Sign in with
Sign up | Sign in
Your question

Geforce 280 GTX strange errors/crashes, please help!

Last response: in Graphics & Displays
Share
August 16, 2008 2:52:50 PM

Hi there,

I recently purchased an EVGA Geforce 280 GTX along with an OCZ 1010W PSU for my system, to replace a damaged 650W PSU and geforce 8800 ultra.

After installing the PSU, all voltage readings to the motherboard seemed to be correct, so i went ahead and installed the 280. Everything was working perfectly, however at random times while playing a game, the game would suggenly lock up (screen freeze) for around 5-10 seconds, then unfreeze, but with highly reduced FPS (60 -> 20 ish), and sometimes graphical glitches. For example, it happened twice while playing crysis; approaching a village, and when the vtol on the assault level opened. At both of these times, i noticed graphical glitching, such as what i can only describe as a 'red veil' over everything.

Oddly enough, this could be fixed by tabbing to windows, and then back to the game. The game would, however, freeze up for a few seconds before letting me resume play, but the glitching was gone and the FPS had returned to normal. The problem thereafter was that upon quitting the game and entering windows, windows did not respond to any actions, and suddenly the video froze. I could hear an error pop up, but could not see it as the video was frozen..... the strangest thing is that windows was still fully operational, but i had a frozen image on my screen. I could even shut it down with [start -> up -> enter -> right -> enter].

This is an extremely strange problem, the likes of which i have never seen before. I'm doing everything i can think of to try to find the cause and remedy, but i'm adept with computers at best, and starting to draw blanks :( 

System info:

Intel Core 2 Duo E6600
Asus Commando motherboard (stock BIOS ver.): http://www.asus.com/products.aspx?l1=3&l2=11&l3=307&mod...
Corsair Dominator ram PC6400 DDR800 (4x 1gb)
OCZ PSU 1010w: http://www.ebuyer.com/product/127855
EVGA Geforce 280 GTX: http://www.ebuyer.com/product/147809

Additional info:
I installed nvidia driver 177.41 after installing the 280, and the errors occured while running this driver. I have just downloaded and installed nvidia driver 177.83 from the evga driver site, and will test this tonight to see if it has solved the problem (i hope to god its only a software fault). If not, i'm pretty much fresh out of ideas :( 

Someone had mentioned checking the voltage rails to the graphics card, but i have no idea how to do so, and have been unable to find out how thus far.

any help would be greatly appreciated, as i only have a few days to return this card if it is somehowbuggered.

Many thanks in advance.
August 16, 2008 3:00:44 PM

A quick addition, i am also running 6x SATA devices, 4 hard drives and 2 DVD-RW drives.

August 16, 2008 3:06:59 PM

Another addition as i am not allowed to edit posts:

A quick addition, i am also running 6x SATA devices, 4 hard drives and 2 DVD-RW drives.

I doubt that this is a heat problem, as the nvidia performance monitor indicates that the GPU has peaked at 76 degrees C. while playing crysis, with a threshold of 110 degrees. I also had the same problem with World of Warcraft, with the GPU peaking at 55 degrees.

So far, these problems have only occurred once a day, with around 7 hours operation time over the last few days.
Related resources
August 16, 2008 3:19:38 PM

Operating System: Windows XP SP2 32 bit
a b U Graphics card
August 16, 2008 3:29:58 PM

Could be your mobo. You said youve switched everything (psu , gfx card) but kept the mobo. What happened previously that you had to do this?
a b U Graphics card
August 16, 2008 3:39:41 PM

Zalgradis said:
Another addition as i am not allowed to edit posts:

You can edit your posts


As to the issue, its hard to tell, could be defective. or the motherboard as jaydeejohn suggests.
August 16, 2008 3:43:15 PM

The reason i upgraded was fairly minor, not exactly urgent.. but a ball bearing in the PSU fan became loose, and started grinding so loud, i thought somebody was using a powerdrill in my room.

