Sign in with
Sign up | Sign in
Your question

Computer Loosing Power randomly

Last response: in Components
Share
February 5, 2013 6:42:31 PM

Hello, I Built a computer about a year ago, specs at www.kingzcraft.net/ss/ . This has happened a few times over the last few months, but twice in the last two days. What happened is the computer starts freezing, then it will just drop power all together and then restart.

Most recent: What just happened is I was copying a file from a archive (win rar) then it froze, Windows became unresponsive, CPU load went down alot, then the sound skiped and powered down.
a c 1176 ) Power supply
February 5, 2013 7:16:52 PM

Have you tested your UPS' ability to properly switch over to battery operation with your Corsair GS800 power supply during a power interruption event?

X-bit labs tested the GS800 with their APC SmartUPS SC 620 that outputs a stepped approximation to a sinewave during battery operation and found that the UPS wouldn't switch over to battery operation even when the PSU's power draw was well within the UPS' load capability.
m
0
l
February 5, 2013 7:57:28 PM

I don't think it has anything to do with the UPS, its never switched to battery.
m
0
l
Related resources
February 5, 2013 10:58:43 PM

Could it just be a bad PSU?
m
0
l
February 5, 2013 11:08:31 PM

Also I think I should add that its on 24/7 running TeamSpeak3 server, EasyPHP Apache, MySQL, 3 Minecraft servers, and a bunch of other random
m
0
l
February 5, 2013 11:31:23 PM

Does Windows display a message after logging back in saying it recovered from an error? If so, you're getting blue screens but you have the 'Automatically restart on system failure' setting enabled preventing you from seeing the blue screens. If it's always a restart though and no error message when logging back in, could be the hard drive.

Or like you said, could still be the PSU. But testing it would be difficult without another PSU to try for a period of time. I just had to RMA my PSU, but mine was easy to diagnose because it only failed when trying to play newer games or running furmark and I could replicate the problem consistently without fail. And I had another lower wattage PSU that would work perfectly.
m
0
l
February 6, 2013 1:43:49 AM

arson94 said:
Does Windows display a message after logging back in saying it recovered from an error? If so, you're getting blue screens but you have the 'Automatically restart on system failure' setting enabled preventing you from seeing the blue screens. If it's always a restart though and no error message when logging back in, could be the hard drive.

Or like you said, could still be the PSU. But testing it would be difficult without another PSU to try for a period of time. I just had to RMA my PSU, but mine was easy to diagnose because it only failed when trying to play newer games or running furmark and I could replicate the problem consistently without fail. And I had another lower wattage PSU that would work perfectly.



No blue screen, or error. Just shuts off and restarts. Just tried looking for the MEMORY.DMP and there is none. Does that mean its not blue screening?
m
0
l
February 6, 2013 2:32:44 AM

If I remember correctly, if you set Windows to not use a pagefile or a really small pagefile, it can prevent Windows from being able to create a memory dump. But, if you didn't do that, then by default Windows uses a large enough pagefile to create memory dumps and if no memory dumps exist, then you probably aren't experiencing blue screens. Either way though, it should still be recorded in Event Viewer under Windows Logs. The event would either be in Application or System with an event ID of 1001 I think.

My PSU is still in RMA so I'm being forced to use my MacBook Pro from work and I can't verify anything for you completely. If you find no events in event viewer though, then I'd check the hard drive. Do you hear any loud clicks or similar sounds when these shutdowns happen? Anything like a metal hitting metal type of sound? Run an error check on your OS drive. Right click my computer > properties > tools tab > click 'check now' under error-checking. Check both check boxes under 'check disk options' and click OK. Then it'll tell you that a reboot is required. Upon reboot, it'll run the chkdsk process and it'll take a couple of hours probably, the larger the hard drive the longer it takes.

