Major compatibility issue with NVidia 400/500-series GPUs & mainboards?

Ruoh said:
You may be a candidate for that EVGA power boost thing.
And sure enough, this cleared up the glitching issue I was running into. Took all of 5 minutes to install and $20 to solve the problem. Thanks again for the suggestion Ruoh!
 
And sure enough, this cleared up the glitching issue I was running into. Took all of 5 minutes to install and $20 to solve the problem. Thanks again for the suggestion Ruoh!

Awesome! What we experienced makes me think that the card just needs some extra voltage off the PCIe bus for some reason. I was able to get it working in 16x mode as a single card by upping the bus voltage in the bios.
 
That's almost certainly true in this case. Remember the GTX470 only comes with two 6-pins, which means it's rated for a max of 225W. Yet the card draws substantially more power than that, thus if the PCIe ports are power-limited, the card is being underpowered, which will cause it to artifact.
 
That's almost certainly true in this case. Remember the GTX470 only comes with two 6-pins, which means it's rated for a max of 225W. Yet the card draws substantially more power than that, thus if the PCIe ports are power-limited, the card is being underpowered, which will cause it to artifact.

6-pin PCIe power connectors provide up to 150 Watt. They're no different from the 8-pin PCIe connectors. Only way I can see it being an issue is if the videocard checks whether the two extra pins are grounded and limits itself based on that. But that'd be stupid, right? :)
 
They can provide 150W. They could provide 1000W before they melted. But they're rated for 75W. I'm well aware they don't have any more conducting rails than the 8-pins, but there's a reason both connectors still exist, they're to fix limits on how much power a graphics card draws by specification. The power draw limit won't be at the PSU end, but what if it's on the graphics card itself?
 
They can provide 150W. They could provide 1000W before they melted. But they're rated for 75W. I'm well aware they don't have any more conducting rails than the 8-pins, but there's a reason both connectors still exist, they're to fix limits on how much power a graphics card draws by specification. The power draw limit won't be at the PSU end, but what if it's on the graphics card itself?

The conductors for 6- and 8-pin PCIe connectors are rated by the PSU manufacturers for 150 Watt as they use the same wire gauge. It's just a physical difference at the side of the GPU.

If the GPU wasn't designed to pull the power it needs, it's a failure. Period.
 
Exactly, and I'm saying the GTX470 is a failure for that reason (The HD4870X2 and GTX295 are also technically a failure by the same rights)
 
That's almost certainly true in this case. Remember the GTX470 only comes with two 6-pins, which means it's rated for a max of 225W. Yet the card draws substantially more power than that, thus if the PCIe ports are power-limited, the card is being underpowered, which will cause it to artifact.

The only change I made was going from 900mV to 1000mV. Seems like a pretty small change, but I guess 10% made the difference. A few days later I swapped out the 5 year old 500W Antec for a new Seasonic 620W, so that will help.
 
I know GPUs aren't exactly ohmic, but an 11% increase in volts to an ohmic device increases power draw by 23.5%, and that's quite a lot. If you use 250W to start with (which the 470s can use, and exceed at a push), then add that 23.5% increase and you get 309W, or in excess of. That's a lot.
 
I know GPUs aren't exactly ohmic, but an 11% increase in volts to an ohmic device increases power draw by 23.5%, and that's quite a lot. If you use 250W to start with (which the 470s can use, and exceed at a push), then add that 23.5% increase and you get 309W, or in excess of. That's a lot.

I my case, it was a GTX460.
 
That's not so bad then, as they're similarly sized cards with similar coolers and the same pair of 6-pins. 460s do have their fair share of compatibility problems though.
 
Thinking about getting a GTX 580, though this thread has basically convinced me to go with 6970 instead. My motherboard (M4A79 Deluxe) is definitely on the bad list.

Even if there is a chance it will work, just don't feel like taking the risk with a $500 card. This is the same reason why I skipped the AMD 58xx series (GSOD).
 
