Bad GPU or PSU?

December 14, 2010 at 17:30:10
Specs: Windows XP
Hey, I think I may have just made a mistake by ordering a new GPU over a PSU. I've been having problems lately trying to play World of Warcraft. My screen will just freeze and go distorted, and I have to hard reset the comp every time.

I ran a memtest and it came up good. I put new thermal paste on the gpu and it was still heating up pretty quickly in game. Anytime I would get in an area where my fps would drop it would freeze up shortly after.

The concern is in speedfan I noticed my voltages were weird. All of them check out good except for my -12v. It's running at -1.42. Does this basically guarantee that my psu is bad?

Others seem normal though
3.3 = 3.28
5 = 5.03
12 = 11.97
-5 = -8.78

Card is EVGA GeForce 8600 GT 256MB
PSU is ENERMAX Liberty ELT400AWT 400W ATX12V


See More: Bad GPU or PSU?

Report •

#1
December 14, 2010 at 17:49:39
I think I would have looked at event logs and maybe used perfmon to run a log. Could have used maybe gpu-z for more ideas.


-8 on a -5 is way off. Not sure if you even use it. Might look at any ac component on the dc line too if caps are going out.

Why did it take me over a year to phone in a problem to ATT?


Report •

#2
December 14, 2010 at 18:22:29
-5V was only used in boards with ISA slots and I highly doubt yours has one. Disregard that completely as your powersupply doesn't even have a -5V rail. http://www.enermaxusa.com/catalog/p...

According to specs it has dual +12V rails rated @ 20Amps each with 80% efficiency so in sustained use situation you should not load up either of the +12V rails to more than about 16Amps. It might be worth checking whether you have the load correctly balanced across both rails. Since Speedfan uses the motherboard sensors to check the voltages I suspect your -12V reading is an anomaly. There is little that uses this power anyway with only a 0.6Amp rating on that rail. Does speedfan give you the graphics card temperature? how are you determining what "heating up pretty quickly" means?

Found this review of your graphics card:
http://www.mail-archive.com/gnhlug-...
Indicates power requirement as min 350W with 18Amps on +12V and your PSU seems to meet this as long as the load is well balanced.

Goin' Fishin' (Some day)


Report •

#3
December 14, 2010 at 18:50:21
My card is idling at around 45 but whenever I go in game it will continue to go up into 60 or so territory within about 5 minutes. This didn't seem to hot to me, and the thermal paste dropped it to about 55. I don't think its overheating, but I believe the ram on it has gone bad.

Sometimes the game will freeze for a second and instead of completely locking up, I'll get random images in places where they shouldn't be in game. Or tree's will turn black for example.

How do I go about balancing my PSU load?


Report •

Related Solutions

#4
December 14, 2010 at 19:05:13
Just check what devices such as motherboard main connectors, drives etc are connected to the PSU rails and deternine the power requirements of each. As far as GPU temps go 55-60degC under load is well within normal range so it isn't currently overheating but the damage might have already been done.

Goin' Fishin' (Some day)


Report •

#5
December 14, 2010 at 19:48:04
Everything seems in the right spot on the psu. Although there are PCI-Express slots on there that I'm not using. However I don't think there is a plug on the mobo for pci, I think it just runs off the mobo power. I'm ignorant when it comes to PSU's though really.

This is the mobo I'm running.
http://www.newegg.com/Product/Produ...


Report •

#6
December 14, 2010 at 21:11:46
I switched PCI Express slots and it seems to be working pretty well now. Running about 65 degrees in game and hasn't crashed.

Report •

#7
December 14, 2010 at 21:48:32
SpeedFan makes assumptions based on the hardware monitoring chip it detects on the mboard.
It assumes sensors are connected to all inputs - on some mboards not all of the inputs have sensors connected to them - in that case, SpeedFan's reading is incorrect and it never changes. In other cases, the mboard maker did not connect the sensor to the correct input specified by the chip maker.

If SpeedFan detects a hardware monitoring chip, then almost always you can go into the bios Setup and find the current readings listed in the bios - the readings for inputs that are not hooked up to sensors will NOT be listed there, and what reading is for what will be correct, even if the mboard maker didn't connect them to the chip manufacturer's spec input. Also, the bios has almost always been tweaked to make the readings as accurate as they can be.

In other words, if there are temp, voltage, and rpm readings in the bios, don't bother using SpeedFan unless you tweak it to not show invalid readings (no sensor is hooked up) and adjust the readings so they're ~ the same as in the bios.


Report •

#8
December 14, 2010 at 23:43:12
I cannot get the newegg link to work. Can you post the motherboard model number so I can look for a manufacturer's link?

Goin' Fishin' (Some day)


Report •

#9
December 15, 2010 at 05:36:10
Try this link:

http://www.newegg.com/Product/Produ...

It's an oldie but was good in it's day.


Report •

#10
December 15, 2010 at 12:33:23
I still think it has little to do with your hardware.

Why did it take me over a year to phone in a problem to ATT?


Report •

Ask Question