PC freezes constantly

[NeilManlius]

Hi there!

I've got a very big problem with my PC and I just don't know what to do about it. Sometimes, my PC comes to freeze and restarts with a blue screen or no blue screen at all appears and I have to push the reset button. Sometimes the screen becomes black or the screen becomes black, then the image reappears and the notice "Display driver stopped responding but has recovered" shows up, but mostly the PC freezes right after that.

These crashes are not reproducible although they seem to occur more likely when the system load is quite high or if the video card temperature rises above 160 °F.

My system specification:
CPU: AMD Athlon II x4 630 4x2.8 GHz
Mainboard: Gigabyte GA-M720-US3
RAM: 2x2048 MB Kingston ValueRAM DDR2-800
Video card: Point-of-View Nvidia GeForce 9500 GT (Driver version: 306.23)
Chipset: Nvidia nForce 720D (Driver version: 15.56)
HDD: Seagate ST3500418AS
Power supply unit: Tagan 500 W
OS: Windows XP Home SP3 32-bit & Windows 7 Professional SP1 32-bit

If a blue screen shows up, it mostly bears STOP code 0x00000116 and a message related to "nvlddmkm.sys" which, as I found out, is part of the Nvidia display driver.

Crashes occur when Win7 is running as well as when WinXP is running.

I assembled the system about 3 years ago and the crashes started to come up a few months later. I've also done numerous driver updates and reinstallations of Win7 but then problem continued to exist. Also a change of the power unit (from NoName 420 W to Tagan 500 W) did not improve the situation. First I though about a chipset driver error and updated the chipset driver. After that, the problem didn't show up again for at least a few months until the next crash came up.

I thought about a video card error, so I made a test and replaced the video card by my old ATI Radeon X550. After that, the PC only crashed once in two weeks but still in the same way as described.

Then I checked my RAM using Memtest86 v4.0 and Windows Memory Diagnostic. Both Memtest and WMD found some errors but when I ran Memtest again, no errors were found at all.

By the advice of some other hardware experts, I checked the voltage values of my PC and found out that the RAM voltage, which should be 1.8 V, was always at about 1.952 V. However most RAM manufacturers only guarantee correct function of their RAM modules within 1.8 V +/- 0.1 V, so my voltage is already 0.5 V above the maximum tolerance. Nevertheless, other people told me that occurred frequently with Gigabyte mainboards.

I also tried Linux Mint 13 x64 from a Live CD but everything worked fine.

So I find this situation quite paradox:
- The blue screen messages indicate a video card error but the problem persists(?) with another video card.
- The freezes imply a memory error but why does the system work fine under Linux Mint?
- That the system works fine when running a Live CD OS might imply a hard disc error but why does the blue screen come up with a video card driver file then?

Does anybody of you know what causes this problem or has anybody ever experienced something similar?

THX in advance.

See you,
NeilManlius

 

[TheGamerZ]

If memtest found errors at any point, reduce the ram to the specs on their sticker, return the whole system to stock and test the memory one stick at a time.

 

[NeilManlius]

Hi there!

Thanks for your answer.

reduce the ram to the specs on their sticker

I am not overclocking. I left all BIOS settings at their defaults. Is that necessary despite that?

Some other hardware guy told me take out the RAM modules and swap their places. After that, I ran Memtest for nine hours but didn't find any errors. However that didn't solve my problem.

test the memory one stick at a time

If I recall correctly, I've done that some time ago but it showed up very contradictory results. Sometimes it finds numerous errors, at times it find none...

return the whole system to stock

Could you explain to me what that means (English is not my native language)?

See you,
NeilManlius

 

[Dangman]

If I recall correctly, I've done that some time ago but it showed up very contradictory results. Sometimes it finds numerous errors, at times it find none...

Test the RAM seperately in another PC.

Could you explain to me what that means (English is not my native language)?

It basically means that if you are overclocking, turn off/disable the overclock and set the PC back to default speeds.

 

[Cheetoz]

I am not overclocking. I left all BIOS settings at their defaults. Is that necessary despite that?

sometimes, yes.

 

[XenIneX]

Do a clean install of ALL of the nVidia drivers: video, chipset, audio; everything.

 

[NeilManlius]

Test the RAM seperately in another PC.