The GSOD wasn't a board incompatibility issue, it was just a bug with early drivers solved around January 2010, and the subsequent GSODs were a load of failed hardware that people expected to be driver faults but they weren't. If you got a GSOD with 10.2 or later, if you RMA'ed the card and got a replacement, 9 times out of 10 plus, you wouldn't have GSODs any more.
 
Unfortunately it's an older Abit IP35-E which only has options to change the PCI-E frequency. Given your suggestion, would a glitch such as the one above be caused by a flaky or old PSU?

I'm running an Abit IP35 Pro, and my Gigabyte GTX 460 worked great until it just died today. Trying to get it warranty replaced. :(
 
I had issues from the first day with MSI OC GTX570 because my system would not post. My specs are in the sig. I've wasted 2-3 weeks trying to RMA my Asus mobo (sitll under 3 warranty), but just before I send it to ASUS I thought I might just buy newer mobo (SATA6/USB3) so I went with brand new GIGABYTE GA-880GMA-UD2H (880G/SB850) along with 8Gb of DDR3. After spending couple of hours of swamping mobos into my case, the system with GTX560 still won't boot/post. Asus and Gigabyte use different chipsets and different BIOSes, so I don’t know what a hell is going on? I've tried my old 8800GT PCI-E card in the new gigabyte mobo and it boot fine All I know is I”ve been almost a month without working GPU after spending $350….

I did RMA the GTX570 with online retailer and I did ask them to try it on their AMD based systems, but it seems to be lottery so they might/might not reproduce issue. Funny thing is GTX570 seems to be fine on my old HTPC Intel based e2160 (Gigabyte P35 mobo) and my friends 2500k - sandy bridge (Gigabyte P67 mobo), in both cases systems boot fine. However I cannot boot/post with the GTX570 in either my current Asus AMD mobo nor Gigabyte mobo (which I did end up returning and paying restocking fee).

So I guess both mobos must have hw compatibility issues with GTX4xx and GTX5xx cards. (most likely power states….). I did test about 5-6 GTX460 prior to buying GTX570 and I remember having to reboot few times with some cards in order to be able to POST, so maybe this is good that this happen, as it seems there is a good chance I would run into problems sooner or later.

So what do you guys think should try another GTX570 or ask for ask retailer to swap the card for 2Gb HD6950? I guess I might have no choice but to go with AMD/ATI….

UPDATE:
I was told power state is controllered via nvidia drivers, since I cannot even post it must be different issues. Perhaps I'm not getting full 75 watts on PCI-E, however I find it hard to belive it does happens on both AMD motherboards. So I guess I guess it is faulty GPU. I just hope my retailer online will be able to reprdouce the issue.

UPDATE 2:
- I was able to succesffully test one of my buddies EVGA GTX570 in my rig, so it looks like a faulty MSI GTX570
- Also I just got email from online retailer that the card is falulty as it couldn't pass 3DMark test, they will give me full refund....

So should ask for another MSI GTX570? or buy a XFX HD6950 2GB for the same price?
 
Last edited:
wow i read this tread a while back but dismissed it because you know how people are.... if it's not happening to you, it's not your problem. but I think i encountered the exact same thing with my sli 580s last night after pulling my cpu from my computer to change TIM (that's it no other change). The bios detects it as a new CPU (even though it was the same) but I had my profile saved. reload it, sucessful post... boot in to windows and the system just hard locks every time at the welcome screen like WTF?! The screen goes slightly dim for a second when it locks and eventually the hard drive spins down. When it's in this locked stated, I can power off the PC with just a quick jab of the power button, not needing the 4 second hold like it should be. Change back the mobo bios to default settings with no OC, doesn't help!.

