Been having some dramas with my rig recently. I am running 2 x asus GTX 660 Direct OC II in sli. playing any game causes it to crash and recover the display driver after a few minutes. I did a clean install of the latest Nvidia divers and removed all overclocks that were set in MSI after burner but it still does it. I have removed and reseated both cards and ensured all connections are good. I can run Furmark for an hour in SLI with no issues at all. Remove SLI and the games run just fine without issue.. Could this be a power issue? Its a 650w OCZ modular PSU and when I first ran SLI I had no problems at all.
Chances are it's a bad driver if running windows 10 with both Nvidia and AMD having issues on that front. 2 660s will not be drawing a lot of power and if it benches fine under furmark then it's not a power issue.
it's not just Win10 - i had a very similar issue with one of my machines & Win8.1 - & simply couldn't find any workaround... ...whereas the self-same build & drivers would work perfectly with Win7 & Win8 - & subsequently with the later drivers & Win10. So it's not to say that Win10 'might' not be the issue here, as there's no details of the OS - but there's certainly something odd that's been affecting a reasonable number of machines (as there's bunches of posts on the NVidia forums & whatnot) with Win8.1... ...but, along with trying alt drivers, it 'may' be worth trying an alt version of windows to see if it makes any difference before looking at the PSU or the cards potentially being at fault.
I've been having problems with my GTX 660s in SLI, and games crashing and display driver crashing. They did seem to be working, then I swapped them around as I wanted my gigabyte card as the primary as it's quieter, and since then, they've started messing about. Most of my crashes seem to be screen goes black, sound goes all screwy and then the PC reboots itself, with windows then telling me it recovered from a BSOD. Going to try switching them back around and see if that helps, can't understand why it should though, This is on Win 7 and I'm running a Corsair 750W (which is more than enough).
It's worth switching around the PSU power cables. Can solve SLI issues if it's PSU related. Also no cost involved!
spec page for the that card lists. Power Consumption up to 150W1 additional 6 pin PCIe power required https://www.asus.com/Graphics-Cards/GTX660DC2O2GD5/specifications/ Lets say that is 300 watts, that leaves another 350w. I believe it is possible that you lose some of the top end of a PSU, which could bring it down to 325>? 300? that's still 300 watts to drive your disks, cpu, system board. What would be nice is if you could borrow at 750watter and test, no more crashes means it was power related.
I don't think it's power related, my issues are pretty much exactly the same, and I'm sure I've got plenty of juice spare with only an i3. At least I hope it has, , shouldn't be dying as it's not very old. FurMark and Unigine both seem to run fine too IIRC, (may have had a crash from FurMark after 30mins can't remember), but games don't want to play ball atm.
I had a bit of time last night so decided to do some checks. I made sure its running the latest chipset drivers etc and I also ran memtest from inside windows to see what would happen and left it overnight with 4 tests open and running 1024mb at a time. Woke up to a BSOD from the display driver. Pulled the side off the pc this morning and swapped the GFX cards around and reseated the memory. Booted into Memtest and currently running that now to see what happens. I may bring some Ram home today to help rule that out as well. I dont have access to another power supply ATM so I will do everything else again to see what happens! This machine is running W7 ultimate still. Havent jumped to 10 on it yet as I want to see how it plays out first as I know gfx drivers have been troublesome so far for some people. Thanks for the suggestions so far peoples!
do we update the bios of the system board here, I think we do, then repeat the fresh driver install. testing SLI gaming across different games is a nice idea to see if its only one game that does it.
seems NVidia are having increasing issues of recent with drivers - threads everywhere regarding baddrivers a straw poll of ocuk shows over 30% of users under windows 10 and NVidia are having some sort of issue.
Already running the latest bios for this board. Well memtest from boot ran 4 passes without issue. It would crash with a few games I tried. Witcher 3, The long dark and Crysis 3. Switch to 1 GPU and no issues at all. Will do some more testing tonight I Think.
I just ran unigine again and got a crash during the second run, bccode 116. Did a bit of googling and decided to run a memtest, got 20 errors so far and only halfway through the second pass, assuming that means I need to RMA my ram. Errors only in test 13(hammer). Results aren't entirely clear,
Had a play again last night and same issues. - Memtest passed 4 times without issue. - As the cards are swapped over, running without sli they are stable. I havent physically removed a card as I assumed disabling SLI is the same as.. or am I wrong? - clean installation of drivers. Both current and I installed an older one to test. Computer is working fine in every other aspect and games play fine in non SLI. Think I may just replace the power supply and see what happens.
I'm back to assuming it's all poor drivers form nvidia. Like you, I have no crashes if I disable SLI or set the global profile to only use 1 card. And I had no issues before I got the second card. I did have trouble setting it all up, but did manage to get it working eventually, but swapping cards over and updating drivers seems to have stuffed it up. I also find that if I have a crash and it doesn't BSOD and just CTDs, then if I try to run a game again without restarting, fps will be terrible as only one card is really used. What sort of voltages are monitoring programs reporting? Might confirm PSU fault if they look a bit low.
Sounds very similar! As for voltages, what should they be approx and I will test it. I have no idea what they should be if I am honest so I wouldnt know if they are low or not.
12V, 5V and 3.3V, you want them all as close to those values as possible. If you use MSI Afterburner, you'll be able to monitor the GPU voltages, and see what happens with them, if you log it too, you'll be able to see what happens at the crash. I've just switched my cards around, and reinstalled drivers, and hey presto, everything seems to be working. Ran Furmark for a little while, no problems, ran unigine for 1.5 passes, then benched it, and no problems. I must have a very dominant graphics card that insists on being in the primary slot,
ran a test tonight for the voltages etc. Used HWINFO64 to log the min/max/average stuff if someone wants to have a look. Click the link to download an excel csv spreadsheet it creates. https://www.mediafire.com/?gdowiwpthmxd2bh went back to non sli and the game ran without issues. Its annoying me now! lol
Nothing there looks obviously wrong. Temps look really good. Have you tried swapping the cards around? Wouldn't usually suggest it (as it shouldn't make a difference) apart from the fact that it worked for me, Maybe try the second card on it's own in the primary slot to see if that card is dying/unstable. Are you using any screen capture software? Or MSI Afterburner OSD or steam OSD for FPS etc? These can sometimes cause problems. Maybe try running FurMark and prime95 to stress the whole lot, see if it crashes then, suppose that might indicate PSU, as it's all right with just Furmark. Do all games cause it to crash? Or are more simple/easier to run ones ok?
thats what I thought when I looked at it. I hope it isnt a card failing. I have swapped the cards over and out of sli I have no issues with either. Its only when I run SLI I have these problems.
