I am having a weird reboot problem with Raven 2012 pc in sig and wanted to know what others think. During certain games and Heaven stress test, my PC will reboot. This has happened in BF3 and GW2. Other games such as dead island, torchlight 2, etc. run fine. During these games and general Internet browsing, the PC runs fine for days. During Heaven benchmark and GW2, I always get a reboot at random time between 10 minutes and an hour. I have noticed if I turn SLI off, I do not have this problem. Here are the troubleshooting steps I've taken
Disabling Overclock - I returned cpu to stock clocks
Heat - I have used core temp and evga precision x to log temps up until point of reboot. These logs show no abnormal temps for cpu or gpu
Ram- thinking it may be ram, I reseated my ram sticks and ran Memtest86+ all night with no errors
Driver - I updated to latest drivers. After the driver update, I noticed there were no errors. Then I realized I had not re-enabled SLI. Upon re-enabling SLI, errors occurred again. I then disabled SLI, and physically removed GPU1. I ran tests on just GPU2 and again everything ran fine.
At this point the only things left I can think of are PSU or Motherboard. This is 1K watt PSU which is well above the necessary wattage for 680 SLI.
PSU- I tried another PSU (same model) however, issue is the same.
It seems like the only thing left at this point is the motherboard.
Is there a way to try a different SLI Bridge?
Also have you disable the onboard video from the CPU when you disable/enabled SLI, I know this sounds dumb but it might hurt discrete gpu performance.
I was gonna say it was a PSU problem but 1000 watts WILL have no problem with your system.
Another thing to look at is the GPU's itself, does you motherboard detect having two GPUs connected like in bios?
Have you tried GPU's in the PCI-E slots?....
Driver - I updated to latest drivers. After the driver update, I noticed there were no errors. Then I realized I had not re-enabled SLI. Upon re-enabling SLI, errors occurred again. I then disabled SLI, and physically removed GPU1. I ran tests on just GPU2 and again everything ran fine.
Is ur mobo grounded properly. My friend had a simmilar problem where his pc wouldnt boot sometime because he didnt screw in the mobo completely. Check for frayed wires. Maybe ur processor is hitting its capacity n it shuts down because it cant handle it?
Is ur mobo grounded properly. My friend had a simmilar problem where his pc wouldnt boot sometime because he didnt screw in the mobo completely. Check for frayed wires. Maybe ur processor is hitting its capacity n it shuts down because it cant handle it?
Pretty sure it's not a grounding problem. It's too consistent and exactly reproducible.
It always reboots fine. It always runs fine except for:
certain games AND SLI being enabled. If I disable SLI it runs fine. As I said in op, cpu temps are fine. I've also stress tested just cpu with no issue.
I even tried blowing desk fan into case to see if there was something on motherboard that I don't have sensor for overheating., Didn't help.
At this point I am thinking there is obscure motherboard defect or obscure graphics card problem that causes only SLI to malfunction. I will update this after trying new motherboard.
I checked temps best I could with built in thermal radar software. Its hard to verify without ability to alt tab at crash. There was no unusal temperature spike leading up to that point.
I suppose it's possible that "thermal armor" cover on motherboard could have blocked desk fan enough that it was still overheating.
I will try swapping my known working 580 SLI into NEW pc and put questionable 680 SLI into known working OLD pc and see what happens. (both PC have necessary requirements for both SLI setups)
I'll also try checking mobo temps with HWmonitor.
Thanks for ideas. Keep them coming! I will post updates later tonight.
I put old SLI setup(gtx580) into new pc and new SLI setup into old pc. Old PC ran fine, new pc still restarted. I then installed motherboard fans thinking it was obscure temperature issue. It still restarts. Finally, I swapped out motherboard with replacement. This didn't help either.
I have done some more reading and noticed my RAM isn't on QVL. I've come full circle and am back to RAM issue. This is the only thing really left that I know to swap out. I find it odd that RAM that tested fine with memtest86+ could be the issue, but I seem to recall vaguely where I had a similiar issue with a PC about 10 years ago- where memtest showed fine, but pc still failed in game. iirc, a friend had the same type of ram, let me borrow his and this corrected it.
IF it's a reproducible problem with certain games and ONLY with SLi, then it's a game/driver issue. have you gone to the website of the game in question and see if similar problems are happening to other people?
it's possible that the chipset drivers for your sabertooth is not compatible with the game when activating SLi?