Then we can check event viewer for the log of the chkdsk results and see if any bad sectors were found or not. Windows will automatically reboot when it's done so it'll be waiting at the logon screen when you return. If you already knew how to do all of this, then my bad lol.

m
0
l
February 6, 2013 3:01:35 AM

arson94 said:
If I remember correctly, if you set Windows to not use a pagefile or a really small pagefile, it can prevent Windows from being able to create a memory dump. But, if you didn't do that, then by default Windows uses a large enough pagefile to create memory dumps and if no memory dumps exist, then you probably aren't experiencing blue screens. Either way though, it should still be recorded in Event Viewer under Windows Logs. The event would either be in Application or System with an event ID of 1001 I think.

My PSU is still in RMA so I'm being forced to use my MacBook Pro from work and I can't verify anything for you completely. If you find no events in event viewer though, then I'd check the hard drive. Do you hear any loud clicks or similar sounds when these shutdowns happen? Anything like a metal hitting metal type of sound? Run an error check on your OS drive. Right click my computer > properties > tools tab > click 'check now' under error-checking. Check both check boxes under 'check disk options' and click OK. Then it'll tell you that a reboot is required. Upon reboot, it'll run the chkdsk process and it'll take a couple of hours probably, the larger the hard drive the longer it takes.

Then we can check event viewer for the log of the chkdsk results and see if any bad sectors were found or not. Windows will automatically reboot when it's done so it'll be waiting at the logon screen when you return. If you already knew how to do all of this, then my bad lol.



So I checked Event Viewer and found this http://pastebin.com/P1tSv4ZL I checked out the Microsoft site, but no idea what to do. There are about 20 mores of those error, they are getting more and more frequent.
m
0
l
February 6, 2013 3:54:40 AM

Yea, those events are created every time Windows is instantly shutdown as in a loss of power. When my PSU first started going out, it seemed random and then eventually got to the point where I could recreate the incident easily. But mine only happened when my GPU would was under heavy load like gaming or benchmarking. It never got to the point though of shutting down just doing desktop sh*t. Also, my system never restarted. It would reset but remain powered on. I had to hit the reset button or hold the power button to power the system down and turn it back on to get back to Windows.

Ok, there's 2 things I think of right now to check.

1. This will be the easiest to check and probably the least likely to be the problem. Your APC BR1300G UPS and your Corsair GS800 PSU with active PFC. I've been doing some research lately because I've been looking into a new UPS and I've come across statements from UPS manufacturers and the general public alike that some UPS's that don't produce a pure-sinewave waveform can cause issues with some active pfc power supplies. I can't get the manual for your UPS from APC's website because it's not functioning correctly at the moment when I try to search products but newegg's listing for your UPS doesn't dictate that it's a pure-sinewave UPS. But from what I read it's only a problem when the UPS switches to battery. Easy to test as all you need to do is unplug your UPS from the wall while your system is running and see if your system stays running properly when the UPS switches to battery. Or if your UPS has some sort of test button/function to switch to battery without unplugging it from the wall. If your system stays running correctly when the switch to battery happens then this is not the problem.

2. This one is much more probable I think. It's your overclock. Your system may be overheating or your overclock isn't stable for some reason even though you are water-cooled. Which would mean there's an issue with your loop. I would think Asus has a utility that can monitor CPU/system temps and I know AMD's catalyst control center can monitor temps of your GPU. Considering your system last restarted while copying a file from a .rar, I would lean more toward your CPU than GPU. Run prime95, or another CPU benchmarking tool if you know one, and monitor your temps and see if they elevate too high. Or if your CPU benchmarking tool reports any errors. If they do elevate too high or report errors with your overclock, go back to default settings and see what happens.

If temps are fine, we'll think of some more sh*t to troubleshoot. And the only way to test if it's your PSU at this point is if you have another PSU to test in your system. A PSU tester won't detect anything at this point.

