PDA

View Full Version : Evga GTX 280 FTW crash problems - not overheating


Pages : [1] 2

darkrider01
06-22-08, 11:55 PM
Sorry for the novel-like post, but I'm desperate now.


Problem: Start a game - after a few minutes, screen goes black, PC shuts completely down.

I have experienced this problem with Age of Conan, Crysis, Guild Wars, and Eve Online. I can try other games, but since none of these will run more than 5 minutes, I see little point in trying. rthdribl also crashes after 10-15 minutes. Temps in rthdribl usually peak at 69C.

Card runs at 40-45C idle, 68-75C under load. Ambient case temps are around 25-30C.

For 2 + days I have been trying to root out the cause of this problem. Below are some of things I have tried in no particular order:

1. Installed newest BIOS
2. Reinstalled video drivers (from disk)
3. Uninstalled video drivers - ran driver cleaner - reinstalled drivers
4. Downloaded and installed latest drivers from Nvidia's site - same process as #3.
5. Removed all my RAM, tried a single stick - did this with all 4 sticks (2x1GB and 2x2GB)
6. Moved connectors around on my PSU (modular)
7. Turned off everything in room except for PC and monitor.
8. Updated all DirectX components
9. Cried, got mad, and gave myself a headache.
10. Underclocked my RAM down to 800MHz.
11. Ran my processor at stock clocks - with RAM at stock and underclocked.
12. Ran video card at reference clocks.
13. Set fan speed to 100%.
14. Tried every combination of BIOS settings I can think of.

Anomolies:

I had Crysis run once for 15 minutes. I stopped playing, did a little dance, and sat back down in my chair and started up Age of Conan. Within 5 minutes, my PC was at a black screen again and then shut down.

My XFX 8800GTX ran flawless in this rig. None of the problems came about until the 280 GTX was installed.

1000W Ultra X3 PSU with 70A on the +12V rail. PSU is equipped with both 6 pin and 8 pin PCIE connectors.

Plans:

I'm ordering a new motherboard tomorrow. Probably going for a Gigabyte GA-X48-DS4 or maybe the Gigabyte GA-EP45-DQ6. I've been looking at the Abit X38 Quad GT, but customer support horror stories have me concerned. I'm done with MSI motherboards forever.

I may send my RAM in for RMA and see if patriot will send me 4 matched sticks. All the sticks I have now are all DDR2 1066 with 5-5-5-15 timings, but all matched might make a difference.

Is there anything else that I'm missing? Short of reinstalling windows, I can't think of anything. Could my board being PCIE 1.0 instead of PCIE 2.0 have anything to do with it? Anyone have any suggestions?

Evga tech support said take it to a shop and have them test it. I think my testing methods have been far superior to and more exhausting than anything a shop would do.

jAkUp
06-23-08, 12:42 AM
Try running the 6pin to 8pin converter in the box, sometimes this can avoid overloading on particular rail.

spajdr
06-23-08, 03:23 AM
Seeing that mobo, gfx card produce alot of heat that raise hot air into chipset cooling, did you tried to put gfx card into second PCIE slot, more away from northbridge cooling? I guess you have plugged 8pin ATX connector into motherboard?

darkrider01
06-23-08, 09:01 AM
Try running the 6pin to 8pin converter in the box, sometimes this can avoid overloading on particular rail.

I have not tried this yet, but I will tonight. I had thought about doing this, but thought the PSU should be able to handle it. It is definitely worth a shot.

Seeing that mobo, gfx card produce alot of heat that raise hot air into chipset cooling, did you tried to put gfx card into second PCIE slot, more away from northbridge cooling? I guess you have plugged 8pin ATX connector into motherboard?

The 8pin mobo connector is plugged in. I have a fan blowing directly on the Northbridge cooling fins. The GTX 280 doesn't cover the Northbridge any more than the 8800GTX did. I have not tried the card in the second PCIE slot as it is a 4x slot.

spajdr
06-23-08, 09:12 AM
Pretty weird for sure, if nothing overheat then it must be some mobo problem, or PSU is not compatible with GFX card, it happened before with some 1kW PSUs and 8800Ultra. I bet though its motherboard.

hell_of_doom227
06-23-08, 09:18 AM
I was thinking between regular clock speed or super overclocked eVga and i figured i will stick with stock speed. A little extra Mhz is not worth troubles...
I never overclocked these 8800GTXs i'm still running. I'm against overclocking video cards. I think you lose more then what you actually get.

jAkUp
06-23-08, 11:32 AM
Well, since the PC completely shuts off, to me it sounds like a power issue.

mailman2
06-23-08, 11:40 AM
Does it crash at any other point or just when pushing the GPU? A power issue would be random and usually will reboot, not shutdown unless something in the PSU is tripped. Sounds to me more like a heat issue than a power issue. If its only while pushing the GPU and if the machine isnt rebooting it's more than likely an issue related to the GPU. Time for an RMA.

mullet
06-23-08, 11:40 AM
Try running the 6pin to 8pin converter in the box, sometimes this can avoid overloading on particular rail.