@cliffro: thanks, i have been trying that, but after posting i was given an error "you cannot edit this post" :( 

As far as i can tell, my motherboard seems fine, atleast it was with my 8800 ultra.
a b U Graphics card
August 16, 2008 3:48:25 PM

It could be a power issue, as it demands more, it seems to start to fail, from your comments. Thats why I asked about your mobo. It also could be a driver issue. Did you wipe all your drivers before installing the 280?
August 16, 2008 3:56:33 PM

@jaydeejohn: I did uninstall my previous drivers yes, and deleted the temp folders on my C: drive. As for the power issue, as far as i can tell my motherboard is recieving sufficient power, but i dont know how to check the geforce card :( 

I have been playing crysis for an hour now with driver 177.83 (previous version 177.41, which is when i experienced the issues). I can tell that there is a slight decrease in performance.. but so far no crashes, though i would have to play for atleast 10 hours to see whether or not the new drivers have helped.
a b U Graphics card
August 16, 2008 4:13:54 PM

Hopefully solved then. I know my nVidia drivers since the 169s have been a tad dodgy. Maybe thats all it is. I thought your psu had gone up in smoke and had effected your mbo w/o your knowledge. Hope it holds with these drivers. GL
August 16, 2008 4:19:53 PM

Do you have your 8800 still? If so, toss that in there and see if it works. If so, you may have a defective 280. I had a similar problem with 2 Asus 280s. After two RMAs I went to eVGA for a 280 and things have been fine since.
August 16, 2008 4:32:58 PM

Thanks John.

@pbridgido: I'll test games with driver 177.83 for a few days, and if the problem repeats itself, ill slap the 8800 back in. thanks for the tip.

Ill update this thread as soon as something happens ><
a b U Graphics card
August 16, 2008 4:40:40 PM

The unit has multiple 12v rails. try using the molex to 8 pin adapter. grab molex connectors from different lines. Also make sure you fans and HDD are on seperate lines. This sounds like a load balancing issue.
August 16, 2008 4:55:27 PM

@psyKhiqZero: You may well be right, though the system has only two lines with 3x SATA power connectors on each, but the two DVD drives (almost always idle as i never use them) are both connected to a single molex to dual sata connector. That same line powers a few fans in the system. Other than that, three HDDs are powered by one of the sata lines, and the fourth is powered by the last. The PCI-E cables are completely seperate, and the PSU has 4 in total.

With this in mind, do you think that putting two HDDs on each sata line (would require a lot of annoying fiddling in this small case), or swapping say, PCI-E 2 (current line used for 8 pin) to PCI-E 4 would help? or possibly using the unused Molex line (total 3 or 4 connectors) to solely power the 8 pin connector on the graphics card with a convertor?
August 16, 2008 4:57:41 PM

Edit: that message i mentioned in the original post (the one i couldnt see due to screen being frozen) may well be one regarding insufficient power, as i remember getting those messages in my older system when i had a load balance issue with my 7900.
a b U Graphics card
August 16, 2008 5:16:18 PM

yeah i would spread the HDD between both sata lines. Also use the molex line to power the 8 pin via the adapter (you mentioned that you only had 1 molex line but if there are 2 try using 2 seperate molex lines. and use switch between the 6 pin connecctors if the system proves to be unstable.
a b U Graphics card
August 16, 2008 5:22:13 PM

If possible return the unit and get the corsair 1000 watt unit. It only has 2 12v rails but each can supply 40amps, your less likely to get a load balancing issue with that unit.

Your unit as 4 12v rails rated at only 18 amps. You need to try to get 3 of the 4 rails from your PSU to your GPU to provide the 40+ amps it takes to power the GPU.

It really sucks they don't label the rails. Good luck and god speed my friend. lol
August 16, 2008 5:38:51 PM

Update on the drive spread:

My PSU has 4 power lines for devices other than the PCI-E cards;

2x Molex, 2x Sata (each has 3 respective connectors).

The current device spread is:
Molex 1: Powering 1x DVD drive and 1x Hard drive
Molex 2: Powering case fans
SATA 1: Powering 1x DVD drive and 1x Hard drive
SARA 2: Powering 2x Hard drive.

PCI-E 1 and 2 are in use on the graphics card, and seem to be on seperate lines.

@PsykhiqZero; thanks for the feedback on the PSU. I cant believe such a highly rated PSU would have such a glaringly obvious flaw.. what a disappointment.. Upon checking the PSU manual i can confirm that the PCI-E rails are only 18 amps :(  I'll test out the system for a few more days.. and will consider returning the PSU.
August 16, 2008 5:44:53 PM

After double checking, it seems my version of the PSU does actually supply 20.0 amps to each rail, which does reach the 40 amp requirement.. barely..

