Notices

Overclockers Forums > Hardware > General Hardware
General Hardware Discuss your new build and gather upgrade suggestions. And if your question doesn't fit anywhere else...it goes here.
Forum Jump

Guidelines for Thorough Stability Testing

Post Reply New Thread Subscribe Search this Thread
 
 
Thread Tools
Old 10-12-04, 11:04 PM Thread Starter   #1
felinusz
Senior Overclocking Magus

 
felinusz's Avatar 

Join Date: Feb 2003
Location: Taiwan

 
Guidelines for Thorough Stability Testing


INTRODUCTION: THE CONCEPT OF STABILITY


PREAMBLE


As of late, I've been seeing more and more of my fellow overclockers tossing the concept of stability to the wind, and running their un-tested machines carefree, unaware of the perils and pitfalls which may await them.

As I write this, I'm blasting some music on my computer, running a graphics card stress tester in the background just for the heck of it, and burning a CD. I can do all this on my overclocked machine with complete peace of mind knowing that my music won't stop playing, my stress testing won't crash, and my CD won't get messed up. And yeah, it's pretty likely that Windows won’t go and pull a blue screen of death on me either.

How is this possible? Aren't I overclocked? Well, I've also stability tested my computer thoroughly, and know it to be stable.

I'm writing this guide for everyone, even those of you who already know what is involved in thorough stability testing (you know who you are!) – if these words are used as reference, or are helpful in any way, then they’ve served their purpose. This isn't an almighty rulebook or anything, just a few extremely good guidelines to try and follow when stability testing your computer.

My goal is for every single overclocker in the world to know how to thoroughly test their machine for instability.

Let's get right into the guts of this.


WHY SHOULD I CARE ABOUT STABILITY?


Your computer is a tool, made to aid you in performing a variety of tasks, whether it's playing a favorite game, burning a CD, or writing up your twenty page Philosophy term paper.

When your computer is running, it's stressing itself; it's usually working pretty hard, how much depending of course on what you're doing with it. Your tool is seeing some use; as a hammer gets bashed and swung, your computer is performing all kinds of processes. If your hardware is unstable, it's bound to make a mistake sooner or later - your very expensive hammer is going to either crack, or fall apart and hit you in the face.

Without trying to scare anyone; that mistake could happen while you're defragmenting your hard-drive, or it could happen just as you're saving that previously mentioned twenty page philosophy term paper on Post-Modern Existentialism that you've been slaving over for the last month (I hope you backed it up).

That little mistake could cost you everything on your hard-drive.

In short, if your computer hasn't been stability tested, it's entirely possible that it isn't stable, and is going to mess up on you at just the wrong moment. It’s Murphy’s Law, and it happens all too often for my liking.

If you’re still not impressed, go to your local store, and spend lots of money on a tool. Then do something to it so it doesn’t work properly. This tool is your computer.

Quote:
From Prime95 Readme

Today's computers are not perfect. Even brand new systems from major manufacturers can have hidden flaws. If any of several key components such as CPU, memory, cooling, etc. are not up to spec, it can lead to incorrect calculations and/or unexplained system crashes.....

.....For these reasons, both non-overclockers and overclockers need programs that test the stability of their computers.....

OVERCLOCKING AND STABILITY


Stability isn’t just important for us overclockers, it’s important for people who use stock hardware as well. However, we overclockers are at far greater risk because of what we do to our hardware.

When we overclock, we push the limits of how fast our hardware can run; we add strain to our hardware in day-to-day use, and we drastically increase the chances of having an unstable setup, even if everything looks peachy on the surface (or desktop if you will).

Almost all of us have had some experience with instability: trying for that extra speed step, and getting a reboot, or getting artifacting after clocking up our video card. On the surface, more voltage, less speed, or better cooling makes the problem go away. But underneath the underneath, a potential for instability still exists; you can never be sure just how solid your setup is until you’ve tested it thoroughly and properly.

This potential for instability is a very important concept that many of us disregard – many people are of the mindset that if it looks O.K. on the surface, and if it runs O.K. for the odd light gaming or benching session, then it probably is O.K.

This mentality is nicely comparable to that of someone looking at quicksand – quicksand looks solid on the surface (surface = desktop). Unknown to the observer however, quicksand is all mushy, to some degree, underneath the surface (underneath = a processor doing some complex math). The dangerous quicksand and the perfectly safe sand-by-the-beach both look the same from a surface perspective.

Quote:
From Prime95 Readme

"Why run a stress test if you are going to ignore the results?" These people want a guaranteed 100% rock solid machine. Passing these stability tests gives them the ability to run CPU intensive programs with confidence.

WHAT CAN I DO?


You can take a little bit of time, and test your machine thoroughly; you can poke at that suspicious looking sand with a big long stick, and see if it’s safe for you to walk on or not. This is where we move away from my awareness-raising scare tactics, and extremely clever metaphors, and move towards actual techniques involving three really awesome pieces of software that you can use to thoroughly stability test your machine. In the next section I’ll deal in-depth with some effective stability testing methods, as well as some of the specifics of stability testing that everyone should know about .

