Intel Burn Test Utility

Speaking of how far down the rabbit hole... Why even bother saying that? This isn't a "better" program in terms of what I can see you are saying, it's better for the fact that it's quicker. If I run my PC with P95 I don't leave it running for hours on end, that for me is wasting my valueable time! If something can make it crash in 10 minutes as opposed to 10 hours, big thumbs up from me. My PC was stable at 3.2GHz with P95 but crashed with this. I let P95 run for about an hour though, if it crashed on this, fair enough, I went back to 3GHz and am dialing that in first.

For me, conveniance is what we are going for, speed is what we desire, the overclock does that, this program does that.
 
Y'know, another good benefit of this is with something like Arctic Silver 5 which requires running the CPU at a high load and temp to get it to cure. If it makes CPUs run as hot as reported, it could help shorten curing times alot while also testing stability.
 
Got any comparisons to Core Damage?

http://damage.vigilantesoftware.com/

Does it get your cpu hotter than linpack? Got any more info on this app? That page is short on details. Where did the program come from?

IntelBurnTest uses linpack which comes from Intel. Who's behind vigilante software and what qualifications do they have to be able to create a better stress testing tool than Intel?

I haven't used CoreDamage so I'm interested in comparisons too.
 
CoreDamage doesn't sound like something I would run. Ever. Simply due to the name. I compare it to an auto manufacturer producing a car called "Deathtrap". Would you drive a "Deathtrap"? No. So why run a program named "CoreDamage" from some unreputable company?
 
All I saw was my the realtemp shoot above 70c and then my motherboard shut off my pc. My fans didn't even get a chance to turn on to maximum! (my fans are mobo regulated). This isn;t gonna make me change my oc though: i've been gaming and using the pc non stop for a year and have had 0 problems.
 
Stock Q9450 @ 1.238V (laptop) and 65C was my highest on normal fan speed. At least I know Im not overheating. :)

Idle temp is 47C
 
This program crashes everytime I use maximum stress option.

WIth no error checking, I get:

CPU frequency: 2.400 GHz
Number of CPUs: 4
Number of threads: 4
Parameters are set to:

Number of tests : 1
Number of equations to solve (problem size) : 17241
Leading dimension of array : 17241
Number of trials to run : 5
Data alignment value (in Kbytes) : 4

No runs is allowed for LinData\x86\linpack32.exe: not enough memory


With error checking, I get a 0xc0000005, which is some sort of access violation error.



This computer is stable thru OCCT and prime95. Is this a bug?


Q6600 B3 2.4ghz @ 1.12v
4gb PC6400 Patriot 5-5-5-12 at auto
Gigabyte GA-EP35-DS3R
 
This program crashes everytime I use maximum stress option.

WIth no error checking, I get:

CPU frequency: 2.400 GHz
Number of CPUs: 4
Number of threads: 4
Parameters are set to:

Number of tests : 1
Number of equations to solve (problem size) : 17241
Leading dimension of array : 17241
Number of trials to run : 5
Data alignment value (in Kbytes) : 4

No runs is allowed for LinData\x86\linpack32.exe: not enough memory


With error checking, I get a 0xc0000005, which is some sort of access violation error.



This computer is stable thru OCCT and prime95. Is this a bug?



Q6600 B3 2.4ghz @ 1.12v
4gb PC6400 Patriot 5-5-5-12 at auto
Gigabyte GA-EP35-DS3R

If you aren't running 64 bit you need to manually set how much memory the program can use
With my 4gb I had to set it to 2200mb

For those curious:
My max temps - Orthos: 54c, Core Damage: 58c, linpack: 60c
 
just curious here guys, but....

I'm running this on my P4 northwood, and... when I go to the task manager, only 1 thread is fully loaded (total load is 50%)

This is like when you foolishly run Prime95 on only one thread (or one core, in the case you of people with these fancy dual/quad cpus lol :p )

I mean, it knows I have hyperthreading:

Code:
----------------------------------------------------
Executing Intel(R) Linpack 32-bit mode...
----------------------------------------------------
Intel(R) LINPACK data

Current date/time: Fri Sep 05 23:27:05 2008

CPU frequency:    3.003 GHz
Number of CPUs: 2
Number of threads: 2
Parameters are set to:

Number of tests                             : 1
Number of equations to solve (problem size) : 8461
Leading dimension of array                  : 8461
Number of trials to run                     : 5
Data alignment value (in Kbytes)            : 4
So, are you guys running multiple instances of this program? Or what?? Cause this shit is not getting my P4 as hot as 2 instances of Prime95 - that's for sure. It's not loading both threads...

edit: yup, gotta run 2 instances of it! I hope you guys are doing that, cause if not, you're probably not stressing your CPUs "enough."
 