Do be prepared to do a lot of testing unless you're lucky and we find the problem early. I tested my bastard of a computer for like 2 weeks, including a reinstall of Windows, every ****** video driver available for my Sapphire 6970, lengthy RAM tests and testing all of my components in other known working systems, before I was finally able to test another PSU and everything worked fine. It was frustrating and I felt like kicking a puppy and punching a baby in the face by the time I was done.
m
0
l
February 6, 2013 3:23:02 PM

arson94 said:
Yea, those events are created every time Windows is instantly shutdown as in a loss of power. When my PSU first started going out, it seemed random and then eventually got to the point where I could recreate the incident easily. But mine only happened when my GPU would was under heavy load like gaming or benchmarking. It never got to the point though of shutting down just doing desktop sh*t. Also, my system never restarted. It would reset but remain powered on. I had to hit the reset button or hold the power button to power the system down and turn it back on to get back to Windows.

Ok, there's 2 things I think of right now to check.

1. This will be the easiest to check and probably the least likely to be the problem. Your APC BR1300G UPS and your Corsair GS800 PSU with active PFC. I've been doing some research lately because I've been looking into a new UPS and I've come across statements from UPS manufacturers and the general public alike that some UPS's that don't produce a pure-sinewave waveform can cause issues with some active pfc power supplies. I can't get the manual for your UPS from APC's website because it's not functioning correctly at the moment when I try to search products but newegg's listing for your UPS doesn't dictate that it's a pure-sinewave UPS. But from what I read it's only a problem when the UPS switches to battery. Easy to test as all you need to do is unplug your UPS from the wall while your system is running and see if your system stays running properly when the UPS switches to battery. Or if your UPS has some sort of test button/function to switch to battery without unplugging it from the wall. If your system stays running correctly when the switch to battery happens then this is not the problem.

2. This one is much more probable I think. It's your overclock. Your system may be overheating or your overclock isn't stable for some reason even though you are water-cooled. Which would mean there's an issue with your loop. I would think Asus has a utility that can monitor CPU/system temps and I know AMD's catalyst control center can monitor temps of your GPU. Considering your system last restarted while copying a file from a .rar, I would lean more toward your CPU than GPU. Run prime95, or another CPU benchmarking tool if you know one, and monitor your temps and see if they elevate too high. Or if your CPU benchmarking tool reports any errors. If they do elevate too high or report errors with your overclock, go back to default settings and see what happens.

If temps are fine, we'll think of some more sh*t to troubleshoot. And the only way to test if it's your PSU at this point is if you have another PSU to test in your system. A PSU tester won't detect anything at this point.

Do be prepared to do a lot of testing unless you're lucky and we find the problem early. I tested my bastard of a computer for like 2 weeks, including a reinstall of Windows, every ****** video driver available for my Sapphire 6970, lengthy RAM tests and testing all of my components in other known working systems, before I was finally able to test another PSU and everything worked fine. It was frustrating and I felt like kicking a puppy and punching a baby in the face by the time I was done.



So just realized that my overclock settings have reset, so we can rule that out. And ive always had CPU widgets show temp and stuff. The temperature has always been ok with the liquid cooling 60ºc-ish max. What Il try to do is recreate the crash it at will?
m
0
l
February 6, 2013 8:45:18 PM

If you can figure out how to recreate the crash at will it will definitely expedite the process. Stress testing your computer using CPU, RAM and GPU stress testing programs might help you find a way to recreate the issue as well. GPU stress testing, like with Furmark or something else similar, will probably be the highest PSU load test. Or running all three stress testers simultaneously maybe. And I just realized you have SSD's. You definitely don't want to chkdsk (Error checking) your SSD's. That's only for HDD's.
m
0
l
February 9, 2013 7:06:59 PM

I think its a Java Memory leak. Java process uses up max ram its allowed to use then it freezes the rest of the server.
m
0
l
February 9, 2013 7:07:30 PM

I have no time to stress test it on the weekends, will do it durring the week.
m
0
l
February 9, 2013 7:41:05 PM

Maybe there's an update to Java?
m
0
l
February 9, 2013 10:17:05 PM

arson94 said:
Maybe there's an update to Java?


And there is... Updating now.
m
0
l
February 10, 2013 4:00:36 AM

Is it just the runtime environment or do you have more running on that server? Like the SDK or something else.
m
0
l
!