Pillowfight

My AMD rig has been shutting off randomly without warning (not right away, usually after an hour or more of use). At first I thought this might have something to do with my battery backup which had been warning me for months was in need of a new battery. I should have known better, but I bought a new one. Anyway, same thing, still random shutoffs.

My suspicion moved to the power supply. It's only four months old or so, and is an expensive Seasonic S12. I know even expensive ones can fail, but it's supposed to be very good. It replaced an Ultra that died after 2 years--which exhibited the same symptoms I am getting now, btw. That was in November or December.

Nothing in system logs to indicate thermal issues with CPU or anything.

Tested PSU with a $20 PSU Coolmax tester and everything has come up perfect on that, but Speedfan is indicating some of the rails are way off (look at the negative ones):



I have no idea what to think now. So now, age-old question...PSU or motherboard? I need to get this right because I don't have tons of money to throw into this blindly.

P.S. Yes I checked the caps on the board, they appear fine...

Tumbleweed36

First of all, that power supply is most likely the best made power supply in that class, so I seriously doubt if you have a bad power supply. Seasonic is considered the Cadillac of power supplie in most circles.

The second issue is the readings on Speedfan, they may not be correct. Many of those aftermarket programs don't read the sensors correctly on all boards. BTW, your voltages are fine according to what you posted.

Let's do two checks; First of all, go to the bios and read the temps and voltages and report them back here. Then, download and run SensorviewPro and report the temps and voltages as soon as you turn the unit on. Then, put the computer under stress like a game or something and then report those temperatures given to you by SensorviewPro. Post those also and we can take a look at your issues with a better understanding of what is going on with your rig.

Pillowfight

I did check the voltages in the bios and they align very closely with the expected specs, nothing unusual there.

I showed the above screenshot to two different techs in two different shops and they didn't like those negative numbers. I told them it was a Seasonic and they said it doesn't matter, it could still be junk. I'm inclined to believe you, however, because the PSU is so young and yeah it is supposed to be the Caddy of PSUs. But I am not an expert in voltages and rails.

As for stress testing it...I have no games...the most stressful thing I have is Photoshop CS3. Suggestions?

Tumbleweed36

Just make it work hard and any program that puts it under stress will help us to see. We don't know at this point what to compare those numbers with and that is why I asked for the bios and Sensorview. Therefore, I will reserve my comments on the techs concern until I see some more comparative figures on those stats. Sensors are some times just not correct. My current concern is the temperatures and the voltages in those programs.

BTW, I own a 600 Seasonic just like yours and also a 700 watt Seasonic. Those are the only brand I use in my personal computers, although have sold many others in my shop until I sold my shop a few months ago.

That is not to say it could not be the power supply, all of them from time to time can die. However, you don't have to worry if that is the issue, because Seasonic is great with replacing anything that is faulty. One of the best companies you can work with IMHO.

Post the bios figures with the Sensorview (both temps and voltages) so I can get a complete picture.

wrench97

What is reading 126c?
Try using Sensorsview program to see if it ID's it.

Pillowfight

Another screenshot coming. The Seasonic I have is a 550W, and I have three SATA drives plus several external and a load of peripherals.

linderman

I personally would try running a memory testing program like Memtest86 (free)

you can find how ot make the bootable disk here

How to Burn a Bootable ISO File


when testing memory test one stick of memory at a time, use only the slot closest to the cpu and run the test at least 2 hours per stick


you can do the same thing with your boot drive / test it with the hard drive manufacturers diagnostic utility

Pillowfight

The bios.

Re: the RAM, I've heard of bad RAM doing all kinds of horrible things but not total system shutdowns?

linderman

yes; defective ram can cause shutdowns


another test you could perform would be to run Orthos (free download) and work your cpu for at least an hour ......... watch your cpu temps with "core temp" also free; abort your test if temps reach 60C

Pillowfight

Okay, here's the "somewhat stressed" test

linderman

somewhat stressed ? please elaborate

linderman

does it do the auto shutdown while you are using the computer or only when totally idle ?

Pillowfight

When I'm using it, and seemingly at random. "Somewhat stressed" = me running Photoshop a bit, I don't have any games to really stress test it.

linderman

download orthos to stress test (free)

http://www.techpowerup.com/downloads/385/

also download core temp (free) run that at the same time as orthos; if the cpu temps gets to 60C abort the test

http://www.alcpu.com/CoreTemp/

Pillowfight

I ran those and after 6 minutes the most it went up to is 50C?

Pillowfight

I just tested all the RAM and that came up fine too.

Pillowfight

I have been running rig for four or five hours now and it hasn't punked out on me again yet...but I am nervous about it. Do you think I am somehow overtaxing the PSU? Doesn't seem likely since it's been running for at least a year with everything more or less the same, but I don't know.

SensorsPro is saying the video card temp is around 55C...

grimx133

Nah, that isn't a real power hungry card. As Tumbleweed36 mentioned, SeaSonic are great power supplies, it's not going to sweat the load you're putting on it. 55 is fine for a video card, they run much hotter than cpu's.