What is stable system? Is Prime the best method? - Post your thoughts here

[luihed]

I think the word overclock and the word stability have different meaning to different people... When I show a max screenie of an overclock, it never fails that somebody will say " but is it stable at that speed"?... My answer to that is "yes it is" because my goal at that particular time is to get a screenshot and obviously my computer is stable anough to achieve that goal.......I know thats not the point of this thread but I just thought id throw it in there.....

Now for 24/7 stability, I think it really depends on how important it is to a person that his computer doesnt crash.... Testing Orthos for 8 hrs is something I dont understand because If stability really means taht much to a person, why not just clock it at a lower speed?..... Does one really have to go through such test to make sure his computer wont crash ever?....

Now for the stability perfectionist out there who likes to question other people's overclock for its stability, I have only one question... " Is your current overclock stable enough that you would put a love ones life on it"? meaning if your brothers life depends on that computer would you still have that same overclock?


BTW, I cant remember the last time I test for stability, if my computer crash, I turn it back on at a lower speed and life goes on.......

[cadaveca]

Now for the stability perfectionist out there who likes to question other people's overclock for its stability, I have only one question... " Is your current overclock stable enough that you would put a love ones life on it"? meaning if your brothers life depends on that computer would you still have that same overclock?

If you are crunching, then you should be able to say yes, IMHO.

[Jobeo]

i agree with running a variety of tests to prove it stable... as you're clocking up higher and higher the short tests are probably good enough but once I arrive at a decent speed i'm gonna verify with as many programs as possible...

[WiCKeD]

I think the word overclock and the word stability have different meaning to different people... When I show a max screenie of an overclock, it never fails that somebody will say " but is it stable at that speed"?... My answer to that is "yes it is" because my goal at that particular time is to get a screenshot and obviously my computer is stable anough to achieve that goal.......I know thats not the point of this thread but I just thought id throw it in there.....

Are you deceiving the person who expected to get the same overclock and run their system stabley? Yes. Your questionable ethics ruin expectations for the rest of the community and that is totally selfish.

Would I risk a love one's life? Yes, my system is as stable as Joe Blow's system on stock cooling.

Let's see more respected members step up to the plate and answer this question...

[railmeat]

Are you deceiving the person who expected to get the same overclock and run their system stabley? Yes. Your questionable ethics ruin expectations for the rest of the community and that is totally selfish.

Would I risk a love one's life? Yes, my system is as stable as Joe Blow's system on stock cooling.

Let's see more respected members step up to the plate and answer this question...

"If I had to pick a definitive benchmark that can be run in a reasonable amount of time and that stresses the system substantially it would be OCCT. I would be extremely happy if everyone were posting a screenshot of this with their "OMG LOOK AT THIS L33T0RZ OVERCLOCK!!!" Then we will see just how much overclocks change..."


dude amazing quote by you!!!and VERY true.there is way to many ppl with screenshots of amazing clocks and thats it?!?!?!?!look at me and my 15- pi test?!?!wow??but is it REALLY stable and whats your reaL like.i only respect a few ppl here and im not saying names but there ppl who give us sane borderline insane overclocks with dual prime etc for the realists to enjoy vs the suicide screenshots kids claim to be 24\7 stable with no backup.i can wack 3.960 9x400 at 1.575 vcore,is it stable?...no its not,but the next guy claims that his is...sigh

bottom line- i agree %100 occt is brutal,hi cpu+ram chosen.occt is the first test of choice for me.30min give a real good baseline.

2gigs ocz sli 1066 ram 4-3-3-10 90min
http://s76.photobucket.com/albums/j1...ableoczram.jpg

4 gigs gskill ddr800 4-4-4-12 60min
http://s76.photobucket.com/albums/j1...hourHItest.jpg

[Timbosan]

Proving absolute 100% stability is not possible, it is relative to the length of time the test is run. If it passes 24hrs orthos, its 100% stable for 24hrs, it may have failed just after it is stopped, there is no way to ever know.

That said, Orthos small FFT is a good initial test of stability, with more focus on cpu stress, even on blend.
IMHO gaming hard on a high FPS game like NFSU2 is the best test, especially for ram stability I can set my ram at 4-4-4-4 at 980MHz @ 2.14v, and it will pass 8 hours of small or large FFT or blend orthos. However it will crash to desktop in less than ONE LAP of NFSU2. I had to raise Tras to 12 before game stability is achieved at that speed & vdimm.
For a quick stability test you can loop 3D Mark 2001 Lobby low & high a few times.

[Soulburner]

I find my CPU overclock with a 30min OCCT run then 30min Orthos SP2004 Large FFT's for RAM.

