many crashes with graphics all over the screen

reggiew · 30 Oct 2010

Hello,

Ok ive tried to figure out over a month now what the problem is but cant seem to solve it.

My system specs: My system is about 2 years old.

Nothing is overclocked !

Windows 7 ultimate x64 clean installl on formatted drive. With most updates installed ( i did a restore after i did a recent winupdate coz my pc would crash alot after)
Intel quadcore 2.5ghz q9300
Asus p5q deluxe motherboard (all latest drivers installed including ATK0110 ACPI UTILITY but also tried without ATK0110)
The very latest bios update
Ram: a-data technology 2gb stick (dqve1b16) and 2x 1gb stick nanya technology (nt1gt64u8hb0by-25d). Memory is running at 333.3mhz on 5-5-5-15 44 clocks, command rate 2t. Its all set on auto. On 1.80 volts
Harddrives: 2 x sata harddrives samnsung hd321kj & samsung hd502ij (didnt try yet with just 1 drive)
Acer h2d5h monitor ( on pnp and not running the acer driver coz i cant find any win7 drivers)
Graphics card: single asus gtx 280 (tried win7 default installed drivers, tried asus own latest drivers. Tried nvidia 258.96_desktop_win7_winvista_64bit_international_whql and 260.99_desktop_win7_winvista_64bit_international_whql)
Sound: using onboard soundchip in HD audio quality.
Windows DVD station: TSST corp CDDVDW SH-S223Q ATA) (didnt try yet without dvd drive)
Mouse: g5 laser mouse.
Hardware case: coolermaster haf 932 with thermalright cooler on cpu with 2 fans on it.
Powersupply: coolermaster realpower m700

On idle temp my

About the crashes:

The crashes, pc unresponsive and have to reset the computer, im getting are that my screen is filled with collored squares, also the mouse pointer turns in to a white-ish square. At least they look like small squares but could also together form diagonal lines if you look closer.
I've never seen weird graphical artifacts in games before crashes.
I never get a blue screen. When my pc crashes my screen is just filled with , most of the time green, small squares and becomes unresponsive. Sound plays on till it gets garbled and my pc also never automatically reboots. I have to reboot it myself even though i have my pc configured as creating a minidump file and then reboot itself. It never really creates any minidump files though, just kernal dump files.

This started to happen over a month ago mostly when i played the game dark age of camelot. However playing bad company 2, pretty graphic intensive game, and civilization V i didnt seem to have that problem. At that time my pc needed some serious cooling and was overheating alot back then. I was running vista 64 bit and ordered my coolermaster haf 932 case with thermalright cpu cooler etc. There was also loads of dust in my case.

When i took apart my pc and installed all my hardware in my new haf 932 case i cleaned out everything from dust first with compressed air. Also cleaned old cooler paste from cpu and applied new paste.

Later the problem started to spread. Also at times giving me the same crash in bad company 2 but so far only occasionally and so far never in civ 5. Even worse. after i'd start up the pc it would start to give me the same crashes when i'd open up the msn window, strange, or start VLC mediaplayer to play video's.
I'd have to reset the pc like 4 times before the crashes would finally dissapear. At times somehow powering down the pc and then restarting it seemed to help ?
I could then use my pc for hours without a problem till i would start a game and then i'd crash again.

Other times the crashes would dissapear for 2 days and then come back as often as they appeared before.

Action's ive taken:

Used Verifier.exe but after a bunch of reboots it never gave me any error.
Did a memtest86 run over night but never gave me any error.
Ran OCCT stresstests but never gave me any error.
Ran Furmark stresstest and left it running with my gtx280 at max temps for 15 mins, about 105 celcius, but never gave me any error.
Ran prime95 overnight, for over 6 hours, but never gave me any error.
Installed the mentioned nvidia etc drivers a bunch of times , used driversweep in safemode.
Disabled TDR in registry but still get errors. Tried to set TDR on 8 sec but still get crashes. With TDR disabled i do notice i get a blank screen at times for 2 seconds while playing bad company 2, just it doesnt crash.
Reset pagefiles but still get crashes.
Ran Harddrive diagnostics (everything fixed if there was anything to fix) on both game partition and C: but still get crashes.
Did a SFC.exe /scannow but this didnt give me any results, no problems.
Tried disabling AERO but still getting crashes.
Used the directx webupdate but still get crashes.
Removed alot of programs like asus utilities, daemon lite etc and removed the files/services or disabled them.
Renamed nvlddmkm in system drivers dir to .old and put in, and expanded, the one from the NVIDIA dir but still get crashes.
Uninstalled old graphics drivers and installed/reinstalled new ones using this guide
nVidia Driver Cleanup Guide !! « Captain Debugger still get crashes.
Went into bios and set cpu voltages and whatever settings manually, sorry cant find that post again, to be sure that nothing is overclocked but still get crashes. I also get the crashes in default bios settings.
Removed ram memory sticks to see if that would solve the crashes but still getting the crashes.
Updated bios to latest version but still get crashes. Old version, default version, gave crashes too.
Set monitor timings etc but still get crashes. I have 2 dvi ports on my graphics card but when i plug my monitor on 1 dvi i loose display alot, goes blank, but on the other port it doesnt happen.

Actions i've not taken:
Use some meter to check voltages.
Open my gtx280 to check its paste or whatever dust etc.
Replace my graphics card with another.
Use a stresstest for my powersupply.

My system runs idle at 30 to 40 celcius on all cores. Maybe reaches 50 celcius under full load.
My graphics card runs idle at 47ish celcius. Reaches about 80ish celcius under full load, for instance playing bad company 2 on full detail settings. Fans turn on to 100% then.

