eVGA 7800 GT CO Locks Up

[Kaelaan]

Hi.

I was hoping that anyone had any additional insight into my current situation. I have been dealing with this system for the past week and a half and I am about to cut my losses and make a dramatic switch in hardware.

Problem: The computer locks up anywhere from 2 minutes to 2 hours during any game play or 3D benchmark. This never occurs during memory testing, CPU stress, idle or 2D applications. The screen is completely frozen and occasionally, the sound card will “pop” and/or repeat the current sound that occurred during the lock up itself. The computer doesn’t respond to Ctl-Alt-Del and the computer must be reset. If the power button is held for 5 seconds while the lock up occurs, the computer will shut off and never turn on again, until the power is completely shut off at the PSU itself.

The Computer (simple):

Opteron 165, 2GB memory, single eVGA 7800 256MB, 2 WD Raptors (RAID 0), adequate PSU


The Computer (verbose):

Case: Cooler Master Centurion case (decent airflow, CompUSA had a special)
Case Fans: 3 MadDog 60 CFM 24dBA Fans. Located at front (passing through the Hard drives), rear and in a drive bay aimed at the RAM itself.
PSU: Sin-Tek WIN550XSPX-X 500W Power Supply (SLI capable)
PSU: BFG BFGR600WPSU 600W Power Supply (SLI capable)
NOTE: Currently using BFG supply. I swapped supplies to see if I got a different result. I did not. I didn’t bother putting the Sin-Tek back in until I solve the problem as I spent 30-40 minutes wire tying the darned thing.
Motherboard: DFI LanParty UT nF4 SLI-DR Expert
CPU: Opteron 165 Dual Core 1.8GHz (0550)
CPU Heatsink: Thermaltake 4-in-1 Big Typhoon
RAM: OCZ5002048ELGEGXT-K PC4000 2G(2x1024) DDR500 EL GOLD GAMER EXT XTC
Video Card (setup #1): 2 eVGA e-GeForce 7800 GT OC (256-P2-N517-AX) in SLI mode
Video Card (setup #2 and 3): Single eVGA e-GeForce 7800 GT OC (256-P2-N517-AX)
NOTE: I was getting blue screens left and right even when the computer was idle while using SLI mode. Turns out, as I discovered in Video card setup #3, that one of the video cards had faulty RAM. When I plugged the bad card in by itself, the BIOS prompt had funny characters all over it.
Hard Drive (setup #1): Western Digital 36GB Raptor 10k RPM SATA
Hard Drive (setup #2): 2 Western Digital 36GB Raptor 10k RPM SATA (RAID 0)
NOTE: When originally setting up the computer I was setting it up as a RAID 0. During a SuperPi 32M test the computer locked up. When I rebooted one of the hard drives was clicking away. I suspect that the drive had an electrical problem which caused the computer to crash. Needless to say, while I waited for my replacement I had the same symptoms with a single drive set up. I am now, once again, running a RAID 0 setup.
Sound Card: Creative X-Fi XtremeMusic
Sound Card: On-board A97
NOTE: The X-Fi itself was giving me some blue screens in Battlefield2. Turns out that there are some issues with the X-Fi card. So, instead of trying to troubleshoot two problems at once I removed it and used the motherboard sound. Both configurations have the same lockup problem though.
Network Card: On-board Network
DVD/CD Burner: Currently unplugged
Floppy Drive: Currently unplugged
Keyboard: Zboard originally. Now using standard MS keyboard (PS/2)
Mouse: Logitec G7 Wireless Laser mouse originally. Now using standard MS 3 button mouse (PS/2)

Configuration:

CPU: Both 1.8GHz (stock) and 2.5GHz (overclocked). Currently 1.8GHz
Memory: Both 200MHz/DDR 400 (below stock) and 252MHz/DDR504 (slightly above stock, in spec). Currently 200MHz/DDR400. Memory is running in Dual channel 1T.
LDR/FSB dividers: I have tried all options that keep me under a total of 1000MHz.
Voltages: Tested both at spec and slightly above spec for extra stability. Both give same results.
PCI-e Motherboard setup: I made sure that when in SLI mode 8/8 speeds were selected and when not in SLI mode 16x was selected.
On Board devices: Originally, I had all of the devices enabled. However, as of right now, they are pretty much all disabled except for nVidia LAN, nVidia RAID and the on-board sound. USB is even completely disabled. No difference.
BIOS: Latest and greatest BIOS version for both DFI motherboard and eVGA graphics card.
MB Drivers: 6.66 (came with MB) and 6.70. With and without SW drivers, IDE drivers and Network Firewall software. All resulted in same problems.
Video Drivers: 81.85 (came with Video card) and 81.98. I noticed in the release notes for 81.98 a section about instability with Dual Core processors and the latest nVidia Forceware drivers. The work around disables the multiple threads of the Forceware drivers. I have tried both with and without the Dual Core workaround for both OpenGL and DirectX.
Additional Info: AMD Driver and MS Hotfix for Dual Core CPUs have been installed.

