what do u consider stable?

[alibubba]

Hey fellas,

I bought my 2500+ on newegg a week ago or so, I dont remember the stepping, just remember the week, it was 0350 if anyone's interested.

My question is, what would u guys consider a "stable" system?

prime95 likes to crash at 11x220 1.85v, (42-43 load? can't remember, and it was also a cold night last night, so made my room even cooler than usual.)

now there's only 1 benchmark i like to run to consider before i declare my system "stable" and that's my online game, the evil Everquest. Everquest is just a hoarder when it comes to cpu usage, and if my game doesn't crash/reboot itself, i consider that victory and have peace in mind that nothing could go wrong with my system...(knock on wood).

ANYWAYS, the question really is, if prime crashes, what is the likelyhood that something else would go wrong? ala data corruption, random reboots, a nice BSOD...etc.

some settings for the curious: (i haen't updated my sig yet...)
nf7-s rev 2.0 bios 18
2500+ 0350
3200 Giel
water

11x220 1.85v 2.7v / 2.8v ram
Prime = crash / crash
eq = good / good
43-45 load.

11x220 1.8v 2.8v ram
Prime = crash
eq = good
43-45 load.

11x220 1.95v 2.7v ram
prime = random errors here and there
eq = didn't test
47 load (a little high for my liking)

11x215 1.825v 2.7v
prime = crash
eq = good
42 load

idle temps range from 36-38.

[blkgti]

To me, stable means that it can run anything without a problem. I usually test with at least 24 hours of Prime95 (no errors, etc), several hours of 3DMark03, and at least 8 hours of running HOTCpu, Prime, Toast, and Sanda burn-in simultaneously. To be perfectly honest, I wouldn't consider your machine to be stable.

[bafbrian]

Originally posted by blkgti
To me, stable means that it can run anything without a problem. I usually test with at least 24 hours of Prime95 (no errors, etc), several hours of 3DMark03, and at least 8 hours of running HOTCpu, Prime, Toast, and Sanda burn-in simultaneously. To be perfectly honest, I wouldn't consider your machine to be stable.

Exactly what he said. For me, I like to use CPU Burn-In at around 2.06Volts at the stock clock. Burning-In the processor should allow you to use less voltage so that you can run the proc at 11*220. I have an extra Barton here and I have it at 11*220=2.4 with only a voltage of 1.75(1.775 in bios, NF7-S undervolts it). But it is ROCK SOLID STABLE with Prime95.

Before you start really OCing it, I would suggest that you burn it in at whatever voltage you feel comfortable. Burning-In has allows many a member here to increase their and also lower their overall voltage needed for the OC. Burning-In that proc "should" allow you get a STABLE OC of 11*220. But, YMMV depending on setup.

[felinusz]

Real stability means 24 hours or more of Priority Ten Prime95 on Blend, 8 hours or more of 3DMark01 at "High" Priority, 24 to 48 hours of memtest 86 on test # 11, and 4 hours+ of Toast.

I also like to do a mixed test using an N64 emulator with FPS limitation off (PJ64 and Conker's Bad Fur Day :attn: ), Prime95 on Priority 5, and HotCPUTester Pro.

The times I wrote about are just "guidelines"; Try to test for as long as you have time to test for (if you use your rig a lot just leave it on overnight :)).

An unstable overclock isn't any good to anyone, problems do, and will occur (which is why 48 hours of memtest test # 11 isn't overkill) if your machine is unstable, and the feeling of accomplishment which comes with a Crazy Hardcore Overclock that also happens to be absoutely stable is well worth the time it takes :).

[alibubba]

Alright guys, i'll try what you suggested, I burned it in but it was only for 6 hours at 1.9 and another 6 at 2.0 using only prime (i got excited and wanted to play with the new cpu) but i dind't set the priority. I'll have to find a copy of 3DMark01 but i've never heard of some of these other programs before.

i got nervous setting my vcore at 2.0 just because i've never set it that high before but now that i think about it, my temperatures were 39 load so i think i'll give it a go tonight.

thanks.

