IntelBurnTest - The new stress-testing program

[rockfella]

I gave up on this stress test! lol too stressful, no 24 x 7 application(for home users) will ever stress the hardware like this tool does we have today....

[ragingclue]

I was actually having an issue with this test last night. I've got an e8400 @ 4.01 GHz, and I ran 10 IntelBurn Tests on it. It passed them all with flying colors; however, when I looked at the RealTemp indicator of processor load, it wasn't above 60% for the tests. All previous tests had me over 98%, and I would get up to 69C in temp. With last night's tests, I was only hitting 52. I had lowered the voltages in the BIOS to see how low they can be at 4GHz (I think I'm at about 1.3xxx vCore). I ran AoC for a while on very high settings, and did a lot of other stuff, and everything seemed to work fine. Even Prime was just fine. What would be causing the CPU to run only at 60% during the burn test? Memtest is fine as well.

[omiez]

For which OS is the WinPE IBT ?

What is the difference between the 2 IBT's on first page ?

[Falkentyne]

I answered this in the other thread you made.

You really should ask unclewebb about this; same thing happens when I overclock my cpu, but shows 99% at stock speed.

[7im]

Looks like a good tool. How does IBT compare to the much older StressCPU testing app? (which put Prime to shame several years ago by your Folding@home team) And Prime stable has never been F@H stable. :P

StressCPU v2.0

[Razmatazz]

I started using this when I got my E8400, it works great for testing the CPU but temps start to scare me when they get into the 70s. And you need something else to test the memeory/nb, I passed 30 rounds of ITB and prime blend failed instantly until I lowered the fsb.

If prime is stable, you dont need to increase your vcore. It's unnecessary if everything is stable under prime and all other applications, but not under linpack. There isn't a thing that can bring your computer down if all your other programs work fine.

I'm rock solid under prime@4 ghz and 1.3625v (1.352v load), but linpack needs 1.375v (1.360-1.368v load) for stability. That's only because linpack heats my CPU up to around 69-71C. Nothing in the world can heat the CPU that high..do you see the point? But I would never need to run my qx9650 @ 1.375v @ 4 ghz with my TRUE (my tuniq couldnt handle the heat) as nothing would need this vcore except linpack. So 1.3625v is fine for me. (I actually use 1.35v if im not using a heavy stress game).

To keep your cpu running healthy, lower it back to what you needed for prime. But you can rest assured that when you are linpack stable, that cpu will never crash under any circumstances (as long as your RAM/northbridge is stable).

Remember the reason you crash is because linpack heated up your CPU higher than what Prime could do. No program can cause that type of stress...

What has some of us baffled is how linpack could be so stable yet prime blend will fail in just seconds until fsb lowered. I have run into this as well. This has been seen with both duals and quads. Anyone have any insights into why this is occuring?

[truehighroller]

What has some of us baffled is how linpack could be so stable yet prime blend will fail in just seconds until fsb lowered. I have run into this as well. This has been seen with both duals and quads. Anyone have any insights into why this is occuring?

I beleive this is an issue with the motherboard / BIOS, as I used to fail Prrime all the time and yet I would be fine with burntest. I automatically just thought ok something is worng with prime95 and how it works with my system and all of a sudden sure enough, I was right because Asus came out with a BIOS that supposably addressed issues with prime95 erroring out and other issues with C1 stepping CPUs with their P45 motherboards... I think honestly that the burntest is the most solid tester I have come across yet.

[Giux-900]

hi to all !

I have a little problem with ibt on my pc:



This error is not oc or settings related, defaut or oc is the same...
the only work metod is to set memory size with option n4...

my os is winxp 32bit with only 2GB ram

is a software bug.. i think

[omiez]

hi to all !

I have a little problem with ibt on my pc:



This error is not oc or settings related, defaut or oc is the same...
the only work metod is to set memory size with option n4...

my os is winxp 32bit with only 2GB ram

is a software bug.. i think

Hello Giux.

You have to enable Swap File to test in Mode 1.

Ciao Giux

Devi abilitare lo Swap File o IBT ti da questo errore in Modalita' 1.

[Giux-900]

Hello Giux.

You have to enable Swap File to test in Mode 1.