Its a 70A single rail.

Pointguard1122
06-23-08, 04:45 PM
I have a similar issue.

I have an EVGA GTX280 on a EVGA 780i mobo with Vista x64 Ult, 4GB RAM, Corsair 750W PSU. At random times the screen will turn off. The video card seems to stop displaying video. The fan on the card also turns off. EVGA really has no idea what could cause this. I was told either:
1. faulty card
2. PSU issue
3. driver issue
A member on the EVGA forum said that nvidia knows of this and is working on a driver fix. I am going to try another PSU tonight to see if it continues to happen. The same system had no trouble with a 9800GTX or a HD4850. I might take this card to a friend to test in his system. I will post my results later in the week.

turdhat
06-23-08, 05:36 PM
Checking your event view for mem dumps or indications that the macine is bluescreening but you are not seeing it ? Do you have it set to shut down on bluescreen instead of it showing you the error. Check those options because if you are missing a blue screen and could get an error message it at least might help you troubleshoot.

darkrider01
06-23-08, 10:48 PM
Well, I'm extremely confused by all this, but here is some additional info.

The computer wouldn't just shut off - it would go to a black screen. The sounds would continue for a couple of seconds, then it would turn off. Almost like the computer was shutting down.

The event logs just told me that "The previous system shutdown at 11:51:37 PM on 6/22/2008 was unexpected."

Try running the 6pin to 8pin converter in the box, sometimes this can avoid overloading on particular rail.

Its a 70A single rail.

Here is the confusing part - I replaced my 8 pin PCIE connector with 2 x 6 pin -> 8pin adapter - now it's working great. I had Crysis, Eve Online, and Guild Wars all running at 1680x1050 windows on my desktop, and it didn't die. I ran rthdribl and Furmark in full screen at 1920x1200 for over 30 minutes each. Temps were hovering in the mid to high 80's. I've mailed Ultra support to see what they say. I'm gonna reboot and try to OC my CPU back to 3.2 GHz and see what that does.

Why would running the 2 to 1 adapter make a difference if there is a single 70A rail? It looks like the the wires may be too thin to carry the required current for the card if this does actually correct the problem.

Maverickman
06-23-08, 11:03 PM
I had this happen once when I played COD4 with dual cards enabled. I used to do this when I had SLI. The computer just completely shut down after about 5-10 minutes of playing. When I set dual cards back to "no," the problem didn't happen again. I've played Crysis at 1920X1028 with most settings on Very High with no problems. I even played more than an hour. One time the screen froze for a couple seconds during a firefight, but the game then continued with no problems. I'm using one PCI-E six pin and a PCI-E 6-pin to 8-pin adapter. I have a 750W power supply.

darkrider01
06-23-08, 11:30 PM
Also, This thread (http://www.nvnews.net/vbulletin/showthread.php?t=115281) has provided me with some info I was unaware of. It may be that the combination of my current board only having PCIE 1.0 (1.1?) and the PSU using a 6 pin to 8 pin cable was starving my card for power.

I guess I'll find out tomorrow / Wednesday if that is in fact the case. I ordered a Q9300 and Gigabyte GA-X48-DS4. (nana2)

mullet
06-24-08, 12:24 AM
darkrider01, glad its running good now. Weird about the wires but you may be right if there thin.