Unfortunately, I don't have another PC with the same type of RAM, so I can't do that.
But maybe I could ask a friend for some different modules and see if the problem still occurs?

Do a clean install of ALL of the nVidia drivers: video, chipset, audio; everything.

I also thought about that. When I assembled the system, Win7 had been released only a few months before and there were no Win7 drivers available for my mainboard, so I used the Vista drivers instead. But the problem also exists under WinXP and the board was shipped with native WinXP drivers.

If make a clean install of all drivers, how long will that take? I'm currently working on my thesis and I really need the PC to be working.

But if it's only a driver problem, what about the errors found by Memtest?

See you and thx for your feedback,
NeilManlius

 

[XenIneX]

Unfortunately, I don't have another PC with the same type of RAM, so I can't do that.
But maybe I could ask a friend for some different modules and see if the problem still occurs?

Try pulling and reseating the memory. Poorly seated memory is not an uncommon cause of intermittent errors.

I also thought about that. When I assembled the system, Win7 had been released only a few months before and there were no Win7 drivers available for my mainboard, so I used the Vista drivers instead. But the problem also exists under WinXP and the board was shipped with native WinXP drivers.

If make a clean install of all drivers, how long will that take? I'm currently working on my thesis and I really need the PC to be working.

45 minutes, maybe? To my knowledge, Nvidia's current driver installers all have a "clean install" option, which does the job. Just uninstall everything Nvidia from the programs control panel, reboot, and run the latest driver installers from Nvidia.

Also, make sure you deselect the "Network Management Tools" in the chipset drivers package. I never had anything but crashes with those.

But if it's only a driver problem, what about the errors found by Memtest?

Didn't say it was the only issue, but I would expect memory errors to be a little more catastrophic than intermittent crashes attributed to a single driver.

 

[NeilManlius]

Try pulling and reseating the memory. Poorly seated memory is not an uncommon cause of intermittent errors.

I've already done that numerous times and the error always came up again.

Didn't say it was the only issue, but I would expect memory errors to be a little more catastrophic than intermittent crashes attributed to a single driver.

Which is indeed quite strange about that, especially if you take into account that the PC seems to be working correctly when running Linux (although I also remember similar problems with Linux Mint 12).

Could it be that my hard disk is also broken or at least not working correctly? (Because I ran Linux from a live-DVD, so no hard disk was involved.

But I've also experienced a crash with the old Radeon installed although just one within 14 days compared to crashes every few hours with the GeForce installed.

Might a new video card solve the problem?

Kind regards,
NeilManlius

 

[That_Sound_Guy]

I had nearly the same issue you have and it turned out to be over heating of the motherboards chipset. May want to check the HSF on the chipset bridges are intact. My machine would be fine until it experienced load, then it failed with the display driver not responding deal.

 

[NeilManlius]

Yes, I also thought about that because the temperature of the MCP is often above 140 °F whereas the CPU temperature is always at 120-125 °F. The GPU temperature sometimes rieses as high as 180 °F. However I read that Nvidia GPUs can take up to 180 °F.

Are 140 °F too high for chipset temperature and if yes, what can I do about it? Can I install a fan?

See you,
NeilManlius

 

[XenIneX]

Yes, I also thought about that because the temperature of the MCP is often above 140 °F whereas the CPU temperature is always at 120-125 °F. The GPU temperature sometimes rieses as high as 180 °F. However I read that Nvidia GPUs can take up to 180 °F.

Are 140 °F too high for chipset temperature and if yes, what can I do about it? Can I install a fan?

See you,
NeilManlius

Just grab a spare 80mm fan and mount it in front of the thing to see. If you don't have a mount, make one with some tin snips and roof flashing or pipe strap from your local hardware store. Doesn't have to be pretty; it's just for testing.

 

[sinclair-zx81]

If memory tests are failing, it really does not matter if it passes once, if it fails more than once and you have reseated the memory, I really don't see how it can be anything other than memory or motherboard. The memtest you are running.. is it from a boot CD so it runs without your OS? If it's DDR3, that is dirt cheap now... Always nice to have a spare pair of RAM sticks around anyway. If you want to try something else, maybe try undervolting the ram to bring it into spec, or just as a test, under clock the bus. It should normally run at 200MHz, try backing that off to 180 or something in BIOS and rerun the memtests a few times. It could be the RAM has been overvolted for long enough to cause breakdown, if it's that sensitive. Or, if it is breaking down, you may have to overvolt it a bit now to get a hair more life out of it, but don't do this without realizing it may put your system in the dirt permanently.