With a Q6600, this program pegs all 4 cores at 100% running one instance. Perhaps the hyperthreading is throwing it off?

My Q6600 at 3.4 Ghz with a Xig S1283 hits about 66, while P95 hits 60 @ 23C ambient. So, it's not quite the 10+C that others boast. I believe that the temps greatly depend on the heatsink and how well it scales.
 
With a Q6600, this program pegs all 4 cores at 100% running one instance. Perhaps the hyperthreading is throwing it off?

My Q6600 at 3.4 Ghz with a Xig S1283 hits about 66, while P95 hits 60 @ 23C ambient. So, it's not quite the 10+C that others boast. I believe that the temps greatly depend on the heatsink and how well it scales.

That's about the same results I get with my Q6600.

And whoever said you need to set the memory manually if not running a 64bit OS, well, that's just hogwash. I set it to use max memory and never had a problem.
 
My q6600 @ 1.275 and 3.0ghz is hitting mid 50's with my Xigmatek 1283 and Lian K7B (2 x 120 intake, 1 x 120 exhaust)

Heh... I seem to suck at overclocking past 3.0 though :(

*Prime95 sits around 50 under 100% load.

So it is a bit more stressful, I guess mah cooling is just too uber :eek:
 
With a Q6600, this program pegs all 4 cores at 100% running one instance. Perhaps the hyperthreading is throwing it off?

Good to hear. Cool- that's good that it automatically tests all cores. I'll definitely use this program when I upgrade... talk about a time saver over prime95... lol

My P4 probably made the program confused.

Time saving is probably the only reason I would use this, in practicality...
 
i used 1/2 stress test and it passed, that good enough? Used only 768 ram or something. And i have 2gigs of ram. Earlier I overclocked my e2160 from 2.4ghz at 1.38v (system auto voltage), to 3ghz at 1.36-1.38v (system auto voltage), and wanted to test it :eek: Previously before seeing this thread i did run prime95 for 45 minutes with no errors. Seems like everything is stable.
 
All I saw was my the realtemp shoot above 70c and then my motherboard shut off my pc. My fans didn't even get a chance to turn on to maximum! (my fans are mobo regulated). This isn;t gonna make me change my oc though: i've been gaming and using the pc non stop for a year and have had 0 problems.

I am not surprised, I know a lot of people running Q6600 @ 3.6 have backed off from their original "stable" OC as time goes on. These things can take a lot of abuse and take a good while to give out when you start pushing them.

Heck when summer came some Q6600 owners backed down from 3.6 when they saw what CPU temps looks like just because the room went up 5C in temps compared to the lower Spring ambient temps :p
 
i used 1/2 stress test and it passed, that good enough? Used only 768 ram or something. And i have 2gigs of ram. Earlier I overclocked my e2160 from 2.4ghz at 1.38v (system auto voltage), to 3ghz at 1.36-1.38v (system auto voltage), and wanted to test it :eek: Previously before seeing this thread i did run prime95 for 45 minutes with no errors. Seems like everything is stable.

Just do the full test and like 20 passes. Prime95 for 45 minutes really isn't a lot of testing, normally people stretch that out for 8+ hours, it can take Prime95 a long time to find a problem.
 
Does it get your cpu hotter than linpack? Got any more info on this app? That page is short on details. Where did the program come from?

IntelBurnTest uses linpack which comes from Intel. Who's behind vigilante software and what qualifications do they have to be able to create a better stress testing tool than Intel?

I haven't used CoreDamage so I'm interested in comparisons too.

http://www.ocforums.com/showthread.php?t=546405

I'm using inline __asm code in my C++ programs. Visual Studio 2003.

The Core 2 Duo can handle two SSE instructions per clock cycle along with other ALU instructions. The trick is to write a tight loop and interlace ALU and SSE instructions like this:

Code:

addps xmm6, MathData + 48
mulps xmm1, MathData + 64
xor esi, esi
fld qword ptr [edi+10h]
fstp qword ptr [edi+20h]


The instructions need to be aligned on 16-byte boundries and should be using memory operands that hit cache memory. As few bubbles in the pipeline as possible keeps the processor hot.
 
Just do the full test and like 20 passes. Prime95 for 45 minutes really isn't a lot of testing, normally people stretch that out for 8+ hours, it can take Prime95 a long time to find a problem.

