Weird freezes... bad/failing video card?

[Tomec]

Every now and then while playing games, and seemingly only while playing games (important note as I also do a bit of rendering, photoshop, 3d modeling, etc), my computer will totally freeze. When I say totally, I mean totally. This isn't like a chug or program hanging. The screen freezes whatever the exact frame was, and (weirdly enough for me) the AUDIO freezes on the tone or whatever being played right now. Creates the WORSE noise, I don't even know how to describe it lol. Anyways, the only way to fix this is to hold the power button to power off the PC.

This has been happening for a little now, but as of late it's gotten worse. I've gone through all the normal steps to try to trouble shoot it. Re-installed drivers. Dusted thoroughly. Checked cables/seating of things. Possibly more that I can't remember right now lol.

But it's still happening and happening seemingly at random. Sometimes I'll go days without it, sometimes it'll happen multiple times in a day/session. So if anyone has any ideas...

Here are system specs for reference

Asus Rampage 4 Extreme
i7 3930k @ 3.2 GHZ
G.SKILL Ripjaws Z Series DDR3 16GB 9-9-9-24
Antec "High Current Gamer" 900Watt PSU
EVGA GeForce GTX 570 Superclocked 1280 Ram

Besides the video card, all parts are just over 2 years old. Video card is about 3.

Note on HDDs, for games/system I have it split between some SSDs and a WD Caviar Black. However, games on both drives will freeze and it seems like drive makes no difference on whether it freezes or not.

[megabeat]

I blame the ram. Test it with HCI memtest, its small non-install program that checks ram. Run it for 20 minutes, but in my xp if RAM is bad, it will find errors in 10 seconds.

[Tomec]

So digging deeper and starting some more tests (mainly stress tests), I'm now thinking it could be CPU related.... I can pretty readily make the computer freeze (not sure if it's same kind since I wasn't playing music while doing this) by running Prime95 (quick note, just found out it was a way outdated version as of writing this post) Blend Test for about 5-10 minutes. Since that's more a Ram/CPU test I pulled out my copy of memtest64+ and let it run for just under 2 hours with no errors. So I don't know.

I just updated my Prime95 and will see if I still freeze into it... but I have a feeling Ram is OK.

[megabeat]

Memtest 64+ is inferior to HCI memtest. Ive seen Memtest passing tests where HCI finds errors.

[Cyrops]

Not saying I rule out RAM fault, you mentioned: "Note on HDDs, for games/system I have it split between some SSDs and a WD Caviar Black."
How exactly is it setup? How many SSDs? Do you have pagefile on? if so which drive is it set on?

[Thunderball]

It's either RAM or a CPU issue. I've had a simillar problem when I ran my i7 2600k on 3.6 constantly. Ever since I've downclocked to 3.4 I've been getting it only about once a month. I've suspected it to be a combination of faulty PSU and not enough voltage, but tests with different setting and another PSU didnt give anything. My best guess right now is that it's some kind of software issue with virtual cores.

[Tomec]

megabeat: I did a quick run (can't do a full 20m run atm due to things) of the HCI memtest and it showed no errors.

Cyrops: I have windows on an SSD, with offloading things like user settings and desktop and stuff to the WD. Pagefile is ON, on the WD, with a 2-20 gig size. Current allocation as of post, 2114.

Thunderball: I'm currently leaning more towards CPU also. Why, I'm not sure though. I really have no OC on this (actually... isn't 3.2ghz the default clock speed...) and a 900w cpu? Granted could obviously be going bad or something, but I don't know.

It's also very odd that I can do other things that max the CPU but wont freeze at all. Render high quality/production level things and stuff, fine. Run a game or prime95, I'm going to throw a fit.

[Cyrops]

megabeat: I did a quick run (can't do a full 20m run atm due to things) of the HCI memtest and it showed no errors.

Cyrops: I have windows on an SSD, with offloading things like user settings and desktop and stuff to the WD. Pagefile is ON, on the WD, with a 2-20 gig size. Current allocation as of post, 2114.

Thunderball: I'm currently leaning more towards CPU also. Why, I'm not sure though. I really have no OC on this (actually... isn't 3.2ghz the default clock speed...) and a 900w cpu? Granted could obviously be going bad or something, but I don't know.

It's also very odd that I can do other things that max the CPU but wont freeze at all. Render high quality/production level things and stuff, fine. Run a game or prime95, I'm going to throw a fit.

It's your choice about pagefile, but on my old 4gb ram I had it off, and now with my new rig with 16gigs I am never turning it on. There is a posibility of the WD acting up, have you tried running disk check?
Have you tried the HCI memtest megabeat suggested?
i7 3930k is 3.2gigz default speed.
Have you monitored system temperatures?

[Tomec]

Now that I'm done with finals for school.. back to figuring out this.

Disk Check: Shows nothing. I also have Crystal Disk Info and it shows all drives in "good" condition, both SSDs and WD.

HCI memtest: Ran for close to an hour, had about 200% coverage. 0 Errors.

Temps: CPU never goes above 40c under load. Current temp, which is pretty much idle besides having this web browser open, is 25c. GPU is about 45 idle or so (depending on fan speed/speed curve), and I think it goes to close to 80 under load. (Going to test the video card load temp after this post. Along with looking closer at other temps like the PCH and "Motherboard" sensor.)

Edit: Ok, so temp wise the GPU got to 85c after about 5m of the stress test. It seemed to peak around 65-70 ish pretty quick, then slowly kept building up to 85 when I decided to stop the test due to temp. Neither the PCH or "Motherboard" sensor changed at all from idle. I also ran a quick Prime95 just to boost temps (didn't want to freeze again while editing) and CPU didn't go above 40, in fact it was 38. Again PCH/Mobo didn't change.

