Constant restarts - out of ideas

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
I've recently updated my PC with the following hardware:

- QuadCore Intel Core 2 Quad Q9550
- 2x Corsair Dominator 2 GB DDR2-800 DDR2 SDRAM
- Asus P5Q Deluxe

At the same time, I installed Windows Vista 64-bit, instead of the XP I'd use before.
The trouble started when I started playing games - the computer would restart itself all of a sudden. I've tried several games (Far Cry 2, Fallout 3, Red Alert 3), and they all cause the computer to restart. Sometimes I can play for about 2 hours, other times it happens within 5 minutes.
It doesn't happen when I'm just using windows or surfing the net.

After a bit of trial and error (and several restarts) I noticed the temperatures for the four cores were way too high - two of them reached around 90C while playing games.
I quickly bought and installed a new CPU-cooler (Zalman CNPS 9700NT), and the temperatures are fine now.

This didn't fix the restarting problem however.
I've tried fiddling around with drivers for my bios (my bios is now updated to latest v. 1406) and GPU.
I've formatted my drive and installed XP Professional 32-bit, but the problem's still there. Windows event viewer only says, that the game I'd been playing suddenly stopped.
I've tried running the games with only one block of ram, and then the other, and moving them to different slots, but it still happens.

I haven't seen my GPU's temperature rise above 79C, and from what I can understand that's acceptable..? Also, the card is a few months old and haven't caused any problems before.

I'm am not overclocking in any way.


I'm at my wits end here - I'm having a hard time finding the problem, since I've upgraded so much all at once.

Could my CPU have been permanently damaged from the heat? ... as I said, the temperature's fine now, but it did get pretty hot several times before I noticed it.


Any ideas would be appreciated - thanks in advance!


My specs are:

- the aforementioned cpu, motherboard and ram.
- OS: Windows XP Professional 32-bit
- GPU: NVIDIA GeForce GTX 280
- 2x harddisk (ST3120026AS ATA Device (120 GB, 7200 RPM, SATA))
- PSU: Thermaltake ToughPower Cable Management W0116 (750W)
- Soundblaster Audigy 2 6.1

CPU-Z report that may be useful:
CPU.jpg

Cache.jpg

Mainboard.jpg

Memory.jpg

SPD.jpg

 
Well, right off your tRAS is set to 18 instead of 15... but that should not be an issue.

You should run memtest for several hours or all night.b You could also test one stick at a time.

It could be your PSU. Your Thermaltake is not a bad model supposedly, but I don't trust the company. They sell some bad PSUs.

Check your temps using Real Temp.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
Thanks for the reply!


Ok, that's news to me - I'm not that tech-savy... do I fix this in the bios? (yeah, I know that you said, that it shouldn't be an issue, but I might as well)



I've already tested one stick of ram at a time - gonna' try the memtest, when I have the time.


Any way to test this?


... not sure what for..? My CPU's temps are fine, now that I've installed the Zalman.



Keep the ideas coming!
 

roadrunner197069

Splendid
Sep 3, 2007
4,416
0
22,780
Real temp is more reliable then PC probe.

Asus makes great hardware, but their software sucks and is very unreliable.

I have seen 100s of post about blue screens on Asus, and 90% of them are caused from the Asus software. Uninstall all the Asus software and see what happens.

If it still happens feel free to use the software, but I bet uninstalling Asus software will fix your problems.

I use Asus exclusive and have experienced the BSODs from Asus software first hand.
 
There is no way to test a PSU at home. It takes some pretty sophisticated equipment to determine if a PSU is providing the quality of service required to run a stable rig.

This is one of the reasons I'm such a PSU elitist.

You should always find out what your actual temps are. Your hardware doesn't stand up and say, "50 degress Celsius!" All your sensors are very simple things, that have to be interpreted correctly. Picture a mercury thermometer with no numbers on it. That is what your CPU sensors generate.

