2x GTX 295's drivers not responding, crashing

[STITuner]

Aigomorla recommended i move this question here.

Ikonik 1200W PSU

I've tried each 295 alone and i dont get the error. all controllable fans are at 100%. the driver that stops responding is "nvlddmkm". I uninstalled all the drivers and reinstalled them with the newest and i'm still having the issue. I had this issue with my last computer too. it had a x1950 & 600W coolermaster PSU.

I'm also getting a BSOD because of nvoclk64 when i click on "system information" or "nvmonitor" in ntune. And got this BSOD a few times when clicking "agree" on the "you are going to blow up your computer EULA in ntune" but it eventually let me do it.

It will run SLI all day on the desktop or internet surfing no problem.
It will play C&C Red Alert 3 all day with no problems
Unreal Tournament 3 - sometimes all day, sometimes 20 minute intervals between crash & recovery
Crysis Warhead - 5 minutes, usually a hard crash needs rebooted
FarCry 2 - 10 minutes sometimes hard crash sometimes not.

To give you an idea of the load on the PSU
computer has all things listed in my signature +
3 8in fans
1 ~4in fan
1 80mm fan on CPU
backlit keyboard with LED
USB corded mouse
Xbox 360 controller (seems to take quite a bit of juice for a controller, if you have any info on a peak wattage usage number for this controller let me know)

GPU temps are at max 160F
CPU temps stay south of 110F
no other temp reading raises a flag

link to old threads.
http://forums.anandtech.com/me...id=37&threadid=2291199
http://forums.anandtech.com/me...id=32&threadid=2291606

 

[HOOfan 1]

I don't see how a driver error can be blamed on the PSU...your PSU is more than capable enough.

 

[STITuner]

well nvlddmkm not responding is a very generic error message. My data indicates that it throws this error whenever there is any videocard error on a nvidia videocard. its so common it has its own .com http://www.nvlddmkm.com

 

[HOOfan 1]

I guess you could see if Ikonik would RMA your PSU, but the Ikonik Vulcan 1200W is certainly up to the job when it is functioning properly. If the PSU were messing up I think you would see other problems other than a driver error though. It could be your motherboard's fault or it could be one of the videocards. EVGA is known for their customer service, try and have a chat with them. Maybe bring up the problem at slizone.

 

[rarebear]

How long ago did you install the OS?

If its been awhile it maybe an OS issue..
I have did reinstalls and tried a hardware setup and the problem fixed it self...

Try unstalling drivers and tr a registry cleaner trial then do driver install that may work..
I had a issue with the newest ATI drivers crashing on install and that fixed my problem..

 

[STITuner]

Originally posted by: rarebear
How long ago did you install the OS?

If its been awhile it maybe an OS issue..
I have did reinstalls and tried a hardware setup and the problem fixed it self...

Try unstalling drivers and tr a registry cleaner trial then do driver install that may work..
I had a issue with the newest ATI drivers crashing on install and that fixed my problem..

computer is brand new. Everything is brand new. just built it last month.

This is about the 7th computer I've built in my life

but i'm no expert, just a guy who thinks rationally about things and has some IT experience

(except the mouse, lol, can't part with my microsoft 5 button optical from ~2002, best mouse ever)

 

[aigomorla]

this is the main arguement i had on why i think its a bad psu:

Originally posted by: STITuner

It will run SLI all day on the desktop or internet surfing no problem.
It will play C&C Red Alert 3 all day with no problems
Unreal Tournament 3 - sometimes all day, sometimes 20 minute intervals between crash & recovery
Crysis Warhead - 5 minutes, usually a hard crash needs rebooted
FarCry 2 - 10 minutes sometimes hard crash sometimes not.

if you notice he's crashing a lot faster on the multi gpu supported programs which would load his gpu up to the max.

He pulls 1 gpu, and he's in heaven...

[really sounds power related, otherwise he would have problems with 1 card solo]

the 295 is SLI'd internally so SLI is enabled with 1 card,

This is why i think its not SLI drivers, unless quad SLI requires its own set of drivers.

 

[STITuner]

yeah, I think the problem with my old computer was that it didn't have enough amperage on the 12volt rail for my videocard. so i understand how a 1200W PSU might not have enough power for 2 GTX 295's.

calls for 46 amps on a 12V rail 680W power supply. for one card.

according to rwlabs.com the ikonik 1200w power supply has +12V3@38A & +12V4@38A. this is the most powerful most high quality PSU I've ever seen. If i need something more i dont know what to do.

Also if this is the most powerful PSU on the market how did nVidia or EVGA test the GTX 295 to prove that it is SLI capable?

and i understand the idea that its already in SLi because its two videocards in one. and agree with your statement.

