Okay, I've posted this in another thread, but I think it'd be better in it's own, so that more kind people can tell me whats wrong.

My screen looks something like this (below), and SpeedFan's voltage monitoring informs me that the -5V supply is only providing between -0.08 and -0.68 volts at any given time...

Would the lack of -5V be the cause of my screen problems? Does AGP work off the -5V supply? My card is a Ati All-in-Wonder 9800 Pro, and as such has the extra power connector (like a floppy drive power connector) on it.

My list of suspects is currently the gfx card itself, the AGP slot or the PSU. Any further thoughts would be appreciated. Will try another graphics card, another monitor (you never know), and another PSU later to see if it cures it (or makes it slightly better).

if your -5v doesn't exist it's very bad for your motherboard and can definately cause such screen "effects". BUT, I suggest you check the voltage from the bios configuration screen (usually DEL or F2 at the initial boot screen). Some extenral utility might read a wrong sensor.

But anyway if that's the problem then replace your PSU asap.

If the GPU core has been damaged by the heat though, then artifacts like those would stay on the screen.

I had exactly the same problem with a 9800XT that overheated and then it was like that in windows, and whilst booting there was weird characters all over the screen.

However back to your problem, try the PSU as others have said. That would be the easy fix

Thinking about it a bit more, you might be able to get a new one under warranty maybe if its still in the time limit. If not then you can't go wrong putting some arctic paste on there. I put some on my 6600GT and temps dropped a lot, and that was getting stupidly hot with the normal stuff. All is not lost yet

2 things, I would say, but definatly related to your GFX card.

I had a GFTI4600 do this, and that was because the memory was Crap, but this also done this on the POST screen.

My current FX5500 done this, due to the HS not attached properly, thus as they say up there, the GPU is overheating. Get some Artic Silver, remove the HS, and clean the GPU. Put some new Compound on and re-attach.

Done the trick for me.

Fordie

Hey Tristan ,

This happnes all the time ? even during bootup?

Have you ever or are you clocking the system or GFX card ?

have you tested at the motherboard 20 or 24pin plug for the -5v with a multimeter?

does the prob go away if you remove the side of the case? or whole lid case thing (shuttle)

awaiting more info Tristan

SD.

Many suggestions:
1) Check monitor cable
2) reinstall graphics drivers
3) check your ram chips. If you have many just try them separately

EDIT: 4) are all the fans spinning?

Okay, took PC home. Hooked up to parents screen (17" TFT) and the artifacts still there, so NOT monitor.

Then tried my old PSU for the shuttle (I upgraded from 220W to 250W SilentX just over a year ago). Made no difference, so I don't think the PSU is/was the problem. The old PSU also reported very low -5V readings, so I'm suspecting this PSU just doesn't use a -5V supply and the reading in SpeedFan is a false reading.

Next thing to try is another gfx card. My brother _should_ be bringing an old GF4 MX440 to try, but as yet I haven't tested another card. If that cures it we know it was gfx related.

If it doesn't cure it, we know it's probably Motherboard related... gulp...

Some kind soul has also PM'd me to inform me my VIA drivers might be up the creek, but as they have never been touched, and no software installed in the last few days that would cause them to change, I have to assume it's not the drivers. I'll give them a try when I've tried the other gfx card. If the alternative gfx card works, I'll try arctic paste on the gpu. My PC has never been under or over clocked, and all voltage adjustments have been left at Auto in the BIOS.

Thanks for your help guys.

Now, be honest Tristan: did you hacksaw anything briefly before this problem started occuring?

Quote from Hyperactive :

Many suggestions:
1) Check monitor cable

if its a monitor problem, how can he take a screenshot of it?

Quote from dUmAsS :

if its a monitor problem, how can he take a screenshot of it?

What... are saying? I just thought the cable is little loose...

Apparently it's not monitor problem because changing the monitor didn't help. I bet my my money (none) for the memory and some for the motherboard...

looks a lot like a broken fraphics card to me

but anyway on the -5v thing ... you might want to check this with a multimeter just to make sure its a false readout (but if it is it might hint to a motherboard problem)

I would say it looks like a GFX card problem. Had this kind of stuff happening few years back when I was a overclocking maniac and the GFX card picked up too much heat. Although it never happened in windows, just in games. But as shotclass said, get a multimeter and check the -5v line.

Quote from Hyperactive :

What... are saying? I just thought the cable is little loose...

Apparently it's not monitor problem because changing the monitor didn't help. I bet my my money (none) for the memory and some for the motherboard...

you cant take a screenshot of whats on the monitor. the screenshot is whats coming out the gfx card.

if i unplug my monitor, my screenshots wont turn black will they

anyway... the problem looks like an overheating GPU or failing ram

if the picture is corrupt at boot then its failing ram. if this happens after a few mins or a game then overheating gpu.

so things to do...

1. check fan is turning
2. see if picture is corrupt at boot

if its corrupt at boot then...

1. reseat graphics card
2. get an RMA number and return for replacement

Mainboards are cheaper than vidcards, tristan... So you better hope it's the mainboard

My guess it's either the AGP power line (which incidentally is -5V, that's why there's a 12V molex on those GeForces (actually it's 3.3v but is tapped off of the -5V from the PSU)... Or a bad connection on the AGP Bus...

If you want to be sure it's not the vidcard, put this one in a different PC. You can't tell if it's the vidcard if you put another one in this PC unless it's the exact same card (I fell for that one once). Oh! And make sure you install all the drivers for the mainboard and videocards... I had a perfectly good videocard once, until you installed the drivers. It went crazy because of a small flooding problem (ie, my LC blew giving the internals a nice shower). After that it never worked with drivers (perfectly without them though).

Testing:

Tried different monitor - not the monitor at fault (and I wouldn't be able to print screen it if it was).
Tried different PSU - not the PSU at fault, and the -5V line not present in either PSU.
Tried Geforce2 MX - no artifacts after over an hour of use.

Therefore graphics card at fault. Trip to Anglian Internet (best place in Norwich if you want to know) for a 9600 Pro 256MB and a £25 PCI TV card. PC working fine, TV in many ways better (ATi Multimedia Center is software from HELL!), a few less fps that I'm used to, but can still run AA and AF in LFS without any problems...

Thanks guys for your help and suggestions. If it turns out it WAS the m'board, I'll let you know so you can point and laugh at me...

card not in warranty still?

Didn't ATI give those cards a 3 year warranty?

/me checks

a -5 Rail is not required... Nothing today runs off it... Most new Power supply's dont even provide one.

and the ****ed up screen.. i get that when i overclock my video card to much... check the clock on yours and make sure they are at the default... if they are then check the heatsink to make sure its dust free and not overheating.


EDIT: it could also be bad drivers, try a newer or earlier version.\

EDIT2: here is my speedfan. My old Ultra X-Connect 500W did not have a -5v rail.. my new psu does



Also, just plug the floppy drive connector into the card.. You HAVE to have it plugged in for it to work right.

Quote from Roadie :

You HAVE to have it plugged in for it to work right.

no, you HAVE to have it plugged in for it to boot