7900 GTX cards keep dying

Wow, thanks for all the info, I knew I wasn't crazy.
My cards did exactly what you described. Both BFG
I did RMA them but I wonder if crossfire might be the way to go
until this issue is worked out. ..performance seems pretty similar.
 
you know, one has to wonder why EVGA, XFX, BFG, etc.....are still shipping cards even if they know many of them will be coming right back to them

i understand they cannot just stop selling product...well...ok, they can, but they won't

but in the end they will end up with a ton of used cards that are defective, which could have been tested before shipping out to avoid all the bad publicity, but instead they ship them out, let us pay for shipping both ways in some cases, and let us be the testers, and in turn let us be the ones that are inconvenienced

i feel it would be more responsible to come out and say that all future orders are being placed on hold until Nvidia works out the issues, but I also understand why the board partners would be cutting off their own heads by calling Nvidia out on this matter

kind of like the company i work for, they ship out units that they know have issues all the time, then just pay someone to fix it after it is installed in the customer's place, but the part i dont like about it is the bad press we get as a result, i would rather have to wait on a reliable product myself vs. get a faulty one now that costs me downtime and headaches.......i say do it right the first time or dont do it at all

obviously the board partners like EVGA, XFX, and BFG are not stress testing their products before shipment...i know it takes time they dont have because of demand, but it appears as if they are simply shipping out product in masses in hopes that only 20% come back......kind of a shotgun approach
 
HeavyH20 said:
No HDR but I believe SM 3.0 is used.
Try raising the memory clock back above 650 on the issue cards, and the instability returns.
Click to expand...

I thought the memory on these cards are rated for 700, am I wrong? so they should be stable at 700 at least right?
 
1.1ns GDDR3 is actually spec'd at 900 MHz at default timings. NVIDIA tightens these timings a tad, so 800 Mhz is their new spec. So, it is no surpise that a lot of cards will do 900 MHz.

The GT cards use 1.4 ns GDDR3 that is spec'd to 1400 or so. I think that is where you are getting the 700 MHz number.
 
thanks,

but to underclock to 650 in order to get stability, it's kind of ridiculous
 
HeavyH20 said:
Two were EVGA GTX EGS and two were GTX SC cards. Different machines, PSUs, motherboards and CPUs.
Click to expand...

Is your EVGA purchase as the over clock version? Or Did you buy the regular speed video card and run it as a regular clock speed?
 
HeavyH20 said:
If you really want to test your card, simply loop the last test of 3Dmark06 exclusively for 50 loops. Does not take long, only about 2 hours.
Click to expand...

If I do this test will it determine whether or not my card is good or bad?

I ran 3dmark06 immediately after installing my card and everything ran/looked great on it.

eVGA 7900GTX EGS (stock speeds)
 
HeavyH20 said:
Giving up on NVIDIA. Maybe there is a nice Crossfire system in my near future. :D
Click to expand...


Bwaahahah!!!!!. And you think that will run better? Hope you enjoy troubleshooting.
 
Wow... this worries me.

I replaced a perfectly good ATI1900XTX a week ago with a new BFG7900GTXOC.

For the first day it was great. Played Oblivion, SW Galaxies and the Tomb Raider Demo. Then (2) days into it I started getting artifacts in Tomb Raider Legends. From then on i noticed accasional little colored dots in my 2D Windows just surfing the net. Tomb Raider wouldn't play for more then a minute without crapping out and requiring me to reboot. I tried 3 different drivers trying to fix it to no avail. Then, 24 hours ago it started artifacting & going to hash every 30 minutes just using Windows XP - no games. Finally tried a fresh install and the screen would goto hash when Windows tried detecting it during the install. It was totally dead :(

I RMA'd it to to Newegg this mornign but they were sold out of the BFG cards. I "settled" for a plain EVGA 7900GTX (Non-OC'd) card which will get here tomorrow (so says my e-mail anyway). From the sounds of it, that may not last either?!

This is very unacceptable for a $550.00 card.

