System lock up (no BSOD)

[theimmortalone]

Dear gentlemen,

please assist me in determining the source of my problem.

The problem: computer locks up (freezes), stops responding completely. No BSOD, no black screen (just the still image of the last frame before the lock up). No buttons work, only power down helps.

When can the problem be reproduced: The PC was build in December 2011, but the problem first appeared after a week of playing Diablo 3 (May 2012). However, the problem can be reproduced by: 1. Running 3dmark 11 (sometimes it completes successfully, but generally locks up at Combined test), 2. Running Furmark 15-minute stress test (always a lock up after about 15% complete, short Furmark test is successful), 3. Playing Diablo 3 for about 15-20 minutes.

What I already checked:

1. Memtest left overnight, no problems detected.
2. Checked temperatures for overheating: in Furmark the lockup occurs at 65 degrees Celsius on one GPU and 58 on another (overheating seems unlikely).
3. All peripherals are at default values: no overclocking ever attempted.

My system:

Case: Silverstone FT-02
Motherboard: Gigabyte G1 Sniper 2
CPU: Intel I7-2700k
Memory: 4x4 sticks of Samsung 10600
GPU: Nvidia GTX 590
Audio: onboard Creative chip
Drives: SSD Intel 160 GB (OS drive and all programs are installed here) and WD 1.5 TB (solely as storage, no programs installed on it)
PSU: Chieftec APS-1000C
OS: Windows 7 64 bit

CPU and GPU are liquid cooled via single coolance waterloop and 3x section radiator placed on top of the 3xair penetrator fans of the Silverstone FT-02. Since the PSU is located at the top of the case where hot air is rising, the intake of the PSU is obviously facing outside the case (to get cool air) and then the air comes out from top of the PSU.

Overheating should not be the issue, since (despite single loop and not liquid cooling-friendly case) idle graphics card temps are 38/35 degrees Celsius, and around 65/58 degrees Celsius at max load.

I am puzzled and thinking about replacing the PSU. Or maybe it is some problem with SSD?

Thank you for reading this. Your help will be truly appreciated!

Anton

 

[mubin]

So you cant boot at all for now? Do you hear any post beep code? Is is freeze in post screen or after boot?

 

[Raiddinn]

Try to borrow a PSU and try it in your computer.

Straight up freezes during gaming are usually because the video card is trying to suck down more power than the PSU can give.

 

[theimmortalone]

So you cant boot at all for now? Do you hear any post beep code? Is is freeze in post screen or after boot?

Kindly thank you for the response.

I apologize for not providing enough details at first.

I can boot every time, Windows loads every time as well and the system is stable and I can play other games/watch videos unless: (1) I run Diablo 3 for 15-20 mins (system freezes in-game), or (2) I run 3dmark 11 or Furmark (3dmark - freeze before Combined test, Furmark - freeze after about of 1 min of stress test). Maximum temperatures of the GPUs right before the system locks up are 65 degrees Celsius.

Thank you for your time!

 

[theimmortalone]

Try to borrow a PSU and try it in your computer.

Straight up freezes during gaming are usually because the video card is trying to suck down more power than the PSU can give.

Thank you for the comment!

I will try to borrow (or get a new PSU straight away), probably Thermaltake TPX-1475, and see what happens.

On a side note, I would like to clarify (as it is probably not evident from my original post, sorry for that) that the system boots, I can login into Windows, watch videos, play other games, but the lock-up only manifests itself when I do one of the 3 things I posted above (D3 for 15-20 mins, 3dmark before combined test and Furmark after 1 min of stress test). Again, temperatures seem fine at full load (65 degrees on the hottest of gpus).

Thank you for looking into the issue!

 

[Raiddinn]

I would not get a Thermaltake as a replacement, its generally a pretty bad brand.

I would suggest you get an XFX instead if you are getting a new one.

That being said. restarts or lockups during gaming and stress testing usually mean that the amount your PC is using went 1w past the amount your PSU is capable of delivering.

Often it has to do with not having good enough cooling, because high temperatures inside the PSU (nothing monitors this) reduce the amount of power a PSU can deliver. Every 1c the temperature is higher in it directly reduces the power by a %.

The more power you pull in relation to how much the PSU can deliver the more inefficient the PSU runs. The more inefficient it runs the more power it pulls from the wall to deliver the same to the parts. The difference between what is pulled from the wall and what makes it to the parts shows up as yet more heat inside the PSU, so it has the potential to be a vicious cycle.

The best way to counter this effect is to have a PSU that stays efficient and works well at high temperatures and to have a case with a bottom mount PSU rack.

