"Burning-In" my processor

[Actaeon]

I have a few questions regarding "Burning-In" my processor. Some say its a Myth, some say its fact. I'm not sure what to believe, but I am pretty sure it doesn't hurt to try.

I want to achieve maximum stability for my AthlonXP2100+ TbredB @ 185x13, 2.4ghz. I am running at 1.85v, my mobo overvolts a little to 1.87v. I rather not move the voltage any higher, as in my opinion, its high enough already. I don't wanna experience sudden Tbred Death Syndrome. My temps are pretty good, I use a high-end air cooler, the Alpha PAL8045 w/ 52CFM fan. Temps are around 47C full load, and since my AS3 has yet to settle, I assume it'll go under 45C Full Load within several days. My memory is Kingsmax PC3200. Nothing extremely impressive, not a very highly regarded piece of memory. I am running it at the most aggressive settings, but CAS@2.5 w/ 2.8v (overvolts to about 2.85v).

Once thing I noticed was that I could Prime95 it completely stable while surfing the web, chatting, posting here, listening to music ect. It would still be running, then out of no where, a hard crash, locking up the computer completely. I am not 100% sure if it crashes on the same test, or at the same time, but I do recall around the same length of time, a bit longer than 30 minutes. I read several people had a problem where they would fail on the same test, the same time, every time. They recommended reinstalling Prime95. So I reinstalled Prime95 earlier today, but havn't got the chance to run it. So I am not 100% sure its a hardware, or Prime95 problem. But after my SiSandra burn-in, I'll run Prime95 to check.

Earlier today, I left my computer running SoB overnight, when I woke up, I recieved an error, and it had to close down the program. It happened perhaps 2 hours earlier.

Currently I am at school, but and I'm running the SiSandra Burn-In test. Testing out my Memory bandwidth, CPU Arthimatic and Media. How long do you suggest I run this for? After that, I plan on running Prime95. Then perhaps, Memtest as well.

I understand the fact it might not be stable, and I'm not using enough voltage, or I have it too high, or my memory might be stressed. But what I am most curious about is why Prime95 would run just fine for 30+ minutes, then just crash hard with no signs of warning. I am also considering reformatting, maybe that could help. Anybody have any suggestions why it would crash hard like that? Could it be a problem with Prime95?

Thanks
Jonathan

 

[AkumaX]

read this thread

go to about 1/2 way down, look for pm's post

 

[Actaeon]

Thanks for the reply. Very interesting post.

Hrm... I'm not sure what to do now. I was hoping after running several stress programs, that my cpu would perhaps run somewhat more stable. Not achieve higher overclockability. 2.4ghz is fine for me .

Any suggestions on how to get this thing a bit more stable?

 

[rogue1979]

Put the speed down one notch to obtain absolute stability. Run Prime95 for a couple of days total time. Run you system hard with real games (looping 3DMark is not as good) for about a week, maybe crunch some Seti in there too. All this time make sure you don't use higher than normal overclocking voltage, and keep an eye on the cpu temps. Then try putting the speed back up to 2.4GHz. Many times I have been "not quite stable" at a specific speed or voltage. One or two weeks of breaking or burning in has always allowed me run a little faster or at a slightly decreased voltage.

 

[THUGSROOK]

Originally posted by: AkumaX
read this thread

go to about 1/2 way down, look for pm's post

 

[DAPUNISHER]

Originally posted by: AkumaX
read this thread

go to about 1/2 way down, look for pm's post

I linked to pm's post in the "burning in memory" thread started a few days ago, I do however maintain that burning-in will remain a self-perpetuating myth though

 

[chizow]

I much prefer burn-out. In this day and age with something newer and faster coming out seemingly daily, and with current pricing, there's no point in holding back IMO. Not a knock on anyone who suggested stressing/running tests, but its a waste of time IMO as life is too short (both ours and the PC parts) to be doing anything other than maximizing performance in the here and now and returning to productivity. Once you've got it running stable, set it and forget it. By the time you remember you're OCing and possibly killing your CPU, you'll have placed another order with Newegg

Chiz

That was a great read btw, where was it originally posted? Highly Technical?

