PDA

View Full Version : Please Check My PSU Testing (8800GTX SLi Instability)


Hurin
11-17-2006, 12:40 AM
Hi All,

My 8800GTX SLi rig is crashing during "Firefly Forest" in 3DMark06 nearly every time. It seems okay during the other tests.

I figured it might be power-related. I have an identical system here that I'm building for a friend but it only has one 8800GTX (no SLi).

Both systems use Seasonic M12 700W.

So, In order to rule out the power supply I:

*Put the systems side-by-side.

*Unplugged one of my 8800GTXes from my power supply.

*Ran a PCI-E cable from my friend's power upply to the PCI-E power connectors on my 8800GTX.

*Powered up my friend's system and waited for it to get to the Windows Desktop -- so its 8800GTX would be idle.

*Powered on my system - thus drawing power for one 8800GTX from my internal power supply, and drawing power for the second one from my friend's PSU.

*My rig still hung up/restarted in Firefly Forest. Thus ruling out power issue?

So, here are my two questions:

1. Was this a good test? With my friend's sitting at idle, I assume that his 8800GTX was leaving plenty of power in reserve on his system for my second one.

2. I had a momentary scare when I realized that I didn't know which computer to shut down first. If I shut down my friend's first, one of my 8800GTXes would probably complain about lack of power. Didn't want that. . . so I shut down mine first. . . but then I have a "turned off" 8800GTX connected to a PCI-E cable that was still sending juice. Could that have damaged anything?

Thanks for any feedback! I'm hoping that these video cards are just so new that there is a stability problem with the drivers. . . and its not my video cards, or my motherboard, etc.

Best,

H

Tazz
11-17-2006, 06:23 AM
You test the cards individually for stability?

jonnyGURU
11-17-2006, 06:24 AM
1. I don't think it's the PSU. EVeryone is so paranoid about G80 cards needing so much power that the first thing they assume when there is a problem is not enough power. Could still be a bad (defective) card (see below.)

2. No. Although the lead was live, it's not like the card was drawing any power.

You have SLi: Try the build one card at a time (as Tazz suggests.)

If one of the cards are bad, it might lock up or BSOD by itself too. Also could be a bad motherboard or just a driver issue (I couldn't play Quake4 on my XP 64 rig until just recently and the only thing that changed were a few Windows Updates.)

Slartibartfast
11-17-2006, 08:54 AM
It could be a heat issue: by the time the system gets to that scene is could be too hot inside.

Hurin
11-17-2006, 12:38 PM
Actually, I didn't think it was likely to be power. But seemed like the first (and easy) thing to rule out.

Thankfully, not only do I have SLi, but I actually have a third 8800GTX as well in my friend's system that I'm building. Right now, it actually looks like two of the three cards are defective as only one of them can loop 3DMark06 for more than a few hours. Though, I'll only state that with confidence when the "good" one makes it through a 24-hour loop. (It's at six right now. . . the bad ones crash around 30 min to three hours).

Thanks for the input guys!

Edit: I think I ruled out heat via the "Big house fan blowing into the side" method. But ya never know.

jonnyGURU
11-17-2006, 02:19 PM
Could be that problem with the incorrect resistor.

You know that BFG 8800GTX I had?

Bad resistor and had to get RMA'd.

Hurin
11-17-2006, 05:06 PM
Could be that problem with the incorrect resistor.

You know that BFG 8800GTX I had?

Bad resistor and had to get RMA'd.
Been wondering about that. But if it is, it's likely that they either screwed up the fix or missed it in testing. . . because the anti-static bag was "double sealed" with a note that said the product had been tested for quality assurance purposes.

H

jonnyGURU
11-17-2006, 06:14 PM
. . because the anti-static bag was "double sealed" with a note that said the product had been tested for quality assurance purposes.

So? Mine was too.

They're all tested for "quality assurance," yet thousands of units shipped with the incorrect resistor.

You need to call whoever made your card, give them the serial number, and get see if it's one of the cards w/ the resistor issue.

CAD4466HK
11-17-2006, 07:22 PM
. . . because the anti-static bag was "double sealed" with a note that said the product had been tested for quality assurance purposes.

