the vcores 1.304~1.312 or even 1.320v. That's way to high, and the developers of Prime need to relax or it'll be totally impossible before too long.

So!, I bought myself a copy of AIDA64 Extreme Edition, and I can get x44 with little fuss, and presumably higher. I have read on the asus website that AIDA is their recommendation for Ivy Bridge, as prime95 is classed as being too synthetic in real world stress testing, and I agree. This is what my volts and temps were after just over 2hrs on AIDA64:

http://img803.imageshack.us/img803/706/45518501.jpg

So to sum up, you couldn't pass the most popular stress test so you used a less strenuous one instead?

Prime is real world software not synthetic however things have to be kept in context, to a person with no computer all software is synthetic, to a person who only uses word and outlook everything above the Windows Experience Index is synthetic, Prime is a real world application however people who have no use for it run it as a synthetic stress test.

At the end of the day stability is subjective, some people demand absolute stability whereas others (like you) are happy with a system that is not completely stable but is stable enough for their means.

klepp0906 · 26 Jul 2014 at 19:23

zepzeppelin said:
Recently Stress testing with prime95 has, (in not so many words) become a real pain in the ass for Ivy Bridge users, especially with the recent version 27.9 which kick the hell out of a 3570k and makes it near impossible to get stable with a x45 Ratio. I've had temps reaching 90+ in less than an hour, and unable to get an hour out of Prime unless the vcores 1.304~1.312 or even 1.320v. That's way to high, and the developers of Prime need to relax or it'll be totally impossible before too long.

So!, I bought myself a copy of AIDA64 Extreme Edition, and I can get x44 with little fuss, and presumably higher. I have read on the asus website that AIDA is their recommendation for Ivy Bridge, as prime95 is classed as being too synthetic in real world stress testing, and I agree. This is what my volts and temps were after just over 2hrs on AIDA64:

http://img803.imageshack.us/img803/706/45518501.jpg

as of 28.5 i stopped using it. If i go from 24 hours 27.9 stable to 24 minutes 28.5 stable.. im done. Its archaic and completely writen off as of that point.

It was already so stressful that you'll never see loads or temps like that no matter what you use your pc for. Now they bump it up another level which i was left with.. either lower my overclock (which has ran untouched for over a year with 0 crashes or lockups in anything) or move on and i moved on.

what happens when they bump it up agian, all you prime fanboys gonna lower your overclocks i assume? lol makes me giggle a bit.

anyhow, IBT for a quick n dirty/max temp check. Hyper PI to dial in memory timings. Aida or preferably x264 for a real world stability test that doesnt insta shoot you up to 100c.