Do I Have A Power Supply Issue?

[malmsteenisgod]

I've been experiencing some unexpected shutdowns while playing games, or sometimes just watching a video. For a while I thought it was my 4850 overheating, but that doesn't seem likely, even though I've seen it around 80C on load sometimes. But, lately my computer has shutdown while playing less graphics intense games, like Oblivion and Ghost Recon.

 

My power supply is an OCZ Gamexstream 700W

http://www.newegg.com/Product/Product.aspx...N82E16817341002

 

and it's only about 4 months old, so I'm wondering if maybe it was a faulty PSU. What do you guys think?

[damian]

Load your video card using overdrive or ATITool and tell us what the temps are at load. If this has never happened before its hard to say whether its the power supply or not. It could be a number of things. The last thing you want to do is RMA your psu. Do you have a spare one you could use to compare?

 

Does your computer shutdown every time you are using it?

[Crazy_Nate]

Load your video card using overdrive or ATITool and tell us what the temps are at load. If this has never happened before its hard to say whether its the power supply or not. It could be a number of things. The last thing you want to do is RMA your psu. Do you have a spare one you could use to compare?

 

Does your computer shutdown every time you are using it?

 

Furmark is excellent for loading up graphics cards.

 

Will this computer pass Prime95 or OCCT for a long period of time (say, 12 hours)?

 

The 700W GXS is a 4-rail power supply right? Are your rails relatively balanced? Too much load on one rail might trigger shutdown...

[malmsteenisgod]

Load your video card using overdrive or ATITool and tell us what the temps are at load. If this has never happened before its hard to say whether its the power supply or not. It could be a number of things. The last thing you want to do is RMA your psu. Do you have a spare one you could use to compare?

 

Does your computer shutdown every time you are using it?

 

The highest I've seen my cards temp get under load in a game is around 80C. High, I know. It's really god-awful warm in my dorm. But, it doesn't shutdown everytime I'm using my computer. It's really strange, actually. I used to encounter shut downs and BSOD's while playing LotRO, but I never had problems while playing Crysis. To solve that problem I just put my Q6600 back at stock and it ran fine from then.

 

Oh, and sorry I don't have an extra PSU to try out.

 

Furmark is excellent for loading up graphics cards.

 

Will this computer pass Prime95 or OCCT for a long period of time (say, 12 hours)?

 

The 700W GXS is a 4-rail power supply right? Are your rails relatively balanced? Too much load on one rail might trigger shutdown...

 

I can't say I've run it for 12 hours. I think the most I've run it is 6, because, well frankly I'm not that patient. But yes, it has run fine under prime95 for 6 hours or so, never saw any temps on my CPU over 60C. Yet, for some reason I'll occasionally during a pretty weak game like GRAW2 or Oblivion I'll encounter shutdowns and eventually put my processor back at stock. I don't see how it could run fine for 6 hours in Prime95 and then fail after 2 hours of one of the aforementioned games.

 

It is a 4-rail power supply. I don't know much about about PSU rails, so could you direct me to something that shows how to test the load on the rails?

 

Thanks for the responses.

Edited November 10, 2008 by malmsteenisgod

[malmsteenisgod]

Ok, I put my 4850 at 675/1150 and ran furmark for about 10 minutes and saw about 95C. I stopped it there, paritally because I don't like seeing that temperature, and partially because it didn't seem like it was going to go any further. But I never saw any corruption or artifacts. But I think I can rule out the card.

 

So, I'm thinking either it's my PSU or CPU. But from what I've seen in Prime95 they seem to run fine. Perhaps when I'm playing a game and my CPU and GPU are loaded up it's loading the PSU too much? That still wouldn't explain why I can run Crysis fine with my Q6600 @ 3.0ghz but I can't run Oblivion for more than a few hours at 3.0ghz.

[scr4wl]

That still wouldn't explain why I can run Crysis fine with my Q6600 @ 3.0ghz but I can't run Oblivion for more than a few hours at 3.0ghz.

 

Cyrsis is more on the GPU then the CPU so it doesn't strain your cpu

 

Oblivion uses the CPU so it gets hot because you need to do a more stable overclock...

...i think

[Waco]

You've got more power than I do with less equipment....it's got to be something faulty or mis-configured.

[DLS2008]

OK, download Riva Tuner, CPU-Z, Core Temp, Prime 95, and Furmark. Run Prime 95 for at least 12 hours with your CPU overclocked to where you want it. It isn't truly stable unless it can pass 12 hours. Run it 18 or 24 hours if possible. While it's running, watch the temps in Core Temp. See how high they get and let us know. Also, what are your voltages in the BIOS? Next, run Furmark and watch the GPU temperatures in Riva Tuner. Let us know how high they get. Also, use Riva Tuner to increase the fan speed and see if that lowers your temps. Finally, run Prime 95 and Furmark together and check the temperatures. Let us know as much as you can. I would suspect it's in your CPU. What is your VCore? Also, do you have your components spread across your rails? There's no way that setup will draw more tha 700W. Is your GPU on its own rail like it should be? The more details we know, the better our help is.