I know I should not laugh but, "quality assurance", what's that:confused:
Seems to be a typo on your "double sealed" anti-static bag,
It should read as follows; "lack there of quality assurance";)

Hurin
11-17-2006, 07:37 PM
Maybe I wasn't clear. . . the static bag had been sealed. . . then the seal was broken. . . and a second seal was put over the first seal. I assumed this was done when they got the cards back to check and/or repair the resistor issue.

Having said that, I could definitely be wrong!. . . if I can't get this sorted out soon, I will indeed be sending these cards back.

Thanks!

H

P.S. I got them from Newegg. I'd have thought that BFG would have been careful with their Newegg stock! :(

jonnyGURU
11-17-2006, 07:52 PM
Oh....! I get what you're saying....

No no no.... They didn't find out about the resistor problem until two days before launch. Even if those cards got sent back on Monday, there was NO WAY they could have been fixed and sent back to the retailer in a week. I M P O S S I B L E.

Re: PS: Why would BFG treat Newegg any differently than any other retailer?!?! Give Newegg the "known good stuff" and everyone else the shaft? That's good business practice... NOT!

CAD4466HK
11-17-2006, 08:05 PM
Maybe I wasn't clear. . . the static bag had been sealed. . . then the seal was broken. . . and a second seal was put over the first seal. I assumed this was done when they got the cards back to check and/or repair the resistor issue.

Having said that, I could definitely be wrong!. . . if I can't get this sorted out soon, I will indeed be sending these cards back.

Thanks!

H

P.S. I got them from Newegg. I'd have thought that BFG would have been careful with their Newegg stock! :(


Thanks for your sake for clearing that up;)
See that mobo in my sig?Got it from Newegg. First one had a nick in the CPU area on the back of
the mobo, second one had a nick on the back also, but on the memory instead, which still made it DOA:mad: The warehouse guys must take lunch on the merchandise, after they gank a few SATA cables and SLI bridges
on their 10 min breaks:p

See that 7900GT KO? First bought it from Newegg, they woulden't even deal
with me on the first RMA, they said it was eVGA's fault, which I thought would be better, right? Hell no, at least Newegg is pretty good at RMA's;) most of the time.:o

Hurin
11-17-2006, 10:29 PM
Oh....! I get what you're saying....

No no no.... They didn't find out about the resistor problem until two days before launch. Even if those cards got sent back on Monday, there was NO WAY they could have been fixed and sent back to the retailer in a week. I M P O S S I B L E.

Re: PS: Why would BFG treat Newegg any differently than any other retailer?!?! Give Newegg the "known good stuff" and everyone else the shaft? That's good business practice... NOT!
Well, launch was on Wednesday. They actually found out about the problem and initiated the recall on prior Friday. I have a feeling a lot of the stuff that had not yet been ship to retailers was inspected and/or fixed and then shipped. But you're right, the odds of getting cards back to BFG (or anyone), fixed, and back to the retailer in only four business days is unlikely. I have a feeling that they replaced on-hand stock as much as possible, but you'd hope they'd yank un-tested/un-fixed stock from the shelves. But, maybe I'm too optimistic.

I've now got one stable system. . . last test I moved the RAM, video card (1), and Audigy (only because it was easy to move) to the other computer, and the instability followed those parts. So now, it's time to narrow it down from there. I'm hoping it's the RAM!

Bbq
11-18-2006, 12:12 AM
Well, once I got a higher-end board: Two opteron 2/8xx's, two pci-e, some pci-x, in eATX. It was great, except that it was missing 2 ram banks. And it had a QC sticker an inch away from it.

Hurin
11-19-2006, 02:59 PM
Update. All three cards have the bad resistor on them. Though, this doesn't explain why one card is totally stable.

I'm sending all three back. Been on the phone with BFG. . . debating whether I have better odds of getting a good (non-refurb/non-repaired. . . just good from the git' go) new card from BFG or by returning them via Newegg. To be honest, I'm a bit perturbed at Newegg for not following through on the recall and selling cards that shouldn't have been sold. They should share a portion of this pain. :P

Here's my thread (http://www.hardforum.com/showthread.php?t=1122072) on [H] with more details.

H