I still find it highly odd though that I can do things that push the PC components just as much, but it wont freeze the computer.

[Sturmbringe]

The only time I have ever gotten freezes was when I tried to overvolt my GPU through EVGA PRECISION four notches.

Try to reduce your GPU's OC and see what happens.

[glo]

Now that I'm done with finals for school.. back to figuring out this.

Disk Check: Shows nothing. I also have Crystal Disk Info and it shows all drives in "good" condition, both SSDs and WD.

HCI memtest: Ran for close to an hour, had about 200% coverage. 0 Errors.

Temps: CPU never goes above 40c under load. Current temp, which is pretty much idle besides having this web browser open, is 25c. GPU is about 45 idle or so (depending on fan speed/speed curve), and I think it goes to close to 80 under load. (Going to test the video card load temp after this post. Along with looking closer at other temps like the PCH and "Motherboard" sensor.)

Edit: Ok, so temp wise the GPU got to 85c after about 5m of the stress test. It seemed to peak around 65-70 ish pretty quick, then slowly kept building up to 85 when I decided to stop the test due to temp. Neither the PCH or "Motherboard" sensor changed at all from idle. I also ran a quick Prime95 just to boost temps (didn't want to freeze again while editing) and CPU didn't go above 40, in fact it was 38. Again PCH/Mobo didn't change.

I still find it highly odd though that I can do things that push the PC components just as much, but it wont freeze the computer.

Well, one thing is for sure, your CPU temperature sensors aren't reporting accurate numbers as those are essentially impossible, especially under Prime95. Reporting incorrect values points to problems on the chip or the motherboard itself. I'd at least verify with RealTemp to make sure you see the same values.

[Faithh]

PCH is the chipset temp which is useless, the fan starts to spin at 80° by default. Don't use AI suite 2 to monitor your CPU temps, coretemp shows here 60° on cores and ai suite 2 is displaying 46°. The current version of coretemp (RC6) is kinda unstable so use Realtemp instead. Do not use any turbo profiles or CPU level up oc stuff.

Have you even monitored the VRM temps of the GPU? Leave gpu-z running in the background and select "log to file".

[Tomec]

So it looks like the AI Suite I was using was only reporting the LOWEST core temp (really...?) which under Realtemp is about 40 while Prime95 is running. HOWEVER, my hottest core peaked at 52 during it. I don't think I mentioned it before, but I have a custom water setup on my CPU. Probably overkill since it was a leftover cpu/mobo/vid card rig from an old build... So I never really thought about the temps being odd. Especially in winter.

I'll double check whether I use turbo profiles or level up in the bios, but I remember pretty much turning off a lot of stuff and lowering speeds to default clock levels. I really don't mess around much with that stuff anymore.

As for the GPU stuff. I definitely haven't messed with the clocks or voltages in that. So the whole "turn down your OC" thing differently isn't the issue here. I tried looking for VRM temps on the GPU with GPU-z and logging to file, but I'm not seeing anything about that. I even downloaded a fresh copy to be sure.

[glo]

Here, try running this and see if you get any errors. It's normal for your screen to bug out and revert to a super low resolution during testing:

http://mikelab.kiev.ua/index_en.php?...ROGRAMS/vmt_en

[Tomec]

Here, try running this and see if you get any errors.

No errors found, but it couldn't do a couple tests due to saying it wasn't supported... I don't know if this was age of video card, driver issue, etc though.

Here is the whole log thing

Changing video mode to 640x480x16...OK
[12/12/2013 9:25:39 PM] Test started for "Primary Display Driver (NVIDIA GeForce GTX 570)"...
Trying 16bpp RGB:565 mode...OK
Trying 16bpp RGB:555 mode...OK
Trying 16bpp BGR:565 mode...NOT SUPPORTED (Code: 80004001)
Trying 32bpp RGB:888 mode...OK
Trying 32bpp BGR:888 mode...NOT SUPPORTED (Code: 80004001)
[12/12/2013 9:36:44 PM] Pass completed (0 errors found)

[glo]

No errors found, but it couldn't do a couple tests due to saying it wasn't supported... I don't know if this was age of video card, driver issue, etc though.

Here is the whole log thing

Changing video mode to 640x480x16...OK
[12/12/2013 9:25:39 PM] Test started for "Primary Display Driver (NVIDIA GeForce GTX 570)"...
Trying 16bpp RGB:565 mode...OK
Trying 16bpp RGB:555 mode...OK
Trying 16bpp BGR:565 mode...NOT SUPPORTED (Code: 80004001)
Trying 32bpp RGB:888 mode...OK
Trying 32bpp BGR:888 mode...NOT SUPPORTED (Code: 80004001)
[12/12/2013 9:36:44 PM] Pass completed (0 errors found)

Well, your VRAM is still good, so that's a good thing. Unlikely that the GPU core is bad as that's quite rare. So now you can rule out your GPU/CPU. Next up, let's run a HDD S.M.A.R.T. test to make sure you don't have a failing drive causing lockups:

http://crystalmark.info/software/Cry...o/index-e.html

Take a screenshot of the entire window and post it here. If you have more than one, post them all.

[Cyrops]

Have you checked transistors on MB?

[George]

What brand/model is your SSD?

[Tomec]

Here are the links to the drives in crystal disk info...

Corsair Force 3 (C, windows sys files and other things I want to load quick)
http://imageshack.us/a/img571/6066/e193.jpg

WD (X, other programs, games, windows user profiles and stuff
http://imageshack.us/a/img35/5722/n3rk.jpg

Kingston HyperX (G, games)
http://imageshack.us/a/img594/3270/yi05.jpg


As for transistors on mobo. Just visually to see if anything looks crooked or something. I wouldn't know what/how to do anything besides that.