- Edit - Clarity

 

[theimmortalone]

I would not get a Thermaltake as a replacement, its generally a pretty bad brand.

I would suggest you get an XFX instead if you are getting a new one.

That being said. restarts or lockups during gaming and stress testing usually mean that the amount your PC is using went 1w past the amount your PSU is capable of delivering.

Often it has to do with not having good enough cooling, because high temperatures inside the PSU (nothing monitors this) reduce the amount of power a PSU can deliver. Every 1c the temperature is higher in it directly reduces the power by a %.

The more power you pull in relation to how much the PSU can deliver the more inefficient the PSU runs. The more inefficient it runs the more power it pulls from the wall to deliver the same to the parts. The difference between what is pulled from the wall and what makes it to the parts shows up as yet more heat inside the PSU, so it has the potential to be a vicious cycle.

The best way to counter this effect is to have a PSU that stays efficient and works well at high temperatures and to have a case with a bottom mount PSU rack.

- Edit - Clarity

Thank you once again, Raiddinn!

I guess the PSU is indeed struggling in my case, with all the heat coming up.

I've arranged for a new Enermax Platimax 1200 to be delivered tomorrow, which, according to the reviews I've seen, does not suffer from continuous load under above average temps. Do you reckon this would be an acceptable replacement for my current PSU?

Thank you for your time.

 

[Raiddinn]

It sounds like massive overkill, but it should definitely work.

The full system with a 590 uses like 500w at load, so a quality 850 like an XFX unit would have probably been more advisable.

Ideally you want to aim the regular use to be about 40-50% load and the max load to be about 66% of the PSUs rated wattage.

That is if you are getting a serious brand like XFX, Seasonic, or Enermax.

Brands like Thermaltake are misleading about their wattages so you can't trust what they say. Getting one of those you want at least double the real wattage you will use, if not triple.

Even then, the lower wattage real PSU is better than the higher wattage generic one that is better used as a paperweight. At least the good brand is highly unlikely to damage your parts.

The poor brands tend to have very bad internal operations and leave your computer wide open for hardware damage due to bad power delivery (dirty power).

- Edit - Clarity

 

[theimmortalone]

It sounds like massive overkill, but it should definitely work.

The full system with a 590 uses like 500w at load, so a quality 850 like an XFX unit would have probably been more advisable.

Ideally you want to aim the regular use to be about 40-50% load and the max load to be about 66% of the PSUs rated wattage.

That is if you are getting a serious brand like XFX, Seasonic, or Enermax.

Brands like Thermaltake are misleading about their wattages so you can't trust what they say. Getting one of those you want at least double the real wattage you will use, if not triple.

Even then, the lower wattage real PSU is better than the higher wattage generic one that is better used as a paperweight. At least the good brand is highly unlikely to damage your parts.

The poor brands tend to have very bad internal operations and leave your computer wide open for hardware damage due to bad power delivery (dirty power).

- Edit - Clarity

Thanks again for your expertise.

I will (hopefully) test the new PSU tomorrow and reply with the results. Cheers!

 

[theimmortalone]

Today I got the new PSU, Enermax Platimax 1200W. Such a beauty! All hooked up, and after the first launch I managed a full 3dmark 11 test and complete 15 minutes of Furmark stress test.

However, after the end of Furmark stress test (100%), when switching back to window mode, the PC froze (as before). After reset the problem came back. I tried rolling back GPU drivers (from the newest back to and including 285.62). But nonetheless, I am unable to successfully finish a 3dmark 11 test or a Furmark 15-minute stress test (a short Furmark test is fine).

As for the temperatures, by the end of the first Furmark 15 min test, 1 of the GPUs showed 69, another showed 62 degrees Celsius. No hot air seems to be coming from the PSU.

For additional info, Intel HD 3000 are disabled in BIOS, so no conflict with the GPUs should take place.

I would be really grateful for any additional advice.

Thank you for your time.

 

[Raiddinn]

Take everything out of the case and put it on a wooden table then try to run your stress test on it.

 

[mubin]

Do you hear any beep code? Try reseat ram and gpu again.

 

[Raiddinn]

If he can run stress tests, he probably doesn't hear beep codes. Just sayin.

 

[mubin]

Just want to sure if there is any hardware issue. Or else, maybe driver causing hang.

 

[theimmortalone]

Do you hear any beep code? Try reseat ram and gpu again.

Thank you for looking into this thread, mubin!

