IntelBurnTest - System Stress & Stability Testing

[jolliffee]

This is a first pass draft, please move if not in the right place


IntelBurnTest : Stress & Stability Testing
________________________________________
If you are worried about the stability of your system but have found life is too short to use Prime95 for hours and hours, there is a new kid on the block for Intel CPU’s; IntelBurnTest

While Prime95 possibly remains the most popular CPU stress testing program, better even than OCCT, because OCCT doesn't stress the system as hard as Prime95.

If your system is Prime95 stable for 8 hours? 12 hours? or even 24 hours is it actually stable? Maybe not as I found out! I had a fully 24 Hour Prime95 5 Hours MemTest stable system but IntelBurnTest failed on the first run ouch!

What is IntelBurnTest:
IntelBurnTest is a front end to the utilities Intel use to test their CPU’s, these utilities (LINPACK) have been around for an age and use a command line interface (CLI) which was time consuming and unfriendly,

This new interface has been written and supported by AgentGOD. So the actual programme that doe the stressing is not new, it is the interface that is new; this is important because LINPACK has many years of development and credibility in stressing Intel CPU’s


Why is IntelBurnTest Important to Over Clocker’s?
Do I think IntelBurnTest better than Prime95? YES – I DO!
1) It takes no more than 30 Minutes to perform a full stress test
2) The system is stressed much harder than Prime95

Summary:-
Linpack:
1. More accurate than Prime95 Small FFTs/Blend.
2. Takes less time to tell if your CPU/RAM is unstable than Prime95 (usually
something like 8 minutes Linpack vs 40 hours under Prime95).
3. Use the same stress-testing engine that Intel uses to test their products
before they are packed and put on shelves for sale.

IntelBurnTest:
1. Simplifies usage of Linpack.
2. Direct output of results to the screen.
3. Simplifies the process of selecting a test size to use.
4. Better appearance.
So if you want to know if your system is stable a quick 5 to 20 passes of IntelBurnTest will do it in around 30Minutes (Depends on how much memory you have)
So if your system is Prime95 stable for 24 hours, it still has a high chance to fail in IntelBurnTest. Try it??

Before I moved to Vista 64 I was fully Prime95 & IntelBurnTest stable, after moving to Vista 64 it took no more than 5 minutes of IntelBurnTest x64 to totally destroy my smug stability tests.

A word of caution! Your core temperatures will most likely go higher than they have before so if you are on the edge be very carful and run RealTemp to monitor.

My Own Experience:
I’ve had to increase the CPU voltage and reduce the multiplier to keep temp down. I’ve also noticed Vdroop is higher which indicates it is stressing the the CPU much more

Using IntelBurnTest!
IntelBurnTest is so easy to use I don’t need to say much, other than if your ouput numbers are not the same it has failed. Also there are a couple of test options to reduce the stress of the test.

32-bit or 64-bit?

As I mentioned ealier I found the 64 bit version stressed the system more than the 32bit version. I fully expected this because 32 bits do not drive the system anywhere near as much as a 64 bit system, goes without saying really!

AgentGOD says it is better to run IntelBurnTest under a 64-bit OS as it gives more accurate results. This is most probably because certain parts of the CPU are inactive during 32-bit operations.

Also, under 32-bit Windows, you won't be able to let IntelBurnTest use more than 2GB memory because Windows (32-bit XP) does not allow allocation of more than 2GB memory to a program. But it is still a great test program, and all the points I have made still apply!

My Thoughts
When is a system truly stable; Umm! A big question with many answers and opinions. For me a system is stable, when it doesn’t crash during the normal use I put it to, and you are aware of the limits of your systems stability. But that doesn't necessarily mean the system is not producing any errors. As an example, I was Prime95 24Hour stable but still failed IntelBurnTest.
Do I want the extra heat from increasing voltage or am I going to look elsewhere for the stability?

I’m truly thankful to AgentGOD for producing IntelBurnTest, it has saved me hours of Prime95 runs.

But please remember this is not the magic program, and it will stress your system so be careful.

[Shanara Red]

Sounds interesting ... I do not however understand how it's possible, that you are geting higer temps on your CPU of the bat ... Prime 95 is supposed to fully load you cpu, so exactly what this wunderkind does to stress it even further is not clear to me. (let's leave out the 32b vs 64b debate for now)

And secondly ... are you sure, that a 30 min max-uber stress test is a better way of testing stability than an extended, several hour run under Prime 95 max load? Even tough you system is unlikely to be stressed out anywhere close to what Prime does to it in everyday use, it does give the benefit of running for an extended period of time. Maybe in such a severe stress test that you are suggesting, errors that would not normally come up during regular use could just simply appear because of over-stressing?