This gets me approximate stable settings. Then I use the system for everything from forums to gaming and over the next few days will tweak things slightly if I encounter instabilities.

[Spoudazo]

If it is for my own personal use then stable is:

(1) running all apps I use
(2) not BSODing, CTDing, etc.
(3) CPU isn't roasting
,etc.

If the overclock is for someone else then,
Intel TAT, Orthos, or whatever running stable, cool, etc. for hours on end.

[ShawnTRD]

Is Prime95 the best way to test my C2D? And if so is Blend the best test to start with?

[The Legend]

Prime95 will only stress 1 core.

Download Orthos. Run Small FFTs. It will stress both cores.

[crazyea]

Prime95 - p95v255a tests all 4 cores.

http://www.mersenne.org/gimps/

EDIT: or two if thats all you have. :P

[003]

Prime95 can test both cores at the same time. Open two instances of it, then go to advanced > affinity and have one set to 1 and the other set to 0.

Then go to torture test on both of them and select custom, and in the box where it says mb of memory to test, divide it in half for both of them and off you go.

You can configure it to run and test as many cores as you want.

[graysky]

Right... version 25 of p95 tests all cores by default... just be sure you enable error checking or else it's useless :p

[crazyea]

or just download the newer version, it asks you how many cores to test......

[linflas]

Crunch numbers for Team Xtremesystems in WorldCommunityGrid, if you can crunch 24/7 without errors or failure, you are good to go, and you help a good cause to boot.

[yonton228]

Well I think this is an important issue (not really) but to some. I think we should have a poll and have people vote what the standard means of testing and for how long to test for, for some of the overclocking databases. Example:

Prime 1-4 cores (depending on CPU) what settings, small FFT's for 12 hours.
Dual Orthos (insert same guidelines as above, or whatever gets the most votes)

However this recommendation would be for "databases" only not discussion threads or anything of that matter.

PS: Yes I realize that there is a database thread already however I think that it may need to be updated IMO.



-yonton228/timmy

[JAG87]

Right... version 25 of p95 tests all cores by default... just be sure you enable error checking or else it's useless :p

thanks for that tip buddy. cheers

[cadaveca]

Prime95 small FFT for cpu testing, FFT size to the size of cpu cache(4096k for core2duo/quad). The larger FFT size will test ram, cache in large hits, and memory controller.


No need to set affinty or anything as long as you use the latest version.

No other stress test with error checking will generate more heat than this on Core2, so you'll test stability within the limits of your cooling. This also means that generally cpu will be stable doing other things, including games, etc, however, you must run this test for more than 12 hours without erros to ensure full stability. The perfect example is BF2...prime can run for 2 hours, but BF2 may still crash....obviously not fully stable, then, with just two hours.

[Glorfindel]

Hi,

Yes there is some prog that load your CPU at 100% like prime. I'm a seasoned OCer, but i like computer chess, in fact i bought high end system to compete in computer chess tourament. My Rybka (strongest computer chess prog) mp-64 will load all four cores at 100% all the game long.

I'm also a teacher, and i'm using 3d cad software (CatiaV5) and CFD software (fluent) those will load quite good the cpu too.

[cadaveca]

but do they provide error-checking, and tell you what exactly the error was? If so, pls provide a link so I can add this to my overclocking arsenal. I'd love a new tool for testing!

[Glorfindel]

but do they provide error-checking, and tell you what exactly the error was? If so, pls provide a link so I can add this to my overclocking arsenal. I'd love a new tool for testing!

No they dont, they just crash ;-) but they load all my cores big times ;-) so for me, Stable mean, able to play a 3 hours chess game with the four cores of my quad at 100% and keeping the cpu temp under 70C.

When i loose a game i'm loosing rate points or tourament, i dont want that. So i take Prime or ortho or occt and if i can run them about 3-4 hours without crash, it's Chess proof for me!

I think it depend of your need and goal.

[KTE]

I posted a bloody long reply and the forum timed out and I lost my message. Freaking fire am I grieved or what!

I will try again once I can recompose myself...

[Demo]

Orthos is using Prime95 engine right ?

[maxwolfie]

I've noticed that most people do at least 2 hrs Orthos small FFT. However this doesn't really test your RAM.

Is there a general consensus for both CPU/RAM/anything other components that are affected by overclocking the CPU?

[STEvil]

No.

Orthos, Memtest, Super-Pi, WCG (or favorite distributed computing program). All systems ram and chipsets react slightly differently.

If its not stable what you do with it all the time, its not stable. If it is, it is.