What is wrong with my video card?

dderidex

Supreme [H]ardness
Joined
Oct 31, 2001
Messages
6,328
Not a gripe post, but I'm looking for thoughts.

I've got something VERY wrong with my system, and I want to get opinions on what I should be looking at to try and figure out how to solve it!

Is it power related? Heat? Cabling? Defective card? WHAT?

Anyway, posting a pic, but PLEASE READ THE DETAILS BEFORE POSTING!!

omgwtf.jpg


Okay, you see the double groups of multi-colored garbage on the screen (top and halfway down)? And then the black lines of corruption around the time values? That's actually EXACTLY what I see on the display....and they stay put, kinda changing patterns over time a little.

But they don't appear all the time!

In fact, at 1024x768 with 4x anti-aliasing and image "post-processing" (bloom) turned on....the image is 100% fine!

At 0x anti-aliasing, and with "post-processing" turned on, I'm fine all the way up to 1680x1050 (my monitor's max resolution). No image corruption anywhere.

A 2x anti-aliasing, and with "post-processing" turned on, I'm fine all the way up to 1680x900 (the res option in 3dMark03 just below 1680x1050) - no corruption anywhere. At 1680x1050, I just get one single band of the corruption at the very bottom of the screen, and it's all black (no multi-colors - like you see above, just without the 2 upper groups of corruption).

At 4x anti-aliasing with "post-processing", the corruption starts at 1152x864 - that's actually what res the pic I snapped above is. If I turn off "post-processing", then I get just the black bars of corruption at the bottom up to 1680x1050, at which point I get the multi-colored lines of corruption back again.

The card I'm using is a BFG 6600GT OC. Only, it's not even at stock clocks for the pic above, I actually ran it at 370 core / 370 mem (DDR740) just to try and eliminate variables. Max core temp was 61C during the test - and, although it runs hotter at stock clocks (over 80C), the image it generates looks the EXACT same - no better or worse.

Anybody have any thoughts on where this problem could POSSIBLY be?!
 
Hmmm....*might* just have figured this out myself.

omgwtf.gif


I bet you'd never guess how many times I ran 3dMark03 in that timespan? Unless, of course, you guessed '3', in which case I'd have to assume you were some kind of psychic or something.

Boy, does that give me a warm, fuzzy feeling about Shuttle's power supplies. That's a stock SN25P PSU running a stock Athlon64 X2 '4200+' system with 2gb of ram - nothing overclocked or overvolted or anything.

Niiiiiice.
 
Just before you blame the PSU, put in another video card. I've experienced the exact same kinds of voltage hops that you're observing, and it turned out that the GF3 that my client was using was the cause. All the other cards that have been put in the system worked fine, but any system we inserted the GF3 into led to massive voltage dips and hops. Shuttle does put decent PSUs in their rigs, and while you may indeed have a bad PSU, check the videocard first.

 
Well, keep in mind, this is the THIRD video card I've used in a week.

I started out the week with a freshly-received-from-Newegg BFG 6800GT OC. Unusable, as the driver would keep interrupting 3d games with a message 'the nVidia sentinel has detected this card is not receiving enough power, and is dropping clocks to a safe level' (words to that effect). The resulting clocks were 200mhz on the core in 3d mode, so I sent that back (thinking it was the card) and ran up to Best Buy and picked up a 6600GT.

That ran fine in 2d mode, but I tried firing up a 3d app - started running it - went to take a shower and came back to find the PC dead. Wouldn't boot up! Wouldn't even power ON! Including no LEDs....even on the motherboard. Or the video card. Hmmm...this makes me suspicious, so I pull the video card out and....it boots fine. Put it back in, no boot. Take it out, boots fine again.

Odd, so I take that back to Best Buy and exchange it for the same type, and that's the card I'm running on now the results above show.

I think it's safe to say this PSU sucks.
 
What PSU exactly? What system specs? And what are the multimeter readings?
 
Spectre said:
What PSU exactly? What system specs? And what are the multimeter readings?

It's the stock SilenX 350W PSU that comes with the Shuttle SN25P.

System is:
AMD Athlon64 X2 '4200+'
2gb Corsair XMS TwinX PC3200 (2x 1024mb)
1x 200gb 7200rpm Seagate SATA150 hard drive
1x Lite-On CDRW/DVD-ROM optical drive

And that's it.

No overclocks on anything. No overvolting, no CCFLs, no exotic fans (just stock fans, in fact), nothing.
 
dderidex said:
It's the stock SilenX 350W PSU that comes with the Shuttle SN25P.

System is:
AMD Athlon64 X2 '4200+'
2gb Corsair XMS TwinX PC3200 (2x 1024mb)
1x 200gb 7200rpm Seagate SATA150 hard drive
1x Lite-On CDRW/DVD-ROM optical drive

And that's it.

No overclocks on anything. No overvolting, no CCFLs, no exotic fans (just stock fans, in fact), nothing.