I'm no pro, but that would seem to be the case, if your system was 24 hours stable under Prime ...

[affiliate13]

I usually use a mix of stress tests, i have seen folk running Orthos, Prime95 and OCCT for several hours no problems and then crashes in games cured simply by a touch more voltage.

It is nice to see another way to stress test, but it should perhaps be viewed as part of a tool kit rather than a definative method.

[Kelainefes]

Woha it gave me like 6 C higher temps compared to Prime95 small FFT!

[Clunk]

TAT used to do a similar thing in that it could produce around 20% more heat than prime on a quad core CPU.

Maybe TAT used linpack?

[jolliffee]

Hi Shanara

Originally Posted by Shanara Red

Sounds interesting ... I do not however understand how it's possible, that you are geting higer temps on your CPU of the bat ... Prime 95 is supposed to fully load you cpu, so exactly what this wunderkind does to stress it even further is not clear to me. (let's leave out the 32b vs 64b debate for now)

You'll need to try it and make your own mind up, I know these are the tests performed on Intel CPU's before they ship so have some credability.

There is no debate I'm afraid; 64 bit environments will put more strain on your system, since you'll be using twice the data width through the CPU etc.

Originally Posted by Shanara

And secondly ... are you sure, that a 30 min max-uber stress test is a better way of testing stability than an extended, several hour run under Prime 95 max load? Even tough you system is unlikely to be stressed out anywhere close to what Prime does to it in everyday use, it does give the benefit of running for an extended period of time. Maybe in such a severe stress test that you are suggesting, errors that would not normally come up during regular use could just simply appear because of over-stressing?

Like most people I use a number of tests but life is too short to wait 24 hours to find an error.

I'm no pro, but that would seem to be the case, if your system was 24 hours stable under Prime ...

Systems degrade through time so stability is only a point in time, I guess?
What I do know is that if you run some tests that give you the confidence to use your stsem for work, and then that system does not error the probability is you have a stable system for your particular envirnment.
Also we all know that a new BIOS can blow a hole through that stability in an instant!

[Kelainefes]

Originally Posted by Shanara Red

Sounds interesting ... I do not however understand how it's possible, that you are geting higer temps on your CPU of the bat ... Prime 95 is supposed to fully load you cpu, so exactly what this wunderkind does to stress it even further is not clear to me. (let's leave out the 32b vs 64b debate for now)
.........

It is possible, ever noticed how while you play a game cpu goes 100%? Well when i game core temps are around 42-45 C, when i stress small FFT with Prime95 they go 52-56 C and all the way up to 60-62 C with IntelBurnTest.
Same cooler, fixed fan speed, same room temp.
Don't exactly know why, but i think that everybody here will tell you the same happens in their systems.

[kinnsy]

I think it may have something to do with integer calculations and floating point. As they have separate units within the CPU that handle those calculations. Doing FFT is not going to put much stress on the integer unit but lots on the FPU.

I can remember when FPU was done on a separate chip, 8087, 80287 and 80387. These cost way more than the regular CPU too. From the 80486 onwards the FPU was included on the the same chip.

[Kelainefes]

Yes, and now we have SIMD units too (MMX, SSE, SSE2, SSE3,3DNow, etc) but they are not indipendent, ie if you are using a part of a pipeline you may not be able to use the other parts, and depending on how optimized the code is you will use more or less.
It's kind of the same with video cards, in the new ATI's GPUs the 800 cores are grouped 5 by 5 so if the code is not optimized you will use only 160.

[jolliffee]

Originally Posted by Kelainefes

It is possible, ever noticed how while you play a game cpu goes 100%? Well when i game core temps are around 42-45 C, when i stress small FFT with Prime95 they go 52-56 C and all the way up to 60-62 C with IntelBurnTest.
Same cooler, fixed fan speed, same room temp.
Don't exactly know why, but i think that everybody here will tell you the same happens in their systems.

I took a closer look; It uses Intel's Linpack math libraries. While Prime95 uses 100% of your CPU, there are still lot's of wait statements involved. The linpack libraries are designed to get the data you need as fast as possible so stress it more

Also noticed it stresses the PS too, I've been struggling to get IntelBurnTest to pass at 1600memory clocks, my PS was drawing 344 watts. Normaly in the 240 watt area.