I am wondering if I should regret selling the 1900XTX.
 
im betting a driver will be released to clock down that Z roster thingy that everyone seems to be blaming, and then perhalps ocing will work a tad better.

just outta curiosity what were your temps when you started seeing the artifacts?
 
I was hoping it was a driver related item. But, one card would work and another would not. Slowly the cards degrade down to the 650 MHz memory clock. Not sure a driver would create that behavior.

My cards artifacted at 38C load.
 
darkjedi said:
I am wondering if I should regret selling the 1900XTX.
Click to expand...
Hindsight is a wonderful thing. I ain't selling my XT....at least until G80/R600 anyway :D
 
rincewind said:
Hindsight is a wonderful thing. I ain't selling my XT....at least until G80/R600 anyway :D
Click to expand...

It's pretty pointless to switch from a X1900XTX to a 7900GTX anyways as thier performance is very simular, I only got 2 7900Gts but they don't seem to have been affected with the problems of the gtx.
 
(cf)Eclipse said:
i have two pieces of advice that everyone should follow:

1. make sure ALL the ram chips are well cooled. preferably no more than "warm to the touch" when stress testing
2. put heatsinks on all mosfets, and be sure that there is plenty of airflow over them
Click to expand...

very good advice.. especially for the 79GT crowd.. I already have a lot of air flow over and around my cards, but I am going to sink the memory and mosfets when I get back from my travels.. reading all of these posts makes me leary of even doing a vmod now. lol. I guess I will sink the mems and vregs and run the hell out of the cards for a while before trying that.. I have less than 24 hours of use out of them at current.. man.. been away for two weeks and all I can think about now is geting home to test these cards more. whew.
 
dR.Jester said:
If I do this test will it determine whether or not my card is good or bad?

I ran 3dmark06 immediately after installing my card and everything ran/looked great on it.

eVGA 7900GTX EGS (stock speeds)
Click to expand...


Just one run? I had plenty of those with no issues. Try running just the last test 50 times. Not the whole run with SM 2, CPU and SM 3 tests. That will work the card.


Is your EVGA purchase as the over clock version? Or Did you buy the regular speed video card and run it as a regular clock speed?
Click to expand...

The EGS are regular clocked (650/1600) and the SC versions are factory overclocked (690/880).


Bwaahahah!!!!!. And you think that will run better? Hope you enjoy troubleshooting.
Click to expand...

Already set up a X1900 XT Crossfire system for someone else. No harder than SLI. Works great. Not the buggy solution that was introduced some time ago. Oh wait, SLI and Crossfire were BOTH buggy at intro. Either way, both are baked solutions, now.
 
yeah, crossfire is supposed to be pretty solid now.. and that RD580 chipset is a scorcher.. the DFI offering with that chipset is supposed to be /very/ nice.
 
So I got my new card this morning... the EVGA 7900GTX (non-OC'd) version.

I put it in my PC and fired up Windows XP. It found the new card, i installed the driver and the PC immediately displayed the hash again. Vertical lines with horizontal squiggles so nothing is visible. This was a fresh install of windows that i did last night to get it ready for the new card. Very bad.

I just wasn't willing to accept that I got a bad card, so I began trying different things... BIOS setting changes, different power supply... no effect.

Then I decided to try the card in the 2nd PCI-E slot. Guess what, it worked!

So now I realize that I have a bad MoBo. So I have to wonder if the BFG card was actually bad of if it was my MoBo. However, it worked fine a week ago with the ATI card. Very wierd. I went down to Fry's and snagged a new ABIT AN8-32X and all seems to be working as it should. Although i have not played any major games on it yet. I have just run 3DMark06 to make sure it works. It does.

Do you think the BFG card killed my Asus board's PCI-E slot?
 
BTW, I was using the 84.56 driver, already. Didn't fix anything since it is a hardware fault.

And, a video card will not take down a PCI-E slot. What card was previously working in the slot?
 
I have had both of my 7900GTX's since launch week. I haven't had any problems with them at all.
 