Tests:
MemTest86: Test 5 (5 complete sequences), Test 8 (5 complete sequences), Tests 1-8 (10 complete sequences), Tests 1-8 (24 hours) – All completed successfully with no errors.
SuperPi 32M: Completed the tests multiple times on Core 0, Core 1 and on both Core 0 and Core 1 simultaneously. Only once crashed, however, at that moment one of the hard drives had an electrical failure. The hard drive no longer worked after and was replaced. Since then, I never had an issue with SuperPi 32M again.
Prime95: Small FFTs, In-Place Large FFTs and Blend ran successfully multiple times on both Cores.

 

[Kaelaan]

At the moment, I am starting to run out of things to disable on my computer. I am tempted to buy a non-eVGA 7800, an ATI based card, a single Core Opteron and another motherboard for testing. I wish I could disable one of the Opteron cores for testing purposes.

EDIT: I forgot to mention that I am running Windows XP SP2 with all the latest service packs. I have also went into the services and made some changes using msconfig to my Startup items. However, even though I prevented a bunch of additional useless services and processes from running, it doesn't seem to make much of a difference either.

I'm starting to wonder if I have another dead card. Makes me wonder how many cards I have to buy before I get a decent one. The other possibility is a bad motherboard (PCI-e slot).

 

[Kaelaan]

Oh yes. Forgot to mention that I have been monitoring the temperature settings. I have yet to see any problem temperatures arrise just yet. Even though the BIOS/motherboard can result in inaccurate temperature readings, I have to believe them since I have so much extra ventalation and no problems with the computer at all until I run a 3D application/game.

 

[Rattle]

what mobo, power supply do you have and what are your max temps...

 

[Kaelaan]

what mobo, power supply do you have and what are your max temps...

Read the above post. I give all the specs of my system short of the screws that hold the case together.

I believe I have identified the problem. DFI LanParty UT nF4 SLI-DR Expert motherboards will eventually (anywhere from 5 minutes to hours) lock up in any 3D applications in if an eVGA 7800 GT CO is in the 16x slot. The work around that I have found is to change the slot to an 8x in the BIOS.

I have a feeling that other motherboard manufacturers using the nForce4 chipset may be experiencing this problem as well. I have heard of suggestions from ASUS motherboard owners to place the card in the 2nd PCI-e 16x slot. Most people don't realize that even though the 2nd slot is the size of a PCI-e 16x, it only runs at a maximum of 8x. In DFI's case, 2x unless you enable 8x-4x-nc-8x (SLI mode) in the BIOS.

 

[jpersinger]

Well you can see the specs of my rig in my sig; I have had no problems with my EVGA 7800gt in the first 16x slot. Although I do not have an SLI board, it technically is SLI capable and I am in the first 16x slot. What sort of temps are you getting on that 7800? I have had temp problems at stock settings; also I guess I would not be surprised if this is just one of the intricacies of utilizing a DFI board.

 

[Kaelaan]

Well you can see the specs of my rig in my sig; I have had no problems with my EVGA 7800gt in the first 16x slot. Although I do not have an SLI board, it technically is SLI capable and I am in the first 16x slot. What sort of temps are you getting on that 7800? I have had temp problems at stock settings; also I guess I would not be surprised if this is just one of the intricacies of utilizing a DFI board.

I assume you are referring to SLI capable if you "hack" the board. (As posted elsewhere on ocforums)

Not all people are having these problems, but the new DR Expert board that I have, quite a few people have complained about the same problem. Also, I have the eVGA 7800 GT CO model. Pretty much the same thing, just a slightly higher clock speed on the core and memory. The trouble is that it is intermittent and difficult to reproduce. It could be just a few minutes before you get a lock up or it could be hours. When it happens there are no BSODs, bugchecks, event log errors etc to help diagnose the problem. I am also sure that some games that aren't too 3d intensive will never show any symptoms.

My temps seem to never exceed the high 40s when under load and I am at 37 currently while idle. I'm currently using the stock eVGA fan. However, I will soon be switching to the NV Silencer 5 (rev 3) for noise control. Also, I know it definately isn't the PSU after trying two sufficient units. I also make sure that I have all three power connectors on the motherboard connected (yes, 3 .. the latest DR Expert doesn't have the 4-pin HD molex connector as the previous versions had) as well as the PCI-e power connector directly from the PSU.

The problem I have been experiences does stop occuring when switching to 8x mode. I have been playing straight for the past day with no lock ups or crashes. So, I will leave it in 8x for now. I will be getting my second card back sometime soon anyway. Which means I will need to use 8x/8x for SLI mode. Unless of course I switch to the new ASUS motherboard that supports a true 16x/16x SLI mode.

