prosser13

Electomigration will occur a lot faster if your using the CPU constantly at 100% load as opposed to enabling SpeedStep. Even if an overclock is completely stable on the first day, after 3 months of 24/7 usage it is possible that due to electomigration the overclock will no longer be stable

I guess what I'm trying to say is that the CPU or GPU's overclock doesn't necessarily have to be unstable - even a stable overclock will degrade over time.

__________________

Riou

In theory, you are stressing your components. If there are any weak parts, those might fail sooner compared to a good quality one.

T3h_Ch33z_Muncha

Others beg to differ...
I'm quite certain my 9600GT was stable.

__________________
Please don't call me Muncha, as cheese is not the only insinuatable munchable object; Ch33z will suffice. Oh, and i almost forgot to express my love for Heru
Planned upgrades (kind of in order): 22" Monitor, Goliathus Omega (mousepad), new chair, Windows 7, CM 690, new desk, Ideazon ZBoard MERC, Sennheiser HD201, bong, DX 11 card(s)+necessary PSU
Epic things happening in 2010: Toy Story 3, new H.I.M album, Supreme Commander 2, MechWarrior, The Elder Scrolls V, Mafia II, Star Wars: The Old Republic, Bioshock 2, Crysis 2, Iron Man 2

Clockadile Dundee

Well, either folding caused your 8800gts to run at over 100% load...
Or it was just one of those bad apples, and was not stable at stock settings, or it was not getting enough power, or the temperatures were out of control, or something, but I think we can all agree, it doesn't ask the gpu to run at over 100% load. Otherwise, what are you guys suggesting, exactly?

__________________

Clockadile Dundee

What makes you so sure of that? I am curious about this, please don't interpret this as any form of contention or even tension. I am happy, and I love you all, so much...

The only thing I love more than animals, is appropriate font sizes...actually I love animals more.

Sorry.

__________________

Clockadile Dundee

Yes, you are absolutely right, and you stated it more accurately.

However, consider this : how can everyone be so sure there overclock is stable? With a cpu, I know, a different animal, but you really need to run at least a few different stability programs, for a combined total of like 24 hours, or over, to be reasonably sure it is stable. What lengths, do the average person go to, to ensure stability, after oc'ing their card...maybe 5 minutes of one program, if its lucky (the gpu)? I'm not really sure, but that kind of thing seems to be the status quo, even on here.

I'm sure people get cards, brand new in box, which are degrading out of box...just because it is BNIB, at stock settings, even then it could be degrading...because its not actually stable? So a poor example will do worse.

I have non-verbal learning disability, a neurological disorder. One feature, is randomly telling strangers personal information, for example.

__________________

prosser13

Electromigration is not an instant process - it occurs over time. To simplify it, the walls between the paths within the processor are broken down as they are smashed into it. Higher voltage means the things smashing into them smash into them harder, so they break down faster, however this process is always happening and always occurring. People may say they have not seen a difference, but that is because they just haven't seen the change yet; it may take one card months and another years to show any ill effect but electromigration, but eventually it will happen.

It's a rather simple analogy sorry

__________________

Clockadile Dundee

No, that is spot on, the explanation, which most simply conveys the idea fully, is the most perfect; not the most complicated.

__________________

prosser13

Regardless of whether it is stable or not, it may still break down quickly, and then show signs of instability.

Using another analogy:

Think of a road, at the end of which is a solid wall. That wall is when the card becomes so unstable it stops working - the end of the road for the card, so to speak.

Every card has its own road, with its own pot holes; some cards may display signs of instability from the beginning, like artefacts, but last for a long time, while others may be perfectly stable and then suddenly show serious instability.

Another point is that every card's 'road' is a different length. As every chip is individual, so is its life span. This is obvious in DOA and cards which are received faulty.

However, what I'm trying to say is - some cards may last for a lot longer than others, even when run at 100% load. Others may show long term signs of instability before they die, etc. Just because 3 other people have come into this thread and said "I've run F@H for x number of weeks/months/years and my card still seems fine", the impact of running F@H will be different on each card.

I hope you can understand what I'm trying to say

__________________

T3h_Ch33z_Muncha

Stress testing. Simple as that.
I'm not denying that; chips simply cannot last forever. But folding is/was the means for that 100% load.

__________________
Please don't call me Muncha, as cheese is not the only insinuatable munchable object; Ch33z will suffice. Oh, and i almost forgot to express my love for Heru
Planned upgrades (kind of in order): 22" Monitor, Goliathus Omega (mousepad), new chair, Windows 7, CM 690, new desk, Ideazon ZBoard MERC, Sennheiser HD201, bong, DX 11 card(s)+necessary PSU
Epic things happening in 2010: Toy Story 3, new H.I.M album, Supreme Commander 2, MechWarrior, The Elder Scrolls V, Mafia II, Star Wars: The Old Republic, Bioshock 2, Crysis 2, Iron Man 2