When you mention a dual molex to 8 pin convertor, do each of the molex inputs have to come from seperate lines? and if so, will it make an enormous difference if one of those lines is powering a HDD and a DVD drive, or would i have to rewire my entire system and spread all 6 drives purely between the two SATA cables? If i read correctly, these lines carry 30 amps; would this bring the total 'ampage?' of the 8 pin connector to 60, thus solving the possible load problem?

Many thanks for your help!
a b U Graphics card
August 16, 2008 6:21:31 PM

About the dual molex to 8 pin adapter, it's hard to say for sure wheather using 2 molex from seperate lines will make a big difference, as the rails are not exactly labled. I would try to use different lines just in case though. It should make that much of a difference if the lines are used to power a HDD and a DVD drive.

Try this setup
Molex 1: DVD, HDD, GPU (via dual molex to 8 pin pci-e)
Molex 2. Fans, GPU (via dual molex to 8 pin pci-e)
sata 1 and 2 leave alone:

Do you have a dual 4 pin molex to 6pin adapter and did your gpu come with a dual 6 pin pci-e to 8 pin adapter?

IF you have the adapters try this
Molex 1: HDD, GPU (via daul molex to 6 pin pci-e)
Molex 2: Fans, GPU (via dual molex to 6 pin pci-e)
PCI-e 1: GPU (via dual 6pin pci-e to 8 pin pci-e)
PCI-e 2: GPU (via dual 6 pin pci-e to 8pin PCI-e)
Sata 1 +2: leave as is.
August 16, 2008 6:30:45 PM

Thanks for the feedback once again.

I do have two connectors: Dual molex to 8 pin, and dual molex to 6 pin.

I could also try to purchase Dual 6 pin to Dual 8 pin connectors, but would it not be better to connect those from the PCI-E lines alone?

Ie: PCI-E 1 and 2 6 pin -> 1x 8 pin via adapter
and PCI-E 3 and 4 6 pin -> 1x 8 pin via adapter

I also found this posted on : http://web.aanet.com.au/SnooP/psucalc.php

"Even amongst brands, some models are better than others, so reading up reviews on candidate PSU's is a very good idea to ensure you the PSU you end up bying can run reliably at 100% load. For example the OCZ gamexstream 1010w has out of spec voltage ripple past around 50% load, so may not run systems reliabily even though this psu calculator may indicate it could"
a b U Graphics card
August 16, 2008 7:04:17 PM

With load balancing issues it's going to be trail and error before you can find the right setup to get your psu to work. Do what you can with what you have before buying new stuff.

Using just the pci-e connectors may or may not work. You would hope that ocz would have put each of the connectors on a seperate rail but you never know. According to the article you may just be better off returning your psu and getting a different one instead of going through the pain of load balancing.
August 16, 2008 9:07:13 PM

Thank you for the feedback. I'm going to play a few games for the rest of the night and see if the problem persists with driver 177.83, and if it does, ill use my [2x molex -> 6 pin] and [2x 6 pin to 8 pin] adapters to see if it helps.

Out of curiosity, does anyone know of a useful software program to benchmark a graphics card regardless of its system stats, to compare against usual benchmarks?
a b U Graphics card
August 16, 2008 9:27:23 PM

Thats a reason why PCP&C does a single rail
August 16, 2008 10:31:16 PM

Using driver 177.83, another epic fail. This time, i managed to screenshot it:

http://img107.imageshack.us/img107/424/crysisfailar2.jp...

I'm going to try PsyKhiqZero's second setup suggestion now and go back to 177.41.... but to be honest, i'm thinking i sohuld just return the PSU, as the main reason i bought it was to SLI 280's some point in the future, and even with that setup, SLI wont be possible. Sigh.

@Jaydeejohn: I don't quite understand that comment, could you please elaborate?
a b U Graphics card
August 16, 2008 10:36:07 PM

PC Power & Cooling uses a single rail on their psus. It eliminates not having enough power on any given line out, having all the power needed using the 1 rail
August 16, 2008 11:59:51 PM

Ah, i see. Maybe i should get hold of one of those.

