IntelBurnTest - The new stress-testing program

[AgentGOD]

Readme

IntelBurnTest v2.4
by AgentGOD
----------------------------
The original. All others are copycats.

Home Page: http://www.ultimate-filez.com
Donation : http://www.ultimate-filez.com/?page=donate

Description:
	A program that simplifies the usage of Intel(R) Linpack. Linpack by
Intel(R) is an extremely stressful program that will put even the most pow-
-erful X86/X64 CPU in the world at its knees. Load temp under Linpack will
be up to 22*C higher than the competing software Prime95. This program will
make usage of Linpack easier and more practical.

Keep in mind, use this program at your own risk. By using this program, you
agree that neither I nor Intel shall be responsible for including, but not
limited to: burned up CPU, fried motherboard, spontaneous room temperature
increase, hair loss, or mental stress.

Xtreme Stress Mode (Right-click the "Start" button) allows unprecedented
improvement of testing accuracy.
	Note: Xtreme Stress Mode may cause the CPU to heat up more than it
		already does. During testing, Windows may not be usable.


Compatibility:
	Works with Intel(R) and AMD CPUs. Tested Intel(R) Core 2(tm) Quad
	Q6600, Intel(R) Pentium(tm) III 550 MHz, AMD Athlon 64(tm) X2 4200+,
	and AMD Opteron 165 (Working with AMD processors since v1.7). Also
	tested on AMD Phenom II X4 940 Black Edition.

	Tested on Microsoft(R) Windows XP Professional SP2/SP3, XP Professional
	X64 Edition SP2. Tested on Microsoft(R) Windows Vista(tm)/7 X86/X64
	by many awesome users like yourself ;)

Benefits of using Linpack:
1. More accurate than Prime95 Small FFTs/Blend (under x64 OS).
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.

Benefits of using IntelBurnTest:
1. Simplifies usage of Linpack.
2. Real-time output of results to the screen.
3. Simplifies the process of selecting a test size to use.
4. Better appearance.
5. Intuitive interface.
6. Real-time error checking.
7. System status acknowledgement.

Requirements:
 - An x86/x64 CPU
 - At least 512 MB of system memory. Recommended minimum is 1 GB.
 - A modern Windows-NT based operating system (XP and above). Both x86 and x64 are supported.
 - Microsoft(R) Visual C++ 2008 Runtime (x86), can be found here:
   http://www.microsoft.com/downloads/details.aspx?FamilyID=9b2da534-3e03-4391-8a4d-074b9f2bc1bf
 - Microsoft(R) .NET Framework v2.0 (minimum), can be found here:
   http://www.microsoft.com/downloads/details.aspx?FamilyID=0856EACB-4362-4B0D-8EDD-AAB15C5E04F5


How to test CPU and RAM stability (summarized):
Pre. You need to extract everything from the archive to a single folder, while
	maintaining the original directory structure.
1. It is best to use a 64-bit OS with the 64-bit mode for the most accurate
   test result.
2. Use the most available RAM possible (IntelBurnTest can now do this on
   its own automatically). The more memory it uses, the more accurate.
   However, most people can use "Standard" mode as it should be sufficient.
3. Select the run # (should use at least 5 at minimum, no more than 20).
4. Start running. Output should look something like this:
	Time (s)	Speed (GFlops)		Result
	0.269		31.7314			3.684559e-002
	0.260		32.7829			3.684559e-002
	0.257		33.1546			3.684559e-002
	0.264		32.2425			3.684559e-002
	0.266		32.0700			3.684559e-002
   NOTE: The time it takes to run each iteration may vary from test size, and
         so will the speed and the result. But if the results do not stay
 	 consistent, your system is unstable (specifically CPU).
5. Above indicates that the test ran fine so far without instability, but
   below indicates instability (the Residual(norm) values aren't the same
   throughout running the test):
	Time (s)	Speed (GFlops)		Result
	15.245     	43.7434			5.380399e+003
	15.250     	43.7303			8.675501e-002
	15.162     	43.9816			3.570760e-002
   NOTE: Instability doesn't necessarily have to happen at the beginning, it
	 can happen in the middle, or at the end. It could be just ONE of
	 them that is inconsistent, but that DOES indicate instability.
	 BSOD or freezing are signs of instability as well.
 *As of v2.0, the program will now have real-time result output, as well as
  real-time result checking for system stability based on the consistency
  of the outputted results. The testing will automatically stop if the pro-
  -gram detects such inconsistency. During testing, you should see an ani-
  -mated flame under "Freeze Test". If at any time during the testing you
  see it stop moving, it is most likely that your PC is frozen.