First of all, since the problem was getting out of hand, bofore dismantling the system (which is a pain with all the tubing (since it's liquid cooled), I checked the SSD by doing a clean install on my HDD, and the problem repeated itself.

After that I formatted the SSD and did a clean Windows 7 install. After updating all drivers (motherboard, graphics card, etc) I took out the graphics card. After reinstalling the Device Manager was happy to show me only 1 GPU (instead of 2x GPUs, as it should when displaying a GTX 590). I started playing with RAM memory modules, replacing them 1 by 1, loading the system with only 1 and 2 modules. Still one GPU was displayed. I ran Furmark, and it completed successfully, though with just one

I thought that one GPU might be dead, but still the other one seemed operational. So I loaded all 4 x 4GB modules back, and Windows is now showing 2xGPU's again. Ran Furmark, and it crashed again.

Then I removed 2x RAM modules, and Windows shows 1 GPU, but Furmark ends successfully.

I apologize for this poorly structured wall of text, but I thought it was important to provide the information "as is", without omitting anything.

Does the above point to a dead GPU or problems with Motherboard?

Thank you for your time.

 

[Raiddinn]

Download a program callled MemTest86+, use an image copying program to deconstruct the ISO into regular files and copy them to the CD, then run it on each of the sticks of RAM individually.

You will have to take 3 out and then rotate the one that is in.

If something is messed up with your RAM, that will usually find it.

Ideally, you will want to go for at least 8 passes. I have seen errors on the 7th pass before. There is a chance an error would show even later than that, but I personally haven't seen it.

 

[theimmortalone]

Download a program callled MemTest86+, use an image copying program to deconstruct the ISO into regular files and copy them to the CD, then run it on each of the sticks of RAM individually.

You will have to take 3 out and then rotate the one that is in.

If something is messed up with your RAM, that will usually find it.

Ideally, you will want to go for at least 8 passes. I have seen errors on the 7th pass before. There is a chance an error would show even later than that, but I personally haven't seen it.

Raiddinn, thank you for your continuous assistance.

I ran memtest 4 times, on each of the RAM sticks separately (that took a while!). 8 passes and 0 errors every time. So I guess it's reasonable to assume the problem lies elsewhere.

Does this point to the GPU or motherboard?

Thank you for your time.

 

[Raiddinn]

It is good to have a better idea where not to look, although we should be careful not to take the RAM test results as gospel. For now I would call RAM temporarily off the table, though.

So, if I get this right... The computer will crash any time you try to stress test with the two 590 video card chips installed. If you disable one of them it will not crash, but if not then it will always crash.

At this point, I am going to have to assume that it is a bad PSU. If you do get a new one, I would suggest trying to get one from Enermax instead if they are available. If not and you still want to get a PSU, List the brands you can get and I will advise between those.

 

[theimmortalone]

It is good to have a better idea where not to look, although we should be careful not to take the RAM test results as gospel. For now I would call RAM temporarily off the table, though.

So, if I get this right... The computer will crash any time you try to stress test with the two 590 video card chips installed. If you disable one of them it will not crash, but if not then it will always crash.

At this point, I am going to have to assume that it is a bad PSU. If you do get a new one, I would suggest trying to get one from Enermax instead if they are available. If not and you still want to get a PSU, List the brands you can get and I will advise between those.

Raiddinn, thank you for the comment!

If you would kindly look a few messages back, you will see that I have already purchased and installed the Enermax Platimax 1200W. And, unfortunately, the problem persists with the new PSU as well.

Perhaps it is an indication that the problem lies elsewhere?

Thank you for your time.

 

[Raiddinn]

Sorry, I am kinda busy with stuff and didn't have time to look backwards in the thread. Just trying to give you something quickly. I am getting ready to move in RL.

I can't see the problem being the motherboard, because the test completes with 1 video card enabled just fine (IIRC).

Maybe the video card is damaged somehow, or the drivers aren't working for it. You could try a different driver.

 

[theimmortalone]

Sorry, I am kinda busy with stuff and didn't have time to look backwards in the thread. Just trying to give you something quickly. I am getting ready to move in RL.

I can't see the problem being the motherboard, because the test completes with 1 video card enabled just fine (IIRC).

Maybe the video card is damaged somehow, or the drivers aren't working for it. You could try a different driver.

Raiddinn, many thanks for all the help you have provided! Thanks to you we have managed to narrow the search to just a couple of devices. I shall see whether I can get a replacement GTX 590.

Thank you very much for spending your time looking at this thread, even though you have your hands full already. Cheers!

 

[Raiddinn]

Let me know if anything new develops, I might still be able to chip in on it.