Having Stability Problems? Try these first before you give up...

[curio]

Here are some tricks that I have learned over the years to help make your machine the most stable it can be while Folding.

First of all, remember that dust is one of your biggest enemies! The first thing you should do is get all the dust out of your computer. Remember to keep those computers clean - because when you are running your CPU at full load most of the time, it doesn't take that much restriction\insulation to start giving you problems.

I've personally seen the CPU temperature in a very dusty computer drop 40 degrees celcius just by cleaning out the dust, albiet it was a very dirty computer.


Ok, now that your computer is clean, follow the steps below:

1) First, check your machine's capabilities with the torture test on Prime95 or try out StressCPU. Either one of these programs will find weaknesses in your rig better then Folding could. (In other words, if you pass these tests, then Folding shouldn't be an issue). These programs are a must if you overclock.

-Click here to Download Prime95
-Click here to Download StressCPU

Instructions for using Prime95 can be found on the link below, but to put it simply, just run the torture test:

http://www.mersenne.org/freesoft.htm

You should also test your memory for instability. Here is a good program to do so:

Memtest86

...and make sure that your power supply unit (PSU) is not failing! A faulty PSU can cause many headaches.

2) Check your BIOS! If you are overclocking, make sure your voltages are set correctly, and if you have to, just decrease your CPU/FSB a little. You'd be surpised how much stablity you can gain from decreasing your FSB a couple MHz. Make sure to check your memory timings also! Even though Fah is a CPU dependent program, tight memory timings can easily give you lockups. Try relaxing your timings a bit - it could save you a lot of problems.

3) Another thing I'd like to add is to make sure and check your voltages and FSB speeds in REAL TIME. I've found that many motherboards will tend to undervolt what you set in the BIOS, and many will also raise the FSB by a MHz or two. These factors can easily cause system instability, especially when your machine is running 100% 24/7 in extreme environments.

My current Abit motherboard undervolts my CPU by up to .1 v, and also sets my FSB about 2MHz higher then what I set in the BIOS. Remember that your voltage and FSB/CPU speeds vary, as it is a current and a frequency - so you want to get your average in real time and do not rely on what you set in the BIOS.

Try a couple different programs from around the internet and compare them to each other to get a good idea of your real time voltage and FSB speed. One good program is CPU-Z:

Click here to Download CPU-Z

4) If you still are having stability issues while folding, you can always set Folding to use less of your processor. Just delete your 'client.cfg' file (found in your FAH folder), restart the Folding program, and re-enter your data again. Say 'yes' to 'Change Advanced Options' and this time around set your 'Use CPU Speed 0-100' to around 85 or 90. This will tell Fah to only use 85-90% of your processor. This is a quick and easy workaround for those that are having stability issues.


5) If you added any extra switches to the Fah program (like -forceasm), try removing these switches from the Fah command line.


6) Always check your Fahlog! It is found in the directory where the folding program is, and it will tell you what kind of errors you are getting. Follow this link for some common errors and their explanation: http://forum.folding-community.org/viewtopic.php?t=4824


Overall, it comes down to achieving the most stable machine that you can get, and then worry about the performance you can get from it second.

Remember - a processor running only at 85% 24/7 is better then a processor at 100% that is locked up for 10 hours of the day. Anything you can do to get your machines running 24/7 without lockups will produce better and give you less headaches in the long run.

The goal here is to get people not to give up so easily just because their computer crashed when running Fah. I hope to help be a part of a cure someday, and the more computers that we have running stable, the closer we are to that goal.

Take care and Fold on!

-Troy

[Origin]

very good suggestions from the General.

[Lupine]

Good info sir curio!

This should be a sticky.

[curio]

Another thing I'd like to add is to make sure and check your voltages and FSB speeds in REAL TIME.

I've found that many motherboards will tend to undervolt what you set in the BIOS, and many will also raise the FSB by a MHz or two. These factors can cause system instability for sure, especially when your machine is running 100% 24/7 in extreme environments.

My current Abit motherboard undervolts by up to .1 v - which is a lot. Remember that your voltage and FSB/CPU speeds vary, as it is a current and a frequency - so you want to get your average in real time and do not rely on what you set in the BIOS.

Try a couple different programs around the internet and compare them to each other to get a good idea your real time voltage. One good program is CPU-Z:

http://www.cpuid.com/download/cpu-z-120a.zip

[curio]

I was wondering if a mod (jima / Jason) could put a link to this thread in the 'Folding@Home quick links and info' thread?

There is no need to make it a sticky, but I think it is helpful and might deserve to be in the quick links post.

Thanks
-Troy

[citywok]

will do

[jima13]

Originally posted by Parisioa
will do

TY