Last edited by felinusz; 05-03-07 at 10:51 PM.
felinusz is offline   QUOTE Thanks
Old 10-12-04, 11:05 PM Thread Starter   #2
felinusz
Senior Overclocking Magus

 
felinusz's Avatar 

Join Date: Feb 2003
Location: Taiwan

 
PROPER STABILITY TESTING METHODOLOGY


PREAMBLE


To be perfectly honest with everyone, stability testing isn’t all that complicated.

Awareness is way more than half the battle.

There are three programs which I live by when it comes to stability testing, two more so than the third, but I digress.

These three programs are Prime95, 3DMark, and memtest86. All three are small programs that are completely free and easily accessible to anyone running a Windows machine with internet access. Respectively, these programs focus on stress testing your CPU, your GPU, and your RAM.

There are also tons of other stress testing programs which are effective to varying degrees – when it comes to these it’s all about using what you feel is necessary, credible, and thorough. Some of the many stress testing programs I’m not going to mention or talk about are fairly effective substitutes for the ones which I am going to talk about.

However, if you skip any of the "big three" - CPU, GPU, and RAM - you really aren’t being thorough enough, to put it bluntly. Three programs, one of each focusing on CPU, GPU, and RAM, do need to be used, as individually they are inconclusive.

One important thing for you to remember is that this is not an opinion. This is a fact that is backed up by a lot of experience, most of which isn’t mine; rather that of the millions of other overclockers out there.

Of further importance, is the factor of general system stability at manufacturer settings. Before overclocking or tweaking your machine, one should ensure that all hardware is operating properly and with integrity. Overclocking an unstable machine then stress testing it, finding errors, and spending hours troubleshooting a problem inherent to some faulty component, isn't something anyone wants to put themselves through! Sometimes a component is simply faulty to begin with, and needs to be replaced or switched out.


WHY THE FOLLOWING 'RECCOMENDED' TESTING PROCEDURE IS THOROUGH, BUT NOT DECISIVE

Stability is also not always compromised by CPU, GPU, or Memory processing errors as a result of overclocking or heat. Instability can also be a result of other components failing or causing conflicts, most notably a machine's PSU (Power Supply Unit), Motherboard, and internal add-on cards.

When overclocking, we add strain to our computer PSU, requiring higher power output from it. This can compromise the consistancy of a unit's voltage outputs, causing it to deliver inadequate power to the computer. Especially when paired a high power draw overclocked machine, low voltage rails often compromise stability, even when no processing errors or faults are occuring with the computer's other hardware. Such intstability will manifest itself through errors during the following 'reccomended' testing procedures, and might even appear to be CPU GPU or RAM specific! If instability is experienced, it is important to rule out one's PSU by testing it with a load - using a voltmeter to read the PSU's output voltages while stress testing the computer with the same test that previously generated errors. If readouts are normal, if they don't fluctuate or droop when the test load is applied, the PSU can generally be ruled out as a source of error.

Motherboard integrity is more difficult to pinpoint, but can usually be ruled out by researching known issues with one's specific board, or by testing the machine's responses to the sort of changes that would compromise the motherboard, without the simultanious application of CPU GPU or RAM overclocks. These variables include things such as overvoltages, PCI AGP or PCI-E frequencies, newer/older/aftermarket BIOS revisions,and general frequency changes independant of CPU GPU or memory clockspeeds.

Sometimes an external device, such as a wireless USB adaptor or a printer, can create hardware or software conflicts that will compromise a computer's stability. This sort of instability is usually quite easy to pinpoint - for example: if you add a new device and your machine starts crashing every five minutes, if the errors go away when the new device is removed... you can usually infer the cause of the errors through simple application of logic and common sense



THE MOST IMPORTANT TESTING GUIDELINE: RULING OUT ALTERNATE SOURCES OF ERROR

When stability testing the various components within your machine, it is a good idea to try and test them more or less 'one at a time'. If we test our processor while running an unstable memory overclock, for example, we might attribute instability present in our RAM to an otherwise perfectly stable and rock solid CPU overclock.

To remove alternate sources of error, one should clock up and overvolt one component at a time while stability testing, focusing testing on the altered component's integrity while the overclock is in place. By focusing on one component at a time, and keeping other component's running at confirmed-stable manufacturer's specifications, we make it a lot easier to rule out alternate sources of error. This allows us to easily tweak the source of error (whichever overcocked component it might be) back into a stable setting before proceeding with other hardware.

By ruling out alternate sources of error, we test with the confidence that any errors we discover are a result of the single modified component that has been configured with settings of unknown integrity.

For the same reasons, it is important to confirm a machine's integrity at stock settings before overclocking it, and important to ensure the integrity of a PSU if errors are discovered.

Ultimately, it does us no good knowing the machine is unstable, if we do not know which part of the machine is unstable. It is impossible to take appropriate measures towards ensuring stability, if we don't know which component needs adjustments.

Of course, testing components alone is no measure of an entire system's stability. If everything within your computer is aggressively overclocked and overvolted, some tests should be run with all your alterations in place, to make sure that none of the adjusted settings are causing conflicts with each other that would not normally manifest themselves at purely stock settings, or partially overclocked settings.






