• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Artefacts without overheating this time...

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Anjow

Member
Joined
Jan 12, 2005
Location
UKbiquitous
Artefacts + crashing without overheating this time...

You've probably seen my 8800GTX saga, I had 2 cards from XFX and the fan didn't start on either of them, leading them to overheat and crash. I have since got one from Sparkle and things seemed to be going well until this.

I was playing Oblivion with RivaTuner running (had had a couple of artefacts/crashes before so I wanted to diagnose) and I started noticing some geometric artefacts - it was a series of parallel floating lines in a forest which slowly moved up and down, as though with the waving of a tree. I looked at RivaTuner (on my second monitor) and saw that the temperature was at 68 celsius (what the card had idled at prior to increasing the fan speed) so it looks as though it wasn't caused by overheating.

Attached is my RivaTuner log for reference. The peak in temperature is when it was occurring - I then quit the game.

What else could be causing this? I am using the 97.44 drivers - are there some others that I should try? If so, where would I get them? I hadn't noticed any on the nvidia drivers page of their site.

Edit: I just started up the game again and reloaded my save. I noticed that the artefacts were still there, exactly the same ones. I am using a couple of mods for the game, most notably Oscuro's Oblivion Overhaul - perhaps this one is something to do with that? Regardless, I have still suffered other artefacts which have caused crashes - in Oblivion with the chequers appearing all over both my screens, and in Dark Messiah of Might & Magic with massive geometric artefacts.
 

Attachments

  • rivaTunerLog.zip
    3.3 KB · Views: 111
Last edited:
Now I have had an artefact-then-crash scenario. I was again playing Oblivion when I got the chequered-screen artefacts for half a second then my machine automatically rebooted. Upon reboot I got a BSOD with a stop error, a photo of which I've attached. I've also attached the RivaTuner log, though it doesn't show anything interesting - still no overheating.

Can anyone help me please?
 

Attachments

  • IMG_0364 (Large).jpg
    IMG_0364 (Large).jpg
    88.8 KB · Views: 134
  • rivaTunerLog.zip
    11.1 KB · Views: 88
If anything your card is bottlenecking because of your CPU and RAM(taking your using whats in your sig)....
 
Looks like a dirver issue.

First off are there any overclocks? If so do you get the same thing at all stock?

If not then follow these steps:

  1. Uninstall your gfx drivers.
  2. Reboot.
  3. Run Dirver Cleaner for nVidia. You can download Driver Cleaner here.
  4. Reboot.
  5. Install fresh copy of the latest drivers.
  6. Reboot.

You may want to do the same with your chip set drivers. If this does not work then you could always try a reinstall of windows.
 
There's no overclocks. Reinstalling windows is not an option. I'll try the driver cleaner shortly. Thanks for the response.
 
Okay, I did that and I am still getting the spiky artefacts ingame.

Not had a crash yet though.
 
Anjow said:
Okay, I did that and I am still getting the spiky artefacts ingame.

Not had a crash yet though.

A few things..for starters what computer PS are you using.

Make sure your computer case has decent ventilation and air flow through it. Poor case ventilation
equals hot video card.

There are 8800 driver issues with all drivers later that the 97.02's when it comes to setting the
8800's memory clocks. The 8800GTX has 768Mb of memory setup in three domains of 256mb's
each. Each memory domain has it's own clock generator.

At boot the 8800's memory clocks are set a bios default of 398mhz on each domain. When the
the Windows drivers are loaded at OS startup the memory clocks are increased to 900Mhz...or
at least they are supposed to be and that is where the problem comes in.

With drivers later than the 97.02's only the lowest of three memory domains clocks up to 900Mhz.
The other two higher domains stay at the 398Mhz bios boot up default resulting in the first 256Mb's
of memory running 900Mhz and the remaining 512mb's of memory running 398mhz. When textures
are mapped across the asyncronuous clocked memory divide artifacts can be the result.

The 97.02's clock all the memory domains correctly which eliminates any async memory clock
issues but being older drivers they may/will have other game specific bugs that were/are
corrected in the later drivers. All you can do is give the 97.02's a try and see if that helps your
issues.

Another way, with the later drivers anyway, is to reduce the amount of memory the game is using
for textures. You can do that by keeping the resolution at 1024x768 and running without AA
enabled as you will never use more than the first 256Mb's of video memory that way.

You may be able to run higher resolutions and/or with some AA dialed in depending on the game
and the textures in it. You just want to keep the memory usage in the first 256Mb's of properly
clocked memory until NV again pops a driver out without the 8800 memory domain clock bug
built in.

Viper
 
Last edited:
Immortal_Hero said:
Looks like a dirver issue.

First off are there any overclocks? If so do you get the same thing at all stock?

If not then follow these steps:

  1. Uninstall your gfx drivers.
  2. Reboot.
  3. Run Dirver Cleaner for nVidia. You can download Driver Cleaner here.
  4. Reboot.
  5. Install fresh copy of the latest drivers.
  6. Reboot.