So I tried different ram and same issue. Quick recap:
PC restarts / freezes during Multiple different games when SLI is enabled. SLI disabled works fine
I have monitored for CPU, GPU, and motherboard temps with nothing overheating
Prime95 and memtest testing came back fine
Here is hardware I have tested:
PSU (tried replacement)
RAM (memtest fine, tried replacement)
Motherboard (tried replacement)
GPUs (Moved whole SLI setup to other pc and it ran fine)
I think I've pretty much ruled out all hardware.
I think I will try a clean install of gpu driver and possibly update intel chipset driver. My understanding is that there is version of it newer than what it on ASUS website. If that doesn't work I guess I'll have to try reformat.
Definitely doesn't make any sense.
edit: Tried clean install of gpu drivers and also tried running ccleaner to "clean up" registry. Neither task helped.
Try force feeding the PSU some air with an external fan. My PSU fan recently died and I had some similar fail at load conditions.
Is it SLI related or just both GPUs loaded. Try SLI diabled and load it with Folding@Home.
Try force feeding the PSU some air with an external fan. My PSU fan recently died and I had some similar fail at load conditions.
Is it SLI related or just both GPUs loaded. Try SLI diabled and load it with Folding@Home.
I will try fan on psu. This psu also has a low fan mode enabled. I'll try high fan mode. I don't think it's defective. I RMAd PSU already and new one does same thing.
I can try loading each card with F@H with SLI disabled, but how does that help me determine what the problem is?
I will try fan on psu. This psu also has a low fan mode enabled. I'll try high fan mode. I don't think it's defective. I RMAd PSU already and new one does same thing.
I can try loading each card with F@H with SLI disabled, but how does that help me determine what the problem is?
This is a good idea. It never dawned on me that I can transplant my whole SLI setup to my old pc to test (which currently has working 580 SLI)
I will try logging feature in hwmonitor pro
I checked temps best I could with built in thermal radar software. Its hard to verify without ability to alt tab at crash. There was no unusal temperature spike leading up to that point.
I suppose it's possible that "thermal armor" cover on motherboard could have blocked desk fan enough that it was still overheating.
I will try swapping my known working 580 SLI into NEW pc and put questionable 680 SLI into known working OLD pc and see what happens. (both PC have necessary requirements for both SLI setups)
I'll also try checking mobo temps with HWmonitor.
Thanks for ideas. Keep them coming! I will post updates later tonight.
Also wanted to mention you might wanna be sure your drivers are up to date on that motherboard too. So head out to ASUS website. Maybe you need a bios update? Normally you don't need a update unless you have a certain issue related to the motherboard. Did you overclock anything?
Hmm.. I was having an issue in BO2 and BF3 where my PC would just go black screen randomly, but Torch Light 2 and BL2 were not affected. Running GTX 580s SLI at stock speed with the latest drivers. Tried everything to fix it. The only solution I could find was to up the volts in MSI AfterBurner and my problems went away. Or to run as you did with SLI off. Well after a while of my GPUs being unnecessarily over volted in SLI, I decided to reinstall my Acronis backup of Windows 7. Then deleted the old Nvidia drivers I had on the backup and installed 314.07 drivers. Now, no crashing no black screens and no more overvolting.. Everything runs as it should.. So, you can always try a fresh reinstall, hopefully you have a decent either full or incremental backup you can load from.. Just my .02
Connections have been checked, PSU. Motherboard, sli bridge, and ram have been replaced. Both GPUs have tested fine in other system. All drivers are up to date. BIOS is up to date.
Quote:
Originally Posted by ep45-ds3l
Hmm.. I was having an issue in BO2 and BF3 where my PC would just go black screen randomly, but Torch Light 2 and BL2 were not affected. Running GTX 580s SLI at stock speed with the latest drivers. Tried everything to fix it. The only solution I could find was to up the volts in MSI AfterBurner and my problems went away. Or to run as you did with SLI off. Well after a while of my GPUs being unnecessarily over volted in SLI, I decided to reinstall my Acronis backup of Windows 7. Then deleted the old Nvidia drivers I had on the backup and installed 314.07 drivers. Now, no crashing no black screens and no more overvolting.. Everything runs as it should.. So, you can always try a fresh reinstall, hopefully you have a decent either full or incremental backup you can load from.. Just my .02
I have been wondering about this and may try adjusting volts / stocks on cards to see if that gets me anything.