Now that we've covered some basic principles of stability testing, and some common problems we might encounter independant of CPU GPU and RAM, lets go over CPU, GPU, and RAM specific testing in three easy to follow steps .

We'll start with the CPU.


STEP 1: CPU STABILITY - PRIME95


There is one program that is very widely viewed as the CPU stability-tester to rule over all CPU stability testers, for good reason, and that program is Prime95.

There are many alternatives to Prime95 for CPU stress testing, some of which are highly effective and very easy to use. Although Prime95 is a well-proven "tried, tested, and true" stress test program, some of the alternatives are by no means inferior. I choose to keep Prime95 here, as the 'recommended' CPU stress tester, due to its time tested effectiveness. Prime95 is not definitively 'the best', I only recommend its use because I know that it is highly effective.

Please keep in mind that Prime95 can error even with a completely stable processor overclock. Unstable memory, or other system problems can also cause erroring in Prime95 - your processor is not the sole potential cause of Prime95 testing errors.


Quote:
From Prime95 Readme

This program is a good stress test for the CPU, memory, caches, CPU cooling, and case cooling. The torture test runs continuously, comparing your computer's results to results that are known to be correct. Any mismatch and you've got a problem!
Prime95 puts your processor through a very rigorous "math test", and immediately checks your chip’s answers for any mistakes.

The work which Prime95 makes your chip do will bring your processor up to a near-peak load temperature, which in turn helps stress your processor as much as possible, while also conveniently giving you an idea of what your load temperatures are. Any potential for instability that’s present will be found by Prime95 after a sufficient amount of time, given that you use the software properly.

You can download the latest version of Prime95 here: http://www.mersenne.org/freesoft.htm

HOW TO USE PRIME95 EFFECTIVELY



First, install the program as you would any other. Next, run it, and go to the ‘Advanced’ tab – select ‘Password’. Type in the password 9876 and enter it. Now go back to the ‘Advanced’ tab, and select ‘Priority’. Set the priority level to 10.

This effectively gives all of your system resources to Prime95 when it is running – now any processes running hidden in the background won’t be able to steal work time from Prime95, ensuring the most effective stress test possible.

At the very bottom of this post, a picture has been uploaded to further clarify Priority Ten setup for Prime95.

When you want to stress test your processor, run Prime95, go to the ‘Options’ tab, and select ‘Torture Test’. Run the Torture test at the default settings.

Make sure that you have turned off any screensavers, and closed all other applications when Prime95 is running.

Prime95 should never be run in tandem with any other stress testing programs, period. Prime95 is known and proven to be most effective when run by itself, and is less thorough when used with something else running at the same time. If you use Motherboard Monitor Five to monitor your temperatures, you should turn the interval time way down – 60 seconds is appropriate.

If you have an Intel processor with Hyper Threading, you need to run two instances of Prime95 for complete effectiveness. This is proven fact; in that two instances of Prime95 will catch instability that one instance won’t, on an Intel machine with HT. In order to run two instances simultaneously, simply install a second copy of Prime95 in a different folder, and run it in tandem with your original. Priority ten should be used for both instances of Prime95 in this case.

When you are stability testing with Prime95, you want to run the Torture Test for at least 24 hours. Why 24 hours?

There is a very common misconception that if your machine can pass Prime95 stability testing for, say, four hours, your machine will be able to run stable, regardless of what you are doing, for four hours as well, without issue. This is simply not the case.

Prime95 often finds errors in its 16th - 20th hour of testing, a potential for instability that wasn’t found after only four hours of testing. After only four hours of Prime95, the potential for instability still exists. 24 hours is widely viewed as a sufficient time period to catch any instability that may be present, but by all means test longer if you are able.

If 24 hours seems like an extraordinarily long time to leave your computer on, keeping your machine unusable because of the processes it’s doing, try running Prime95 overnight, and then through to all day while you’re at school or work.

For CPU specific testing, a Large FFT Prime95 test is an alternative to the more "system-stress" oriented Blend test that runs by default. The choice is ultimately up to the end user - the Blend test is reccomended in these guidelines because of it's qualities as both a processor and system stress test.

That’s all there is to Prime95 – 24 hours of Prime95 at Priority ten is "certified stable", and ready to rock for 24/7 use.


EXCELLENT PROCESSOR STRESS TESTING ALTERNATIVES



Distributed computing

A lot of people run distributed computing programs 24/7, which constantly keep their CPU at full stress load while helping to aid important scientific research.

Running a program like Folding@Home or SETI after passing 24 hours of Prime95 is a superb way to constantly keep an eye on your system's stability, while also aiding Medical research, or the search for extraterrestrial life.

It is imperative that one does not run a distributed computing program on a machine that hasn't been stability tested, as mistake-filled results do not in any way help the effort.

For more detailed information about these distributed computing programs, check out the ocforums SETI and F@H team forums.

F@H team forums: http://www.ocforums.com/forumdisplay.php?f=21
SETI team forums: http://www.ocforums.com/forumdisplay.php?f=18

StressPrime2004 Orthos

A new program called StressPrime2004 Orthos, based on the Gromacs core (which is used with F@H) and designed for CPU stability testing, is a highly viable alternative to Prime95. The Gromacs core is known to be very stressful on an overclock, this program should be very effective.