Ciao Giux

Devi abilitare lo Swap File o IBT ti da questo errore in Modalita' 1.

hi, thanks for fast reply !

this is my file swap:



you can
explain me the exact setting size ?

grazie

[xyrico]

I love this program. It's great for quickly dialing in settings without having to Prime for 8 hours only to have it fail at something like 8 hours and 15 minutes.

Something I've noticed that seems to be a bit unique to my rig is that I very rarely see an inconsistency in the numbers, even if I'm unstable. 32-bit IntelBurnTest will either just crash or give a blue-screen, but the 64-bit version has an odd tendency to print out the "completed successfully" message after just a couple of passes, even if I set it to do something like 200. Either way lets me know I need to do more tweaking.

My personal record for longest time to detect a problem is the time I had 1 inconsistent number about 160 passes into a total run of 173. This took 3 hours, but I could have probably been running Prime for days before it failed.

[InfinityG35]

hi to all !

I have a little problem with ibt on my pc:



This error is not oc or settings related, defaut or oc is the same...
the only work metod is to set memory size with option n4...

my os is winxp 32bit with only 2GB ram

is a software bug.. i think

I have the same problem, even at stock. Just on a clean XP SP3 install, did not mess with swapfile. How to get it running normally?

[Brandito]

anyone else tested this in windows 7 yet?

i'm trying it out right now.

[spiro]

I would recommend that anyone doing this test do it only 5 runs. It apparently puts a lot of pressure on your capacitors. I had 3 motherboards fail (2 MSI Neo's and 1 Intel EE board) after running an overnight stress test. I had thought that it was something else, but now I am beginning to think that maybe running this application in an overclocked environment for a long time could be the culprit. Best to stay prudent and run it a few runs only. It usually can find if your system is not 100% stable within the 5 runs anyway.

[AgentGOD]

Updated to v1.92!
- Updated Linpack binaries to the latest (April 2009)
- Added check to make sure Linpack binaries were properly placed

Also: I've done some more personal testing... and found that a custom test memory size of 1024 MB and 5 iterations is sufficient enough for fast stability testing. Tested on Vista Ultimate X64 SP2.

For those of you with Core i7 processors, please let me know if these new binaries correctly utilize HyperThreading.

Thanks!

[WaterFlex]

Good news, AgentGOD

[AgentGOD]

*Off-topic: Wow, that is very fast DDR2 memory you got there

[Kougar]

Updated to v1.92!
- Updated Linpack binaries to the latest (April 2009)
- Added check to make sure Linpack binaries were properly placed

For those of you with Core i7 processors, please let me know if these new binaries correctly utilize HyperThreading.

Thanks!

According to the Task Manager, only 49-50% of the CPU is being utilized. The graphs are all crazy, one graph shows 0% in use the rest are hovering anywhere around 20%, 50%, to as high as 80%.

[AgentGOD]

According to the Task Manager, only 49-50% of the CPU is being utilized. The graphs are all crazy, one graph shows 0% in use the rest are hovering anywhere around 20%, 50%, to as high as 80%.

Well then, it's what Intel intended to happen on Core i7.

[Leeghoofd]

Well then, it's what Intel intended to happen on Core i7.

????

[003]

Well then, it's what Intel intended to happen on Core i7.

True, but it is possible to load all 8 threads on i7. LinX does it by launching two instances of the linpack executable simultaneously.

[Pyr0]

True, but it is possible to load all 8 threads on i7. LinX does it by launching two instances of the linpack executable simultaneously.

LinX doesn't use 2 instances of linpack anymore
LinX loads all cores 100% using one linpack exe now since version 0.5.5.2 in Jan

[AgentGOD]

Intel themselves said to disable HyperThreading before running the program.

[Kougar]

Intel themselves said to disable HyperThreading before running the program.

Probably because running 4x verse 8x instance of Linpack will cause Linpack to run twice as slow. Since all 8 threads are trying to use the exact same execution units hyperthreading can't improve upon the performance.

I have no plans to disable HT because I will be running virtual machines and loading all eight threads... so I don't care if Linx runs 2x as slow. I need to ensure the system is 100% stable with the exact settings and features I plan to be using.

And belatedly, thanks for the update AgentGod!

[AgentGOD]

And for those of you who still care...!


Sneak peak of v2.0 beta! It even has an animation to let you know if your system is frozen or not, AND it outputs results as Linpack runs while it checks for any errors in the results produced. Stress level selection is now also much simpler.