Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I recently upgraded the core components of a computer I built a few years ago. I replaced the motherboard, memory, and CPU and have been running into issues since

The problem I started running into (after my upgrade) is that after some period of time (a half hour to an hour) the computer will shut down. There are no warning messages or flags, it just abruptly shuts itself off. While it has happened under "normal" use (web browser, a few applications), it is almost guaranteed to happen during gaming.

1) I do not believe it to be a heat issue. I initially had problems with heat but recently re-installed the CPU heat sink using Arctic Silver 5, and have been obsessively monitoring heat since. Similarly, the GPU heat has also been fine. I've run Intel Burn Test and Furmark concurrently and neither CPU nor GPU exceed 70° C.

2) I do not believe it to be a PSU issue. I only know this because in an attempt to resolve this issue I just installed a brand new PSU that should have plenty of headroom for my system (600W, 40A single 12V rail). So the shut-down issue has happened with both power supplies.

3) I do not believe the memory is an issue - as I write this my computer is running Memtest86+. When I left for work it had been running for 9 hours and had completed 4 passes with no errors. I decided to let it go another 9 or 10 hours to get through a full 8 passes just in case, but I would be surprised if it finds any at this point. It's also worth pointing out that I checked in BIOS and the memory is currently running at its rated speed of 1600.

My question is: what's left? The RAM and GPU seem to be properly seated, Windows is up to date. What's left to check?

My specs are as follows, note that nothing is overclocked:

share|improve this question
    
Did you check Event Viewer? Sometimes hardware failures are reported there before an actual shutdown/BSOD/etc occurs. –  nevets1219 May 26 '11 at 17:42
    
@Michael Are you connected to a known good UPS? –  P.Brian.Mackey May 26 '11 at 18:10
    
I have not checked Event Viewer, good suggestion. Also Mr. Mackey: I am connected to a generic power strip. –  Michael May 26 '11 at 18:12
    
@Michael - I would get a UPS of sufficient VA rating to exclude any power issues. Especially considering you have already thoroughly tested for common problems and still experience intermittent issues. Also, plugin just the PSU and dont do any heavy load stuff (games, video editing etc) and let it run for a while. Dont plugin the monitor and stuff on the UPS. If the problem is gone then you have a power problem. –  P.Brian.Mackey May 26 '11 at 18:14
1  
@Mackey: The symptoms are of buckling, heat damage/aging, or component failure. Michael is using a power bar, and it would be very evident if the power bar were unable to provide sufficient power to the computer. Before power management kicks in, components generally take full power at system start, a power bar issue would be evident immediately. And at $100 and up for a decent UPS, his money really ought to be spent elsewhere to more likely causes of the problem prior to checking the UPS. –  music2myear May 27 '11 at 13:57

5 Answers 5

up vote 5 down vote accepted

In this case, I would follow some basic troubleshooting guidelines.

Step 1) Determine a reproducible process for testing the issue. You need a set of steps to take that will make the system crash each time you do it. The best process is one where you can walk away and do something else while it is running. the best processes don't require anything more than windows, but that can sometimes be a challenge. You will be doing this process constantly until you determine what it wrong. Before you move on, make sure that you can perform the test on a known working machine. You don't want your test to crash all computers, just the one having problems.

Step 2) Reduce the number of variables down to the smallest set. This is where you start removing hardware until you have a bare bones system. You can start with reducing down to 1 stick of ram, removing extra peripherals, and disconnecting hard drives. This is also a good time to reduce the number of running programs down to the fewest possible.

Step 3) Eliminate the OS/Programs from consideration. Get a known good blank hard drive and reinstall windows onto it. Add the latest manufacture drivers and any software needed to test for your error.

Step 4) Start swapping components with known good ones. This is the most time intensive and difficult step to complete. Find compatible hardware and start changing things out. I'd go in this order:

  • Video Card (sometimes good to go with a different chip-set series)
  • Memory
  • CPU
  • Motherboard
  • Power supply
  • Case

If you can't figure out what is wrong after doing all of the above, then you may just need to trash the whole system and build a new one from scratch. What you currently have should work together as long everything is configured correctly.

Hope this helps

share|improve this answer
4  
Remember this question? Yup, it's several months later and had I followed this advice more closely I would have figured this out a looong time ago. The biggest lesson learned is that memtest is NOT the only test for RAM. When I finally broke down and started testing the physical sticks of RAM, I found the issue. The computer is completely stable with 1 or 2 sticks, but once 4 get put in it crashes. I tested this quite thoroughly. I believe this is due to undervolting, so I am going to try and adjust that in my BIOS. –  Michael Oct 7 '11 at 12:57
    
Memtest is an interesting tool that only checks to make sure that the sectors on your ram sticks are good. This test does not put a heavy load on the RAM so you won't catch performance issues with this. I think you are on the right track. This sounds like a motherboard bug. Overvolting the RAM might solve the problem, but you may also fry your memory. If you still have problems, it may take a new/different motherboard to solve your problem. –  Doltknuckle Oct 12 '11 at 18:21

One possibility is a short-circuit between the motherboard and the case. Check the clearance between the motherboard and ensure that there are no contact points between the motherboard and the case except for the mounting points. You might try getting some spacers to raise it slightly higher just as a test.