 

[NeilManlius]

is it from a boot CD so it runs without your OS?

Yes, I'm running it from the Ultimate Boot CD. Do you know it?

maybe try undervolting the ram to bring it into spec

Unfortunately I can't undervolt my RAM. There is no BIOS option to lower the voltage, just one to increase it but it's already too high at default settings.

if it fails more than once and you have reseated the memory

I've already done some memory testing some time ago and some errors were also found back then. But strangely Memtest also tests memory ranges above 4096 MB although I only have 4096 MB installed. Back then, errors were found in that area, so I went to the PC store but people told me it was a false positive.

Anyway, you say that if Memtest fails more than once (and WMD also did), the RAM definitely is down?

Or could it be that it's a mainboard error because the MCP is getting too hot?

THX in advance
NeilManlius

 

[NeilManlius]

Hi guys,

what do you advise me to do after all?

If I asked a friend to lene me some other RAM modules and my PC worked fine with them, would that be some kinda proof that my current RAM modules are broken?

Or does that fact that crashes occur frequently with the one but almost never with the other card that I should get me a new video card?

THX in advance.

See you,
NeilManlius

P.S. Please don'i be upset about my follow-up posting.

 

[Dangman]

If you're seeing errors at all in Memtest, then just RMA the RAM.

 

[NeilManlius]

Hi Danny!

then just RMA the RAM.

What does "RMA" mean?

See you,
NeilManlius

 

[Dangman]

Return merchandise authorization

In other words, contact the manufacturer, tell them you have issues, send the RAM back, wait 1-2 weeks, and you'll get new RAM from that manufacturer.

 

[NeilManlius]

send the RAM back, wait 1-2 weeks, and you'll get new RAM from that manufacturer.

I think that the warranty for my RAM has already expired as my PC is already more than 2 years old.

A "funny" thing I noticed is that with the old Radeon installed the image occasionally freezes but then my PC makes a little beep like from the internal speaker and I can continue to work. So it seems I'm also having some issues with the other graphics card.

All that makes me think that it's not simply the video card but also (or instead) the memory.

But why does my PC seem to work fine under Linux then?

Can you exclude a driver error or a hard disk problem?

Can anybody confirm my point of view?

Best regards,
NeilManlius

 

[Dangman]

I think that the warranty for my RAM has already expired as my PC is already more than 2 years old.

RAM has a lifetime warranty.

But why does my PC seem to work fine under Linux then?

Can you exclude a driver error or a hard disk problem?

It might be an OS issue. Did you try a fresh install of Windows? On the other hand, it could also be how the OSes handle hardware differently.

 

[NeilManlius]

RAM has a lifetime warranty.

Only in the U.S. or also in different countries?

It might be an OS issue. Did you try a fresh install of Windows?

I've installed Win7 numerous times but always using the same drivers. I've done a chipset driver update though. However the freezes don't only occur when running Win7 but also when running WinXP although less frequent. That was what made me think that it's unlikely to be a driver error.

However, could it also be a hard disk problem?

See you,
NeilManlius

P.S. THX very much for your help so far.

 

[larrymoencurly]

Some other hardware guy told me take out the RAM modules and swap their places. After that, I ran Memtest for nine hours but didn't find any errors. However that didn't solve my problem.

Even the best memory diagnostics can't find all errors, so try a couple more tests, including MemTest86, MemTest86+ (the two don't always give the same results, despite being based on the same test methods), and Gold Memory. Gold Memory was the only diagnostic that beat MemTest86 in RealWorldTech's very recent 12-year-old evaluation and didn't cost tons of money (i.e., PHD's RST programs)

In my limited experience, Kingston seems to have been pretty reliable since DDR2 first became popular, but I doubt they test to the same standards as Micron or Samsung. If you contact Kingston, try to speak with higher level tech support and be kind of technical when discussing your results because that seems to cause Kingston to provide better quality replacement memory. I'm not saying you'll get their higher cost memory but the same model, only screened more thoroughly.

 

[Dangman]

Only in the U.S. or also in different countries?