You need a program that will come along and say "Ah, this is an X processor with the Q revision, so this is where the numbers go on the thermometer."

If you compared the two programs, you would have a better picture of how high your temps actually got, and whether that might have been a problem.

You can manually change your timings in BIOS, yes. Should not be an issue though.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
Perhaps I wasn't clear enough in my previous posts - I have been monitoring my temperatures for the last few days, and not through PC probe.

I used SpeedFan at first, and then I used (and still use) Real Temp.

The cores temperatures seems to be fine - the only temperature-related problems I think I could suffer from, would be the GPU (79C) or maybe if the CPU took damage during the first two days (before I replaced the stock cooler).

@roadrunner
The only thing I've installed, that was motherboard related, was an update for the bios, SoundMAX ADI audio drivers (I removed my normal soundcard, just to see, if that conflicted with anything) and Marvell Yukon Gigabit Ethernet Drivers (so I can get online).
I haven't installed anything else, like EPU-Six Engine.
Also, I haven't experienced a BSOD - the machine just restarts completely.
 
There is an update on the Marvell drivers that is available from MS but not ASUS, at least for my P5Q-E.

Ooops. I missed this before. Your memory is overclocked. Sure, it has the EPP #1 setting at 533Mhz, but it's NOT RATED for that. It's 400Mhz RAM running at 533.

Take the settings as they are now, but set your DRAM frequency down to 800Mhz or below. Your BIOS will give you a list of options, one of which will be in that ballpark.

I saw the information in the SPD tab and thought your RAM was 1066, but as you listed in the beginning it's 800Mhz. That would be a problem.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
I let XP update my Marvell drivers automatically (is that what you meant?).
I also set my DRAM to 800Mhz.

But it still restarts... The only thing that I can think of doing, that I haven't already done, is using memtest-86 (like Proximon advised).

I doubt I can find the time in the next few days, so if anyone has any other idea, please let me know.
 
Yes that's what I meant on the drivers... the original drivers had issues. Not that I thought that was the problem, just thought I would mention it.

What BIOS version is it? If you got hold of an older board, the BIOS may not be playing well with your CPU/RAM.

It wouldn't be much of a gamble to carefully flash it now... you are getting near the point of RMAing the board, and it's not like it's working atm.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
I've updated my bios from Version 0610 to the latest version (1406) a few days ago.

I saw somewhere that the older version had problems with quadcores, but that should have been fixed in the latest version.


What about the cpu? As I've mentioned, it ran far too hot several times before I noticed.
At first I thought that the restarts were due to the heat - sometimes during startup, after a restart, I would get the message that the CPU was too hot and that the computer couldn't restart, and if I quickly went into the bios, the CPU temps would be 90+C.

So, was this an entirely different problem? The restarts are the same as when the CPU was too warm, only now the temps are fine.

Could I have fried the CPU or maybe some censors?
 

Priv

Distinguished
Oct 20, 2008
135
0
18,680
Heat should not be a problem for the cpu, I do not think that is the problem. A cpu shuts down as soon as it gets a too high temperature. Now the temperatures are fine the heat should not be the cause. Also because the crashes only happen during playing games that are intensive for your computer it is most likely a stability issue. Have you tried running prime95 and memtest yet?
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
I did use prime95 a few days ago, but that was before I noticed the cooling problem, so it might be best to take it for a spin again.

I'll start experimenting a bit with them right now.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
Quick update:

I ran Prime95 for 3 hours last night before I turned it off, and there was no errors. I know that you're supposed to let it run for 12-24 hours, but this was all I had time for.

Afterwards I played a game for about 25 minutes before the computer restarted (just to check if there was any change; there wasn't).

Here's the odd thing: I turned off the computer after the restart, before it could start up Windows again, and when I turned it on this morning, it didn't start up - instead there was a message (still in DOS) from American Megatrends saying "Overclocking Failed".