You can download StressPrime2004 Orthos here: http://sp2004.fre3.com/beta/beta2.htm

SuperPi

SuperPi, a Pi calcation program, is another excellent and widely proven CPU stess tester. The '32M' length benchmark is an easy-to-use indication of stability, and is usually quite accurate. A SuperPi software mod that can be looped indefinitely is in the works - such a program will be ideal for long-term processor stability testing.

SupePi can be downloaded here: http://files.extremeoverclocking.com/file.php?f=36

PiFast

Another Pi calculation benchmark program.

PiFast can be downloaded here: http://numbers.computation.free.fr/C...ram/pifast.zip


CPUBurn

Another processor stress tester like 'Toast', designed to maximize temperature and power draw.

CPUBurn can be downloaded from the project page here: http://pages.sbcglobal.net/redelm/


OCCT

This program is really amazing. Made by a group French overclockers, with the intention of being an enthusiast-oriented stress-tester, OCCT is perhaps the very future of processor stability testing. This program is extremely effective - it will peak your processor temperature at a full workload, and in a relatively (compared to Prime95) brief period of time give a very thorough indication of processor stability.

OCCT can be downloaded here: http://www.ocbase.com/perestroika_en/index.php?Download


Toast

Toast is a temperarure-increasing processor stress tester that will take your chip right up to a peak load temperature and keep it there. Toast is a good indication of long-term stability, and excellent for testing the capacity of one's cooling.

Toast is available here: http://www.majorgeeks.com/download.php?det=867


STEP 2: 3D STABILITY – 3DMARK


When it comes to stability, 3D testing is often overlooked completely. Many people do not overclock their video cards, and as such decide that 3D stability tests are a waste of their time. They couldn’t be more wrong! An increasingly common phenomenon is an overclock which will pass Prime95 and memtest86 for 24 hours, but lock or crash 3DMark in a few minutes – it is for this reason that 3DMark testing is a good idea. And for those who do overclock their video cards, 3DMark is an invaluable video card stability testing tool.

Quote:
From 3DMark 2001 SE Readme

3DMark2001 SE is a diagnostics tool for measuring the 3D game performance of PCs. It is entertaining and easy to use, which makes it "must have" software for all home PC users interested in 3D games. Even a beginner PC user can get a game performance measurement with 3DMark2001 SE. For the more advanced users, 3DMark2001 SE offers a wide range of display settings and testing options for the benchmark run.
You can download 3DMark 2001 SE here: http://www.futuremark.com/download/?3dmark2001.shtml
You can download 3DMark 2003 here: http://www.futuremark.com/download/?3dmark03.shtml
You can download 3DMark 2005 here: http://www.futuremark.com/download/?3dmark05.shtml
You can download 3DMark 2006 here: http://www.futuremark.com/download/3dmark06/

HOW TO USE 3DMARK EFFECTIVELY


Using 3DMark as an effective stability test is a little bit different from using it to bench your machine. There are two main ways to effectively use 3DMark as a stability tester. Running all of the tests looped in order to test your entire machine, or running only the Nature test (from 3DMark01 or 3DMark03) looped while checking visually for artifacting in order to test your video card at peak load temperatures.

Artifacting, or 'snow', is a term for visual erroring, almost always caused by graphics card instability. Artifacting most commonly makes itself apparent through large geometric objects flickering in and out on your screen, texture corruption (checker boarding), and 'texture snow', which appears as many white specks.

After installing 3DMark, run the program, and select all the 'game' tests for use. Now set the benchmark for 'looping', so that all the tests you have selected will loop indefinitely. With looping, the benchmark will not stop after one run, which will enable you to stress test with 3DMark for several hours. Next press CTRL+ALT+DELETE, and go to the Task Manager. Right-Click on the 3DMark icon in the ‘applications’ tray, and select ‘Go To Process’. Right click on the process you are taken to, and select ‘Set Priority’. Set the Priority to ‘Realtime’. This needs to be done every time you use 3DMark for stress testing.

Giving the program 'Realtime' priority effectively gives all of your system resources to 3DMark when it is running – any processes running in the background won’t steal resources from 3DMark while it’s stress testing, ensuring the most thorough test possible.

Make sure any screensavers are turned off, and that no other applications are running, and start running the tests when you’re ready.

You should always run 3DMark by itself, never at the same time as any other stress testers. This ensures that it’s doing the most thorough job that it can.

Because 3DMark will almost always lock within the first ten test run loops, if it’s going to lock at all, 4 hours of looped testing is a more than sufficient test for your machine. After four hours of looped testing, 3DMark has done all it really can, and likely isn’t going to catch any instability.

Quote:
From Above

An increasingly common phenomenon is an overclock which will pass Prime95 and memtest86 for 24 hours, but lock or crash 3DMark in a few minutes – it is for this reason that 3DMark testing is a good idea.


