What would you do in this situation?

[tapir]

OK -- here's the decision/situation I am trying to grapple with. From about 2 months ago I started rebuilding my rig. Once I got an E4300 I decided to get a somewhat modern video card even though I don't have that much time to game just because I like to try the newest games out once in a while... the thing is, I have a 24" LCD so I will push games to 1920x1200 when I can. Anyway I ordered a new 7900gt on ebay, the card was made by Forsa, I figured it would be more than adequate for my needs. When the card came 3 weeks ago, I noticed that it had a huge honking heatsink on it. They sent me a 7900GTX! I thought that was pretty cool and decided not to tell the eBay seller and just keep the card (note to self: not gonna do that again, BAD KARMA, BAD BAD BAD) and I installed the 7900GTX card and tested it out in 3dMARK 06. It seemed to work pretty well so I kept it.

Long story short: I can't run AA at 1920x1200 in any game without eventually getting artifacing. In games it usually takes the form of a ton of neon-colored lights scattered everywhere. It happens in COH and FEAR and in the 2nd HDR test (deep freeze) every time I put AA on. I manually adjusted the fan speed of my card to max and that seemed to help, but I can still recreate the problem if I give it a hard enough test to deal with.

I can't decide whether this is just a power supply issue? I have a 370W enermax noisetaker with 14A and 15A 12V rails for combined 29A? Supposedly this ought to be enough to run this card.

I think I ought to RMA the card based on what I've read about 7900GTX. The problem: FORSA is in Hong Kong, it will take me at least 45-60 days where I will have to use another video card. I don't want another nVidia card based on my experience with their drivers in vista 64-bit. But then again the ATi cards seem to use even more power...?

Any suggestions on whether to seriously consider trying a new PSU, a website with a good return policy / inexpensive options for PSU that would eliminate any possibility that PSU is the fault... should I RMA, should I sit on the card and just not use antialiasing... any advice would be super welcome.

 

[Matt2]

That's very strange that it only happens when AA is on. It's completely 100% stable when AA is off?

The ebay seller could have knowingly sent you the 7900GTX knowing that it was defective.

I dont know what your gaming preferences are, but at 1920x1200 my X1900XTX chokes with AA on so it wouldnt be that big of a deal to me.

What are your other system specs?

 

[tapir]

I'm pretty sure it's stable with AA off. I'll probably do some more evaluative testing. I know it's not a heat issue even though I thought it was originally because I reseated the HSF and put Arctic Silver 2 on and the temps are pretty low and also because I opened the case and threw a big box fan next to it and I can still get the artifacts to show up. One strange thing is that in the 3dmark06 test the artifacts always start at exactly the same time.

 

[Matt2]

Driver issue?

 

[funboy6942]

Whats the chances it was a lesser card, bio flashed to a GTX, slapped a GTX heatsink on it, but because it is a lesser card running at speeds it wasnt intended to, it is giving you the artifacts. maybe if you go back and re-read the ebay listing it may say it is the card you were to get flashed to a GTX somewhere with a GTX HSF?

If that is the case, slim but since it only does it with AA and those cards are bio flashable to GTX speeds, cant rule it out, it would just be a matter of re-flashing it back to what it was supposed to be.

Also if it is a flasshed card to a GTX it will get rejected for the rma once they get it, and send it right back to making you flip the bill, for they will think you were trying to pull a fast one on them.

So I would see if there is anyway to find out if it is indeed a GTX or not, other then by a massive heatsink which can be bought off other forum members or ebay from someone with a GTX that upgraded to a better one.

 

[tapir]

It's possible that it could be a GTO that was bios flashed but I don't believe so. If memory serves me the retail box was still shrinkwrapped and it was not a used item so they would have to go out of their way to do that. I feel like it may be a driver issue since I still haven't had issues with AA turned off that I can remember. I may do some stress testing that doesnt involve AA and see what happens. I have a feeling I'm going to need to RMA the card though.

 

[giantpinkbunnyhead]

Yeah, I'd definitely check the drivers. Also, do you have a program to monitor the temps on that card? Might help to know that too.

And if all else fails... I'm not positive on this, but my feeling is that 370W isn't really enough to drive a 7900GTX without issues. Having 30 amps on the 12V helps... looks like you've got 29, and I've heard that when you have split rails like you have.. the total amps is less than the sum of the individual rails. Again... that's hearsay and I dont' know how valid it is. But yeah... I might try a new PSU before giving up hope, and a new PSU should be helpful for a ways down the road anyway.

edit: I know your card isn't a BFG, but their own req's for this card state a minimum 400W PSU with 26A on the 12V. I would have to think that similar requirements exist for other companies' cards, so yeah.. you're like... right on the edge with that PSU.
Text

 

[apoppin]

looks to me i could be several issues

you definitely need to test it

make sure you try several drivers ... prefer certified

first of all, that resolution is really demanding ... try 1:1 pixel scaling and run it at a lower resolution with AA on and off just to see

is your CPU oc'd? ... if so, return it to stock and then also run your GTX slightly underclocked
[to "somewhat" test the PS and the theory that it may be a 'flashed' GTX]

what was your previous card? IF ati, make sure you use driver cleaner ... and if it was nvidia make sure driver cleaner did not also uninstall your MB drivers [if nvidia]

what is the rest of your rig like? ... OS .. RAM ...