Another possibility (though a slight one) might be a "brown out". If the circuit the computer is running off of doesn't supply enough power for the computer, resulting in an under-power condition, it can cause the PSU to shut off because it is not getting enough power to run the computer.

share|improve this answer
1  
I've seen a short-circuit issue before and with that it doesn't start at all. Since then i've always run the motherboard outside of the case while i'm building it, and that's useful generally too 'cos things are more accessible. –  barlop May 26 '11 at 17:57
    
My UPS has a real time voltage display. Buying one of these would help the OP with the brown out testing: officemax.com/catalog/…^freeText~ups^paramValue~true^refine~1^region~1^param~return_skus^retu‌​rn_skus~Y –  P.Brian.Mackey May 26 '11 at 17:58
    
@barlop - I think that would depend on how much contact the case has with the motherboard. It could be that a very slight shift in the case (possibly caused by a change in temperature) causes enough contact to short circuit the board at times. –  P.Brian.Mackey May 26 '11 at 18:05
    
Regarding power, I am connected to a generic power strip, which is also powering my monitor, speakers, audio interface, and an external hard drive. I did not run into any issues before upgrading, but perhaps my new and more power-hungry system is too much for that setup? –  Michael May 26 '11 at 18:13
1  
You may be able to use your BIOS setup to discount or prove that the power is an issue. There is a setting in your BIOS in the Advanced Menu under APM called "Restore AC Power Loss". If it is selected to [Power On] then it is unlikely that PSU or input power is the cause of the problem because your PC would try to power back on after each power down. If it is selected to [Power Off] change it to [Power On] and see what happens when your PC fails. –  Tog May 27 '11 at 14:28

Memtest is best left to run overnight as a bare minimum, at work, I usually let it run over the weekend, make sure you have the case closed for the tests.
Check the memory voltage settings in BIOS match the memory specs.
I see the memory has heatsinks attached, check they are secure
Make sure you have good airflow across the memory, due to it's location on a some boards it is not uncommon for cables to impede the airflow over it.

share|improve this answer
    
In my experience Memtest usually fails during the first run with bad Mem. He said he doesn't go over 70 degrees C, that shouldn't be enough to trip the heat failsafe (but it is high for an intel, mine runs about 40 on load). He did check the voltage settings. A poorly installed CPU fan would probably go over 70 degrees and I doubt the machine could run more than ~20 minutes. –  P.Brian.Mackey May 26 '11 at 18:21
    
I checked the RAM speed (1600) but did not check the voltages, I will do so when I get home. –  Michael May 26 '11 at 18:30
1  
@P.Brian I have to disagree on the memtest thing. I have had quite a few PCs in a customer's facility that gave several good passes before finally failing, usually fitted with "performance" memory that is running at the spec'd limit. These sticks consistently failed when fitted to a known good system for final test before filing under b1n. –  Tog May 26 '11 at 18:57
    
@P.Brian In regard to your point about temperature, you are talking about CPU/GPU temps, I am talking about Memory temps. Also, as stated by the OP, he has not yet checked the memory voltages. –  Tog May 26 '11 at 19:15
    
After 19 hours of running, there were still no errors in memtest86+, and that constituted 8 full passes on all 16 GB. Checking in BIOS my memory is indeed running at the manufacturer's spec of 1.5V as well as the right timing of 9-9-9-24. –  Michael May 26 '11 at 23:18

Have you tried to update the BIOS on your motherboard? I had the same shutting down issue in my self-built rig and updating the BIOS did the trick.

share|improve this answer
    
Thanks for the suggestion, but it looks like I have the latest BIOS version (1503). –  Michael May 26 '11 at 23:34

When you were upgrading the core components, you could could have flexed the motherboard too much causing shorts inside the motherboard itself (not with the case). I have had the same exact issue with a laptop at the company I work at. Somebody like putting the laptop in a backpack, then they would sling the backpack around, causing the Lenovo T500 laptop to flex a lot, flexing the motherboard. Lenovo sent a tech to replace the motherboard for free (HA HA HA!).

If you are wanting to make sure it isn't incoming power, you could get a nice UPS (Uninterrupted Power Supply) with the digital reading on it that will give you the information you need. You probably need one anyways in case of power outages so that you computer does shutdown incorrectly.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.