Dan_D said:
I have had both of my 7900GTX's since launch week. I haven't had any problems with them at all.
Click to expand...


I am beginning to wonder if I should just go SLI for my 30 incher. I get the feeling my screen is stressing out the single card too much. TRAA seems to be killing my cards.
 
jacuzz1 said:
I am beginning to wonder if I should just go SLI for my 30 incher. I get the feeling my screen is stressing out the single card too much. TRAA seems to be killing my cards.
Click to expand...

If you're going for the 30' LCD, you'll most definitely want SLI. In fact, I'm not even sure how well THAT will be handled with all the effects up, at least not in newer titles.
 
Again, same here, Dan. Luckily every problem I have had was due to user error and poor mobo settings. Additionally, although its probably overkill, I have had the side of my case off since I got the cards (its VERY crowded down there and partly out of worry of cooking my XFi card) with a 120mm fan blowing straight onto them and the southbridge. This cured the temp differential between the top & bottom cards and I have had no stability issues since. I've run the Ati OC tool for atrifacts almost every week with perfect reults and looped 5 3DMark 06s with no probs. These cards *can* run fine but do require attention and tlc.
 
The issues with the cards will depend on usage patterns. Running 3Dmark06 a few times barely works the card, since it is a mix of SM 2.0 tests, a break running the CPU tests, followed by two SM 3.0 tests with the last working the memory the hardest. This last test is the killer. I had a card that had worked without issue for weeks. I decided to test it out before my 30 days was up from the retailer. So, I ran 50 exclusive runs of the last 3Dmark06 test in a loop. The card never made it. It exhibited the same symptoms of the other cards.

And, if there is a fix, BIOS or driver, I would prefer to keep the cards. When they worked, they worked great.
 
HeavyH2O

Did you try any of the suggestions on this board and the evga forums?

-adding sinks to the backside voltage regulators
-underclocking the memory

If so, did you see the same issues?



I just put a new computer together basing it around the 7900GT OC from evga and am having all of the issues people are reporting and worse. I'm hoping the RMA card will work better for me.

Things I've tried:
1. Driver change (newest one on evga's site)
2. Bios change (the one dated 4/28/06 for my card, and one sent by evga's Mario)
3. Tried to underclock the card, but I hardlock every time I touch the setting.
 
Opie said:
That's bizarre and the first I've read anywhere about that many massive failures period, and am boggled that there could be 5 from a single user. There was the vertex clock issue at release, but damn man, that sucks. I guess I've been really lucky, and yes they are touchy beasts, but my BFGs have been and continue to be awesome. I've run them hard since the day the GTX was released.
Click to expand...


umm what vertex issue guys ? i have a pair of ASUS 7900GTX es and would like to know what to look out for ?
 
quadnad said:
If you're going for the 30' LCD, you'll most definitely want SLI. In fact, I'm not even sure how well THAT will be handled with all the effects up, at least not in newer titles.
Click to expand...

It is not a question of going to. I do run the 30. My first 7900GTX lasted from March 9th till the first week of May. My rma card lasted 72hours. They are going to burn in my next card for 24 hours straight (If i am to believe them) on 3dmark 06 before shipping. If this one dies I am not sure wtf i am going to do. My case temps are idle 32c, cpu core idles at 36c, and video core is idle 38c. Temps are not at issue as it relates to the core and memoryThere is something else in play here.

I am more than inclinded to believe H2o and Evga , whose explainations are similar.
 
jacuzz1 said:
It is not a question of going to. I do run the 30. My first 7900GTX lasted from March 9th till the first week of May. My rma card lasted 72hours. They are going to burn in my next card for 24 hours straight (If i am to believe them) on 3dmark 06 before shipping. If this one dies I am not sure wtf i am going to do. My case temps are idle 32c, cpu core idles at 36c, and video core is idle 38c. Temps are not at issue as it relates to the core and memoryThere is something else in play here.

I am more than inclinded to believe H2o and Evga , whose explainations are similar.
Click to expand...

S0 x1900xt is not an option?
I mean you have sufficient reason to ask EVGA to buy the card back...
 