[felinusz]

Vcore isn't dangeroua so long as your temperatures are low (below 40 degrees celcius while burning in, preferrably) - when burning in (a questionable practice - it works for some, but not for others) set your processor to the lowest speed possible (5 X 100, or 6 X 100), and raise the Vcore as high as possible, then run Prime95, and at least one other program that involves 3D graphics - the idea being that you are "working in" all the "parts" pf the processor which "do different things" - again, burning in isn't proven at all (Never worked for me, but I do it anyways).

If you've already got Prime95, all you *really* need is "memtest86" memory stability testing (a google search will net it), and 3DMark01 (Googe it for different servers, the Guru of 3D website has a confirmed uncorrupted file and fast transfer speeds).

[alibubba]

thanks for the feedback felinusz, i'll keep that in mind for tomorrow.

I got a general question for you prime users though, how do i set priority? it's greyed out...

[jonspd]

For me

1. about 2 week's of prime95 going up on vcore form 1.8 to 2.0 in the smalliest increments allowed while uping the mhz of the cpu changeing fsb and muilt on the way then when I get my max I run.

24 hour's of prime95
12 hour's of 3dmark2001
6 hourr's of 3dmark2003
May run toast for a few hours also.

Last but not least prime95 and 3dmark2001 runing high prority for 24 hour's.



EDIT password for prime is 9876

[ChillPhatCat]

The question I pose to anybody that spends so much time testing the stability... what are you using your comp for?

Is it really necessary to run a torture test for 2 days? The processor makes mistakes all the time that we don't notice... it is ineviteable... the question is how many mistakes are acceptable... well unless you've got a database or its some important business machine ie a server then some mistakes equating to an occasional crash every few months is perfectly acceptable. If it's all that important then you probably shouldn't be overclocking that computer anyhow...

2 hours of prime95 is plenty for 99% of the population.

[sQUiRLy]

I'm going to have to agree with ChillPhatCat on this one except for the last comment. It should be known that 2hrs of prime95 is no-way considered stabile. At the minimum, stability tests should be run long enough to bring all the hardware up to maximum temps and then span across the coldest and warmest parts of that particular day. That would roughly be around 13-14hrs. I've errored in the prime95 torture test after 13hrs1min so for me its about 14-15hrs before I declare system stability. :)

[CrashOveride]

I run prime for a lil while if it crashes it's unstable, I run 3dmark a few times, if it crashes it's unstable, then I play a lot of games and if they crash it's unstable. Thats good enough for me.

[simpleman]

ido prime for 12 hrs. normally, but most of the time i will start it before bed then let it run all day wile at work also.
so thats around 17 hrs.

[maxima88]

You know what Alibubba. If you use your pc primarily for games and your pc doesn't crash or reboot while playing games, then you're good to go. It would be nice to pass prime95 for 48hours and run 3dmark2001 & 2003 for 24hours and other proggies to prove 100% stability but in your case, if your pc runs flawlessly with what you use it for, I wouldn't worry about it if I were you.

Enjoy your pc.

[sQUiRLy]

Originally posted by maxima88
You know what Alibubba. If you use your pc primarily for games and your pc doesn't crash or reboot while playing games, then you're good to go. It would be nice to pass prime95 for 48hours and run 3dmark2001 & 2003 for 24hours and other proggies to prove 100% stability but in your case, if your pc runs flawlessly with what you use it for, I wouldn't worry about it if I were you.

Enjoy your pc.

Thats pretty much how the creaters of prime95 put it:
"CAN I IGNORE THE PROBLEM?
-------------------------

Ignoring the problem is a matter of personal preference. There are
two schools of thought on this subject.

It is likely that most programs you run will not stress your computer
enough to cause a wrong result or system crash. A few games stress your
machine and a system crash could result. Stay away from distributed
computing projects where an incorrect calculation might cause you to
return wrong results. You are not helping these projects by returning
bad data! In conclusion, if you are comfortable with a small risk of
an occasional system crash then feel free to live a little dangerously!

The second school of thought is, "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." -prime95 stress.txt

To each his own I suppose.