 

[jpersinger]

I have the CO version as well; high 40 C load temp is excellent. At stock settings and around 20C ambient my card can load in FEAR as high as 70C. I have a CM stacker with around 300cfm intake and 250 cfm exhaust; I suspect that I have a bit of stagnant air underneath the card due to the configuration of the case but even with the sidepanel off my card runs hot. The DFI boards are a little ridiculous; they perform well but there are so many flaws that you need to be aware of. The 5 volt mem jumper and the BIOS under certain circumstances applying the highest Vcore multi come to mind. So I'm not surprised that the Expert has flaws as well; I remember it seeming like the initial batches were eating high end procs. Don't worry about running in 8x rather than 16x; I suspect that there is little to no performance difference as you shouldn't be anywhere near the bandwidth max of 8x.

 

[Joeteck]

Oh yes. Forgot to mention that I have been monitoring the temperature settings. I have yet to see any problem temperatures arrise just yet. Even though the BIOS/motherboard can result in inaccurate temperature readings, I have to believe them since I have so much extra ventalation and no problems with the computer at all until I run a 3D application/game.

It is the POWER SUPPLY!!


You claim it to be perfect, but freezing is the common side effect of an bad or over loaded PS.

Try doing the bare minimum.

Video card, one stick of RAM at 2T, one hard drive and no fans...

And everything stock.

 

[Kaelaan]

It is the POWER SUPPLY!!


You claim it to be perfect, but freezing is the common side effect of an bad or over loaded PS.

Try doing the bare minimum.

Video card, one stick of RAM at 2T, one hard drive and no fans...

And everything stock.

Actually, I don't claim it to be perfect. What I did claim is that I tried three perfectly adequate PSUs. One that is an extremely good quality (Sin-Tek) while the other is .. "good enough" (BFG) and a third was an Antec Neo 480 without a PCI-e connector. I also tested the voltages and on the Sin-Tek (my first choice), they didn't budge a bit on the analog meter under load. All the power supplies gave the same results. I don't think I need to try a 4th power supply at the moment.

Also, during my minimalistic testing, I only had the motherboard with LAN enabled (no USB, floppy, IDE or sound) the CPU, 1 stick of memory (not sure what T mode it was in), Video card, 1 hard drive (while I was waiting for my 2nd sata drive for RAID) and CPU Fan.

Just in case, I was thinking that one of the Fans could be causing some impedance issue or something I did disconnect all the fans EXCEPT the CPU fan which I left on for obvious reasons. I placed a fan that I use for the house aimed into the computer with the case open. (I normally do this to computers that I suspect having thermal problems as well to test case ventalation).

All gave the same results. It will lock up in 16x mode eventually when using any 3d application or game. I also had a lock up about 3-4 hours ago in 8x mode. This was the first lock up in 8x mode with approx. 14 hours of constant gaming. I learned two things from this. First, even with my fix there are still some hardware stability issues. Second, I really suck at Battlefield 2.

Anyway, I have a feeling that I have a bad component somewhere. I know the RAM tests great in Memtest86. I know that I haven't been able to show any flaws in the CPU in Prime95 blended mode or Super Pi 32M. I have tried numerous power supplies and removed all of the extra components of my system (swapped keyboard/mouse with ps/2 models, sound, Drives, fans). The only things I see left are the video card or motherboard.

So, about 2 hours ago I went and bought myself a new DFI motherboard. It is the older non-Expert SLI model. I just finished setting it back up before I noticed this post. Tomorrow, I will throw windows and a couple games on it to see if there are any more lock up issues.

 

[jpersinger]

Is the problem only occurring when playing BF2? If so, that is your problem. BF2 can be very buggy; there was about a week where it would actually hard boot my system randomly like you mention. They released an update that was not compatible with win2k.

 

[Kaelaan]

Is the problem only occurring when playing BF2? If so, that is your problem. BF2 can be very buggy; there was about a week where it would actually hard boot my system randomly like you mention. They released an update that was not compatible with win2k.

No. So far the same problem occurs in BF2, EQ2, WoW and in the 3DMark06 Demo loops. Alright, time to check to see if it was my motherboard.....

 

[KillrBuckeye]

Try underclocking the GPU and video memory and see if that helps. These cards, at stock settings, are overclocked compared to a reference 7800GT card. If underclocking fixes the problem, then you know that your particular card just can't handle those stock clocks. In that case, just request an RMA.

 

[Kaelaan]

Try underclocking the GPU and video memory and see if that helps. These cards, at stock settings, are overclocked compared to a reference 7800GT card. If underclocking fixes the problem, then you know that your particular card just can't handle those stock clocks. In that case, just request an RMA.

Right now I'm running 5 continuous cycles (1-8) in MemTest86 just to make sure my ram works with the new motherboard. I ended up going from the Expert to the SLI-D model. Once I throw windows on, I'll see if it starts up again. If it does, I will underclock the card and see if that helps.

 

[Kaelaan]

After playing a bunch of games for a few hours last night, the new motherboard is looking "hopefully optimistic". I think after a week of no crashes, I will say it has a clean bill of health.

Curious to know how many other user's of the new DFI Expert motherboards are having problems. I'm going to RMA this one regardless.