Well my 3ghz overclock failed. Pc randomly shutdown twice in one day. Like the shutdown process was started. I guess i didn't stress test enough. But does this mean that it was just the overclock that failed? I know when it restarted the last time, when i tried starting it up right after, it started, but didn't show any screen, and than it started, and the bios was back at the default settings. I know i read somewhere in the manual that the mobo resets bios settings when an overclock fails.

Another thing is, hwmonitor was showing me the volts for the cpu, and it would jump from 1.36 to 1.38 sometimes and vice versa. Maybe that causes some problems? I have the voltage set to auto for everything atm. Maybe I could manually set the voltage to 1.38 or 1.39 and run the stress test again. Although are you sure i'd have to do 20 passes? I thought only 5 was necessary.
 
Maybe I could manually set the voltage to 1.38 or 1.39 and run the stress test again.

Normally its a good idea to set voltage manually.

Although are you sure i'd have to do 20 passes? I thought only 5 was necessary.

20 passes really does not take that long, likely a good idea to do 20 instead of 5 if you want to get a larger sample size. 5 tests is good for a quick check.
 
wow so, my computer is still restarting sometimes. Even when it's back at 1.8ghz default speeds. Maybe i shouldn't have tried overclocking at all, or atleast beyond 2.4ghz (that's where i first overclocked).
 
wow so, my computer is still restarting sometimes. Even when it's back at 1.8ghz default speeds. Maybe i shouldn't have tried overclocking at all, or atleast beyond 2.4ghz (that's where i first overclocked).

When is it restarting?
 
Only did it once today though and it's completely random, pc is usually idle. I have a earthwatts 430w powersupply Someone told me to just load fail safe defaults for bios and i also read that you could disable the automatically restart option, that's in the startup and recovery options. Changing bios settings might be a good idea, since i do have a lot of the safety settings off, like the ones that lower the voltage and the multiplier etc. But i have a xigmatech 1283 cooler so temperatures never really get that hot.
 
I like this utility!

No more waiting around for hours for 2 instances of P95 to stress the core when OC'ing.

This baby kicks both of my cores right in the nuts from the start! I have about a +12C jump in core temps over P95, so something is happening.

I'm also pretty sure it's accurate - I've spent days testing and re-testing my OC with P95 and this utility has confirmed it in a couple of hours. Highly recommended.:D
 
This utility sounds awesome from what I've read and I will probably try it out on my q6600 3.6 1.3875 vcore OC; however, I'm in the same boat as those who say as long as the computer does what I need it to do efficiently and without any crashing, it's fine. I look at it this way - My current OC has been rock solid stable and the only time I've ever had BSOD's are when I muck around with overclocking the ram without knowing what I'm actually doing. If I run this program and it turns out I blue screen instantly, I could do one of two things. I could freak out and investigate the source of instability, which might end up being something like a large lack of voltage. Or I could leave everything as it is and enjoy longer component life from not upping the voltage as well as a perfectly functional system. Either way, the end result is the same, except instead of chasing this seemingly intangible and completely relative concept of "stability", I can say "I couldn't care less" and have the same great overclock without worrying about what an app or other people say is stable or not.

But to those who are die hard overclockers and stability chasers, my hat's off to you.
 
5 passes is what I use for quick tests but I have started doing 100 passes now for testing ultimate stability. I've had the first 20 passes go without a hitch only to error on the 21st so I don't even trust 20 anymore. Let me clarify: I wouldn't trust 20 with my life. 100 passes takes around 3 hours on my box and I don't compile anything that takes that long. Compilation is probably the most intensive task my computer does with all 4 cores (how does video encoding stack up I wonder?) so if I can run linpack for 3 hours I can certainly compile for 3 hours without error.

I actually do think that 20 passes is probably stable enough for me but I don't want to leave anything I can control to chance.

Either way, the end result is the same, except instead of chasing this seemingly intangible and completely relative concept of "stability", I can say "I couldn't care less" and have the same great overclock without worrying about what an app or other people say is stable or not.

It all boils down to what you use your computer for. Some of us play and work on our machines and stability is essential (of course we could just not OC, but that's not very [H]ard). A few flipped bits here and there will bite you in the ass eventually, and you don't want to waste time chasing down a non-bug because of an erroneous compilation, or otherwise end up with corrupted files (could be system files, or your VM disk image!).

If your computer isn't what you use to pay rent and buy food and it is stable enough for you, then that's obviously good enough. As you said, it is all relative.
 
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.
Wasn't the person's original point that people who get a "fail" with this program will still call their overclock "stable" erroneously.