Definitely feels like a driver issue (possibly power states as earlier described) because booting in to safe mode and uninstalling the driver allows me back in windows 7. I can reinstall the driver in normal mode but it starts locking up again. After hours of trouble shooting, I got it to work (so i thought), I go to Nvidia Control Panel and enable SLI aand boom, driver crash and then lock. Reboot and it's the repeat of the exact same issue I originally described in now allowing me back in to windows!! :mad: I'm going in circles here for hours trying to trouble shoot this shit. Then finally after suspecting one of my cards as bad, I uninistall the driver, boot back up with just one card installed. Everything seems to work fine! I'm afraid to reinstall my other 580 but I bet there's nothing physically wrong with it after reading this. But I'll try anyways.

Nvidia has let me down. I can't stand ATI CCC and how they handle custom resolutions so my only solution is to find a work around until nvidia fixes this.
 
Last edited:
Perhasps we should go back to Matrox and Voodoo cards :p

wow i read this tread a while back but dismissed it because you know how people are.... if it's not happening to you, it's not your problem. but I think i encountered the exact same thing with my sli 580s last night after pulling my cpu from my computer to change TIM (that's it no other change). The bios detects it as a new CPU (even though it was the same) but I had my profile saved. reload it, sucessful post... boot in to windows and the system just hard locks every time at the welcome screen like WTF?! The screen goes slightly dim for a second when it locks and eventually the hard drive spins down. When it's in this locked stated, I can power off the PC with just a quick jab of the power button, not needing the 4 second hold like it should be. Change back the mobo bios to default settings with no OC, doesn't help!.

Definitely feels like a driver issue (possibly power states as earlier described) because booting in to safe mode and uninstalling the driver allows me back in windows 7. I can reinstall the driver in normal mode but it starts locking up again. After hours of trouble shooting, I got it to work (so i thought), I go to Nvidia Control Panel and enable SLI aand boom, driver crash and then lock. Reboot and it's the repeat of the exact same issue I originally described in now allowing me back in to windows!! :mad: I'm going in circles here for hours trying to trouble shoot this shit. Then finally after suspecting one of my cards as bad, I uninistall the driver, boot back up with just one card installed. Everything seems to work fine! I'm afraid to reinstall my other 580 but I bet there's nothing physically wrong with it after reading this. But I'll try anyways.

Nvidia has let me down. I can't stand ATI CCC and how they handle custom resolutions so my only solution is to find a work around until nvidia fixes this.
 
Asus P6T deluxe is on the failure list multiple times, but I plugged in a new stock 570 yesterday (pny xlr8), and it seems to be working .... for now ... latest 267.24 beta drivers released yesterday
 
Perhasps we should go back to Matrox and Voodoo cards :p

I miss those days actually. :eek: I remember the day I installed my first 3dfx card and was blown away by how much faster Tomb Raider & Decent ran! So much better than my "S3 ViRGE". :p

Btw, the Asus sabertooth motherboard was listed as one of the mobo's that's suppose to work. Leave it to me to somehow make it fail on the "known good". :rolleyes: sigh...

I've been stress testing my system all day and night since. It seems to be working now but to be honest, I have no idea how i "fixed it" or if it's just going to fail sometime later. I guess I'll find out soon enough. I'm also going to be returning my 580s for 570s so that's another change of variables too.

samuelmorris: haven't seen RadeonPro :confused: - will look in to it.
 
Follow up to my post #56:

So I just returned my 580s and got my first gtx 570 in today. Swap out the cards, boots in to windows, generic vga drivers appears to be used while my already installed nvidia drivers do it's thing. It detects the new 570 prompts for reboot. Great! reboot I did and what do you know, upon next boot up, the lock right at windows login. :rolleyes:

This time, deterimened to find out exactly what the cause is, I systematically went through each one of these bios settings suggested as "might work" on the nvidia forum thread following this issue:

<snip>
4. BIOS settings. There are a few settings people keep reporting success with. They are:
Setting PCI-E Frequency either to 100, 103, 105, or 110
Setting PCI-E Slot wattage to 50, 75 or the highest you can go. If you're blessed enough to have this setting, you're in luck! A lot of people say this solves their issue!
Setting your PCI-E slot to run at 8x, 4x or 2x instead of 16x. There's usually not BIOS setting for this, it's all automatic.
Up your IOH Vcore voltage
NB Voltage: Upping it or sometimes lowering it. You'll have to play around with it.
HT Link Frequency: Setting this lower can help.
Zooze seems to have success enabling the USB Keyboard and Mouse Option in the BIOS, so it's worth a shot.