 

[Jeff7181]

The technical explanation isn't even needed... if you know anything about basic physics you should be able to figure out that high voltages degrades the conductor... and high heat degrades the conductor.

 

[DAPUNISHER]

Originally posted by: Jeff7181
The technical explanation isn't even needed... if you know anything about basic physics you should be able to figure out that high voltages degrades the conductor... and high heat degrades the conductor.

Yes, but it's the rate at which it occurs that remains to be firmly established, particularly with the .13micron architecture. That's why some basic data to work with such as Thugs ad Duvie and some of the others provided on the degradation in performance of their northwoods after significant voltage increases is so helpful to others

 

[Actaeon]

Originally posted by: DAPUNISHER

Originally posted by: Jeff7181
The technical explanation isn't even needed... if you know anything about basic physics you should be able to figure out that high voltages degrades the conductor... and high heat degrades the conductor.

Yes, but it's the rate at which it occurs that remains to be firmly established, particularly with the .13micron architecture. That's why some basic data to work with such as Thugs ad Duvie and some of the others provided on the degradation in performance of their northwoods after significant voltage increases is so helpful to others

And of course, even then, we don't know how the Thoroughbreds react to such high voltages.

But thanks for the suggestion everyone, I'm back at home now, after completing the 237th "Burn-In" test of SiSandra (Memory and Cache, Memory Bandwidth, CPU Multimedia, CPU Arithmatic), I think its okay now. Time for Prime95.

Hopefully it is now at a level of stability. But i'll report back to this thread with my Prime95 results.

 

[Jeff7181]

Originally posted by: DAPUNISHER

Yes, but it's the rate at which it occurs that remains to be firmly established, particularly with the .13micron architecture. That's why some basic data to work with such as Thugs ad Duvie and some of the others provided on the degradation in performance of their northwoods after significant voltage increases is so helpful to others

That's not my point... my point is that "burning-in" doesn't improve the conductors, it degrades them... no matter how slowly it happens, it still happens, and the fact that it happens is testament to the fact that burning-in electrical componants does not make them more stable or able to attain a higher overclock.

 

[THUGSROOK]

i think you misunderstood what DP meant...

we are all in agreement here~
there is no burn in ~ only burn out

 

[Duvie]

Maybe here is some data for you....Granted mine and thug's experience is on Intel's .13micron process of the northwood and I do not have experience with the T-bred .13 core....

However lets look at it relatively....


I run my cpu at 1.71v which is .21 higher then spec default of 1.5v....ie that is 14% increase...I have run at this level and tested it again last night now for 7-1/2 months now with no degradation to my ocing ability...

I ran my 1.8a (the one I damaged) at 1.83v actual (and lets make sure we are all talking actual vcore as reported by multiple sources). That was .33 higher then default and represented a 22% increase....This damaged that processor in 1 weeks time. My temps were as good as they are now so heat should not have been a factor.


I think as long as you stay within 10-15% range you should be fine in the since that the likely (and oh yes it is happening) degradation will not appear in the operational period in which you may own it. If most here in this forum upgrade in 1-2 period you would andshould be fine. The lower in that range the better the longevity. and ofcourse that default vcore should last you a very long time.

Ocing is the game we play and death of the chip is the risk we endure....Don't play if you can't take the loss...

 

[rogue1979]

I agree with Duvie, though I wonder if a 15% increase in core voltage might actually be conservative for the .13 micron Athlons. Kinda premature to know for sure yet, but so far they seem more robust than the .13 micron Northwood.

 

[Mikki]

Not to mention that if your cpu does get damaged by voltage to the point that it's not usable or just overclockable, that just gives ya an excuse to upgrade!!!! hehe...

 

[Duvie]

I agree Rogue...that may be a bit conservative for the Tbreds but it is relatively early to tell..A lot of p4 users runing water and vapochill devices didn't start dropping their chips (NSDS) for a few months...

Overall I imagine they would have to be close as I can't imagine that amd's .13micorn gates can be so much more robust against higher vcore then INtel....Maybe a 5% fluctuation for amd, but I wouldn't call it safe for awhile....

 

[daos]

agreed...