Intel Burn Test Utility

Have been using this for a while. Absolutely the best tool for stability and thermal testing.
 
wow, so it seems like it stresses the CPU even more so than prime, awesome for a test that is short.

SilverSilver, can you also run burn test and then prime (within the same time frame) and tell us the temp difference if there is any? thanks.
 
Burn Test versus Prime results in some 20C difference for me. I don't get above 54C in Prime or Orthos. With Burn Test, I was getting 74ish last time I ran it.

What I like about it, is no waiting several hours to test absolute stability with Orthos or Prime. I do a 10 test run with Intel Burn test and it tells me in minutes whether or not I am stable.
 
I really didnt see my temps go that much higher using Burn Test over Prime95. Maybe 5c. Am I using this right? I mean, it's not a complicated piece of software to use either. Selected to use all memory and run 10 tests. Highest temp I saw was 66c. Under prime i'll see 55-60c depending on room temp.
 
whoa, those high temps makes me nervous, I agree it is such an awesome stability test that stresses your core harder for under 10 mins.

Just worries me how the temps are so high, prime runs it at 100%, what can this burn test be doing to push it even further than this?
 
dennisjai said:
whoa, those high temps makes me nervous, I agree it is such an awesome stability test that stresses your core harder for under 10 mins.

Just worries me how the temps are so high, prime runs it at 100%, what can this burn test be doing to push it even further than this?
Click to expand...

It explains in that thread what this does different. It's not that it stresses the CPU "more" -- it stresses it in a more efficient and consistent manner. The technical methods behind this are lost to me though.
 
dennisjai said:
whoa, those high temps makes me nervous, I agree it is such an awesome stability test that stresses your core harder for under 10 mins.

Just worries me how the temps are so high, prime runs it at 100%, what can this burn test be doing to push it even further than this?
Click to expand...

Thing to keep in mind is that the CPU running at 100% does not mean its working its getting completely utilized. It just means that its doing what it can with the current code as fast as possible. And more often than not, that code is not completely optimized for the platform, for example you can write an app that does X=X+1 forever and it can max out the CPU at "100%", however its certainly not making full use of the CPU. Prime95 is written to run on a variety of CPUs, this burn test was designed with Intel libraries, from Intel, so you can presume that this really is getting everything possible from modern Intel CPUs at least as far as Intel engineers are concerned. This is finally a true torture test for a lot of people who are concerned with stability. I tested my CPU for ~5 days straight at 3.4 using Prime95, OCCT, and a couple other things, deemed it stable, then was running it at 3.3GHz for some extra headroom. This burn test bombs out on 3.4GHz before getting through 10 passes, and causes problems at 3.3GHz unless I up the voltage a lot (Prime will run forever @ 3.3GHz, stock volts).

Ive passed this around for a few friends OCing various setups. Almost all of them who thought they were really pushing the limits had this blow up in their face. Also one guy refuses to run this because he "doesn't trust the app" when really he just doesn't want anything to show him that his precious OC might not be stable.

For people that want to really put their system to the test, run this on a 64 bit OS, max RAM allocation, 50 passes. That seems to be a much more significant test compared to Prime95 for any length of time, though both have their places. Also the best way to determine stability in an OC is to use a variety of tool such as this, Prime95, Memtest, Furmark...
 
This is what Intel uses to test their CPUs in the factory (it's a repackaged version of Linpack), so it's definitely the best tool to test your stability.
 
Whoa...

I think I'll stick with Orthos tyvm :p Would like a little more pride in my temps! Well, she's been rock solid for me.
 
Haha, I blue screened 10 seconds in... Dont care though, i have NEVER had a problem in 6 months under normal conditions. That being said, the point of this is again?

btw, my Q6600 is at 9x334 @ 1.29375v in the BIOS. I know its cause my vcore is really low, but like i said, i have NEVER crashed under normal circumstances.
 
Zero82z said:
If your system can survive this, it can survive pretty much any type of load. That's the point.
Click to expand...

So in my situation, the only thing that crashes my computer is that app... so if i just dont run it, im fine. :D easy enough...
 
criccio said:
So in my situation, the only thing that crashes my computer is that app... so if i just dont run it, im fine. :D easy enough...
Click to expand...

Realistically, this will stress your system much more than anything else you might run, so it isn't that big of a deal if it fails this app. That said, for people looking for absolute stability, this is a much better tool than Prime.
 
I have been using this test for a few days now and it has been nothing but great. I no longer have to prime all day, all it takes is a few minutes and im set. Oh and under water (pa120.3, fuzion) my e8600 reaches 67c at 4.2ghz w/ 1.26 vcore.
 
It definately beats having to wait hours for Prime when you're getting down to those last, few changes in your overclock. It sucked having to wait a couple hours because you dropped vCore by .0125 volts.
 