[dubtek]

I had the same problem with prime it would only run 5 minutes then quit with a error never crash just quit the test...I tried changing EVERYTHING the only way it would prime was at stock settings and voltage soooo finally I reinstalled windows and re installed prime and holy ***** its stable I dunno what was causing it...even with it erroring on prime I never crashed and I had it like that for quite a while...

[seadog]

For me, at least 12 hours of prime at priority 10, large FFT (more heat & power consumption) to be considered stable. I use memtest to check ram stability. Usually re-test if the ambient temps increase significantly.

how do i set priority? it's greyed out...

Open Prime, click the Advanced tab > Password is 9876 - this opens the priority option. I think it's in the read-me somewhere.

[DaWiper]

A stable system for me is a system that doesn't lock up no matter what I do.
If your into folding then you NEED a 'prime-stable' system.
My first test is always: Can it pass Aquamark?
You might ask why I don't use3dmark03. Well, It doesn't bench. the whole system(score), and I disslike 3dmark03 and all the problems with nvidia drivers(the cheat stuff)....
If everything goes ok I will run prime95 for atleast 30 mins to get a fell with the max temperatures, to see if any additional cooling is needed. If temps are ok and no crashes then I just use my system as normal...

Old saying:
If it's not broken, then who cares? (or something similar...)

[craneage2001]

I go for the "as-I-sleep" Prime test...I set up my temp probes to log the temps and fire I Prime as I go to bed. If it's running in the morning, it's good to go.

[gungeek]

I consider my system stable when it can run Folding@home 24/7 with gromac cores with sse enabled. No reboots or folding errors allowed.

To test stability, I start with several passes of Memtest86 then an hour of Prime95. If everything looks good, I start folding 24/7 and watch the log files closely for an hour or 2 then daily. Any errors, dropped cores, or reboots means it's not stable. After 2 weeks, I'm pretty confident it's stable.

[alibubba]

Well guys, i did a burn in for 24 hours on prime priority 10 at 2.06v hoping it make my system stable at 11x220, still no go. as for ignoring the problems, i did that with my last cpu (2400+ o/c to 3200+) and i had problems again with prime but it would be stable for games...unless it was a warm day...but this is a new computer, and i guess i'm trying to get a fresh start and do things the right way. :)

i either have to accept the fact it's not meant for the 2400+ mhz or ram or some unknown variable...

don't you just love overclockers and their overclocking toys? they're never happy with that 2354 mhz but they want 50 more mhz! :)

thanks guys

[ChillPhatCat]

Originally posted by sQUiRLy
I'm going to have to agree with ChillPhatCat on this one except for the last comment. It should be known that 2hrs of prime95 is no-way considered stabile. At the minimum, stability tests should be run long enough to bring all the hardware up to maximum temps and then span across the coldest and warmest parts of that particular day. That would roughly be around 13-14hrs. I've errored in the prime95 torture test after 13hrs1min so for me its about 14-15hrs before I declare system stability. :)

This is a good point however, it would seem to me that testing it during the 2 hottest hours of the day would be just as good... the 2 hours figure I just offer up since that's usually a sign that you could do just about anything with the comp and it won't crash... For intense mathematical calculations a couple days wouldn't be over the top...

But like I said... it's all about what you're gonna do with it... I usually re-torture test my CPU when I'm going to do some distributed computing of a different nature ie if I'm gonna be folding for the next few weeks then I'll make sure the speed of the week is stable enough... although folding is not nearly as demanding as Prime.

[Fleck]

I hadn't had to do a lot of testing for stability until I got this new system and was really intent on overclocking. It took me a while to find the spot I'm at right now because Prime95 would always fail out with errors during the first tests after either a minute or two or 6-9 minutes (thereabouts). After enough tweaking I've got the new setup which is bus at 205 (2.37GHz), CPU at 1.9 volts (a little higher than needed but gets to 60C in full prime95 load prolly the regular 92mm case fan i dunno how many CFM it gets), RAM settings at 8-4-4-2.5, and RAM at 2.9 volts. Basically at this point the small FTT (max CPU stressing) test runs past Test 3 (once) and doesn't crash out. NFS UG doesn't crash out afte a while of playing either... this is good enough for me. Right now I'm seeing if my system as it is now will handle past Test 3.