[malmsteenisgod]

OK, download Riva Tuner, CPU-Z, Core Temp, Prime 95, and Furmark. Run Prime 95 for at least 12 hours with your CPU overclocked to where you want it. It isn't truly stable unless it can pass 12 hours. Run it 18 or 24 hours if possible. While it's running, watch the temps in Core Temp. See how high they get and let us know. Also, what are your voltages in the BIOS? Next, run Furmark and watch the GPU temperatures in Riva Tuner. Let us know how high they get. Also, use Riva Tuner to increase the fan speed and see if that lowers your temps. Finally, run Prime 95 and Furmark together and check the temperatures. Let us know as much as you can. I would suspect it's in your CPU. What is your VCore? Also, do you have your components spread across your rails? There's no way that setup will draw more tha 700W. Is your GPU on its own rail like it should be? The more details we know, the better our help is.

 

Ay, I'll run all those tests starting in the morning. My vcore is at 1.325. I'm not sure what you mean by having the components spread across the rails. Could you explain?

[Crazy_Nate]

So...a true multirail power supply will have separate sources of certain voltages (in this case +12V). Each has current limitations, so, essentially, if you exceed those, it will trigger some power protection circuitry.

 

With that said, you want all of your components that use a lot of +12v to be evenly spread out across the rails (in theory, they should be marked on the power supply). For instance:

 

+12V1: CPU

+12V2: GPU 1

+12V3: GPU 2

+12V4: Harddrives / Accessories

 

You can get a rough estimate of how much power your CPU is using by looking at the TDP. The internet usually has power consumption values for graphics cards.

 

If you want to figure out how much power a rail can support, P = I*V

:thumbs-up:

[fire_storm]

Did you try resting everything in your bios back to stock settings and then see if the computer randomly reboots?

 

Also with your video card do you have it set on automatic fan control or manual?

 

I know on my 4870 the automatic fan control never speeds up the fan correctly when it gets hot so I always manually turn it on 100% right before I start playing a game and it keeps my gpu temp 40C-50C cooler then if automatic fan mode was enabled.

[malmsteenisgod]

So...a true multirail power supply will have separate sources of certain voltages (in this case +12V). Each has current limitations, so, essentially, if you exceed those, it will trigger some power protection circuitry.

 

With that said, you want all of your components that use a lot of +12v to be evenly spread out across the rails (in theory, they should be marked on the power supply). For instance:

 

+12V1: CPU

+12V2: GPU 1

+12V3: GPU 2

+12V4: Harddrives / Accessories

 

You can get a rough estimate of how much power your CPU is using by looking at the TDP. The internet usually has power consumption values for graphics cards.

 

If you want to figure out how much power a rail can support, P = I*V

:thumbs-up:

 

Thanks for that lesson. So yes, my graphics card is on it's own rail.

 

Did you try resting everything in your bios back to stock settings and then see if the computer randomly reboots?

 

Also with your video card do you have it set on automatic fan control or manual?

 

I know on my 4870 the automatic fan control never speeds up the fan correctly when it gets hot so I always manually turn it on 100% right before I start playing a game and it keeps my gpu temp 40C-50C cooler then if automatic fan mode was enabled.

 

Yes, I've put my Q6600 back at stock and it runs fine then. So, it is still possible that it's an unstable overclock.

I'm using manual control for my 4850. I'd have to say 100% is too loud for me. I tend to keep my my fan around 65% while playing games.

 

So, I clocked my Q6600 back up to 3.0ghz and started running Prime95 around 11:00 AM Central time and will probably stop it around 11:00 PM. I would love to let it run through the night, however, I live in a one room dorm and trying to sleep with all the fans and lights on while listening to my room mate ###### about the fans and lights would be hell. How long do you guys recommend I run Furmark and how long should I run Prime95 and Furmark together?

 

In the meantime I'll give a rundown of my voltages and such.

 

Q6600 @ 3.0ghz (9x333)

DDR3 1333 7-7-7-20

CPU VID: 1.325

CPU Voltage: 1.325

FSB voltage: 1.325

MCH voltage: 1.325

Ram Voltage: 1.74V

 

In case you're all wondering, I didn't just set them all to 1.325 because I didn't know what I was doing, that was really just how I was able to get it stable when I overclocked it for the first time a few months back. I've heard a lot of people struggle with getting the RAM stable on this board, which may also be my problem. OCZ states that this RAM is to be set at 1.8V, however from everyone I've read with this board, they could not get ram stable on this board with 1.8V. They had to use lower volts, which was also my case. I originally toyed around with this voltage a lot before I found that it would run at 1.74V.

 

A few more things worth mentioning about this board: I haven't figured out how to get my RAM to run at 1333mhz and have my CPU run at 1066mhz. For some reason it doesn't have dividers or something. It has a memory reference frequency. Lowering this frequency just lowers my rams mhz, while raising it only raises it up as high as the CPU's frequency is. So, I can run my RAM and CPU each at 1066mhz, or each at 1333mhz, but I can not run my RAM frequency higher than my CPU frequency.

 

Also, this board cannot be overclocked from the BIOS (don't ask me why, it's retarded). I had to use Intel's Desktop Control Center which they designed specifically for the DX38Bt and DX48BT2 (I think they're also making one for the DX58SO now. But I just thought I'd mention that in case anyone asks why I'm overclocking with this lame program and not from the BIOS.

 

I'll be back later to post results from my tests and such and see where I can go from there. Sorry, for the essay I just wrote.