The reverse is also often true; many machines can run 3DMark for hours and hours without issue, but will fail Prime95 or memtest86 after a short period of time; 3DMark isn’t particularly useful as a system stability test when used by itself.

Other tests in newer versions of 3DMark can also be very effectively used to detect artifacting and video card instabilty, although the Nature test in 01 and 03 is often regarded as the best at getting your video card's core and RAM temperatures to a peak, and keeping them there.


GRAPHICS ADAPTOR STRESS TESTING ALTERNATIVES


ATiTool

A useful program with built in overclocking features, and a built in stress tester. ATiTool is not always entirely accurate, and is known to pass overclocks that are not actually stable when applied to games or 3D Benchmarks.

ATiTool is useful though, especially with ATi video cards.

ATiTool can be downloaded from the project page here: http://www.techpowerup.com/atitool/


STEP 3: MEMORY STABILITY – MEMTEST86 / MEMTEST86+


Memory instability is perhaps the stability-aware overclocker’s second worst nightmare, because of the problems it can cause. Luckily for all of us, this superb little program exists. It’s idiot-proof, really small, and fantastically effective at sleuthing out any memory instability. Because memtest86/memtest86+ tests as much of your RAM as is possible, it’s a stand-alone program, meaning it needs to be run outside of your OS.

Please keep in mind that memtest86 can still error even with a completely stable memory system/subsystem. An unstable processor, or other system problems can also cause erroring in memtest86 - your memory is not the sole potential cause of memtest86 testing errors.

Quote:
From memtest86 Readme

There are many good approaches for testing memory. However, many tests
simply throw some patterns at memory without much thought or knowledge
of the memory architecture or how errors can best be detected. This
works fine for hard memory failures but does little to find intermittent
errors. The BIOS based memory tests are useless for finding intermittent
memory errors.
There are two "versions" of memtest86 out there; memtest86+, and memtest86. Both have been updated and improved on fairly recently, and both appear to be "living" programs that are still seeing fairly regular update and improvement by their respective designers.

The choice is really up to personal preference.

You can download memtest86 here: http://www.memtest86.com

You can download memtest86+ here: http://www.memtest.org/


HOW TO USE MEMTEST86 / MEMTEST86+ EFFECTIVELY


It’s fantastically easy. A quote from the memtest86 Readme:

Quote:
From memtest86 Readme

memtest86 is a stand alone program that cannot be executed under windows and must
be loaded from a floppy disk.

To install Memtest86:
- Extract the files from the zip archive
- Open the directory where the files were extracted and click on "install.bat".
- The install program will prompt you for the floppy drive and also prompt you to
insert a blank floppy.
- To run Memtest86 leave the floppy in the drive and reboot.

NOTE: After the boot floppy has been created you will not be able to read the floppy
from windows. This is normal.
Note that memtest86+ can be run from a CD as well, via a pre-made, bootable, ISO image that you can download from the memtest86+ website, and then burn onto a blank CD. memtest86+ can still be installed on a floppy if one so desires, exactly as above.

After installing memtest86/memtest86+ onto a floppy, simply reboot your computer with the floppy still in the drive. memtest86 consists of 11 different tests (Note: The newest version of memtest contains only 9 tests, with the additional tests having been removed). By default, tests #1 through #7 will run in order, endlessly. For our purposes, this is perfect. Once memtest86/memtest86+ starts running, simply leave it, and let it go for 24 hours.

As with Prime95, 24 hours really is required for a complete and thorough memtest86/memtest86+ stability test, and for the exact same reasons. I’ll quote myself for reference sake.

Quote:
From Above

When you are stability testing with Prime95, you want to run the Torture Test for at least 24 hours. Why 24 hours?

There is a very common misconception that if your machine can pass Prime95 stability testing for, say, four hours, your machine will be able to run stable, regardless of what you are doing, for four hours as well, without issue. This is simply not the case.

Prime95 often finds errors in its 16th - 20th hour of testing, a potential for instability that wasn’t found after only four hours of testing. After only four hours of Prime95, the potential for instability still exists. 24 hours of Prime95 is a slight ‘overkill’, but you can never be too careful. 24 hours is widely viewed as a sufficient time period to catch any instability that may be present, but by all means test longer if you are able.
As with Prime95, it’s easiest for most of us to run memtest86/memtest86+ overnight, and then the following day, so that it’s as un-disruptive as possible.

Many people also use specific memtest86/memtest86+ tests by themselves, to test out a new FSB or memory overclock quickly in order to see whether it’s likely to be stable or not. Tests 5 and 6 in particular are very good for this. However, 24 hours of all the tests on loop is your end-all solution to memory stability testing.


TEST SETUP CLARIFICATION PICTURES


Below I have attached pictures, which should help make clear how to set up Prime95 for priority 10 stress testing, and 3DMark at 'Realtime' Priority.

Some pictures of different types of visual artifacting could be added as well, to give an idea of what to look for during 3D artifact testing. If anyone has a few really clear pictures of different types of artifacting, preferrably from one of the 3DMark builds, please PM me - I'll add them in.
Attached Images
  

Last edited by felinusz; 05-03-07 at 11:11 PM.
felinusz is offline   QUOTE Thanks
Old 10-12-04, 11:05 PM Thread Starter   #3
felinusz
Senior Overclocking Magus

 
felinusz's Avatar 

Join Date: Feb 2003
Location: Taiwan

 
CONCLUSION

I really, really hope that this little guide has been helpful.