Lolento said:
S0 x1900xt is not an option?
I mean you have sufficient reason to ask EVGA to buy the card back...
Click to expand...

The thought has crossed my mind. I have had ATI cards in the past howerver I am in love with the nvidia driver interface and TRAA Supersampled
 
darkjedi said:
So I got my new card this morning... the EVGA 7900GTX (non-OC'd) version.

I put it in my PC and fired up Windows XP. It found the new card, i installed the driver and the PC immediately displayed the hash again. Vertical lines with horizontal squiggles so nothing is visible. This was a fresh install of windows that i did last night to get it ready for the new card. Very bad.

I just wasn't willing to accept that I got a bad card, so I began trying different things... BIOS setting changes, different power supply... no effect.

Then I decided to try the card in the 2nd PCI-E slot. Guess what, it worked!

So now I realize that I have a bad MoBo. So I have to wonder if the BFG card was actually bad of if it was my MoBo. However, it worked fine a week ago with the ATI card. Very wierd. I went down to Fry's and snagged a new ABIT AN8-32X and all seems to be working as it should. Although i have not played any major games on it yet. I have just run 3DMark06 to make sure it works. It does.

Do you think the BFG card killed my Asus board's PCI-E slot?
Click to expand...


reminds me of that song.. video killed the radio star. anyways.. try blasting your "bad" pci-e slot out with canned air.. I have had that fix memory slots and regular pci slots before..
 
r0gue3 said:
HeavyH2O

Did you try any of the suggestions on this board and the evga forums?

-adding sinks to the backside voltage regulators
-underclocking the memory

If so, did you see the same issues?



I just put a new computer together basing it around the 7900GT OC from evga and am having all of the issues people are reporting and worse. I'm hoping the RMA card will work better for me.

Things I've tried:
1. Driver change (newest one on evga's site)
2. Bios change (the one dated 4/28/06 for my card, and one sent by evga's Mario)
3. Tried to underclock the card, but I hardlock every time I touch the setting.
Click to expand...

Tried them all. The second SC had a .06 revision of the BIOS. And, underclocking the card worked fine - set the memory to 650 and had no issues. And, I don't think heat was an issue. Here was what I was using:

80mmfan9iz.jpg
 
yeah, I think it's safe to say that pretty much defines "more than adequate" cooling. :D
 
I've had a PNY 7900 GTX now oc'ed at 690/1760 for nearly a month. Not a single hiccup in Oblivion, GRAW Demo and HL2. Love it with a 24in LCD 2405. :)

Idle temp : 38-40. Max Load temp : 66
 
Well, can't be a driver issue. I got the RMA card installed and it works (as expected) with no issues. In fact, it is faster at the same clocks than the other SC cards (6500 @ default clocks and driver settings in 3Dmark06). Have to do some testing to see how solid this new card really is.
 
HeavyH20 said:
Well, can't be a driver issue. I got the RMA card installed and it works (as expected) with no issues. In fact, it is faster at the same clocks than the other SC cards (6500 @ default clocks and driver settings in 3Dmark06). Have to do some testing to see how solid this new card really is.
Click to expand...

Don't kill it! :D
 
HeavyH20 said:
Well, can't be a driver issue. I got the RMA card installed and it works (as expected) with no issues. In fact, it is faster at the same clocks than the other SC cards (6500 @ default clocks and driver settings in 3Dmark06). Have to do some testing to see how solid this new card really is.
Click to expand...

did you make note of the bios version on the old card vs. new card?

or the memory module version?
 
It is all the same as the prior cards. No changes. Same memory and same BIOS.
 
I've got 2 eVGA 7900 GTX's, in SLI, on a 30", had them since launch, both serial numbers under 100 ;) I run native res 4xAA 16xASF in every game that supports it, or 2048x1536 in ones that do not...

Not a single problem... only problem I have is that SLI just flat out sucks in general for the price.

Hardware wise though, not a single problem with the system since installation.

-F
 
You must log in or register to reply here.
Back
Top