View Full Version : Completely out of ideas...

03-08-10, 04:19 PM

I'm not sure what I'm really looking for from this collective, but I think I've hit rock bottom and totally out of ideas with this PC problem I've been having since May last year with my build. I have exhausted all diagnosing techniques that I know of.

My old rig was a S939 4800+ OC to 3ghz, 550W Antec PSU, 3GB RAM, 4x 500GB hdds (2x striped), Audigy 2 ZS and a trusty 8800GT all on WinXP - it served me very well over the years.

Last year I decided to treat myself to a new platform:
AM2+ 940
Asus M3N-HT Deluxe (nVidia 780a)
2x 2GB Corsair 5-5-5-15
Win7 64-bit (then RC, now RTM)

Since then, on and off I've been suffering badly from the "nvlddmkm stopped responding" issue or simply extremely distorted graphics in games such as the stalker series, GTA SA, GTA4, HL2, FEAR, Mirrors Edge, RA3, 3DMark Vantage etc. The exception is Crysis, which works perfectly.
When I speak of distorted graphics, I mean black flickering lines extending from in-game models, speckles, very colourful pixels across the entire screen, flickering textures and general lines flashing about the render or just hard-locks to some Crash To Desktop due to TDR. It does not happen 100% of the time, but can usually be replicated every time easily with 3DMark Vantage.


Now I have read much on this Timeout Detection Delay business, but I can't say if this really is my issue.

All I can say is what I have tried to resolve it:
Tried my spare 500w PSU and recently bought a new corsair 650w PSU.
RMA the 8800GT 3 times.
RMA the motherboard 3 times (a pain in the backside when you use a V8).
Tried every single version of the mobo BIOS which supports 940/955 CPUs.
New CPU - now using a 955 @ stock
New RAM - now using 4x 2GB Corsair Dominator 8500 @ 800MHz. Tried each single pair and both pairs at the same time.
In the BIOS, I have tried full auto setup, full manual setup, adjusted the CPU, CPU-NB, NB chip, RAM, HT Link voltages. No dice.
Tried the 8800GT on each of the 3 PCI-E x16 slots.
Disabled all on-board devices and removed the Audigy.
Nothing is or has been overclocked.

Currently my machine is completely up to date BIOS, driver and Win7 wise.

I have tried every single version of the nVidia drivers for the 780a chipset and the 8800GT, including the built in Microsoft drivers. I still get this render corruption or TDR issues.

Throwing money at the problem and RMAing has not resolved the issue. The rig runs OCCT for hours perfectly fine, along with MEMTEST with multiple passes and zero errors or failures and is rock solid working on the desktop, it's just games.

I am very, very stuck.

I'm on the verge of ditching this cursed mobo and the corsair RAM and going for an AM3 790FX and 8GB DDR3 instead, but as I have already thrown a lot of money at this issue I wanted to pass it by you guys here first. I see quite a few other people suffer from the "... stopped responding and has successfully recovered", including ATI and Intel users.

The only thing I have yet to try is WinXP, but seeing as I want(ed) to do VM work 4GB is not enough so a 64bit OS is required. Win7 64-bit is still my logical choice of OS.
The ironic thing is that my spare S939 3800+ @2.65Ghz runs my 8800GT and Win7 64-bit without any complaints and is solid as a rock in all of the said games.

Did I simply choose a crap nVidia 780a platform? I see many people out there still using the M3N-HT Deluxe and they seem happy. Am I suffering from some crazy 8800GT incompatibility issue on this board?

If anyone has any advice on what I should do for my next move other than take my mountain bike out of the shed more often, please please do not hesitate to let me know your thoughts.

Thank you for reading.

03-08-10, 06:07 PM
Wow I am feeling bad for you. What is Geforce 8800GT brand and Nvidia BIOS version? Also did you tested onboard Geforce 8200 MGPU to see if it reproduce same issues as on Geforce 8800GT card?

03-08-10, 06:48 PM
It's a BFG 8800GT OC 512mb. The bios revision I cannot say until tomorrow as the rig is now off and the mrs is asleep. I will find out and report tomorrow. What I do remember is that it's an A2 revision which is a slightly revised version of the original model BFG first introduced.

I should add that I have tried underclocking the card(s) and I have a twin 80mm chassis fan providing additional cool air to it and the mobo chipset. Neither are hot. The card tops at around 85c on full load which is historically consistent.

