Solved Random reboots, error reporting calls it BSOD, next gen hardware

[ganjiry]

It is only Watch Dogs at this point now, yes, but see my last post about GPU-Z readings and I have a feeling if I had any other heavy games like this they might perform funny too.

Definitely not running it across all three displays, since it's such a heavy game. When I have my display set for surround (a virtual single display) it disables SLI, which I need to run the game at a higher framerate.

Also... I'm not high... at least not right now.

Sorry mate but I don't know what those Gpu'z readings mean as I've never needed to use it for trouble shooting.
When I first started playing it it was unplayable until I updated to the latest Nvidia driver but that transparency effect was on this latest set. What settings are you running the game at?
I found that the V-sync in Watch Dogs is totally broke and causes a lot of lag spikes/freezing with each of the different settings apart from off. The settings which most affected how well the game ran for me were,
Antialiasing, 2x TXAA ran best for me.
Dof=off
V-sync = off
Max pre rendered frames = 3
Textures
Run the launcher as admin.
Disable desktop composition for the game to save on Vram.
You may have different results than me though. I take it your gpu's are 4Gb cards as you're running 3 screens?

I can't remember what temp you said your cards were getting to but the Kepler gpu's throttle by 13MHz at each of these points 70,80,85,90c. I wouldn't have though that would cause you a problem but it maybe worth checking.
Some info on Kepler boost ~~The GTX 670 Overclocking Master-Guide~~

 

[BotenRedWolf]

aaaaannnddd..... I woke up this morning to find my computer freshly rebooted with the "Windows has recovered blah blah **** you" box... so there's something we still haven't figured out...

edit: I also remembered I fell asleep with a game running...

 

[essenbe]

The dump file is the same as all the others

Loading User Symbols
Loading unloaded module list
........
*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************
Use !analyze -v to get detailed debugging information.
BugCheck 116, {fffffa800d276010, fffff8800fda2d1c, ffffffffc000009a, 4}
*** WARNING: Unable to verify timestamp for nvlddmkm.sys
*** ERROR: Module load completed but symbols could not be loaded for nvlddmkm.sys
Probably caused by : [COLOR=red]nvlddmkm.sys[/COLOR] ( nvlddmkm+98bd1c )

I believe GPUz has a 12V reading, watch that while running your next game. The ATX standards are +/- 5% (11.4 - 12.6) So anything within those ranges should be OK. If you had the same problem with the 670 and now with the 2X 760's it seems to me it is the board or the PSU. However, the possibility still exists that one of your cards is defective.

 

[BotenRedWolf]

Now I'm experiencing it everywhere, even my mouse cursor has problems keeping up with things. I don't quite understand it. It's more prevalent with the mouse cursor and such when I have a virtual machine running (VM is on but inactive and only allotted 4GB of RAM).

 

[essenbe]

And how many CPU cores. I run about 5 VMs but I never leave them running. I run them when I am using them, then shut them down. Inactive still takes away from system resources, which you appear to be having problems with now, for some reason. Shut the VM down and see if that makes a difference.

 

[BotenRedWolf]

And how many CPU cores. I run about 5 VMs but I never leave them running. I run them when I am using them, then shut them down. Inactive still takes away from system resources, which you appear to be having problems with now, for some reason. Shut the VM down and see if that makes a difference.

Single core for most, that one may be two, but again, idle, and host CPU usage was still in the single digits. Shut it off and it works better, but not perfect. It's still strange when there's single digit CPU usage but acts like it's at 100%

 

[essenbe]

If I am not mistaken you have 4 cores and 16 GB of ram. If you gave the VM 2 cores and 4 GB of ram, you gave given 50% of your CPU and 25 % of your ram, plus the 1.5-2 Gb it takes Windows to run. That is really a pretty sizable amount of resources from your computer. That being said, you really should have enough left for normal tasks. Windows should be running on 2 cores and somewhere around 10 GB of ram, which should be sufficient for simple routine tasks, in my opinion. With it running, in windows, click start and type task manager and click the performance tab. It should show you the resources being used and how much you have free. But, again, you should have enough resources for normal usage. But, if you try doing something that is CPU intensive , you may see a big performance hit.

But again, that's why I shut mine down when I am not using it. How much resources you give it matters. A Linux Distro will usually not use all of 2 cores or 2 GB of ram, Windows or a mac install would need more ram and could use more CPU, but I don't give them more than 2 cores.

 

[BotenRedWolf]

Yeah, the VM I use the most currently is a Mac OSX VM, I'm trying to get OSX working on my Lenovo Yoga. Pain in the ass, but would be really really cool. Any ideas on why my machine seems so easily choked, even at lower usage levels?

 

[essenbe]

No, I really don't. Basically your CPU has 4 cores and mine has 4 cores and uses hyper threading so 4 cores/8threads.
That being said, I can't see how you would have mouse lag. You still have ample ram left But have divided your CPU in half. The CPU would take a definite hit in a CPU intensive task, but moving the mouse does not qualify. It should work as normal, I would think. Does it drag on other VMs other than OSX? If not, that may be the answer. Maybe the Mac OS is dragging your computer resources down, even in a VM. And that is only a guess. I have never tried that one. I run XP, 8.1, and several Linux distros, and they do not bother mine that much.

 

[BotenRedWolf]

Typically the other VM's don't affect the mouse, but I've seen that activity at any usage level, even just Chrome with a couple tabs open.