Depends on the country, But U.S, Canada, and UK gets lifetime warranty AFAIK

However, could it also be a hard disk problem?

Yes.

 

[NeilManlius]

I've just done a hard disk test with SeaTools v1.10 and it passed both the long and the short test. So it's rather unlikely that it's a hard disk error, right?

May it be that the errors occur because I'm using the 32-bit versions of WinXP and 7? Should I rather try to use the 64-bit versions or is it rather obvious that it's a hardware error? What should I replace first? The memory or the video card? I'm planning to do a complete reinstall anyway but before that I'd like to know if there's a "perspective" for my PC at all.

See you,
NeilManlius

 

[Dangman]

I've just done a hard disk test with SeaTools v1.10 and it passed both the long and the short test. So it's rather unlikely that it's a hard disk error, right?

The chances of it being the HDD is now lower but still there. Even those tools can sometimes be a tad inaccurate.

May it be that the errors occur because I'm using the 32-bit versions of WinXP and 7? Should I rather try to use the 64-bit versions or is it rather obvious that it's a hardware error?

Did you actually install Linux on that hard drive yet? A Linux LiveCD doesn't touch the hard drive at all. No, it's not related to 32bit or 64bit at all.

What should I replace first? The memory or the video card?

GPU.

I'm planning to do a complete reinstall anyway but before that I'd like to know if there's a "perspective" for my PC at all.

What do you mean by "perspective"?

 

[XenIneX]

Should I rather try to use the 64-bit versions or is it rather obvious that it's a hardware error? [...] I'm planning to do a complete reinstall anyway but before that I'd like to know if there's a "perspective" for my PC at all.

If you're going to be doing a clean install anyway, and have access to the install media, you should switch to 64-bit regardless. On 32-bit you're losing ~20% of your memory to addressing. And, if you decide to upgrade your RAM in the future, this will clear away future roadblocks.

 

[NeilManlius]

Hi there!

Did you actually install Linux on that hard drive yet? A Linux LiveCD doesn't touch the hard drive at all. No, it's not related to 32bit or 64bit at all.

No, I only ran it from a DVD.

Originally Posted by NeilManlius View Post
What should I replace first? The memory or the video card?
GPU.

So you're of the opinion that it's rather a video card problem than a memory problem? (I'm asking that because I also experienced some graphics-related trouble with the old Radeon installed, see above.)

What do you mean by "perspective"?

I mean that I only want to invest anything in my PC when I can see a chance to get my issue fixed at reasonable cost. If I change my video card and memory and make a complete reinstall of Windows right now but my PC comes to crash again nevertheless, I'll have to get me a new mainboard, CPU and memory, so in fact I'll have to get me a new PC. That's exactly the reason why I want to figure out my problem as specific as possible. Meanwhile I'm also busy doing my Master thesis so there isn't really very much time left for me. Please don't be upset about that but I just want to tell you my reason for asking so many questions.

But thanks to you all for your help so far.

Kind regards,
NeilManlius

Edit: I've just done another test with SeaTools v2.23 (from the Ultimate Boot CD) and it also said "Test passed." Also tried Drive Fitness Test but I didn't know if it works with my hard disk, so I cancelled the test.

One thing I have to add is that I once had the graphics card exchanged for some trouble with the fan. I got the same model in exchange but another revision (old one had S-Video output, current one has HDMI). However, IIRC the errors also occurred before the exchange. Does that imply that it can't be the graphics card?

 

[Dangman]

I mean that I only want to invest anything in my PC when I can see a chance to get my issue fixed at reasonable cost. If I change my video card and memory and make a complete reinstall of Windows right now but my PC comes to crash again nevertheless, I'll have to get me a new mainboard, CPU and memory, so in fact I'll have to get me a new PC. That's exactly the reason why I want to figure out my problem as specific as possible. Meanwhile I'm also busy doing my Master thesis so there isn't really very much time left for me. Please don't be upset about that but I just want to tell you my reason for asking so many questions.