If you’ve read through the entire thing, and found it educational, then it has succeeded. If you read only the beginning, and learned a little bit about the dangers of instability, then it has succeeded.

The big goal here is for everyone to be aware of the potential for stability problems, and know how to test for them properly. This is an unrealistic goal, but every person who learns something about stability, and proper, thorough, stress testing, is one less potential victim of stability-related disaster.



I’m thinking about adding a little bit of information on the philosophies of stability, the different schools of thought on the issue, and different courses of action people take. I’m also thinking of making a little list of some of the ‘lesser’ stress testing programs, and where to get them – Some of them are very worthwhile, if inferior substitutes to the ‘Big Three’. I’ve left out really specific stress testing, such as that designed to stress test your hard-drive. The ‘Big Three’ do quite a thorough job of testing an entire system if used properly, and together, and I don’t find it pertinent to clutter an extremely long, but basically simple guide like this with arguably unnecessary information.

One thing this guide is completely devoid of is solutions to instability. This is intentional, as lots of material exists that will aid anyone running an unstable machine in getting it up to par.

If I have missed anything important, worded something really poorly, or not gone as in-depth as I should have on some issue, please PM me.

If anyone has any questions or comments please do take the time to post, or drop me a PM.

Last edited by felinusz; 10-16-04 at 04:32 PM.
felinusz is offline   QUOTE Thanks
Old 10-12-04, 11:15 PM   #4
I.M.O.G.
Glorious Leader

 
I.M.O.G.'s Avatar 

Join Date: Nov 2002
Location: Rootstown, OH

10 Year Badge
 
Nice from what I have read so far, well done. Looks like it took you a while. You've got PM - get on AIM and IM me if you have it.

__________________
The OC Forums Way
We are a team. We are a community. We are a fellowship made strong by mutual respect and shared dedication to the task of enriching all who come here.
The OC Forums Thank You Thread
Put your computer to work for our OC Forum Teams!
Try out our POST TEMPLATES, they save you time answering common questions!

I spend half my money on CPUs, GPUs, and Liquid Nitrogen. The other half I waste.
I.M.O.G. is offline Author Profile Benching Profile Folding Profile Heatware Profile   QUOTE Thanks
Old 10-12-04, 11:23 PM   #5
enduro
Member

 
enduro's Avatar 

Join Date: Mar 2004
Location: Montgomery Alabama Yall

 
Sticky

Good job m8. That was a very good article/ sticky.

Just might add one thing. Prime doesn't work on all systems and before I here gasps of amaze and booing about my having a hard time with a bad overclock, let me explain myself. I "had" , sold now, some ram that would run up to 233 prime stable, 2*512 sticks of EB. Now I have Kingston VR 2*256 BT D43 and it won't pass prime at 220, even though everything else will pass. I can run it stock, and it will still fail. I can up voltages and loosen timings till doomsday, and the time that Prime fails in is always 5 mins.

Given that, I've decided that Prime doesn't like my ram. I've run Memtest, 3DMark 01,03, and 05 perfectly. I've run Burn in Test, Hot CPU Tester, Superpi, and numerous other programs that test for errors, and nothing has come up.

I just wanted to make that clear for some people. Now that doesn't mean that you can throw Prime out the window because you don't like the fact that it errors at you when you are overclocked a lot and it didn't before. I tried that once, and I had to reinstall Windows, and that hdd is still acting really weird. SO don't do it.

Also, I've been pushing my system, which when I can't run prime is difficult. I'm trying to get to 260 right now, so I'm running 260*10, 1.9V, 3-4-4-11 at 2.8V, 1.9V vdd, ACPI off, fast write off, normal stuff off. It will pass everything, but 3Dmark 01, which I use for absolute 3D stability, because it works so well. I get about half way through, and then I get an error message. I've run Memtest and the others to test for errors, but I can't find any, and raising voltages etc doesn't help.

Do you have a faint clue to what my wall is right now? I'm racking my brain to think of different combinations to try, but I can't fine any that work. Thanks man, and this definitely earns you a STICKY. :two thumbs up:
enduro is offline   QUOTE Thanks
Old 10-13-04, 12:13 AM Thread Starter   #6
felinusz
Senior Overclocking Magus

 
felinusz's Avatar 

Join Date: Feb 2003
Location: Taiwan

 
I couldn't help but notice that you have an NF7-S motherboard; have you tried using a different BIOS, and running everything at stock to see if Prime is still having issues with your machine?

I've had a few issues with different BIOSes behaving strangely for me with my board, in particular the various modified D10s (although you aren't using one of these, it's worth a shot). My BIOS-related problems have always been centered around 3DMark, and general 3D stability however, not Prime95.

Have you tried Priming with *everything* at stock, not just your RAM?

If your machine is passing memtest86, the various 3DMarks (2001 SE being the best to use), and multiple other CPU stress testers for appropriately long time periods, I'd be inclined to say that you're stable .