There's a distinction to be made here:

* If you run this test, and Prime95, and test thoroughly, and you never get an error. It remains entirely possible that your OC is 100% stable.

* If you get an error in this test (assuming no other issue), you can no longer say that your OC is 100% stable. You have direct evidence to the contrary.

It's the difference between not knowing for certain but having no evidence to the contrary, and having evidence to the contrary and knowing for certain that an OC has caused a CPU/motherboard/RAM to error.

So there will be folks who used to claim 100% stability via other tests who will deem a failed test with this program as irrelevant or meaningless. That's fine. It's their personal decision. But the fact remains that there is now direct evidence that their OC settings are "too much" in that their system will error where it would not error at stock. The real world significance of this may be nil. But the facts are still there.

I'm one of those folks who doesn't leave an overclock dialed in until my system can do every test just as well as it does at stock. Because a random crash (even if it's the first time in months) while I'm fighting the the Balrog in LotRO can ruin my day and the day of several other people. :p

Anyways, as for me, this progam adds about 12c to my temps over P95.
 
This new program really helped me out.

I was 24-hour prime stable. But if I did my real torture test (Prime95 + 3DMark06 simultaneously), I'd get an application error within a couple hours. Yet, at stock speed, I could run both simultaneously indefinitely. So, obviously, the OC was causing a problem.

Because I was prime stable, I kept assuming that there was an issue with my NB or SB voltage and left vcore alone. After a couple days of testing, I was about to give up. . .

Then I tried this program. It showed errors within two hours (did 20 runs). . . which led me back to my vcore. I upped it two notches and ran it for 20 runs without any errors. Now I'm almost twelve hours into my Prime95 and 3DMark06 torture test with nary a hiccup. And I was able to put all my other voltages back down to where they were.

This program is worth its weight in gold to me at this point. It helped put me back on the right track and saved me days of testing and/or just jacking up all voltages needlessly.
 
I still crash the second I hit "start"... Even at stock.. Even underclocked :p.

I've been folding and gaming on this machine for 23 days now... I think that's stable enough :)
 
I still crash the second I hit "start"... Even at stock.. Even underclocked :p.

I've been folding and gaming on this machine for 23 days now... I think that's stable enough :)

are you trying to allocate more than 2gb of ram on a 32 bit os by any chance?
 
From what I've read over at XS, the test can be inconsistent with other results. For example, one guy had a system that was ORTHOS stable but failed the IBT; he also had a system that was IBT stable but failed ORTHOS.

The point, I think, that using at least two different stress tests is a wise idea if you want to ensure your stability. I remember my first overclock attempt on an X2 5600+; while it was ORTHOS stable for over 8 hours, the system rebooted after about 1 hour of playing Supreme Commander, every time.
 
From what I've read over at XS, the test can be inconsistent with other results. For example, one guy had a system that was ORTHOS stable but failed the IBT; he also had a system that was IBT stable but failed ORTHOS.

The point, I think, that using at least two different stress tests is a wise idea if you want to ensure your stability. I remember my first overclock attempt on an X2 5600+; while it was ORTHOS stable for over 8 hours, the system rebooted after about 1 hour of playing Supreme Commander, every time.

for stress testing, the best results come from using a variety of tools and conditions. I always mix in some load on the video card too while running this or p95.
 
for stress testing, the best results come from using a variety of tools and conditions. I always mix in some load on the video card too while running this or p95.
QFT

My standard P95+3DMark06 loop was giving me an application error (3DMark06). P95 would run indefinitely. So I was being misled by my own testing to believe that it was 3D-related and therefore something other than vcore. But IBT showed errors, leading me to try a tiny bit more vcore. . . great success! :D
 
IBT is a thin wrapper over Intel LINPACK, which is a heavily optimized linear equation solver. Intel offers a free non-commercial Linux version, which is really great because there are not many Linux burn-in tools, let alone ones hand coded by Intel. LINPACK runs 5C hotter than dnetc on my E8400, and more importantly, can detect errors. Nice find
 
...This program is worth its weight in gold to me at this point. It helped put me back on the right track and saved me days of testing and/or just jacking up all voltages needlessly.

So it's pretty worthless then? Being weightless and all...

Just kidding mate - I couldn't help myself :D
 
"Well, I let it run for 20 passes, and that's enough. My OC is 100% stable. :D"

I'm bringing this thread up from the dead. Mine passed 15, failed 35. Got it to pass 35, failed 100 tests. When does it ennnnnnnnnd?
 
Ends when you run everything at stock and pass everything [hopefully].
 
Back
Top