<snip>

Playing around for days with my IOH, ICH, PCI-E, NB voltage... basically every volt setting I could mess with besides CPU related voltage, I incremented up slowly and after determining it has no affect, I restored it to Auto and moved on to the next.

For some reason in my head, I dismissed changing the PCI-E Frequency even though it was first suggested and tried the others first as what I thought was "more likely". :eek: Also in the past since as far back as I can remember, that was always a big no no, don't mess with your PCI-E frequency - quick way to add instability to your system...

Well low and behold, it was exactly that! I changed the PCI-E Frequency to103 and wallah, here I am posting this message to ya'll.

I'll continue stress testing this system overnight benchmarks like 3dmark11 and Heaven Benchmark, once I've deterimined it's good. I'll then add the second 570 for SLI and hope for the best. *gulp* I have 30 days to determine if I'm going to remain on the blue team (alloted return policy), if it starts locking up again, screw it, I'll give ati another shot i guess...



Update #1: That was quick. Same issue is occurring again after cold boot. I can't get in to windows without going in to safe mode. so much for 103 PCI-E frequency fix.

Update #2: OK after trying every fucking thing again to no avail, I switch the card to the other x16 pci-e slot and here i am again, everything seems fine... for now...:rolleyes: So what does this mean? is my first pci-e slot now "incompatible with 4xx & 5xx nvidia cards"? What happens when I get my other 570 tomorrow? I'll have to plug the new card in to the first slot to SLI. Is this random? What if it works for now, for a month even and then goes right back to failing after i have passed my 30 days return window. This uncertainty does not sit well with me. The more this goes on, the more disgusted I am getting with Nvidia. I bet ATI is loving this.

Update #3: Several stress tests, restarts and cold boots later, it *appears* to have fixed the issue. No Bios tweaking necessary, just using the secondary slot 16x slot instead. Tomorrow's shipment of my second 570 worries me though.
 
Last edited:
Is there any general solution other than disabling HDMI audio?

I dont have any issues and I have friends running Nvidia 4 and 5 series cards fine in some of the boards listed as problematic; I just can see any real pattern to these issues.
 
Haven't had any issues with my 480, then again I don't install the nVidia audio drivers or the 3d stereo driver. I install PhysX and the graphics driver only. Best wishes to people having problems. Hope it gets worked out.
 
I know one manufacture suggested it, but I don't think this has anything to do with the hdmi audio drivers or 3d drivers... When I'm experiencing the windows login lock up issue, uninstalling those specific drivers doesn't resolve the issue (i wish it were that simple).
 
Well low and behold, it was exactly that! I changed the PCI-E Frequency to103 and wallah, here I am posting this message to ya'll.

Does that disable a spread spectrum clock possibly?
 
Does that disable a spread spectrum clock possibly?

I always have it disabled. One of the things I tried though was enabling it. Shit I tried every possible bios setting practically. You don't wanna know how many hours I've invested trying different bios settings...
 
In that thread, there's complaints on Asus P7P55D and a EVGA GTX 460, I've got that same Asus motherboard and Asus GTX 460, no problems here. I actually see that several people there report problems with the P7P55D board and various GTX cards, 460, 470. But since I've got no problems, it's obviously not a given thing, seems random, or faulty cards.
 
In that thread, there's complaints on Asus P7P55D and a EVGA GTX 460, I've got that same Asus motherboard and Asus GTX 460, no problems here. I actually see that several people there report problems with the P7P55D board and various GTX cards, 460, 470. But since I've got no problems, it's obviously not a given thing, seems random, or faulty cards.

right. and I'm of the same scenerio as you dude thinking the exact same thing when I first read this thread a while back. Someone who's using a setup that's suppose to work (sabertooth x58 listed as ok) and everything was fine for a few weeks then all of a sudden, I can't seem to get any of my 580s to work or even this brand new 570 without doing odd ball shit I shouldn't have to do. So i'm sure there are other sabertooth motherboard owners going, mine is working, he must have a faulty card. Yah... give it time.