If you are having seemingly sourceless issues with Prime95, which some people do, a little ap that gives very good peak load temperatures called Toast is a pretty good alternative . Be warned however, that it'll get your chip hotter than pretty much anything else will. Here's a link if you're interested: http://users.pandora.be/bull_b2oc/toast.exe


Quote:
posted by enduro

Do you have a faint clue to what my wall is right now? I'm racking my brain to think of different combinations to try, but I can't fine any that work.
You're at a 255 MHz FSB, which is impressively insane for a Socket A setup . Many people find that with extremely high FSB overclocks, things just seem to start messing up for no apparant reason. Whether it's a voltage supply (remember that your board is just as responsible for your voltages as your PSU, and NF7s are notoriously bad in this department) issue, or simply the limits of your board's chipset is up in the air. As an example; my board (NF7-S), chip, and RAM can Prime and pass memtest86 all day with a 225+ MHz FSB and the right BIOS, but anything over 211 MHz invariably locks 3DMark for me. I've tried pretty much everything, with no success.

Sometimes, we've simply walled.

Quote:
posted by I.M.O.G

Nice from what I have read so far, well done. Looks like it took you a while. You've got PM - get on AIM and IM me if you have it.
Thanks for the positive feedback guys .

I've got your PM I.M.O.G., and have hit you back as well


Last edited by felinusz; 10-13-04 at 12:21 AM.
felinusz is offline   QUOTE Thanks
Old 10-13-04, 02:17 AM   #7
DayUSeX
Member

 
DayUSeX's Avatar 

Join Date: Aug 2003

 
awesome job, though the only thing i think u should add, is that the best stability test, once you have primed is fah. Of course for team 32, but i only truley deam my systems stable, when i can have after fx, premier, photoshop, 5 instance of firefox, some music, some chat, ggoing as well as 4 instances of fah and my system dosent crash. It can be on for weeks strait with no problems, that is what i call stable. Some of my amd machiens have passed prime for 24 hours, but did not last past 3 days when being used with fah running full bore. otherwise excelletn job!
DayUSeX is offline   QUOTE Thanks
Old 10-13-04, 07:13 AM   #8
enduro
Member

 
enduro's Avatar 

Join Date: Mar 2004
Location: Montgomery Alabama Yall

 
^ Ditto. I've tested higher fsb's, and I have a cpu usage meter from MBM running. When Windows boots up, and everything is loading, I could watch it hit 100%, and then my computer would crash. It's an awesome test. And thanks for TOAST, I downloaded it, and I'm going to try it here pretty soon.
enduro is offline   QUOTE Thanks
Old 10-13-04, 07:39 AM   #9
jbarket
Member



Join Date: Aug 2004
Location: Little Rock, AR

 
Great thread

Definitely sticky material.

I've only got one suggestion, and that's to use a newer version of 3D Mark for video testing. 3D Mark 01/01SE is very CPU intensive as well. Newer versions like 03 and 05 put a lot more hurt on the video card, and it's a lot easier to see artifacting.

If your card doesn't artifact or completely checker out in Return to Proxycon in 05, I'd be amazed if it had issues.

__________________
Lian Li PC-60
Athlon64 3200+ [Winchester] @ 2.45Ghz
MSI K8N Neo4 Platinum/SLI
eVGA nVidia Geforce 6800GT @ 410/1105
jbarket is offline   QUOTE Thanks
Old 10-13-04, 07:50 AM   #10
musawi
Member

 
musawi's Avatar 

Join Date: Jul 2004
Location: Bahrain

 
^^ Agree with what he says

"Definitely sticky material."
musawi is offline   QUOTE Thanks
Old 10-13-04, 07:50 AM   #11
ColtIce
Disabled

 
ColtIce's Avatar 

Join Date: Nov 2002
Location: 9 months 43202 summer months 98951

 
nice thread.well thought out and organized.should be stuck!
ColtIce is offline   QUOTE Thanks
Old 10-13-04, 08:16 AM   #12
I.M.O.G.
Glorious Leader

 
I.M.O.G.'s Avatar 

Join Date: Nov 2002
Location: Rootstown, OH

10 Year Badge
 
Okay, we can chill on the "sticky" posts. It's currently being worked on; I'm talking with felinusz via PM and you will likely see it stuck before long.

__________________
The OC Forums Way
We are a team. We are a community. We are a fellowship made strong by mutual respect and shared dedication to the task of enriching all who come here.
The OC Forums Thank You Thread
Put your computer to work for our OC Forum Teams!
Try out our POST TEMPLATES, they save you time answering common questions!

I spend half my money on CPUs, GPUs, and Liquid Nitrogen. The other half I waste.
I.M.O.G. is offline Author Profile Benching Profile Folding Profile Heatware Profile   QUOTE Thanks
Old 10-13-04, 12:06 PM Thread Starter   #13
felinusz
Senior Overclocking Magus

 
felinusz's Avatar 

Join Date: Feb 2003
Location: Taiwan

 
Thanks for the suggestions!

I've added some basic information on F@H and SETI, as well as using newer versions of 3DMark as video card stress testors, along with an appropriate link for each program.