You may want to do the same with your chip set drivers. If this does not work then you could always try a reinstall of windows.

Don't want to pledg this thread but I wanted you to know that I was having a similar issue with my 7900's and your solution above solved my issue.... I have finally pushed past my 12200 3dmark05 barrier and the cards are acting like they should.
 
ViperJohn said:
A few things..for starters what computer PS are you using.

OCZ GXS 700W

ViperJohn said:
Make sure your computer case has decent ventilation and air flow through it. Poor case ventilation
equals hot video card.

It does have very good ventilation; 8x 120mm fans - 2 + PSU for exhaust and 6 for intake. If you look at my RivaTuner logs you'll see that I am not overheating.

ViperJohn said:
memory divide artifacts can be the result.

Does that mean the chequer effect in layman's terms?

ViperJohn said:
All you can
do is give the 97.02's a try and see if that corrects your issues.

I'll happily do that; do you know where I can get hold of older drivers?

ViperJohn said:
You can do that by keeping the resolution at 1024x768 and running without AA
enabled as you will never use more than the first 256Mb's of video memory that way.

I could run it like that with my old card :(
 
Anjow said:
OCZ GXS 700W

It does have very good ventilation; 8x 120mm fans - 2 + PSU for exhaust and 6 for intake. If you look at my RivaTuner logs you'll see that I am not overheating.

Does that mean the chequer effect in layman's terms?

I'll happily do that; do you know where I can get hold of older drivers?

I could run it like that with my old card :(

(1 & 2) Okay plenty of PS and case ventilation okay.

(3) That was in laymans terms but we will try. Say you have a particular texture (part of a tree, plant or what ever) that takes, as an example, 400 linear bytes to map into video memory for display (could be displayed for a few or 1000's of consecutive frames dpending on the scene) with 200 bytes of that being mapped at the end of the memory block running 900Mhz and the other 200 bytes being at the beginning of the next memory block running 398mhz. That is the memory divide I am refering too.

When the GPU renders the memory map to the display the part of that texture stored in the higher clocked memory is updating/changing faster than the other part of that texture stored in the lower clocked portion of memory. That can lead to artifacting in any texture that spans that divide as changes occur in real time.

(4) See above.

(5) Yep you surely could have. Keeping the memory use down is just something to try with games where you notice artifacting. It may or may not help with that particular game or scene depending on if the artifacts are cause by textures spanning the async clocked memory domains. Some textures will span no problem and others won't. Scenes rendered at higher frame rates are more likely to have issues than scenes at lower frame rates. Its highly variable but it is something to try...at least until NV gets some later drivers out that do not have async clocked memory domain issues in the mix of normal game bugs to deal with.

Viper
 
I changed to the 97.02 drivers, using driver cleaner in between, and I've played one round of Oblivion - I didn't notice any of the spiky lines in the forests, as I had done before.

Time will tell if this has worked though.
 
Anjow said:
I changed to the 97.02 drivers, using driver cleaner in between, and I've played one round of Oblivion - I didn't notice any of the spiky lines in the forests, as I had done before.

Time will tell if this has worked though.

Awh cool as I love a happy ending lol. Hopefully OB will stay nice and clean now.

Viper
 
Anjow said:
Everything is stock.

The only reason I ask is that is a common display when the OC's are pushed to high. It's hardware caused not driver and is most likely heat related. While your GPU temps are okay it is not the only device that runs warm or can get heat sensitive as the card warms up as a whole. It's just the only temp you can see easily.

Try running temporarily with the case side panel off with a fan moving some air around the card an see if that stops it. If it does then you most likely need more case ventilation...within reason as you do not and should not need a hurricane of air flow.

Viper
 
I doubt having the case side off would make a difference, since the side panel pretty much is 4x 120mm fans, directly above the card. Upon looking at the RivaTuner logs, would you say that the ambient temperatures were alright?

This is the third card I've had (the fans didn't start on the other two, so I spent an extra £50 on this one) and I will be hugely annoyed if I have to return it as well.
 
Anjow said:
I doubt having the case side off would make a difference, since the side panel pretty much is 4x 120mm fans, directly above the card. Upon looking at the RivaTuner logs, would you say that the ambient temperatures were alright?

This is the third card I've had (the fans didn't start on the other two, so I spent an extra £50 on this one) and I will be hugely annoyed if I have to return it as well.

No 4 x 120mm side fans is in the hurricane range. That card has a hardware problem so RMA the dang thing. I know it is a pain is the rear but it is the same with virtually anything computer anymore...especially memory. About 10-15% of what you get will have a problem in one way or another. Quality control and testing costs money which the almighty bottom line bean counters just hate to spend and love to cut down to the barest bone they believe they can get away with.

BTW it is pointless to spend extra money on a given brand of card unless the OEM's warranty makes the extra bucks worth it. They are all built by FlexTronics in China on the same stuffing/assembly lines.

Viper
 
Back