As for the super-onboard 8200 gpu, Yes I could replicate the issue with the previous board, but only tested the current one wirh vantage where it does fail. This I will experiment further tomorrow for a more detailed test.

I will actually buy an M3N79T at the end of this week if we have no luck here. If only it didn't seem like all a gamble.

03-08-10, 07:11 PM
Well it has to be faulty motherboard, ASUS probably returned you the same faulty motherboard 3 times. Best to get a new motherboard and the issues will go away.

03-08-10, 07:59 PM
The only thing I can guess is that there is some sort of incompatibility between your components and a new motherboard is your only solution.

You've replaced EVERYTHING, so defects are almost 100% out of the question.

Just get a new board and be done with it. I'm guessing there is a reason that hardly anyone uses the nvidia 7xx chipsets for AMD.

Kudos to you for sticking with it through so much crap though. That's a hell of a lot of testing and trouble shooting.

03-08-10, 08:12 PM
Did you ever try a clean new install of the win 7 OS?

Install OS first, then after first clean boot, install chipset drivers before any other driver else etc?

What is your general routine?

Also what are your BIOS settings? Snoop Palating or something on? AGP/PCIe settings? Are you overclocking?

03-09-10, 12:44 AM
Regarding motherboard RMA, or any other RMA, I always compare serial numbers. The mobo has come back different each time. I've reinstalled Win7 from scratch many times. There is no PCI-E options really to mess with in the BIOS. I disable the super-onboard gpu when it isn't used, and I have a single 24".

As for my routine, I always run new kit in as stock for a while before playing. This build has not been overclocked once.
When setting up Win7, I install it from disk, no other apps, then chipset drivers first then video then the audigy if it's in. A reboot after each driver. That is generally it as Win7 takes care of our usual worries. I have even tried a spare 320gb hdd and also running vga instead of dvi. All with this issue.

Yes, the chipset has worried me and in a perfect world I would expect a refund or even credit note should I return this mobo as that is the best conclusion as to the cause of all this.

I guess I'm stubborn in some ways as I've always stuck with nForce chipsets since series 2, so series 7 I thought would hold up. I think you may be right in saying get a new mobo and be done with it. It's just a shame really that I cannot turn to away from the cpu maker for a chipset and to turn my back to the 780a.

I will let you guys know if the new motherboard resolves the issue.

03-09-10, 02:40 AM
Dude did you try the nvidia forums?

As it seems that you have tried and done everything. Maybe the forums there might have more insight on nforce chipsets

03-09-10, 06:49 AM
No I've not posted this anywhere else. Perhaps a copy and paste is in order.

An M3A79-T is on its way and will be here tomorrow.

Moved this over to: http://forums.nvidia.com/index.php?showtopic=162318

03-09-10, 02:22 PM
Just tested the onboard GPU - exactly the same as before. Graphic defects in games.

I should have focused on this when I received the replacement mobo a couple of weeks back.

03-09-10, 02:58 PM
I'm glad there was a resolution to this, You are a trooper.

03-10-10, 12:39 PM
For those of you who don't idle in the nVidia forums, the M3A79-T arrived this morning and the issue is still being experienced.

Does anyone have any ideas? I reckon it is the 8800GT series causing the issue after all, though it doesn't explain why the M3N-HT onboard corrupts too.

I'm all ears.

03-10-10, 01:38 PM
For those of you who don't idle in the nVidia forums, the M3A79-T arrived this morning and the issue is still being experienced.

Does anyone have any ideas? I reckon it is the 8800GT series causing the issue after all, though it doesn't explain why the M3N-HT onboard corrupts too.

I'm all ears.

03-10-10, 01:58 PM
do you live below a power line or something weird?

cell phone tower thing??

03-10-10, 02:14 PM
Interesting perspective! Nothing odd about the environment here. No masts nearby as 3G strength is minimal.

The 8800GT works fine in my the spare NF4 board.

03-10-10, 03:11 PM
are you using the 4 pin mobo power plug or the 8 pin?

just asking randomly btw.

This doesn't make no sense at all.

Crysis is still fine?

03-10-10, 07:50 PM
Hmmm that is really very strange issue with 2 modern motherboards but no issue with old motherboard with nForce 4 chipset. Look like BFG 8800GT OC card need to update Nvidia BIOS to the latest version to ensure full compatible with modern motherboards chipsets.