Currently running something similar to PsyKhiqZero's suggested setup:

Molex 1: HDD, DVD, Fans
Molex 2: GPU (via dual molex to 6 pin pci-e)
PCI-e 1: GPU (via dual 6pin pci-e to 8 pin pci-e)
PCI-e 2: GPU (via dual 6 pin pci-e to 8pin PCI-e)
Sata 1: HDD + DVD
Sata 2: 2x HDD

Using driver 177.83

I'll post any glitching if it happens. I still have 10 days to test it until a refund is no longer possible, and wondering if, using this configuration, i can run a second card in SLI (PCI-E 3 + 4 going to 8-pin via adapter, and connecting a dual molex to 6-pin to both molex rails, so essentially one will be powering 3 connectors for the GPU, and the other will be powering 2 drive s+ one GPU connector.
a b U Graphics card
August 17, 2008 3:19:43 AM

Well it's definately looking like a load balancing issue for sure as your situation is getting better as you move the power connectors around. Are you talking about the wiered blue lighting? Can you run like futuremark or something aside from crysis?

Here is the link to futuremark
http://www.futuremark.com/download/3dmarkvantage/

If your running out of time on returning the video card it be best to just swap it also to be safe.
August 17, 2008 4:57:39 AM

From the image posted previously, you can see artifacts (small squares) all over the screen if you look very closely. They were also moving southeast down the screen. Weapons on the floor were also bright (bloom) white :/ 

Update: I updated my BIOS to the latest version, and noticed that i had forgotten to uninstall nTune along with my drivers before installing the 280 >_< im wondering if ntune had been messing with the system.

I test played several games other than crysis, but the problems happened a lot more frequently with crysis. I'm back to driver 177.41, and so far (since uninstalling ntune and reinstalling drivers), the only problem ive had is that call of duty 4 bluescreened very early on. The eror message disappeared mmediately as the system rebooted, but there may be a log of it somewhere. I then reloaded and played the game for a good 2 hours, and the problem did not repeat itself.

I noticed that the voltage on 'VCore' was jumping past its threshold slightly, with the desired value being 1.19, and it jumping to 1.31 (monitored by Pcprobe, an Asus program for my motherboard). I was also running Speedfan at the time of the bluescreen, and i did not run it while playing CoD 4 afterwards... it may have been the cause of the crash. Other than that, 3 hours of gaming in crysis, call of duty, world of warcraft didnt recreate any glitching.

I'll update again after a longer session of gaming tomorrow, and ill post soe 3dmark vantage stats. Thanks again for all of your help.
August 17, 2008 9:19:15 PM

5 hours or so of WoW with no errors, but i have gone ahead and purchased a new PSU, and sending the OCZ one back.

Coolmaster 1000w PSU: http://www.ebuyer.com/product/129034

My friend is using this to power 6 Sata drives + dual 9800GX2's, so it should have no problem with my single 280. Reviews state that the 12v rails provide 30 amps each, and their voltage ripple is massively lower than the OCZ psu.

I am also considering sending
August 17, 2008 9:21:18 PM

I'm also considering sending back the 280 for a replacement, just incase it is faulty, though i have a good 3 years to do that.

Thanks again for your help PsyKhiqZero and Jaydeejohn, without your help i most likely would have shrugged it off and lived with this shoddy gaming PSU.

I'll update the thread once the new PSU arrives.
August 18, 2008 12:22:24 AM

3 hours of Crysis, no errors.

Surely the older version of nTune wasnt causing the problems?
a b U Graphics card
August 18, 2008 4:42:35 AM

Is this with the new PSU?
August 18, 2008 10:55:07 AM

PsyKhiqZero said:
If possible return the unit and get the corsair 1000 watt unit. It only has 2 12v rails but each can supply 40amps, your less likely to get a load balancing issue with that unit.

Your unit as 4 12v rails rated at only 18 amps. You need to try to get 3 of the 4 rails from your PSU to your GPU to provide the 40+ amps it takes to power the GPU.

It really sucks they don't label the rails. Good luck and god speed my friend. lol

I am little confused
GTX280 should not take more then 300W even overclocked and also up to 75W is provided by PCI-e 2.0 bus. Why then need to put another 480W thru external connectors on graphics card.

40A on 12V is sugested to provide enough power for whole system not just graphic card itself.

"A single x16 card may now draw up to 300 W of power (75 W from the slot itself, 150 W from an 8-pin PEG connector, 75 W from a second PEG connector)" small note from specification.
August 18, 2008 12:15:08 PM

@PsyKhiqZero: That is with the old PSU.