Honestly dude, just get a new PC. That system is just too old to put money into. Simply put:
- That PSU is more than likely crap. May not be the direct cause of your issues but still a candidate. Or that your previous no-name PSU caused damage to your PC over time.
- That motherboard uses a Nvidia chipset. As such, it wasn't all that good or that reliable in the first place. If that mobo is found to be defective, you'd basically have to start with a new PC anyway considering the costs of AM2 motherboards.
- Your RAM is DDR2. As such, if you can't RMA that RAM, then you would have to buy new DDR2 RAM. Considering the high prices for DDR2 RAM, at that point it's simply more cost-effective to just build a new PC altogether.

Now if you still want to figure out your exact problem, the fastest way is to simply start swapping in parts. The problem is that your PC is exhibiting issues that are related to the several parts at the same time. The Nvidia related BSODs could be due to the motherboard or GPU. Again, remember that your mobo uses Nvidia drivers as well. The crappy PSU you have may not actually be able to handle the power requirements of the PC hence the BSODs. The Memtest errors also points to a RAM or mobo issue. The fact that you didn't install Linux to the HDD or didn't finish running DFT doesn't rule out the HDD just yet.

 

[NeilManlius]

Hi there!

Thanks for your honest advise. I've decided to make one last trial an get me a new graphics card. I don't need that much power, just some state-of-the-art technology. After all, a new graphics card is still a little cheaper than a new PC and if it doesn't work anyway, I'm gonna get me a new PC and sell the graphics card on ebay or give it to a friend of mine. I'm thinking about that because my PC crashes every few hours when the GeForce is installed but all in all works fine with the Radeon. I don't know if that will fix the issue but at least I'll give it a try.

Anyway, again many thanks for your help. Now I think I've made up my mind.

Best regards,
NeilManlius

P.S. Why do you call my PSU crap?

 

[Dangman]

It's a Tagan 500W. Considering how much the rest of the PC costs, I'm assuming that you didn't spend much on the PSU. As such, that's probably the low-end Tagan.

 

[NeilManlius]

Okay, but even if it is low-end, it should be working correctly.

 

[Dangman]

Okay, but even if it is low-end, it should be working correctly.

Not necessarily.

 

[NeilManlius]

Okay, please don't mind it, but I once had a PC that I'd built myself which was made up of a no-name PSU, no-name mainboard and most probably no-name memory but it always worked fine and never made any problems. In my opinion, components of an established brand don't necessarily work better than no-name ones even though problems may be more unlikely with them. That's at least my point of view.

See you,
NeilManlius

 

[Dangman]

Either way, you still bought a low-end PSU that is more than likely not capable of powering your hardware long-term. When it comes to power supplies, you really shouldn't take any chances with them considering that they can and do cause hardware death. I've seen that plenty of times in my years with PC hardware.

 

[NeilManlius]

Just got me a new graphics card today and when I tried to install the driver, the PC froze again. After that the PC froze at every reboot. Then I tried to reseat the RAM and CPU. When I took out the CPU fan, I noticed the CPU was glued to it. Then I tried to press it back into the socket which didn't work out for sure. A friend of mine told me to loosen the CPU from the fan using a hairdryer. However, some of the pins were crooked and it was very hard work to bend them back into position. You can believe it or not but I manged to get the PC working again as I'm writing this posting on it.

I tried to reinstall the display driver of the new card but every time at startup the PC came to freeze again. Consequently, I put the old Radeon in back again and made a system restore.

Honestly, what can be the cause for that?
(It's not that I'm sticking to this only because I don't want to buy a new PC, it's also because I'm seeing it as a kind of sport now, to fix this error.)

- A hard disk error is rather unlikely because of the connection to the display driver installation.
- Also it can't be the graphics card alone as I've tried a new card which doesn't work either.
- For me it seems that it's either a mainboard error (MCP possibly broken) or a simple driver error (as my PC is working fine under Linux).

Best wishes,
NeilManlius

 

[Dangman]

What was the new GPU?

EDIT: your assessment is still off since you have not installed Linux on the hard drive. A Linux LiveCD only loads info into the RAM. It barely touches the hard drive, if at all. So the hard drive is still a possibility. With that said, my gut says mobo or PSU. Either way, time for a new PC.

 

[NeilManlius]

Hi there!

Either way, time for a new PC.

Yeah, I think so, too. But as long as my old PC is still working, I'll use it until I have finished my thesis. After that, I'll build me a new PC which will be a lot better than my current one.

See you,
NeilManlius

BTW: It's a Radeon HD 6450.