I found 5 BFG 8800GT OC BIOS revisions at Mvktech.net:

BFG 8800GT (OC) PCIe 512MB - Rev 01 BIOS version (http://www.mvktech.net/component/option,com_remository/Itemid,26/func,fileinfo/id,2252/)
BFG 8800GT (OC) PCIe 512MB - Rev 02 BIOS version (http://www.mvktech.net/component/option,com_remository/Itemid,26/func,fileinfo/id,2317/)
BFG 8800GT (OC) PCIe 512MB - Rev 03 BIOS version 62.92.1F.00.18 (http://www.mvktech.net/component/option,com_remository/Itemid,26/func,fileinfo/id,2408/)
BFG 8800GT (OC) PCIe 512MB - Rev 04 BIOS version 62.92.1F.00.65 (http://www.mvktech.net/component/option,com_remository/Itemid,26/func,fileinfo/id,2862/)
BFG 8800GT (OC) PCIe 512MB - Rev 05 BIOS version 62.92.1F.00.65 (http://www.mvktech.net/component/option,com_remository/Itemid,26/func,fileinfo/id,2749/)

You could try flash the latest BIOS to see if it will fix the issue, if it will not fix then guess you need to get a new graphic card.

03-10-10, 09:57 PM
That looks like VRAM overheating to me....good luck man, you've been through a lot.

03-11-10, 02:28 PM
is there much difference in temps between the two systems the 8800gt has been tested in?

04-01-10, 06:37 PM
Not much time recently to post back, and I've put the same text found below on the nvidia forum thread:

Just wanted to post the cause of my issue and to say that ImNuts4 you were right with your thoughts.

It's the 8800GT gaining too much temp too quickly for the cooler to compensate, despite the issue occurring when the temperature is below 75 degrees C. I have to assume that the onboard GPU cooler is also not adequate either so the issue is replicated there too.

The answer: A 2x 80mm fan bracket now facing the 8800GT fan intake - the issue has mostly disappeared. It'll do until I replace it later this year.

At very least I've now been able to compare AMD's 790FX chipset first-hand against the nVidia 780a and I have to say I prefer the nVidia's offerings in terms of features and performance despite the tons of manual options I can tinker with on the 790FX board.
The PII 955 is very slightly quicker on the nVidia platform, but the Corsair Dominators work noticeably faster for throughput plus OC a little better (19700MB/s compared to 23800Mb/s - Blackbox2 figures)
The nVidia RAID controller can detect and utilise AMD HDD sets but the AMD cannot detect nVidia sets. Plus the AMD RAID controller gets confused with the array numbers if you plug in another SATA or even USB hdd, causing the mobo to try and boot from the wrong HDD. Doh!

To be honest I thought AMD's flagship would prove to be an overall better solution, but I was wrong. I'm just glad I didn't move onto the AM3 versions of the 790FX chipsets because I think I would have been disappointed. This does put me off future AMD chipsets based on the 790FX too...

Anyway, thank you to those who shared their thoughts and tried to help. All along it's been a GPU cooler issue (not an overall heat issue). Nothing to do with the chipset of the board, PSU or anything else.

So basically nvnews guys, due to my stubbornness in believing my 8800GT can live forever, it put me through this adventure.

As for your query to the 8800GT temperature differences, my spare machine is in a LianLi case and I would say has very similar ventilation to my current Antec - the temperatures were the same (55c idle and 80 max), but it must must have been making a difference to the card's airflow. The onboard GPU issue was a red herring.

Thank you again guys.

05-15-10, 12:18 PM
An update to the interested:

This issue was nothing to do with the 8800GT. I replaced the 8800GT with a 8800GTS, then a GTX295 and I was still experiencing the issue.

I even replaced the relatively new Corsair 650 with the 850 model - same issue.

The issue is down to a single faulty stick of memory - I tested each pair to find out if I could isolate a pair, then tested a each stick from the pair. Despite passing MEMTEST, the issue of snowflake artifacts and GPU kernel fallovers was highlighted with an individual stick. I should have tested each stick individually a long time ago, but the GTX295 prompted me to go back to basics.

I ASSumed the pair of twin Corsair Dominators would be A1 out of the box. Fool :)

I finally have a stable machine again (it only took me a year). I strongly recommend that anyone who experiences the nvlddmkm.sys BSOD or GPU kernel restarts to check their RAM with the closest of eyes.