Quote:
3D Mark 01/01SE is very CPU intensive as well.
The reason I focused specifically on 3DMark01SE, and not the newer versions, is because of this, as well as the massive file-size of 3DMark05, making it all-but unnaccesable to 56k users .

The other conveniently great thing about 3DMark01SE is the Nature test, which does a better job than anything else out there at getting your video card's GPU and RAM right up to peak load temperatures. Looping the 3DMark01SE Nature test is the Graphics Card equivelent to Toast.

Quote:
though the only thing i think u should add, is that the best stability test, once you have primed is fah.....

.....Some of my amd machiens have passed prime for 24 hours, but did not last past 3 days when being used with fah running full bore.
A very good point there .

Erroring after a 24 hour Prime test period is certainly not unknown, but pretty uncommon. It's also most often attributed to environmental changes over that time, I.E. a slight change in room temperature, or a voltage irregularity.

F@H is a very noble use of clock cycles as well, and everyone who is able to run it, really should be running it

felinusz is offline   QUOTE Thanks
Old 10-13-04, 12:27 PM   #14
jaydog
Member

 
jaydog's Avatar 

Join Date: Jul 2004
Location: Ontario

 
Thanks dude that rocks. I had know Iidea that I was running prime wrong. but I guess if I passed at priority 1,I should have no prob at ten. If thats true maybe I can make it further. This Sticky should me manditory for every member

OK.. soon to be Sticky(I give it an hour)
jaydog is offline   QUOTE Thanks
Old 10-13-04, 12:48 PM   #15
mtb856
Member

 
mtb856's Avatar 

Join Date: Mar 2004
Location: Helping young Meatballs grow...

 
Quote:
Okay, we can chill on the "sticky" posts. It's currently being worked on; I'm talking with felinusz via PM and you will likely see it stuck before long.
:whispers: sticky

Anyyyway... excellent guide felinusz, it's answered a few questions about stability that newbies asked me that I didn't have an answer for, so it's proved its worth to me right there. Keep up the good work

Meatball
mtb856 is offline   QUOTE Thanks
Old 10-13-04, 04:25 PM   #16
enduro
Member

 
enduro's Avatar 

Join Date: Mar 2004
Location: Montgomery Alabama Yall

 
Hey IMOG, any chance that we could link this to our sigs? This would be the perfect thread for it.
enduro is offline   QUOTE Thanks
Old 10-13-04, 05:38 PM   #17
mdcomp
Classic Moderator
Overclockers.com Lead Editor


 
mdcomp's Avatar 

Join Date: Nov 2001
Location: New Jersey

10 Year Badge
 
Quote:
Originally Posted by enduro
Hey IMOG, any chance that we could link this to our sigs? This would be the perfect thread for it.
You can, sure. You can put links to forum threads in your sig.


Mod notified to make this a sticky.
Nice work! Awesome.

Matt

mdcomp is offline Author Profile Benching Profile   QUOTE Thanks
Old 10-13-04, 09:09 PM   #18
jaydog
Member

 
jaydog's Avatar 

Join Date: Jul 2004
Location: Ontario

 
Ok due to this thread(thanks) I ran prime x2 today while I was at school. I came back 3 hours later and I was still on test 1 for both(no crash). Has this ever happened to anyone. Is this normal? Is there something wrong with my ht?
I'm going to run them tonight when I go to bed and see what I get tommorow. very strange
jaydog is offline   QUOTE Thanks
Old 10-13-04, 09:14 PM   #19
enduro
Member

 
enduro's Avatar 

Join Date: Mar 2004
Location: Montgomery Alabama Yall

 
What else do you have running? Are you folding or running some other program? Set Prime to 100 % usage, according to what was said in the guide. Use the code, unlock it, and then set priority to 10. Once you try that, get back to us if it didn't help.
enduro is offline   QUOTE Thanks
Old 10-13-04, 09:35 PM   #20
jaydog
Member

 
jaydog's Avatar 

Join Date: Jul 2004
Location: Ontario

 
nothing else running except asus probe set to poll every 60 sec. I typed the password and set my prority to 10. I ran 2 blend tests neither of which left test 1 in 3 hours so i closed them down when I got home. At least if it stopped or crashed I would know what to do. Any other ideas?
jaydog is offline   QUOTE Thanks

Post Reply New Thread Subscribe


Overclockers Forums > Hardware > General Hardware
General Hardware Discuss your new build and gather upgrade suggestions. And if your question doesn't fit anywhere else...it goes here.
Forum Jump

Thread Tools Search this Thread
Search this Thread:

Advanced Search


Mobile Skin
All times are GMT -5. The time now is 02:18 PM.
Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.
You can add these icons by updating your profile information to include your Heatware ID, Benching Profile ID or your Folding/SETI profile ID. Edit your profile!
X

Welcome to Overclockers.com

Create your username to jump into the discussion!

New members like you have made this the best community on the Internet since 1998!


(4 digit year)

Why Join Us?

  • Share experience
  • Max out your hardware
  • Best forum members anywhere
  • Customized forum experience

Already a member?