Video Card Cooling

Post by **Shonky** » Mon Jul 17, 2006 5:58 pm

SE1Z wrote:Been thinking about this some more....

Is there a program out there that can check the output of my PSU?
I'd like to check what its actually doing....

When I had my AMD cpu, I had a prog that could do this...

Any ideas?

The first place to look would be the manufacturer of your motherboard, check the cd that came with it, almost all mobos that I've seen in the last few years have some sort of app available for checking voltages on each supply rail using an app in Windows.

The other option is to get hold of a multimeter, which is probably preferable in this situation, the voltage sensors on motherboards that are accessible by software are connected after the switch mode voltage regulators on the motherboard, the problem with this is the switch mode regulators on the motherboard can keep the voltage level up (due to the way they operate) even though the actual voltage coming from the power supply could be lower than its nominal voltage.

wild · Post by **wild** » Tue Jul 18, 2006 8:46 am

I have a Zalman silent heatsink on my 7900GT, works wonders. Running at 580/780, up from 450/667.

And as Shonky said, even though you can get voltage readings from the motherboard, use a multimeter instead. I've seen some incredibly inaccurate readings from motherboard sensors in the past.

SE1Z · Post by **SE1Z** » Tue Jul 18, 2006 7:31 pm

seems it was the standard heatsink on the vid card...
I installed the Zalman VF900-cu this afternoon, seems all is fixed

attached is a pic of the errors I was getting before, note the purple messed up texture near the plane.
The see-thru carrier was only a once off, more often than not, the errors were green/pink/purple covered textures, mis-aligned polygons and black triangles in places.

The temps have gone down a LOT also...
here are some logs with the old heatsink/fan:

idle:
2006-07-16 00:30:34 Temperature: GPU: 39.3Ã‚Â°C GPU environment: 32.0Ã‚Â°C
2006-07-16 00:30:44 Temperature: GPU: 39.3Ã‚Â°C GPU environment: 32.0Ã‚Â°C
2006-07-16 00:30:54 Temperature: GPU: 39.4Ã‚Â°C GPU environment: 32.0Ã‚Â°C
2006-07-16 00:31:04 Temperature: GPU: 39.3Ã‚Â°C GPU environment: 32.0Ã‚Â°C

Load:
2006-07-16 21:21:06 Temperature: GPU: 72.8Ã‚Â°C GPU environment: 41.0Ã‚Â°C
2006-07-16 21:21:16 Temperature: GPU: 72.8Ã‚Â°C GPU environment: 41.0Ã‚Â°C
2006-07-16 21:21:26 Temperature: GPU: 73.8Ã‚Â°C GPU environment: 41.0Ã‚Â°C
2006-07-16 21:21:36 Temperature: GPU: 73.1Ã‚Â°C GPU environment: 41.0Ã‚Â°C

Now (with new heatsink/fan)
Idle:
2006-07-18 16:51:15 Temperature: GPU: 29.9Ã‚Â°C GPU environment: 27.0Ã‚Â°C
2006-07-18 16:51:25 Temperature: GPU: 30.0Ã‚Â°C GPU environment: 28.0Ã‚Â°C
2006-07-18 16:51:35 Temperature: GPU: 29.9Ã‚Â°C GPU environment: 28.0Ã‚Â°C
2006-07-18 16:51:45 Temperature: GPU: 30.1Ã‚Â°C GPU environment: 28.0Ã‚Â°C

Load:
2006-07-18 17:03:25 Temperature: GPU: 43.4Ã‚Â°C GPU environment: 35.0Ã‚Â°C
2006-07-18 17:03:35 Temperature: GPU: 43.8Ã‚Â°C GPU environment: 35.0Ã‚Â°C
2006-07-18 17:03:45 Temperature: GPU: 42.4Ã‚Â°C GPU environment: 35.0Ã‚Â°C
2006-07-18 17:03:55 Temperature: GPU: 41.9Ã‚Â°C GPU environment: 34.0Ã‚Â°C

Shaun · Post by **Shaun** » Tue Jul 18, 2006 8:46 pm

fark 73 on load b4, damn lucky u got it when u did, prob fried card eventually at that rate

wild · Post by **wild** » Tue Jul 18, 2006 9:11 pm

Shaun wrote:fark 73 on load b4, damn lucky u got it when u did, prob fried card eventually at that rate

Nah, takes more than that (over 100c generally) to fry these things. However, it isn't good to run at those temperatures for long periods of time.

Shaun · Post by **Shaun** » Tue Jul 18, 2006 9:24 pm

yeh i gathered it woudl take more, but still, if its doing that often, never know really

ysu · Post by **ysu** » Wed Jul 19, 2006 9:40 am

Shaun wrote:yeh i gathered it woudl take more, but still, if its doing that often, never know really

No it's nothing. Some circuits have working temperatures in that range (some AMD processors for example).
It's 110-120 C where the melting problems start.

Shaun · Post by **Shaun** » Wed Jul 19, 2006 10:27 am

70 is alot closer than 40

wild · Post by **wild** » Wed Jul 19, 2006 10:54 am

ysu wrote:No it's nothing. Some circuits have working temperatures in that range (some AMD processors for example).
It's 110-120 C where the melting problems start.