... and look in the nvidia CP and check your GPU temps ... at idle and under extreme load coming out of a game

try *several* games also and make sure they are patched

 

[tapir]

Well I am becoming more and more convinced this is a driver issue guys. To respond to some of the points you've brought up, this is a fresh OS install that I just did a week ago. Interestingly I installed a new set of drivers this morning that were just recently made available. They fixed an issue that I was having that I made another post about where whenever I put my video card under stress, whether it be with a windowed artifact tester or some games (FEAR), the audio would go choppy, but if only the CPU was under stress (prime64 torture), audio would play fine. This was with either my Audigy original or my onboard motherboard sound. So I installed the 101.70s which I think are brand new and intended for 8800 series and they fixed this issue. Now I can put the video card through hell and my Audigy works fine too.

So I ran 3dMARK this morning to see if it would still crap out in Deep Freeze, and it did. But then I ran it with no AA on at 19x12, and it still crapped out. And just now I tried it at 16x12, and that showed artifacts too. And interestingly the artifacts changed with the driver change and now the screen will tint green at times. Weirdly, the artifacts in the deep freeze test ALWAYS occur EXACTLY at the same point in the test, as the camera view changes from one spot to the other near the end. And the Canyon flight test runs with no problems even though that is an HDR test as well, even at 19x12 with AA enabled.

Just now I ran FEAR at 16x12 with 4x AA and 16x AF enabled. And it ran fine. No problems whatsoever. Originally this game exhibited all of the issues I've described: choppy audio when AA was enabled, occasional random neon circular lights everywhere especially with AA enabled... now it seems to be pretty much fine.

However Half Life 2 is odd, I have been running it only at 19x12 because it accounts for FOV unlike FEAR, but if I enable 2x AA or 4x AA artifacts are everywhere. Reading over what I just wrote I guess it would be prudent to test HL2 at 16x12 as well.

I'm really starting to feel that this is all about the drivers. I suppose eventually I'll install winXP 32-bit as a dual boot and see if those mature drivers can neutralize the problems. I know it sounds ludicrous to be using Vista 64-bit and demanding perfect performance, but my DVD-ROM drive is on the fritz and I need to get a new one before I can install another OS... problems like this just get under my skin.

BTW, the rest of the rig:
Gigabyte GA-945P-S3 motherboard
2GB Corsair DDR2-800
E4300 @ ~~2.7GHz, FSB ~~1200MHz
2x 74GB WD740ADFD raptors,
200gb maxtor drive
120gb old WD drive
Audigy 1 Gamer edition

FWIW, I did try clocking down my CPU to stock speed when I was dealing with the audio crackling problem and that didn't help so I haven't tried it again.

As far as underclocking the GPU... I spent about 3 days figuring out how to install AtiTool so I could change the card's clocks, and I finally got it working, and it won't let me adjust the core clocks. Jeez. I really think its driver issues but all of your commentary and feedback is read and appreciated...

 

[apoppin]

oh Vista ... why didn't you say so



if it also happens in XP ... it's not the drivers

 

[tapir]

GAH... spoke too soon about FEAR, my friend just came in here and started playing and sure enough, weird red tiles appear...

I guess I'll turn off AA for now and get a new DVD rom drive to install XP.

 

[apoppin]

i hate to say it but it *doesn't* sound like drivers ... usually the red tile artifacting is a sign your Card has a HW problem

maybe underclock it ... if it's your PS ... but try it in XP and maybe another rig ...

and good luck

 

[JBT]

I doubt 370watt is enough for that card...

 

[tapir]

yeah PS is looking more like the culprit all the time to me, although I still think it's weird that deep freeze blips out at the exact same time every single time. If you were going to get a replacement trying not to break your bank account, what would you look at?

 

[apoppin]

Hot Deals ... there is a 650w OCZ ... lots of nice PSes currently available in many price ranges

i'd get at least 500w with at least 30a on the 12v rail ...

try underclocking your GTX ... that'll give you an 'idea' first is it is the card or PS ... and maybe try it in friends rig before spending big bucks on a card's possible HW problem

 

[tapir]

I tried to underclock it but can't without reinstalling OS. AtiTool is the only option for changing GPU clocks in Vista and I couldn't get it to work right.

I just noticed that my cold cathodes start blinking like they are not getting enough juice when I put the video card under load :surprise:. This worries me and is pretty convincing that it's a PSU issue. I decided to bite the bullet and snagged a 660W Enermax on fleabay that's SLI certified... got a good price on it so we'll find out soon.

I realized that even if the card is fubared and I somehow have enough juice with this PSU I will likely replace it with an x1950xt or possibly an 8800gts, both of which require even more juice than 7900GTX. This was enough to convince me to buy a better PSU.

Also I can't test in a friend's rig because I'm at school right now and everyone here uses macs... lol

Thanks for all the feedback

 

[apoppin]

why not? ... last month i got a 850w OCZ 'overkill' PS for my x1950p... and the future
... i couldn't o/c it at all on my 480w TT PS ...

good luck ... and let us know how it turns out