I am affraid that you have faulty RAM chips on graphic card. Because you dont know the faulty one, you need to replace them all

I came across the same problem with some older graphics cards of mine (various ATI models) over the years.
I always wondered what caused this but, never got around to asking.
Thanks for clearing this up.

Now to figure out how to de-solder all of the RAM chips and where to find replacement chips. Ebay perhaps?

Well, you need a hot air station to dessolder the chips, kapton tape to protect all the surrounding components and of course the ram chips. you may also need some solder paste to solder the new IC chips. You can find all of this on ebay
http://www.ebay.com/itm/20mm-2cm-x-30M-Kapton … =item461db04c79
http://www.ebay.com/itm/3-in1-Welder-Solderin … =item3a97f24b6b

Almost certainly a RAM problem and not a capacitor problem...

The more popular manufacturers only started skimping on the caps GeForce 6 and up AFAIK.

The reference ATI Radeons of that era OTOH (Radeon 9700 pro, for example) used electrolytic Nichicon HC which performed well overall, but did in fact bloat in systems with excessive heat and lousy PSUs (Which was 90% of the systems back then).

Could this be a bad solder joint (on the memory chips)? Perhaps the board was fine when the seller shipped them to you, but a solder joint broke during shipment? If so, all you might need is a reflow

mockingbird wrote:

Almost certainly a RAM problem and not a capacitor problem...

The more popular manufacturers only started skimping on the caps GeForce 6 and up AFAIK.

The reference ATI Radeons of that era OTOH (Radeon 9700 pro, for example) used electrolytic Nichicon HC which performed well overall, but did in fact bloat in systems with excessive heat and lousy PSUs (Which was 90% of the systems back then).

Hmm, I wonder if I should check the caps on my dud 9600XT (Sapphire 256MB DDR, non reference?). The caps aren't visibly bad, but I get this same kind of artifacting. I've already tried a reflow and it just doesn't seem worth it to track down and replace the RAM chips.

Certainly. Sapphire was infamous for their use of lousy caps. Just use quality ultra-low ESR replacements or polymers if you feel like spending a little more.

bestemor wrote:

So, I recently bought a GF4 Ti-4600 and a Ti-4800SE off the bay... (Sparkle and MSI) […]
Show full quote

So, I recently bought a GF4 Ti-4600 and a Ti-4800SE off the bay... (Sparkle and MSI)

Both was advertised as fully working etc.

But I get (almost identical) artifacting on both, from the get-go:
http://i62.tinypic.com/2uh2rh4.jpg
http://i62.tinypic.com/25apyd4.jpg

- Are these beyond rescue, use for 'parts only' ?

Too late for a SNAD (Significantly Not As Described) claim?

I see you tried two different mainboards, you could also try a different PSU. But I agree that this is 98% certainly a defective memory issue.

to the bin with them.

alexanrs wrote:

Could this be a bad solder joint (on the memory chips)? Perhaps the board was fine when the seller shipped them to you, but a solder joint broke during shipment? If so, all you might need is a reflow

Bad solder joint wouldn't cause this so early on boot... I think that the machine just would hang when the chips get hotter. hang because the one pad would loose contact with IC

Have tested a crapload of my other GF4 (and other) cards on the slot1 mobo, partly on request from the seller to investigate some more.

NONE of these other cards showed any sign of artifacting in the exact same hardware and software configuration. (no drivers installed, apart from those included in WinXP SP1)

After some more re-testing, the faulty Ti4600 Sparkle card now has changed behaviour 'somewhat'. Maybe because of all the action/reboots etc heating it(chips) up and/or perhaps 'charging' the caps ?

- At POST, before starting booting into WinXP, the artifacts now don't appear anymore, but as soon as the harddrive initiates windows, there is a (different) pattern emerging, all the way until system tries to show the desktop, then it totally explodes into a massive mess, like a black&white version of the Stars and Stripes. Then goes into an automatic reboot(!) after a few seconds.
If I then leave it unattended, it will repeat this whole process indefinately (while adding a disk check stage because of the reboots)...........

For the record, if then I turn it all off, and just swap the Sparkle GF4 card for any other card, it is all perfectly fine again! (after the obligatory disk check) - I don't have to change or do anyting else, neither in software or hardware.

The easy culprit to blame would be WindowsXP, but... it does NOT happen with ANY of my other cards, so.... back to faulty memory then.
Maybe the previous owner did not test further than POST (?).
But he has confirmed that "the card was working perfectly the day he removed it to photograph for eBay".
Which may indicate it was in an actual working PC (with an OS?) at the time ?

- Oh well, regardless it seems like the local recycling plant is the most likely recipient, as I just cannot get this to work properly nomatter what I do, and all my other similar cards show none of these problems in the exact same setting... 😢

The oven trick could perhaps fix that card, for a while...

I think the memory chips are fine, in my experience its almost always solder joints under the GPU that is the issue even when the result looks like a memory error.

I don't think the oven trick will help any on Anything under Geforce 6 series...

Geforce 4 used leaded solder balls, and the chips weren't faulty, so if you have a defective card, it's most likely a capacitor or RAM issue.

Sparkle was a manufacturer that often used inferior capacitors. I'm willing to bet it's either that or bad RAM...

Solder joint failure can happen on almost anything subjected to thermal stresses. It wouldn't really hurt to try to reflow. But GF4 isn't exactly rare either.

The sellers replies that "Why are you not using a driver? Without the recommended driver installed it is possible to 'overclock' and thus damage the card!"

...which I do not understand, as there earlier were artifacts at first boot, with NO harddrive attached, hence no driver. And even so, winXP SP1 2D drivers from 2001 should not be able to do any sort of 'overclocking' on this GF4, OR... ??? Confused... 😕

As I am just booting to desktop, nothing else, no starting any fancy 3D programs that would require some sort of driver.
And add to the fact that ALL my other cards had no problem without any specific driver installed, I really wonder why this Sparkle should need any... (?)

And HOW would you install the driver the first time around, when you cannot actually see the desktop ??

- Now, not giving up quite yet, I have now installed a generic official Nvidia driver from 2004(while having a different, working GF4 card installed), which should be more than ok for any of these 2002 cards.

But, the only change I see for the Sparkle is that the previous blob of colored blocks never shows - the pre-desktop pattern is UNCHANGED though, and I then get a blank screen, and the whole system HANGS (I think, no picture hence hard to tell)... and no rebooting or anything.
So installing a driver did 'something', but....

bestemor wrote:

The sellers replies that "Why are you not using a driver? Without the recommended driver installed it is possible to 'overclock' […]
Show full quote

The sellers replies that "Why are you not using a driver? Without the recommended driver installed it is possible to 'overclock' and thus damage the card!"

...which I do not understand, as there earlier were artifacts at first boot, with NO harddrive attached, hence no driver. And even so, winXP SP1 2D drivers from 2001 should not be able to do any sort of 'overclocking' on this GF4, OR... ??? Confused... 😕

As I am just booting to desktop, nothing else, no starting any fancy 3D programs that would require some sort of driver.
And add to the fact that ALL my other cards had no problem without any specific driver installed, I really wonder why this Sparkle should need any... (?)

And HOW would you install the driver the first time around, when you cannot actually see the desktop ??

The seller is full of shit.

one or several address(es) in memory chips are damaged... The data stored on them could get corruption, so, you see those artifacts. There is no driver to solve this. You may give up. The proof is on PC boot, no driver loaded no nothing and you see garbage on screen. Bin it and save some time

honestly no harm in trying to bake it to reflow, what can be hurt at this point?