Amuro
06-24-08, 01:01 AM
I'm also using Ultra X3, but have no issues with dual 280s SLI. When did you buy your PSU?

darkrider01
06-24-08, 08:25 AM
I bought it a few months ago, but the thread I linked in one of my earlier posts may give some insight into the problem. From the other thread I linked (This thread) (http://www.nvnews.net/vbulletin/showthread.php?t=115281), the fact that PCIE 2.0 slots deliver 150W could be the difference between why your setup has no problems, and mine does. My new X48 board will be here today and I'll know for sure after I have it installed and running. My plan is to remove the "temporary) 2 x 6 pin to 8 pin adapter when I install the new motherboard.

dancid
06-25-08, 02:09 AM
I'm having a similar problem as far as I can tell. I got the EVGA gtx280 superclocked edition, and i have a 660 watt PSU. I turn on age of conan and after about 5 minutes exactly as other say, my computer goes dead, the screen doesn't go blank, there's a pop in the speakers (which is what my comp does when it shuts off) and all fans everything goes off simultaneously in a split second, it so sudden I don't get sad until about 2 seconds after it's off :( It only happens when i'm playing conan (haven't tried any other games, because i've not installed any). If i install the superclocked editions OC software that comes with it, it automatically clocks it way too high and the temp goes up over 250F. If i shut off the application and it doesn't seem to do it. So i've been monitoring and adjusting fan speeds with nTune. For the sake of power consumptions, i've unplugged two hard drives, which leave me with single OS drive, also removed 1 cd drive, and '2' 80mm case fans. There are people on here with 1000watt PSU's which makes me think this problem may not be universally correctable until nVidia releases the patch. But if you guys have an suggestion for my specific situation, I'd be eternally grateful. I'd like to avoid RMA if I can. For the five minutes it does run conan, it runs it well.

Here's my specs:

one EVGA gtx280 superclocked edition (pci-e slot 1)
Core 2 Duo 3.0 ghz
Modular Raid max rx-630a 630 watt PSU
g skill DDR2 800 5-5-5-15
Asus p5n-t deluxe, triple SLI (only '2' pci-e 2.0, the other is 1 or 1.1 dunno)
nForce 780i chipset (mother board has no fan on bridge, but has a copper pipe system which extends to other parts of the boards, but is based at the bridge with a large copper heatsink. I've seen pictures of this motherboard from manufacturer with a cyclone fan on top of the bridge, however, i got an open-box motherboard from new egg, wish i hadn't. It seems to run faily cool anyhow.)
500 gig seagate serial ATA2
Sony DVD rewritable

I apologize if this post is TMI, i'm new to tech forums.
Thank you all for your help

spajdr
06-25-08, 02:27 AM
dancid : Did your PSU got 8pin PCIEx connector? or you use 2x6pin PCIE connector to 8pin PCIE connector?

dancid
06-25-08, 03:33 AM
Yes i forgot to mention, my PSU comes with two 6 pins and one 8 pin, but my motherboard uses the 8 pin. So i'm forced to use BOTH plug adapters. I also forgot to mention that I'm running vista bus. 64

Lfctony
06-25-08, 03:39 AM
Yes i forgot to mention, my PSU comes with two 6 pins and one 8 pin, but my motherboard uses the 8 pin. So i'm forced to use BOTH plug adapters. I also forgot to mention that I'm running vista bus. 64

The motherboard 8-pin is different than the video card 8-pin. Thank God for you that your motherboard uses the 8-pin EPS and not the 4-pin, or you would have used that and burned your video card... :headexplode:

dancid
06-25-08, 04:00 AM
they put the same plug shapes and pin number on a power supply plug and expect you to differentiate between the two? :eek: Do you think I'll have to wait for a driver than? :thumbdwn:

Pointguard1122
06-25-08, 11:12 AM
Just a quick update on my situation. I took the card to a friend's house and he is having NO ISSUES. My next step is to try using the adapter for the 8 pin instead of the 6+2 pin from the PSU. If that doesn't work, I will try another PSU in my same system. One thing I am noticing from other reports online is that many (if not all) of the reports of the screen going blank are using Vista x64 of some kind. This could support the idea of it being a driver issue.

darkrider01
06-26-08, 08:39 AM
UPDATE:

With my new PCIE 2.0 motherboard and new Processor, I went back to my original 6-pin to 8-pin cable that was supplied with my power supply. It's all working fine - no crashes in Crysis or Eve Online yet, so I'm hopeful. Maybe the extra power coming from the PCIE 2.0 slot is enough.

Also, I emailed Ultra about my PSU. One of their techs responded pretty quickly and said that the 12V rail on my PSU is actually split into 2 seperate rails. They are also going to send me a schematic that depicts which connectors are which rails. If they actually deliver on that promise, I'll be extremely impressed.

I already started with a mild overclock on my new CPU (3.0GHz) - next will be trying this 280 GTX FTW at around 700 core and 2500-2550 on the RAM.

dancid
06-26-08, 09:46 AM
I'm pretty convinced it's a driver issues, but I don't want to go back to vista 32, nor do i want to RMA this beautiful card. hmmm (ack)