Do you have the multimeter readings? I ask because 10.6v's on a 12v you would not be screen capping that image. Software is not accurate.
 
350?? whew.... yeah - that's right in the grey area of "it might work, it might not" depending on the PSU build.. like how some of the same make/model are a little better than others. I think the message you saw saying the video card was not getting enough power should have clued you into that. :) Anyways, some folks have been able to run 6800 and 6600 series cards in that case with that psu from other threads I have read... the 7800GTX runs in it also:

http://www.hardforum.com/showthread.php?t=924941

But I think the psu in your paticular machine is just a little weak and thus you are having these issues... I would see if you can invoke a warranty replacement of that, else buy a new one. Good luck!
 
Have you tried diconnecting peripherals that you don't need, to free up some power, and see if it works then?

 
DamienThorn said:
Have you tried diconnecting peripherals that you don't need, to free up some power, and see if it works then?
Like WHAT? The *only* think you could possibly interpret as "not needing" is the optical drive - I don't *have* anything else on it!
Spectre said:
Do you have the multimeter readings? I ask because 10.6v's on a 12v you would not be screen capping that image. Software is not accurate.
Oh, I wouldn't count on that.

When I fire up a 3d game, there is a *noticeable* drop in the fan speed. I'd guess they drop 25% or so, based on the volume change.

I didn't really pay attention to that until I started suspecting a PSU problem.

revenant said:
350?? whew.... yeah - that's right in the grey area of "it might work, it might not" depending on the PSU build.. like how some of the same make/model are a little better than others. I think the message you saw saying the video card was not getting enough power should have clued you into that. :) Anyways, some folks have been able to run 6800 and 6600 series cards in that case with that psu from other threads I have read... the 7800GTX runs in it also:

Actually, the SilenX 350W are generally considered some of the most solid PSUs around. 250W SilenX PSUs are perfectly happy running GeForce 6800 Ultras.

Anyway, I posted this over on the sfftech forums, and another user noted that on his SN25P with GeForce 6600GT, it shows 11.9v on the 12v rail, and it only dropped to 11.85v when doing the same benchmark I did.

Which makes it sound like I've just got a crap PSU. Too bad they aren't replaceable aside from Shuttle warranty (which, if I understand correctly, I need to send my entire system in for them to analyze before they honor. :mad: )
 
I just wanted to say... I bought a BFG 6800 OC a while back and had similar problems. The card itself was very thin and seemed to not fit in the AGP slot correctly... Tech support had no idea, so I blamed my PSU. I didn't get one and instead returned the card. I picked up an eVGA 6800 and have had no problems since. The eVGA card seemed of a lot higher quality as well. The BFG card was kinda flimsy.
 
dderidex said:
Like WHAT? The *only* think you could possibly interpret as "not needing" is the optical drive - I don't *have* anything else on it!

Take out a stick of memory, and the optical drive. It's not a lot, but it is something. Also, try swapping out all your components, one at a time, to make sure that you don't have another part that malfunctioning and drawing too much power when it's in use.

 
dderidex said:
Actually, the SilenX 350W are generally considered some of the most solid PSUs around. 250W SilenX PSUs are perfectly happy running GeForce 6800 Ultras.

Anyway, I posted this over on the sfftech forums, and another user noted that on his SN25P with GeForce 6600GT, it shows 11.9v on the 12v rail, and it only dropped to 11.85v when doing the same benchmark I did.

Which makes it sound like I've just got a crap PSU. Too bad they aren't replaceable aside from Shuttle warranty (which, if I understand correctly, I need to send my entire system in for them to analyze before they honor. :mad: )

ok - yes, but what I was getting at is that not all psu's of the same make/model are created perfectly equal.. small errors in the fabrication process and/or issues with components used can cause irregularities with power flow/stability under load, etc.. so looks like yours is one of those which made it past quality assurance, but maybe should not have? well, it looks that way... Anyways, I see an RMA in your near future... or something like that. Regardless, I hope you find an easy path to solving this issue. :)
 
DamienThorn said:
Take out a stick of memory, and the optical drive. It's not a lot, but it is something. Also, try swapping out all your components, one at a time, to make sure that you don't have another part that malfunctioning and drawing too much power when it's in use.
Well, I tried pulling a stick of memory, and I actually unplugged the optical drive, floppy drive, and hard drive - figured the +11v on the +12v rail being part of the problem, just booting to the BIOS diagnostic screen should be enough to see if any difference was made.

None, of course. Still showing only 11v on the +12v rail.

GOTTA be the PSU.
 
Well, if that's the case, just grab a multimeter to be certain the PSU is consistantly giving issues, and either RMA it or buy a new PSU.

 
DamienThorn said:
Well, if that's the case, just grab a multimeter to be certain the PSU is consistantly giving issues, and either RMA it or buy a new PSU.
Can't buy a new one - Shuttle uses proprietary power headers on the mobo, plus a proprietary case design.
 
Back
Top