Yep. I most certainly have backups of everything. I will be trying a clean install tonight. (going to be a long night. Will get some vodka on way home.
)
Thanks for all your input. I will update this later.
Update. I had case on it's side to look at PSU fan and found something interesting. With intake fans exposed to open air, and with assistant fans wide open, I do not get lockups. If I turn down assist fans, I get lock ups. If I have tower right side up (with only 1-2 inches of clearance to my desk, I get lockups.)
This leads me to believe motherboard is overheating. Since this is a replacement motherboard and first motherboard had same symptoms, this leads me to believe that this is a design flaw in the motherboard or possibly the thermal armor design. Since I can find no similar reports anywhere, I am thinking it must be the unique combination of my case, motherboard, and GPUs causing it. (ie GPUs spewing hot air into case). When I tested older hotter reference cooler 580s, I still had lockup issue but it was without motherboard fans and took longer.
So it looks like my options are:
1. Make a stand to lift case higher off desk and jack up those annoying little assist fans.
2. Try a more aggressive fan profile on GPUs. (Since I get no lockups with CPU stress test, I am inclined to believe overheating component is more near GPU side of things. My thinking is if GPUs do not get as hot, motherboard will not either.)
3. Remove thermal armor to allow direct air from case fans to pass over motherboard.
4. Get a less gimmicky motherboard.
1-3 are free options, but 4 is perhaps most foolproof. I really like the features of ASUS boards, but if they are not stable using advertised features(SLI, armor , etc), it becomes a moot point. I'm not completely opposed to buying another ASUS or buying something else. What do you guys think? Any suggestions for good Z77 motherboard? (The more USB3 and the best layout for SLI / soundcard preferred).
Have you any space to install extra case fans. Also, what cooling setup do you currently have? Positive or negative pressure? I only skimmed through the posts, so this may have already been suggested.
did you end up doing a fresh install?
Intel drivers are notoriously hard to remove or install if you are using a hard drive from your old build.
I have my old HD plugged in and every once in a while I boot from it and notice how utterly slow it goes on my RIVE 3930k 5GHz compared to the same OS on another drive. Random reboots during gaming were common until I used the fresh install with latest drivers.
SLI also needs a lot of power, check if your motherboard has an extra plug for multi-GPU power. this is definitely required as the cards can draw up to 300w each from the motherboard pci-e 3.0. they take up to 150w ea from the two 6-pin inputs but if your power target is high they will draw from the board under load, and 24 pin ATX cant cut it so Asus and other manufacturers regularly place 4 pin molex, 6pin pci-e or even 8-pin pci-e possibly in future on their boards to supply additional power for multi GPU.
You could also try running your ram at Intel specifications at 1600MHz or less with stock CPU and voltage. anything above 1600MHz usually requires VTT VCCSA and PLL voltage adjustment and that can affect intensive processes.
If available when over clocking use DIGI+ power control to increase power limits for ram and CPU after confirming stability at stock settings.
use prime 95 custom run with 13000MB memory target to test memory stability at 1600MHz or less with stock CPU to confirm before testing sli that your memory isn't to blame, then run heaven benchmark on full detail.
I noticed a trend in thermal radar where VCCSA temp would gradually rise until it hit around 70c and that's when lockup would occur. Efforts to keep temp down with aggressive GPU and case fan profiles seemed to corroborate this. I also noticed even at idle, VCCSA would sit around 50C idle.
I had noticed there is a VCCSA voltage in BIOS. For poops and giggles I changed this from AUTO to its lowest value possible (0.8). To my surprise this dropped VCCSA temp almost in half even without the aggressive fan profiling.
Even more amazing- No more lockups. Even with full load and fans on low, max temp on VCCSA is now only 40c.
I don't know if dropping VCCSA that low hurts anything. I couldn't find anything on google that says it does. I have done tests in Heaven, prime95, and previous games that crashed and everything seems to be working fine.
Amazing to go through all of that and solution be one little setting in BIOS.
Here is temps before change:
Here is after manually setting VCCSA voltage to lowest value
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Related Threads
?
?
?
?
?
Ask a question
Ask a question
Overclock.net
27.8M posts
541.2K members
Since 2004
A forum community dedicated to overclocking enthusiasts and testing the limits of computing. Come join the discussion about computing, builds, collections, displays, models, styles, scales, specifications, reviews, accessories, classifieds, and more!