 *As of v2.1, Core i7 users that wish to test with HyperThreading enabled
  can now override the number of threads for Linpack to execute

 *As of v2.2, if you are having any issues with IntelBurnTest, please en-
  able debug logging by right-clicking the "logging" checkbox to do so, and
  then send me the debug log file. It would also be helpful if you also send
  a screenshot of the issue and steps to reproduce the issue.


Credits:
--------
AgentGOD - Coding this program, duh :)
Cronos (from XS) - for the original the memory size equation
mikeyakame (from XS) - for notifying me of newer binaries
WoZZeR999 (from XS) - memory size suggestion
krille (from XS) - for the constructive criticisms
Rob Williams (from Techgage) - for testing on a Core i7 platform
justin.kerr (from XS) - for testing on a Core i7 platform
Intel - for Linpack and a Core 2(tm) Quad processor ;)

IntelBurnTest, Copyright © 2009 Ultimate-Filez. Linpack, Copyright © Intel(R) Corporation.

Please check the readme for usage instructions
In my own experience, I used Prime95 on my Q6600 with Small FFTs, which showed that it was stable for 10 hours overnight at the current voltage. However, running Linpack with the most memory I had available (I have 64-bit XP, so I see all 4 GB of my RAM), it indicated instability. I had to raise the voltage up a couple of notches to get it fully stable. This will ensure me that it's now at least 24 hours Prime95 stable, if not 100% stable.

Download link:
http://www.ultimate-filez.com/files/IntelBurnTest.zip

Screenshots:


For screen-shot purposes, I used a very low amount of memory to test.

If you like this program, please donate:


Thanks.

If you still wish to use the Legacy version (v1.9x), here they are:
IntelBurnTest v1.92:
http://www.ultimate-filez.com/files/...est-Legacy.zip
Native 64-bit version for WinPE (v1.90)
http://www.ultimate-filez.com/files/...Legacy-x64.zip

Note: for Core i7 CPU owners, please use at least 16 threads if you wish to fully load all physical and logical cores (HyperThreading). * Since v2.4, this may not apply anymore. Please try 8-threads and see if that works for you.

[003]

So basically this is an easier to use linpack? Can it sustain a constant 100% load, or does it drop between tests like the normal linpack? Pretty cool.

[AgentGOD]

Originally Posted by 003

So basically this is an easier to use linpack?

Yep.

Originally Posted by 003

Can it sustain a constant 100% load, or does it drop between tests like the normal linpack? Pretty cool.

That's part of Linpack's stress test. Because voltage varies between having one core loaded, and all four cores loaded, so Intel made sure to test all factors.

[WoZZeR999]

Will this automatically report if there's an error? I remember running linpack the first time, and it was a hassle to figure out if it errored out because you had to actually look at the data.

[003]

The data is easy to look at, because you can scroll through it extremely fast, and if any line is different, you will notice very easily because it sticks out like a sore thumb. But still, it would be nice to have it report if there is an error.

Also, I believe that even if they all are the same, but the number is simply too large, then that also means it's not stable. I'm not sure what the cutoff for stable vs not stable is though.

[AgentGOD]

Originally Posted by 003

The data is easy to look at, because you can scroll through it extremely fast, and if any line is different, you will notice very easily because it sticks out like a sore thumb. But still, it would be nice to have it report if there is an error.

Also, I believe that even if they all are the same, but the number is simply too large, then that also means it's not stable. I'm not sure what the cutoff for stable vs not stable is though.

Number being too large does indicate instability as well, but when that occurs, the other Residual(norm) values will not be equal.

Originally Posted by WoZZeR999

Will this automatically report if there's an error? I remember running linpack the first time, and it was a hassle to figure out if it errored out because you had to actually look at the data.