Voltages are:
cpu vcore 1.26
vin1 1.73
+3.3v = 3.38v
+5v = 5.02v
+12v = 11.65v and lowest i've ever seen it go is 11.60v
vin6 1.20v

I've had my os clean of any stuff running in the background except soundmax audio software and currently have running logitech mouse software, soundmax audio, search and destroy anti spyware/virus and acer display management software.

So far i've only 1 minidump in my minidump folders but do have alot of dump files in my LiveKernalReports/watchdog folder.
Always the error the livekernalreports displays, using bluescreenviewer, has to do with nvlddmkm

Installed the tools for windebug and the debug always gives an error with TDR timeout. Must say though i dont really have a clue if im using the debugger right.

Anyway im not sure which minidump files to attach since i only have 1 old minidump and many kernaldump files.

Ill also attach a CPU-Z report and a CPUID Hardware monitor report. Also attached an older , the only one in the dir, minidump file (102610-32354-01) from 26th october and the 5 most recent kernaldump files.

I would suspect the problem is my graphics card or memory but they withstand the stresstests without any problem or removing ram memory sticks doesnt solve the problem either.

Anyway thanx alot upfront for the help.

Greetz,

Reggie

zzz2496 · 30 Oct 2010

Do you have a spare graphic card? It looks like a dying graphic card...

zzz2496

reggiew · 30 Oct 2010

I could get an old geforce 100 but what i dont understand is why does my card pass all stresstests and never crashes.

I just didnt get the geforce 100 yet coz i love playing games and well thats a crappy old card :).

zzz2496 · 31 Oct 2010

You need to understand, your stress test that stresses the graphic card is only Furmark, yet it's not even stressing the whole graphic card. IMHO, this is either a card going bad, or a bad power supply...

OMG, your power supply - why is it only supplying 11.6 volts??? It's nearing the low threshold of ATX standard. Do you have a spare power supply? Sorry I missed the power supply earlier. A 12v rail that only supplies 11.6v, there's definitely something wrong with it...

zzz2496

reggiew · 31 Oct 2010

Im not sure if there's something wrong with the power supply. I've googled on it and more powersupplies report about 1.65 volts. I just use cpuid hardware monintor to check my voltages btw and im not sure how accurate software is. Prob not.
Isnt it so that +12v voltage can stay within 5% reach ? I think 12-5% is 11.4v.

Also what i wouldnt understand then is why doesnt it crash my video card during video card stresstests or cpu stresstests.
Another thing is why would it not crash for 2 days at times or constantly crash mostly after reboot nowadays.

I'd love the problem being my powersuppply though, its alot cheaper then replacing my gtx 280 which i bought for 600euro at the time lol.

Anyway the only old powersupply i have is one with a 4 pin card connector for power so i wouldnt be able to power my graphics card then

.

Ill go check in bios what bios reports for 12 volt rail.

Ok i just checked my bios and it reports 12.096v on the +12v rail.

For accuracy comparisons;

Bios reports 3.344v on +3.3v
Bios reports 5.016 on +5v
Bios reports 1.256 on CPU Vcore

Idle temp seem accurate also in bios;

Bios reports 14c on cpu while cpuid hardware monitor reports 12c (assuming thats cputin ?)
Bios reports 24c on mb while cpuid hardware monitor reports 24c (assuming thats systin ?)

So the only voltage thats way off in cpuid hardware monitor is the +12v rail.
In hardware monitor the +12v rail never fluctuates though. Its always 11.65v and only at times it goes to 11.60v but never lower.

Also i've got a geforce 100 now. If id put it in my pc would it still require the same amount of power ? Im guessing not ? so is that a way to check if my powersupply is dying if i'd still get crashes ?

Oh yeah and an important thing to take note of is that i never get blue screens. My screen just turns into very small squares which basically form vertical lines. Windows 7, though configured to do so (create small minidump and then restart pc) also never creates a minidumpfile, only kernal dump files, and never automatically restarts. So in short... My screen gets filled with small squares and when i look close they all form vertical lines, and my pc then becomes unresponsive, at first sound just continues to play till sound gets all messy. I then have to restart my pc to fix it.

As a sidenote windows 7 hasnt created a single dump file for 4 days now. My last dump files were on the 27th and were kernal dump files.

zzz2496 · 31 Oct 2010

I've had your problems last year, it was my power supply. Here's what you can do, using whatever monitoring software you have (use one with logging capability), watch your 12v rail, start from doing nothing, then do furmark stress test and at the same time do OCCT. Watch your 12v before and after doing stress test. You have to stress both CPU and GPU at the same time to get an accurate result. A game will tax both CPU and GPU, Furmark isn't (on it's own) neither OCCT. So hit it both, see if the voltage dropped... If it dropped, and your computer crashed, restart it, then read the voltage LOG.

zzz2496

reggiew · 31 Oct 2010

ok thanx. Ill try that. Did you have the graphical small squares also fill your screen or ?

zzz2496 · 31 Oct 2010

I got smears of colors... and bunch of blinking random pixels all over the screen...

zzz2496

reggiew · 31 Oct 2010

ah ok. ill run those tests tomorrow. Any specific settings ill need to know about concerning those stress tests ?

i've always thought maybe powersupply since video card seems to give no trouble under high load. Then again why would i usually get crashes at start up as soon as i load a video etc and then the rest of the day no problems. Weird if thats coz of weak voltages since the rest of the day i play heavy stressing games.

reggiew · 01 Nov 2010

ok i did a test today with furmark running for 10-15 mins with my gpu temp at 104ish celcius and at the same time had occt cpu test running.

hardware sensors monitor and occt saw my 12 rail voltage drop from 12ish volt to 11.93ish volt. Not a big drop.

Oh btw both pcwizard and hardware sensors monitor seem to always report about 12.10 volt on the +12 rail instead of 11.65 like the other program i use to monitor.