 

[essenbe]

The thing we don't know is how many and what processes is windows running in the background. And it almost always has something running, just more at times than at other times. The thing is, you've been having graphics card problems and not these kind. Which makes me wonder if you don't have some type of motherboard problems. If you are using a USB mouse, the Motherboard is giving the mouse power and responds to the movement, just like it gives some power to pcie slots and receives input from the cards. But, there is no test for the motherboard, it is usually a process of elimination.

 

[BotenRedWolf]

The mouse is USB, but wireless (the batteries are new, too). I still haven't gone and tried using only one video card at a time to see if that's causing issues, nor can I get a good reading of my 12V rails under load...


*idea* I have a fluke DMM, wonder if that's precise enough to check my rails.

 

[essenbe]

I really don't know either, but it usually is difficult to see the voltage under load. The best place to get voltage readings is in BIOS, but that just shows idle readings. This is a tutorial on testing power supplies, but I doubt it will test it correctly under load. http://www.sevenforums.com/tutorials/301799-psu-test-dc-output-voltage.html

 

[BotenRedWolf]

I don't see why it wouldn't be inaccurate under load, it will read the same thing the device is seeing. It'll just be uncomfortably warm down there... I'll give it a shot when I get back from class tonight. (Ironically, it's the perfect class to ask that question in, too... "Computer Hardware Essentials" as part of an A+ certification program)

 

[essenbe]

Sounds good. Ask how to determine the power going to the Motherboard and to the pcie slots. I guess the real answer we would like is how much voltage is coming out of the PSU and how much power to each component under load. That is the real thing most of us would like to know. Is the PSU putting out the required power and is the motherboard VRM distributing it properly.

 

[BotenRedWolf]

All voltages were in spec. Measured PCIe cards at their 8-pin connectors on the board (not the connector) and read just over 12V. Load variation was about .05 to .1V with Furmark. Seems like my PSU is more than capable. Just now I did have another BSOD, this time it actually WAS a BSOD saying the driver crashed and the attempt to recover timed out. One thing I did that seemed to help some of the performance issues was enable the iGPU and crank up it's shared memory. Don't know why, maybe because those drivers were never installed? It's confusing. The most recent BSOD was trying to play a movie I ripped a while ago.

 

[essenbe]

I can't speak for the Z97 chipset, but I'm not sure there would be a difference. But, with the Z77 chipset, if you have a dedicated GPU installed, the IGPU is disabled unless you have LucidLogix Virtu MVP enabled and installed.

 

[essenbe]

I have looked over this thread again and run every dump file you've given me and come up with the same thing. Every one was a Video TDR error. TDR means the video driver crashed and tried to recover. Technically it stands for Timeout detection and recovery. When the Drivers crash, they attempt to recover to a functional desktop, in Windows 7. In all previous OS they resulted in an absolute lockup of the computer or BSOD every time. There are a number of issues which can cause this, some of which we have covered. We have tried different drivers, are reasonably certain we have a clean install of the drivers, you have tested the PSU and are reasonably certain that the PSU is putting out what it should and the PSU is plenty strong enough to run your hardware. Your temps seem to be fine, so we know it is not overheating. The one thing that keeps me believing there is something else, is that you say it did the same thing with your 670. So, unless you want to believe that you bought 2 bad cards in a row, It seems to have to be something else. For right now, I would like you to run a short test to check for corrupt or missing files.

Open a elevated command prompt ( click start, type cmd in the search box, right click on the cmd entry and select run as administrator) in the black box that opens, copy/paste sfc /scannow. If you decide to type it, notice the space between the sfc and the /. It is a system file checker which will scan your system files and attempt to correct any missing or corrupt files. What we want are the results to say windows found no integrity violations. If it says files were found but could not be repaired, close the box, reboot and run it again, after opening the administrative command prompt. You may have to reboot and run it three times for it to repair all system files. If it can't repair them after 3 reboots, let us know. Try that and let me know how it goes. Once we get those results we are going to run memtest86+. We'll talk about that tomorrow. Just please don't get ahead of me.

 

[BotenRedWolf]

"Windows Resource Protection did not find any integrity violations."

 

[essenbe]

Good, that's exactly what we would like.

Next I would like you to run memtest86+. It is quite a long test as it needs to be run for 8 complete passes all at the same time. Please download the test from Memtest86+ - Advanced Memory Diagnostic Tool. Although the names atr similar, this testis much better than memtest86. It needs to be run for 8 complete passes or until you get an error. If you get an error, stop the test. In the link above if you will go down to the middle of the page you can download either the ISO file which needs to be burned to a CD by Windows image burner or any disk image burner you choose. Or you can download the auto installer for USB Flash drives to run it from a flash drive. Whichever you choose, be sure to download the zip file and extract it. Each Pass consists of 10 tests, each pass checks a different thing and each pass checks different things. It takes a minimum of 8 passes to completely check the ram, more passes are better. It will take quite a number of hours to do the test, so it is usually better run over night. To run the test, boot from the media you chose and don't press any keys, the test will run automatically and will continue running until you tell it to stop. It will tell you how many passes it has run and how many errors are found. If it finds errors they will show up as red lines at the bottom of the page. It will take somewhere around 9-10 hours to check my computer with 8 GB of ram. If you have more it will take much longer.