One year system randomly restarts, no BSOD

Shodoman

Distinguished
Sep 19, 2012
76
0
18,640
Hello,

Short story: system restarts randomly -full load/no load, tested cpu, ram, video, don't have spare psu or motherboard so i send whole system to service, they send it back with no problems found.

Cpu: i5 3470
Motherboard: Asrock H77 pro4/Mvp
Hdd: WD 1TB blue
Psu: Seasonic S12II-520
Ram: Corsair Vengence 8GB 1600 (4+4)
Case: CM Silencio 450
Video: Sapphire Radeon HD 7870 (had a Radeon HD7770 also from Saphhire)
Windows 7 Ultimate 64

Long story: first problems showed up in spring when in some games the screen would turn black or horizontal stripes and then i had to restart. Thought it was the games or the video card as it did not happen in all games.

In summer these freezes changed into restarts i said ok, it's time to send the video card in for warranty. After 12 days they tell me they changed it with a new one but i refuse it, ask the money back and buy my current video card.

The restarts continue as if nothing happened. So i send the whole system back to the service describing the problem as 3d related because that's what i thought at the moment was happening (very little time for pc you would catch me only in games). They send it back after a couple of days with bios updated - fixed.

I bring it home start a game after 5 minutes bam restart, then 2 more restarts in boot screen, i figure it's a heating problem so i shut it down and go figure, they've left one of the legs of the cpu cooler out of place. 2 day later I've installed a CM T4 cooler as the intel one was crap anyway.

After 3 weeks computer starts to restart again. I go berserk, start monitoring restarts with Aida and still no clue as to where the restarts come from no voltage irregularities no overheating anywhere, nothing. I do some prime 95, some furmark, some of both at the same time. It restarts at one point but then it works for 12 hours with no issues. Then some memtest86 from a Usb stick; 20 minutes into the first 4gb ram it restarts then does 2 passes without flaw on each of the 4 gb.

I reinstalled windows as an act of desperation, needles to say it didn't fix anything. So i send it back to service during the holidays as i was away and after 10 days, last night they said that they couldn't identify a problem so they're sending the system back.

What should i do now? Can it have to do with the usb ports that i use for my mouse and keyboard? Can it be that i use 2 output channels for sound (i use my front audio port for my headset and games while i keep that back for speakers and winamp). Should i buy a new PSU and see if it keeps restarting (i have a 30 day return policy thing)?

I don't know if it matters but I'm from Romania.

Thanks,

Shodoman
 

Shodoman

Distinguished
Sep 19, 2012
76
0
18,640


They already updated my uefi. And i've checked, it's up to date.
I have to buy another psu to test it and was hoping it won't get to that.

Any more help, please?
 

Shodoman

Distinguished
Sep 19, 2012
76
0
18,640
I've went to service to pick it up and asked nicely that they further test the system. Also added to check the psu especially on the 5V lane and also to plug some peripherals (a headset or speakers) because i was thinking this is the only difference between home and service. I still feel it's a long shot though.

Could this have been the issue, or any more ideas on what i should check if this fails?
 

Shodoman

Distinguished
Sep 19, 2012
76
0
18,640
New development after further testing they found the RAM to be faulty but after a few tests with another pair of the same model they said that the RAM model is not compatible.
Doesn't it sound like bullshit... what could be incompatible about them?
 
-Bad RAM will not cause a reboot without making a memory dump on a system configured to save a memory dump on bugcheck.

-if you are configured to make a memory dump and do not get one you have a power problem or your system has lost access to your drive (can not store your memory dump)

- confirm your system is configured to do a memory dump when you get the bugcheck and jut not reboot.
- enable hotswap on your sata port for the harddrive. This will work around conditions where your drive has a error and takes too long to respond, after 30 seconds windows resets the sata port. if hotswap is not enabled in BIOS the drive will not reconnect and you will bugcheck some time later and not get a memory dump.

there are power conditions that will also cause this but the power supply will test out fine.
- too much power being pulled from a line on the power supply
- power supply detecting incorrect voltage conditions can reset a signal to the motherboard that reboots your machine.

also, check for dust in your system, various fans can get stuck (power supply, cpu fan, case fans) the system overheats and causes thermal protection to shut down the system. just blow out the dust with a can of air, and spin the fan will often fix this.

- very common to get this with graphics cards with a system that is slightly under powered by the power supply
mostly you only see it when you play a game for a while. when a graphics card starts to really stress the graphics card with a game it will use more power, produce heat. For this see if you can increase the default fan speed for your graphics card in its controller software. If the graphics card pulls too much power, the power supply will send a signal to the motherboard, and that signal will reset the cpu (just like you hit the reset button, no error logs or memory dump, and the system will reset and start its BIOS routines)




 

Shodoman

Distinguished
Sep 19, 2012
76
0
18,640


Thank you for your answer.
I had already checked for the system to not restart but i have yet to see a bsod.
I also doubt the hdd is at fault as usualy, when such an error occures the program or computer would freeze and it does not.
I cannot test the psu but i've sent it to the service.
I usualy clean the dust filters twice a month, the inside is pretty dust free because of this.
Temps are all good: cpu 50, mobo 40, video max 78 Celsius.
The psu can dish around 480 W on the 12 V rail and i do not oc, shouldn't it be enough?
Anyway, i've chaged the ram and the system crashed. New symptom it didn't start for about 30 seconds after the crash. The power led and fans would be on, but i think the hdd didn't speen and the display would be black.
Any thoughts?
 
with the new RAM, often the settings in BIOS will be wrong when new RAM is added
-I would update the BIOS (to get any fixes and reset all the settings to default, then configure any required settings)
test with memtest86 to confirm settings are correct. (often the Command settings are overlooked, default in BIOS is 1n or 1T but some RAM require 2n or 2T to setup all of the memory addressing commands correctly)


- you have a good power supply but the video card requires (2) 75 watt connectors as well as the 100 watt power from the PCI bus.
I think your power supply is telling the motherboard that the power is out of spec and the mother board resets the CPU. (this is the way it should work) now you have to figure out why the power supply circuit thinks too much power is being pulled on one of its circuits.

-I would look for physical causes, recheck the connections from the power supply to the items that are being powered. unplug them and plug them back in. I would start removing any excess devices and look for any potential short circuits. Even a short or out of spec device like some apple products can cause your machine to reboot through the USB bus.