An interesting article, but only available in German.

The stats are based on numbers from a French shop, which wants to remain anonymous. The data has been collected between March and August of last year. Only cards with >500 sales are included.

GTX 280 doesn't fare too well, whereas 4870 and 4850 are surprisingly solid, considering their high operating temperatures. There have been problems with the 3870X2, which seem to be fixed in the 4870X2.

So it seems rather normal if a GT200 based cards dies every now and then. [the same is true for other cards, the numbers are just a little lower]

MrS

---

Scanning for our furry friends since Jan 2002

Uhuh this is what i have heard with some shops in holland, the statistics seem to are almost the same and represent the returns well for bigger shops.

Based on the 9% failure rate of GTX 280's, I've got that beat with 100% failure rate of my two GTX 280's. Would that be a 200% failure rate, LOL.

Just got my first replacement 280 and I am still waiting for the RMA number on my second. Lets see how long this one lasts!!

Pat

That's where a lifetime warranty comes in handy, not all vendors offer that, only one brand that I know of.

An interesting article, but only available in German.

The stats are based on numbers from a French shop, which wants to remain anonymous. The data has been collected between March and August of last year. Only cards with >500 sales are included.

GTX 280 doesn't fare too well, whereas 4870 and 4850 are surprisingly solid, considering their high operating temperatures. There have been problems with the 3870X2, which seem to be fixed in the 4870X2.

So it seems rather normal if a GT200 based cards dies every now and then. [the same is true for other cards, the numbers are just a little lower]

MrS

I believe it!

I burned my 280 here after less than 6 months of usage.

Now my 4870x2 is still surviving after 8 months of milkyway and folding@home.

I've never crashed a game on ATI while nVidia was a consant nusiance with crashes!

I've had 2 x GTX280's die on me after exactly 6 months of GPUGrid. More than likely manufacturing fault - both were Leadtek, switched to BFGTech since then.

Been an nVidia user ever since I switched to PC's from consoles (Spectrum, Amstrad, Commodeore). All the crashes I've experienced were software (driver) related, so they were easily overcome - reinstall/upgrade/downgrade.

---

Semper ubi sub ubi.

Well I can say I've had 0 failures in what will soon be a year of running GPUGrid.

4x GTX 260 and a 9800 GTX +

[edit] 2x BFG & 2x XFX on the 260's [/edit]

OC'd during the winter, and stock the rest of the year (Arizona has 9 months of summer, just about time to OC them again)

Good ventilation, good power supplies, and a slight bump of the fans on the card (to about 70 - 75%) seem to keep them running quite happily.

---

My 280 finally died and I think it was also a defect. It was a BFG but becuase of my water cooling I have no warranty. I ran gpugrid on it all the time and cooled it with a quad 120mm radiator on a dedicated loop so it NEVER overheated. After a lot of testing I have narrowed the problem down to an on board power issue. Thus the faults are likely caused by something in the power phase system such as a bad capacitor.

I just hope my second one holds together! I think I am going to force a high clock on it to reduce pressure on the phase system by locking it to max at all times as I believe that the problem has something to do with the clock switching between high and low cycles.

The GTX 280 cards used a 65nm fabricated core. I guess these required more juice and stressed the boards power circuitry more. 6 months seems to be quite a common length of time, before these cards die.

Unless you research your cards VERY carefully, you will most likely end up with some sort of unwanted spec. I have made plenty of blunders, mostly incorrect assumptions. You most definitely cannot go by NVidias naming system!

The G200 series seems to contain GT92a/b and GT96 cards, http://en.wikipedia.org/wiki/GeForce_200_Series, and everything from 40nm through to 65nm cores. Of course the code names don’t follow the models either.

Many boards on the other hand look surprisingly similar!
Not a bit wonder they pop. Its a bit like having F1, Nascar and F3 engines inside the same shell. Some wont do so well.

ZEROibis, your problem could be due to the power circuitry overheating. On modern high end cards the backplate of the stock cooler usually touches the hottest parts. This can get problematic if you use an aftermarket cooler, even it blows air directly onto them.

However, if in the case of water cooling you've got no air flow on the front side of the card those chips are likely getting unhealthily hot.

MrS

---

Scanning for our furry friends since Jan 2002

Both sides are watercooled so that is impossible. Also I can monitor the temps of all the independent parts with riviatuner and gpuz so I know that they never got even close to the stock temps. Also it can crash right out of a cold boot with the card not having enough time to heat up so that is not an issue. My backup cards work so the problem is the card. I am going to try the back method to fix it and buy a 285 or 275 if it does not work.