It was a hassle because using Linpack alone did not output the results directly to the screen. With IntelBurnTest, it's one of the features.

[Alex-Ro]

Nice program,i personally find it pretty useful.

[Warboy]

Is this fine?

[AgentGOD]

Yep, looks fine. But for the most accuracy, 64-bit OS is the best.

Also, for that size of memory, I would use something like 20 tests over 5.

Originally Posted by Alex-Ro

Nice program,i personally find it pretty useful.

Thank you for your kind words.

[003]

Yes; however, you haven't run it long enough to be considered a real test. You need to have at least 8 hours to be safe. Also, why are your GFlops so low? I get 50GFlops with my Q9450, which is clocked slightly below yours.

[AgentGOD]

Originally Posted by 003

Yes; however, you haven't run it long enough to be considered a real test. You need to have at least 8 hours to be safe.

That's the old Prime95 rule. Refer to my experience in my first post in this thread. 10 hrs Prime95 stable = failed Linpack in 2 minutes.

Originally Posted by 003

Also, why are your GFlops so low? I get 50GFlops with my Q9450, which is clocked slightly below yours.

Because you have a bit more memory than he does, and the GFLOPs are more accurate when using more memory to test.

[WoZZeR999]

Is there a way to choose how many tests, or only runs?

[Monkeywoman]

Holly dude, that program brought my systems to its knees!!!!!!! like 30sec in BSOD, i had to lower to 3.2 at 1.4v for stability. seriously, very good program.

[Loser777]

I prefer Far Manager.

[WoZZeR999]

Far is something completely different. But some parsing would be nice for 'IntelBurnTest'. If you had it just show important data, with the option of verbose mode, they would be much better I think.

[Loser777]

I use Far to update the Result file in Real Time... AgentGOD's program does the same thing...

[NickS]

Wow, I ran this for four minutes. When do Wolfdales start throttling?

[Leeghoofd]

Thx for the program, another one to add to the list but it's quick and will see if it's efficient enough...

that's one hot CPU there NickS.. better swap that boxed cooler... I get 65°C on the Intel Extreme cooler on my E8400 full blast, 75°C at low setting...

[AgentGOD]

Originally Posted by WoZZeR999

Is there a way to choose how many tests, or only runs?

Well, since you are testing stability, you want to know if your system can keep the Residual(norm) value consistently at one value. Varying the number of tests would not tell you that.

[loonym]

Nice for the simple fact that it's quick. I refuse to spend hours running stability tests.


xfx790i, x3360, 4gb gskill hz

[unclewebb]

Originally Posted by NickS

Wow, I ran this for four minutes. When do Wolfdales start throttling?

Looks like there is finally some software available to test the PROCHOT# section in RealTemp! This info is displayed for each core. The Wolfies should show a check mark in The PROCHOT# - History box (right box) at approximately 93C. Every processor is unique but generally this is the magic number. Throttling will be so minor at first that you might not even be able to detect a drop in MHz in CPU-Z or RealTemp.

If things continue to get hotter then the Status box in RealTemp (left box) will light up along with the History box and then both programs should start to report that your multi is getting cycled down to 6.0 to help keep the temps from getting too crazy. Thermal shut down isn't until 125C so no need to worry just yet.

[Loser777]

Originally Posted by loonym

Nice for the simple fact that it's quick. I refuse to spend hours running stability tests.


xfx790i, x3360, 4gb gskill hz

You do realize that that isn't stable right?

[HDCHOPPER]

anything but a rar

[Leeghoofd]

Glad you told Loonym to look at the scores, thought if it did its passes it would be okay lol...



My parents rig look sweet at 3.6ghz at 1.1volt..

[panheimer]

What the hell?!?!?!
As soon i've read this post, i've downloaded this utility, unpacked it and run a short 5 times test. Well all 5 five test reports the same number, so the system is supposed to be ok, BUT: everest lost nb & sb temp readings, the rampage's lcd poster started flashing and the 3com usb wireless dongle stopped working...
After a system restart everthing goes back normal, but what happended? It may be a beta release issue or is my system not really stable? Don't dare to run it again before someone give me some explanation...
BTW I've primed it for about 10 hours without an error.