That's exactly the problem, you may not have the problem now, but tomorrow, a month, 2 months?, it may all of a sudden become your problem too. Who knows, maybe it's just bad luck but if it were a faulty card, i could go down to frys rent another 4xx or 5xx card from a different brand and it's likely to work right? that's exactly what I did when i ordered my pny 570, thinking maybe it's just a faulty evga 580. Oh look, exactly the same issue.

I dunno about you, but this doesn't sit will for me. I still am within my 30 days of my asus sabertooth motherboard as well. I could go return that, get another presumingly "known good" motherboard... things might be peachy for a while then all of a sudden, I'm here again.
 
Very troubling. A friend of mine bought an EVGA gtx 570 to put on an Intel p45 based Asus P5Q-E. Latest BIOS.

Is he screwed?
 
Very troubling. A friend of mine bought an EVGA gtx 570 to put on an Intel p45 based Asus P5Q-E. Latest BIOS.

Is he screwed?

Is the silence speaking volumes here or is this just one of those stray random issues?
 
Is the silence speaking volumes here or is this just one of those stray random issues?

amd7674 in this thread had this issue with a MSI 570, while an EVGA 570 worked. I don't think its given whether the card works or not with the same MB. I've tried my Gainward GTX 570 on a p5n-D motherboard, which is supposed to have issues according to that list and my card worked fine. Your friend should just try it, since he already bought it. Only way to know for sure I think. :)
 
amd7674 in this thread had this issue with a MSI 570, while an EVGA 570 worked. I don't think its given whether the card works or not with the same MB. I've tried my Gainward GTX 570 on a p5n-D motherboard, which is supposed to have issues according to that list and my card worked fine. Your friend should just try it, since he already bought it. Only way to know for sure I think. :)

Indeed.


I did some calling around about this issue. I called Maingear and I called several of the other outlets like that and I was encouraged by the fact that to a man and place no one was telling me about any serious blowback on this issue which you would think they would if this was a REALLY large scale and big time issue.

Odds favor things being just fine. :)
 
Haven't had any issues with my 480, then again I don't install the nVidia audio drivers or the 3d stereo driver. I install PhysX and the graphics driver only. Best wishes to people having problems. Hope it gets worked out.

Exactly this.

Also on 2 4 series cards and 1 5 series of my friends i got them to uninstall their nVidia audio drivers and disable the HD audio device and they have never had any problems. Luckily they don't use the audio on the card.
 
Indeed.


I did some calling around about this issue. I called Maingear and I called several of the other outlets like that and I was encouraged by the fact that to a man and place no one was telling me about any serious blowback on this issue which you would think they would if this was a REALLY large scale and big time issue.

Odds favor things being just fine. :)

Odds are good. I wouldn't worry before you encounter something that can resemble this issue. Even Nvidia struggles with reproducing this issue according to the thread on nvidia forums, so its not a done deal even if your configuration should match anyone who have this issue in that thread.

I wonder however on another unrelated note why Asus recommends a special Nvidia driver for one of their motherboards:
http://hardforum.com/showthread.php?t=1578865
Looks like the leaked EVGA 266.77 which did wonders against the microstuttering I enountered with the GTX 570 and Black ops. Perhaps people with related issues might enjoy this driver as well. :)
 
[deleted out of date info]
 
Last edited:
Got this issues today with my sli setup... 2nd card works fine
Gonna try uninstalling audio drivers
 
i removed audio and still getting driver kernal recovery error.
Evga sent me this rma card last week and it is doing the same thing.
ram is ok (memtest), corsair tx750w PS with one gtx460.
i install a 8800gt and no issues.
i am at a loss been building pc for over 15years and stumped
 
Back
Top