Note that in nTune SLi is disabled by default. does that mean I"m only using 1 of my 4 GPU's?

like i said it runs UT3 at full detail no problem with SLI disabled in nTune, could 1/2 of a GTX 295 run UT3 at full detail so smoothly?

I think i'm running into issues that no one have ever tried to address before, not even nVidia or EVGA.

Like i said i'm no expert though so dont come down on me hard for my opinion on this.

Thank you for all the discussion i've learned a lot and still am trying to solve this

 

[HOOfan 1]

46Amps is not for the graphics card alone...it is for all of the components that need 12V. GPU, CPU, Hard Drive motors, optical drive motors, fans etc. so 46A is not for a single GPU and thus 2 GPUs will not need 92A....

THE GTX 295 has a 6 pin and an 8 pin PCI-E connectors. 6 pin can supply up to 75W, 8 pin can supply up to 150W...that is 225W total or less than 19A. per a card. The 12V3 and 12V4 rails on the Ikonik have 3 connectors each...so you cannot possibly be reaching the 38A limit on those rails.

As I already stated, the Ikonik is perfectly capable of powering your rig. If your driver crash is a power problem, it is because your PSU has a defect. Also like I said, perhaps you should discuss this matte with EVGA, perhaps they have seen this error before. If they think it is power related, RMA your PSU and have Ikonik replace it.

 

[Beanie46]

What's interesting, HOOfan, is that the test the OP linked to used almost exactly the same setup to test with as the OP has.....i7 965, GTX 295 x 2 in SLI, 12GB DDR3 RAM, a Gigabyte mb instead of his Asus, etc. And the RWLabs tester managed to complete his "testing" of the power supply using a raft of popular game titles, such as Mirrors Edge, Dead Space, Dark Sector, Fallout 3 and Grand Theft Auto 4, without incident.

This sort of makes me suspect he may indeed have some hardware failure going on.....either impending complete failure or intermittant semi-failure. I'd think maybe the mb has issues or a stick of RAM has a failing module or one of his video cards is not working properly. I also am taking into consideration the fact that only sometimes does FarCry 2 crash, and only sometimes does UT3 crash---and sometimes UT3 plays all day.

I'd also think there's possibly an overheating problem somewhere......like his using an 80mm fan on his cpu for cooling?

Just thinking out loud............

 

[Yellowbeard]

Can you provide a link to the specifications for that PSU?

Also, are you OCed or at box stock settings? List all the settings you have made in the BIOS for memory, CPU, and voltages.

 

[theAnimal]

Originally posted by: Yellowbeard
Can you provide a link to the specifications for that PSU?

Here's the jonnyguru.com review.

 

[STITuner]

box settings. all BIOS memory and CPU settings are stock

I wanted to get the bugs out initially before i get crazy

current voltages are: Vcore 1.08 V, +12V 11.4V, AVcc 3.23V

 

[theAnimal]

Is that 11.4V measured with a DMM?

 

[STITuner]

SpeedFan

(i know DMM would be best)

 

[Yellowbeard]

What specific memory are you using and what memory frequency are you running? What voltage do you have set for the memory and memory controller.

 

[WT]

well nvlddmkm not responding is a very generic error message

Had the same thing last year soon after TF2 came out of beta. I had it 20+ times day, moreso when playing games. I finally pulled out my Buffalo Firestix RAM and replaced it with some Super Talent sticks of similar timings, and the errors immediately went away.

Tested the Firestix in another PC using Memtest and one of the sticks was tossing out lots of errors. RMA'd both sticks back to Buffalo (they require both sticks if bought as a dual channel pair) and, once they were back from RMA, I put the RAM back in and have not had that dreaded error since.

My guess ... its not your PSU .. its your RAM.

 

[STITuner]

6 GB [2 GB X3] DDR3-1333 Triple Memory Module Corsair XMS3 Dominator w/DHX technology


more info when i get home and check

 

[STITuner]

BCLK frequency = 134Mhz
CPU voltage = 1.225V
DRAM bus voltage = 1.5V
QPI/DRAm Core voltage = 1.2V

 

[Yellowbeard]

Originally posted by: STITuner
BCLK frequency = 134Mhz
CPU voltage = 1.225V
DRAM bus voltage = 1.5V
QPI/DRAm Core voltage = 1.2V

At what frequency are you running the memory?

 

[STITuner]

just did a memtest, 100% complete 0 errors

 

[STITuner]

RAM Clock:

whatever is stock. I haven't adjusted it. all BIOS settings for everything is "auto"

 

[Yellowbeard]

Check it with CPU-Z, the most recent version.

 

[STITuner]

NB frequency : 2677.9 mhz
DRAM frequency : 535.6 mhz

 

[Yellowbeard]

It should not be a factor but to eliminate variables, try:

DRAM bus voltage = 1.6V