I had an external temperature probe on my Athlon XP a few years back. When the fan failed, I got back to find the reading off the scale. I turned the computer off and started getting readings when it cooled, at about 120c. Waited for it to cool to around 50, fixed the fan, turned the computer on again, worked like a charm.

ysu · Post by **ysu** » Wed Jul 19, 2006 11:08 am

wild wrote:
ysu wrote:No it's nothing. Some circuits have working temperatures in that range (some AMD processors for example).
It's 110-120 C where the melting problems start.
I had an external temperature probe on my Athlon XP a few years back. When the fan failed, I got back to find the reading off the scale. I turned the computer off and started getting readings when it cooled, at about 120c. Waited for it to cool to around 50, fixed the fan, turned the computer on again, worked like a charm.

Wohooo nice

You're lucky too, the tomshardware guys managed to melt down (visibly!) an AMD cpu by removing the heatsink/fan (altho you still had the heatsink) ...

Gizmo · Post by **Gizmo** » Wed Jul 19, 2006 12:27 pm

SIEZ Take the heatsink back off..... Wall Hax without actuall cheating...... Post that on the Steam forums. The 13yr old CSS cheats will love ya!~

Gizmo · Post by **Gizmo** » Wed Jul 19, 2006 12:28 pm

SIEZ Take the heatsink back off..... Wall Hax without actuall cheating...... Post that on the Steam forums. The 13yr old CSS cheats will love ya!~

SE1Z · Post by **SE1Z** » Wed Jul 19, 2006 12:43 pm

Giz: pity it was BF2

Gizmo · Post by **Gizmo** » Wed Jul 19, 2006 2:56 pm

lol my eyes and brain saw bf2 but my hands typed css....lol

r8response · Post by **r8response** » Wed Jul 19, 2006 6:02 pm

Gizmo wrote:lol my eyes and brain saw bf2 but my hands typed css....lol

noob

SE1Z · Post by **SE1Z** » Wed Jul 19, 2006 7:49 pm

spoke too soon

still playing up...

is this any use to anyone to try and get to the bottom of this for me?

Post by **Shonky** » Wed Jul 19, 2006 9:32 pm

Damn dude, get hold of a multimeter to confirm those voltages.

+5.00v rail is well outside the 10% maximum variation. The +12.00 rail is looking a little sad too but is within 10%.

Wtf is up with the -12v rail, -2, -9, +1. That has to be a reading error, or your supply is cactus.

SE1Z · Post by **SE1Z** » Wed Jul 19, 2006 9:38 pm

dammit....
i got nfi what all those readings mean

I managed to get the vid card working well again (no artifacts) - did a clean driver install and upped the AGP voltage in the bios by +0.1

seems to have done the trick, hopefully it lasts untill I can get my hands on a new, decent power supply

Post by **Shonky** » Wed Jul 19, 2006 9:46 pm

If you got no idea what the readings mean have a read of the Power Supply Tutorial

SE1Z · Post by **SE1Z** » Wed Jul 19, 2006 10:06 pm

cheers for the link, will give me something to do at work tomorrow

I keep speaking too soon!!!!

check out the pic.

crazy triangles and colour 'blotches' on the gun....

ysu · Post by **ysu** » Wed Jul 19, 2006 11:37 pm

If those readings are true, you really need a new psu and I'll shut up once and for all about PSUs

Some of those readings are off the scale badly, I wonder why don't you get all kinds of problems frequently - unless the reading is the wrong one.

Otherwise it may be a faulty vid card memory chip, I seem to recall that I've seen some kind of tester for that...but where???

Have you tried running 3d mark ? That may show up with something...

SE1Z · Post by **SE1Z** » Thu Jul 20, 2006 8:35 am

I recall that when I first got the card, I ran 3dMark. I had those colour blotches, but no fancy triangles.

Im fed up with the fcuking thing!

I'd hate to buy a new psu and find I have the same problem......

With the image I posted re: mobo readings, it *may* have been getting its info off the wrong sensor...
I had a play last night, managed to get 'better' readings, but still not great.

*sigh*

thanks for everyones help

I think ill just bite the bullet and buy a new psu.

This one looks the goods - http://www.pccasegear.com.au/prod3935.htm

If that doesnt fix it, the vid cad is porked (and I will be pissed).

Post by **Shonky** » Thu Jul 20, 2006 9:05 am

Don't know anyone you can borrow a PSU off SE1Z to rule it out before buying a new one?

SE1Z · Post by **SE1Z** » Thu Jul 20, 2006 1:02 pm

I bought this at lunch today
http://www.pccasegear.com.au/prod3421.htm

If it doesnt fix the problem, im screwed

Sterling · Post by **Sterling** » Thu Jul 20, 2006 1:17 pm

Bare in mind its not just about buying a PSU capable of more watts, you also need to look at whether the PSU will supply enough amps on one of the +12V rails for your video card.

I was a little suprised when I was doing a bit of research on my new video card how much amps the suckers need.