SLI GTS 250 1GB's and having trouble

[phane7]

Bought a pair of PNY GTS 250's with 1GB each. Ever since installing them my computer has been plagued with problems. PNY Tech Support is apparently non-existent. As far as I can tell, I either have a Windows 7 compatibility issue, an issue with using PCIe 2.0 cards in 16x slots, a power supply issue, a heat issue, a bad card (or two), or any combination of the above.

My specs are:
Foxconn NF4SK8AA-8EKRS (nForce 4 Ultra SLI)
AMD 64 X2 4800+
Antec TruePower Trio 550W (3x +12v rails @ 18A each, max 504W on +12v)
4x 1GB PC3200 DDR (CAS 2.5)
Samsung 1TB 7200 RPM HDD
20x DVD+-RW drive (Generic, not sure brand)
Rosewill RCR-IM5001 75 in 1 memory card reader/USB hub/eSATA expansion
Running Windows 7 Home Premium 64 bit

My power supply is an Antech TruePower Trio 550W, which as far as I can tell is adequate as far as wattage but I've seen mixed answers as to whether or not the 504W max on the +12v is good enough for 2 GTS 250's. From what I can see on reputable review sites it's safe to assume that each card would need a max of 160-170 watts on the +12, but some people on forums are claiming more.

I have an intake fan at the bottom front, an exhaust fan below the PSU, the PSU has a downdraft fan, and my case has side vents by the video cards and a direct duct from the CPU fan. I have no expansion cards at all aside from the video cards themselves. The case itself is in a pretty open area and in the coolest room of the house, so I would assume that temperature wouldn't be an issue, but I'm getting some minor jittery polygon errors in Bad Company 2 that's mostly noticeable in cutscenes (don't know if that's normal or my cards).

I was running two passively cooled 7950 GT's in SLI for years without issue. Ever since slipping these two GTS 250's in there my computer seems to randomly power on and not go back into suspend without crashing. I've also got an occasional (but rare) crash while in game, however sometimes these cards seem to work for hours of intense abuse without a single crash. The time they are most likely to crash my system (as odd as this may sound) is when going into and out of S3 suspend.

I've tested each card individually and can't seem to find an issue with either card (I haven't let one sit by itself overnight while in S3 suspend yet, which seems to be when problems really surface). I've completely uninstalled my driver and used 3D Guru's DriverSweeper v 2.1.0 and reinstalled drivers after that.

Based on the examples in this website I've had my system crash twice with something similar to what they show as the first Video Ram Artifact example and had Modern Warfare 2 lock up once with something like the example of a Bus artifact. The minor polygon errors in BC2 weren't as bad as the examples in this page, someone's cheek might jump out an inch or two at an odd angle for a single frame and then it goes away. It might not have even been noticeable without stereoscopic 3D.

 

[multiscreenz]

You need more power. There's a good chance that's the problem. I'd get at least a 750 Watt PSU.

 

[phane7]

According to various PSU Calculators (Seems like they are all the same) I should be fine with a 495 watt PSU even allowing 30% capacitor aging, and I detailed in my original post how I should be OK on the +12v rail as well. I'd like to know specifically why you would think I'd need more power, as all the solid information I can find says that I shouldn't.

Aside from that, if it were a power problem, wouldn't I be running into frequent crashes under heavy load rather than mostly in S3 suspend (when the power load is actually LOWEST).

 

[builderbobftw]

You need more power. There's a good chance that's the problem. I'd get at least a 750 Watt PSU.

These are GTS 250s, not GTX 480s lol.

 

[phane7]

I went ahead and ordered a TruePower New 750. I figure with 4 25A rails it should be total overkill based on anything I've read anywhere, so that should at least eliminate one potential problem.

Does anyone know if this problem could be related to using PCIe 2.0 cards with an nForce 4 motherboard? I'm worried that my m/b might just be too old and not be able to handle the power management correctly.

 

[megamanx00]

Yeah, a few early nForce 4 boards are not 100% PCI-E complaint. Have you checked to see if there are any BIOS updates for stability issues? Most likely it's just your board not properly handling suspend. The solution of course is just to shut it down rather than suspending it so it doesn't randomly turn on ^_^. I also think that since the cards are right next to each other on your board one of them may be overheating. Even though the air in the case may be rather well ventilated, air may simply not circulate well around the video card fan for whatever reason. Try switching the order of the cards. Also make sure that the SLI bridge is secure, if you have another one try it instead, and make sure that the molex connector on the board to supply extra power to the PCI-E slots is properly secured.

 

[Griffolion]

These are GTS 250s, not GTX 480s lol.

The crowd kindly refers bob to the funny meme detailing a picture of sellarfields nuclear station plugged into 2 480's in SLI.

 

[builderbobftw]

Bragin at a tad over 1,000$

 

[builderbobftw]

Hehe.

But a nice Cisco 1500W+ unit could maybe handle it...

 

[Griffolion]

Haha overkill even for the 480's!

Apparently a 1KW supply will do the trick.

 

[builderbobftw]

Apparentyl, EVGA got 4 480s maybe possibly running on what at first glance appears to be a high-end 1.2kw unit (wich may be a Corsiar style unit, meaning it can do a few hundred more watts than advertised) .

 

[phane7]

Figured I'd update you guys. After fidling around with these for a while I thought I had everything working fine and then one of the two cards just flat out bombed and doesn't work at all anymore. I'm willing to bet that all of my problems were floating around one defective card because my system was having 10x as many problems when the card that eventually died was in PCIe slot #1 and seemed to work fine (until it failed altogether) when it was in PCIe slot #2.

Chalk one up to PNY tech support, they said I need to send *BOTH* cards in under the RMA at my own expense because they can't guarantee that they're going to be able to replace one with the same card for SLI (they claim they won't downgrade me, though). I emailed them the info they claimed to need on Saturday and it's already COB on Monday and haven't heard back on even getting the RMA started...

 

[builderbobftw]

...They would need ot give you a GTX 470 to avoid a "downgrade"

 

[BigBouy]

Hey,

I dunno if you are still having issues here but I have 2 250GTS in sli and I was also having problems with win 7 64bit - The solution for me was to install win 7 32 bit and I was hassle free (pretty much) immediately after... I know 32bit isn't Ideal but its a solution for the moment for me until whatever it is that's causing the problem gets a working driver....

 

[chewrock]

I'm having the exact same problem. Win7 64bit Professional, two GTS 250s in SLI. Fabulous graphics for two months, then the secondary card quit. Primary card is still fine. I'm so upset. This was the best graphics setup I've ever had. I didn't even overclock them. They were running at factory clocks. Can't get PNY on the phone.

Oh, since people are posting their setups, here is mine:

XFX Nvidia 780i, 4gb G.Skill 1200mhz DDR2, Intel Q9400, Zalman 10Flex heatsink w/2 SilenX 11dba fans, Antec P183 case w/3 SilenX 14dba case fans in top chamber and one Enermax Magma in the bottom chamber (power supply & hard drive area), Raidmax 680watt power supply, E-Mu 1212m soundcard, 1x (used to be 2x in SLI) PNY GTS 250 1gb video. Oh, one more thing. I built a custom fan shroud for my chipset so that I could use an 80mm SilenX fan instead of the XFX factory fan, which was a wussy looking 40mm that couldn't blow out a birthday candle, much less cool down an Nvidia 4 series SPP/MCP.

 

[Mousemonkey]

@CleetSR388, start your own thread rather than dragging up and hijacking somebody else's.