Now, I haven't OC'd anything, but is it a possibility that the cpu (... or motherboard, GPU or something else) thinks it's overclocked and is causing the problems? Or could it somehow try to fine tune itself? - most of my settings in bios is set on AUTO...


As always, any advice would be appreciated.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
Ok, this is what has happened so far - sorry if it seems a bit odd; I've just copy/pasted from one of my posts from another forum...



I've gotten my hands on a Nvidia 9700 GTX, and that seems to run flawlessly (or at least much more stable).
I haven't had any restarts after 3-4 hours of gaming - with the 280, I've never been able to play for more than an hour.

This is what I've tried out... (of course I've fiddled with drivers, stresstests, memtests, bios, etc, but nothing helped...)
- GTX 280 in blue PCIe slot: It won't fit all the way in; computer restarts during demanding games
- GTX 280 in black PCIe slot: It fits all the way in; computer still restarts during demanding games
- Nvidia 9700 GTX in blue PCIe slot: It fits all the way in; Computer doesn't restart during demanding games (then again; the computer probably isn't stressed as much, since I have to run at slightly lower settings)


It seems to me, that there could be the following explanations:
* The motherboard is faulty and either won't accept my 280 GTX (whether it's plugged in the blue or black slot), or can't perform well under pressure (the 280 pushes it too far).
* The 280 GTX is faulty and causes restarts no matter what slot it's plugged into. This would explain a lot, but I've been using it for 3½ months without any problems. The problems didn't start until I upgraded.
* The PSU is faulty and can't deliver enough power. The 7900 GTX doesn't require as much power as the 280 GTX, and thus doesn't create any restarts. However my PSU is 750W, and I've only used it for about a year, so it should be enough. Unfortunately it's not easy to measure how much power it really delivers.


So, does anyone have any ideas on how to proceed? Should I just RMA all the parts..?
Also, I would like to know, whether or not my 280 GTX should be able to perform, if it's plugged into the other PCIe slot (the one "in the middle").
 
Hey I'm glad things are getting sorted out :)

On the restart and "OC failure" don't sweat that. It's just an oddity of the ASUS BIOS features.

You can look at your 12V power in speedfan or some such. Run some game in windowed mode with both cards. If the voltage droops significantly in comparison, with the GTX 280, you have your culprit. Not that the numbers in any software app are accurate, but you would have a comparison. I think a .2V droop is too much... perhaps others know better.

If not that, I would just RMA the video card at this point. Seems like it failed.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
Quick question:

Would it be possible to use another PSU exclusively to power the GTX 280?
That is; run my system normally, only instead of drawing power from my current PSU, the 280 would get its power from another PSU (hope it's clear enough...).

It's not meant as a permanent solution, but it would be pretty easy to set up, and, whether the restarts stop or not, would provide an answer to the not-enough-power theory.


Only question is, can it be done..?
 
Just plug in another PSU to the 6-pin connectors eh? I'm sure there is a way to do that. Not quite that simple though, as you have to trick the second PSU.

Forums over at jonnyguru would be the best place to ask.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
Well, I could borrow my friends PC, take some cables from his PSU to my GPU, while still leaving it connected to his motherboard. That way, I'll just leach a bit of power from him.

But if there's even the slightest risk of ruining either system, I'll just leave it be... I don't know enough about PSUs to risk anything.
 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530

Ok, now I've measured both cards 12V in SpeedFan while playing Fallout 3 and while running 3DMark06. I kept a log that updated every three seconds - I think that's accurate enough...

The GTX 7900 kept switching between 11,3V and 11,35V
The 280 GTX stayed stable at 11,3V

I really don't know what to make of these numbers, but it doesn't seem like there's any great drop..?

 

Whateley

Distinguished
Mar 29, 2006
35
0
18,530
Well, now I've tried measuring with PC Probe II:

Idle: 12,4V
Using the 280 while playing Fallout 3: 11,98V (this is in windowed mode - PC Probe can't make a log)