I just completed crysis (last few levels, around 4 hours played) - no errors. It's really starting to look like it was all caused by my not uninstalling nTune >_<

@xrodney: My mobo is not PCI-e 2.0. I'm not sure what it is, but it is atleast 16 speed.
August 18, 2008 12:21:33 PM

Just to clarify: Since uninstalling the graphics drivers and nTune, then reinstalling 177.41, i have had one random bluescreen (i think was caused by a third party monitoring software), but other than that, no problems at all.

The coolermaster 1000w PSU is still on its way, and i have 10 days to test it. The problem now however is that Ebuyer may not accept the OCZ PSU back.
August 18, 2008 12:23:28 PM

I'm also back to using PCI-E 1 and 2 to power the graphics card, rather than he molex adapters.
August 18, 2008 12:23:48 PM

Zalgradis said:
@PsyKhiqZero: That is with the old PSU.

I just completed crysis (last few levels, around 4 hours played) - no errors. It's really starting to look like it was all caused by my not uninstalling nTune >_<

@xrodney: My mobo is not PCI-e 2.0. I'm not sure what it is, but it is atleast 16 speed.


PCI-e 1.x had only 50W output i think but with 2 external connectors to supply enough power it should not be problem.
If i remember GTX 280 should take around 246W under full load
August 18, 2008 8:30:02 PM

system starting to fail (still using OCZ psu, new one arriving tomorrow).

Almost 2 days of no problems, then while playing WoW the screen turned off (no signal), then a few seconds later, the system froze, repeatingf the last 0.1 seconds of sound.

Upon starting the system up again, Pc Probe (asus voltage monitor etc) was setting off warnings all accross the board, mainly rhe DDR and Vcore, SBcore etc. voltages peaking over 4 volts, when the desired is around 1.2 volts - 2 volts.

I shut the system down and turned it on a few minutes later. The readings were all ok, but then all of a sudden (while idle and in windows), same thing happened again..

What the f*ck tbh. Does anyone have any idea at all what might be causing this?
a b U Graphics card
August 18, 2008 8:41:11 PM

4 volts?!?!?!? Something is seriously messed up. Can you check a diffrent PSU?
August 18, 2008 8:55:36 PM

it is indeed... reading might have been incorrect though, as it also said the power fan was at 0 rmp, but i could clearly see it spinning :/ 

note: when this happened while playing woW, the 280 was at 53 degrees, and when idle in windows, 43 degrees.

A friend mentioned that if the voltage was spiking insanely, it may be causing a bios reset, which would make the screen go black and system crash. He said that the system being fine after a system reset also supports this theory.
August 18, 2008 11:20:07 PM

System continuously blacking out now, going to leave it off until i can install the new psu.
a b U Graphics card
August 18, 2008 11:30:04 PM

Definitely wait - you could damage something if the old PSU is that strained.
August 19, 2008 3:26:47 PM

I'm now running the new PSU, though im experiencing something rather wierd..

It 'feels' like i'm getting slightly lower FPS than i did before... but the only item i have changed is the PSU. Could using a different PSU actually lower your frame rate? Or is that an indication that it is not supplying enough power to the GPU?

Thanks again for all of your help.
a b U Graphics card
August 19, 2008 3:38:45 PM

Yes. If your gpu was underpowered itd definately slow down. It was nt getting enough power, and slowed because of it. Have you tested it? Maybe its perceived and not actual?
August 20, 2008 1:16:39 AM

@jaydeejohn: the system does atleast seem stable, as for the underpowered GPU, i don't really know how to test it. The old PSU is packed up ready to be sent back, and i don't have nor know of any software that could check the power the card is recieving.

The slower performance may just be perceived as you said. I would have expected some more serious problems than lower frame rates if the GPU was being underpowered.
a b U Graphics card
August 20, 2008 1:56:50 AM

No, it doesnt. It appears going from left to right, theres a 6 pin and a 8 pin. Then theres two sets each. Top and bottom. If this is so, you need to use 1 top and 1 bottom connector, as youre pulling everything from 1 rail. I could be wrong, but from the pictures, it appears there are 2 sets of connectors, top and bottom. If not ignore this, if so, try 1 top and 1 bottom
August 20, 2008 2:01:34 AM

True, i had wondered if that image was to indicate the "sets of connectors per rail", but there was no information to support it :(  I'll swap one of the connectors some time tomorrow and see if it makes any difference.

I generally thought that, since this PSU was designed to support SLI cards, even if both connectors were on one rail, it would be more than enough to power the graphics card its connected to.
!