Think about your electricity bills if you run prime all the time to get stability. In real life your CPU will probably never get to this intense load, but it is the hard reality. My E8300 at 3.7ghz can render animations at full load for 100+ hours with no issues but this utlity says it is only stable at 3.6ghz. But hack I will risk 3.7ghz.

But ....., it is the best tool to determine the lowest voltages you can use in minutes! Run it around 5 times and it takes 9 mins or so. You can determine your lowest Vcore and get on the road within an hour while prime will take a few days.

I cannot use the 'ful memory' option as it will crash out. I can only assign the max of 2125mb out of the 4 gb I have. Anyone facing the same thing?
 
I get about 12C difference between this util vs P95 and Orthos... can be scary at first... watch those temps the first time you run this util.

kiajoon,
I have 4GB in my system and have no problem running at max. I'm using x64 Vista. I think the util recommends you use it in a 64-bit environment.
 
FINALLY. Exactly what we've been looking for all these years. This is the perfect stability test tool. Going to run it tonight on my rig to see if it's really as stable as it's shown to be throughout 2 years of usage.
 
I really have to laugh -- the ultimate stability tool comes out and it tests overclocks a little "too well" and suddenly, failing OC's become "stable enough." I seem to remember the same arguments when overclocks worked fine 98% of the time, and failed some newfangled app like Orthos or OCCT -- the same things being said.

True, your processor will probably be rock solid 99% of the time even if you fail this test using a very high overclock, that is otherwise reported as stable by other apps. However, I hope these guys don't have "stable" next to their overclock in their sigs...
 
Uh oh.

A problem...

I ran this on my work rig, and now, whenever the CPU gets loaded, it squeals slightly. :(

Edit: NM. It's a whiny fan. Probably didn't like the massive load this program put on the rig. :D
 
nruri said:
I have 4GB in my system and have no problem running at max. I'm using x64 Vista. I think the util recommends you use it in a 64-bit environment.
Click to expand...

Well I guess this is the max memory Windows will allocate within a 32 bit environment. But 2gb of memory usage is good enough to stress out any system anyway I suppose.
 
SilverSliver said:
I really have to laugh -- the ultimate stability tool comes out and it tests overclocks a little "too well" and suddenly, failing OC's become "stable enough." I seem to remember the same arguments when overclocks worked fine 98% of the time, and failed some newfangled app like Orthos or OCCT -- the same things being said.

True, your processor will probably be rock solid 99% of the time even if you fail this test using a very high overclock, that is otherwise reported as stable by other apps. However, I hope these guys don't have "stable" next to their overclock in their sigs...
Click to expand...

Eehhh, define stable. Nothing is ever "stable", it is only as stable as you test it to be. How many hours of Prime equals stable? How many passes of this test equals stable? It's a personal decision because the only person using your computer is you. If you want "stable" then find out what the JPL uses to test chips they send to Mars - and even those fail. Bottom line, stable is what you want it to be. I'm not running any mission critical apps, so as long as my system doesn't crash playing whatever game I'm playing, or lock up re-encoding a video or whatever, then for my purposes it is stable.
 
68c, stock cooling,, stock volts, 3ghz lol. i need an aftermarket cooler
 
Forceman said:
Eehhh, define stable. Nothing is ever "stable", it is only as stable as you test it to be. How many hours of Prime equals stable? How many passes of this test equals stable? It's a personal decision because the only person using your computer is you. If you want "stable" then find out what the JPL uses to test chips they send to Mars - and even those fail. Bottom line, stable is what you want it to be. I'm not running any mission critical apps, so as long as my system doesn't crash playing whatever game I'm playing, or lock up re-encoding a video or whatever, then for my purposes it is stable.
Click to expand...

very true that people oc to their own comfort level. in your situation you are getting an extra ~2% speed for running a setup past its known failure point that's ok for you, but not a lot of other people. usually people are stress testing to make sure that the system does not have a problem down the road when they are doing something important. like you certainly do not want to run into an issue when copying files to a backup drive or something. sure that failure might only happen on the 4th friday of the month when a blind squirell finds a nut during a lunar eclipse, however if a known chance of fail is present, that's too much for a lot of people.
 
ocellaris said:
sure that failure might only happen on the 4th friday of the month when a blind squirell finds a nut during a lunar eclipse, however if a known chance of fail is present, that's too much for a lot of people.
Click to expand...

QFT. Stability is not relative, only people's standards of stability.
 
ocellaris said:
very true that people oc to their own comfort level. in your situation you are getting an extra ~2% speed for running a setup past its known failure point that's ok for you, but not a lot of other people. usually people are stress testing to make sure that the system does not have a problem down the road when they are doing something important. like you certainly do not want to run into an issue when copying files to a backup drive or something. sure that failure might only happen on the 4th friday of the month when a blind squirell finds a nut during a lunar eclipse, however if a known chance of fail is present, that's too much for a lot of people.
Click to expand...

My point is that you don't know what it's known failure point is - because you can't. You can't test every possible part of the CPU, in every possible combination of operations, to find out if there is one particular instance where it will fail. So you run Prime for 24 hours and you call it stable. Then a new program comes out that fails - so now, what, your system isn't stable any more? So you downclock until the new program doesn't fail any longer - is it stable now? What if another new program comes out next week and it fails at that one? How far down the rabbit hole do you want to go?

If you want stable - real stable - then you shouldn't be overclocking, and you should be running ECC RAM on a server board using mission critical tested server components. Otherwise, you get what you pay for.
 
I know. Where do you call the line? I've just seen a lot of posts in the past saying if you arn't Prime stable for 6 months or some absurd number, you aren't stable. I understand the premise, but generally I call it 6 hours Prime, or 15-20 runs of Linpack.
 
In the end, if my computer "has yet to crash" with my overclock, and has been this way for almost a year, why do i need to care about anything else? I can achieve lower temps by running my vCore lower then then what would be considered "stable" by these stressing apps, and yet still never crash... So i have yet to see the upside.

Just FYI, my CPU is at 334x9 @ 1.29375v in the BIOS, i idle at roughly 26/27 degrees, and the ONLY time i have crashed this system is by running this app.
 
Forceman said:
Then a new program comes out that fails - so now, what, your system isn't stable any more? So you downclock until the new program doesn't fail any longer - is it stable now? What if another new program comes out next week and it fails at that one? How far down the rabbit hole do you want to go?
Click to expand...

I think a lot of people people want to test to the current limits of software stability testing. Software changes and normally some newer app comes along that puts a better load on a newer CPU compared to a previous test. SETI used to the "the" test for a stable OC about a decade ago, then people gradually mixed in some 3DMark99 (most of the early 3DMark benchmarks were excellent CPU checks if you ran them at low resolutions in a loop), and eventually Prime95 became the standard because it was simply a better and more effective test at weeding out bad OCs. These same arguments were had throughout that time, "Who cares about Prime95 small FFTs if it can run SETI forever... right?!?" Software apps and games gradually become more optimized over time as well as that PC that could run SETI forever but crashed on P95 then starts crashing on the new game/app/decoder too. So here we have, by most accounts, a better test and at this point people can decide if they want to go with a more strenuous test or just call what they have "good enough". Gradually I think IntelBurnTest (or sometimes similar based on Linpack) will become the standard test app.

If you want stable - real stable - then you shouldn't be overclocking, and you should be running ECC RAM on a server board using mission critical tested server components. Otherwise, you get what you pay for.
Click to expand...

We do this at work, got a bunch of ridiculously overprices P4, Core2Duo and Core2Quad Extreme workstations sitting around simply because they were the fastest available PCs at the time and we needed them for development. But at the same time, I like getting a little extra out of the home PC and knowing its stable to the highest limits I can test it at.
 
Arcygenical said:
It crashes, instantly, even at stock...

Bah.
Click to expand...

Clear out your internet cache, it will fix it.









jk! Just pretending to be A+ certified :D


To be serious though, it could just be the program not working with something, hence why some of these benchmark programs are BS due to incompatibilities somewhere else in the system(typically software related)
 
In my hot ass dorm room where the room temperature is over 31C, running this app on my sig rig caused my two hottest cores to skyrocket to 82C max load! :eek: Naturally it scared the shit out of me, but after 10 runs even at those ridiculous temps, the conclusion was that my OC was deemed stable. :D I definitely like this program a lot better than waiting around for hours on end to test my OC with Prime 95 - quick and easy, just the way I like it! ;)
 
I let it run last night and use all maximum memory at maximum stress level.

Yes, it scared the crap out of me. My temps, which do not normally exceed 57C at the MOST skyrocketed to almost 70C :eek:

Well, I let it run for 20 passes, and that's enough. My OC is 100% stable. :D
 
Forceman said:
What if another new program comes out next week and it fails at that one? How far down the rabbit hole do you want to go?
Click to expand...

As far as is necessary to know that my system is stable. It's not as if it's an impossible target. I don't think it's possible to do that much better than linpack.

Forceman said:
If you want stable - real stable - then you shouldn't be overclocking
Click to expand...

Thanks to rigorous tests like this one, I disagree. linpack stresses my cpu like I never will.

I use computers at home for both work and fun and I want to know that I'm not going to corrupt files here and there or have any problems while I'm working at all. Compilations must be correct and all those sorts of things you think about when developing. I don't want to be pulling my hair out over a bug only to find out the compile erred because of my OC.

Forceman said:
you should be running ECC RAM on a server board using mission critical tested server components.
Click to expand...

My old workstation has ECC memory on a solid board. ;)
 
You must log in or register to reply here.
Back
Top