• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

GPU has problems [SLOVED] originaly: IMC is dying...

It formed this white alloy clump that would not remove. I'll sand it off when I lap the thing eventually. Try using Shin Etsu, I've had good luck with that

I'm guessing you try a detergent of some kind when you got the white clumps because 70% isopropyl always worked perfectly for me. When I was messing around with what was left in one of the brushes I tried what does to aluminum and when I was trying to take it of used dishwashing detergent and ended up with little white spots everywhere. It also ate a chunk out of the aluminum.
 
I'm guessing you try a detergent of some kind when you got the white clumps because 70% isopropyl always worked perfectly for me. When I was messing around with what was left in one of the brushes I tried what does to aluminum and when I was trying to take it of used dishwashing detergent and ended up with little white spots everywhere. It also ate a chunk out of the aluminum.
it bonded into the copper, the white stuff was the alloy
 
it bonded into the copper, the white stuff was the alloy
I just have silver splotches it looks like water spilled on paper. Either way I switched that card over to MX-2 the R9 290X is still running the CLU.
 
Ok so now my R9 290 is having problems the card throws artifacts and memory BSODs before I even get into windows desktop. The damage is definitely physical and not related to drivers so does any one have suggestions on what I could do to fix it because while losing an R9 290 isn't as bad as losing my CPU I'd prefer if I didn't since I don't think that I'll have my RMA accepted after what I did to the card (completely disassembled cooler, replaced thermal pads, CLU left some marks on the core). I think that one of the memory ICs has damaged or shorted contacts because I do get memory errors or it could be one of the contacts from the core I don't think I damaged the core itself because I was being really careful with it. The ICs/resistors/caps and contacts near the core weren't that lucky is some got shorted out by the CLU but there isn't any obvious visible damage.
 
is 8Mpx enough?
Also the MX2 dissolved all the hardened CLU so the core is clean. Seriously I would love to know it what CLU is made of since it has the weirdest reactions to random materials.
PICS:

IMAG0383.jpg


IMAG0384.jpg


IMAG0385.jpg
 
looks like some CLU is still there on a few resistors on the PCIe side and VRM side of the core. I dunno that that is the problem, but otherwise, I don't see anything obvious.
 
CLU should be Gallium but I wondered if its Mercury, the alloy with CLU and Copper I had looked like a Mercury amalgam.

Anyway here is the datasheet:
www.coollaboratory.com/pdf/safetydatasheet_liquid_ultra_englisch.pdf


Edit:

There are possibly small traces of grey coloring on the circuit board around the GPU, in theory this could be conductive enough if it is CLU.
http://blogs.indium.com/blog/carol-gowans/indium-gallium

its really hard to tell from the photo.

I have cleaned the area around the core on my GPU with a very soft tooth brush and 92% isopropyl alcohol. You could try this.

CLU can be hard to clean up, it tends to leave grey smears.
 
Last edited:
If they use mercury it would melt below -49C and they wouldn't be able to sell it in CZ.

I'm going to get some 70% isopropyl and use that.
 
Last edited:
IRQL_NOT_LESS_OR_EQUAL

Raise the Vcore or lower the core clock by at least about 20 Mhz
 
So after I noticed that the MX2 gets rid of the white CLU residue much better than anything I tried I went and covered all the SMD components with it and then stuck the card into my Athlon II 370K rig to see if it fixed the card and it looks like it did I was able to run a full loop of Haven at around 50 to 100FPS (Because I'm using an AMD dual core) so now I'm gonna test the card in my main rig to see if it works fully or it's something weird about the 370K rig that makes it work.
 
Ok so the main rig boots and I can get to windows and I don't see any artifacts but nothing can identify the GPU. GPU-z is completely blank. The CCC says no supported hardware found. Device manager doesn't even display the tab where GPUs would typically show up. I also can't run Haven and the mobo's post code is back to being broken which it was even when the card worked. So any ideas on whats causing this because the card worked in the 370K rig enough to run Haven so it must be an issue with the MB now. I'm running the PCI-e at 2.0 right now.
I also didn't update th BIOS yet so that might have something to do with it.
 
Ok got the latest BIOS installed and damn it boots fast. Also the UI is IMO uglier than the orignal.
The both cards post in the BIOS. The POST LEDs still don't work and card number 2 is still invisible to windows. GPU 2 on it's own also doesn't work.
 
Ok got the latest BIOS installed and damn it boots fast. Also the UI is IMO uglier than the orignal.
The both cards post in the BIOS. The POST LEDs still don't work and card number 2 is still invisible to windows.
I wasn't too keen on the new UI either, but all the settings are in the same place, except for Clock Gen filter, which seems to have been removed.
Some things to make sure of if you haven't already: Rampage Tweak is set to 2 in the DRAM Timing section, PCIe is set to 2.0, both spread spectrums are disabled, blah, blah, blah.
Have you tried the problem card by itself in each of the x16 slots on the RIVE? Could be a bad PCIe slot. Have you tried raising VCCSA a bit with just the 290? Are you still overclocked on the RAM and CPU?
 
could you see the card in the device mgr of the 370k.?? Most likely no...Covering all the SMD components with MX must have fixed 1 of the CLU shorts.. must be another somewhere. Try a full clean.. again.. and reapply to the core only & see if it works. You might need to give it a heat gun treatment.
 
Didn't even think of that, good catch.
Should have checked but since CCC worked on the 370K system I assumed the card was A OK.
Well the 370K no longer boots. I get past the glowing windows logo and when the GPU drivers initialize everything fails.
EDIT: Booted safe mode at 800x600 pixels once I set 2560x1080(native res of monitor) everything failed again I'm using the DP port on the card.
EDIT: Any resolution greater than 800x600 fails. The card does show up in device manager for safe mode.
 
Last edited:
if CCC worked then moving the card from A=B must have F it up again :nutkick: Now for sure your going to have to reclean....
 
if CCC worked then moving the card from A=B must have F it up again :nutkick: Now for sure your going to have to reclean....
Yeah well at least I know there is one state in which it works I just have to get to that state again.
 
You know, you said it started working a little better when you forced PCI-E to run at gen2, for shits and giggles did you try gen1 to see if it made any difference?

I'm going to get some 70% isopropyl and use that.

Why not 91%? I tend to go with the stronger stuff granted I'm not always using it for computer stuff.
 
Last edited:
You know, you said it started working a little better when you forced PCI-E to run at gen2, for shits and giggles did you try gen1 to see if it made any difference?

Why not 91%? I tend to go with the stronger stuff granted I'm not always using it for computer stuff.

The MX2 does a better job at removing the weird white residues than alcohols so I suspect the stuff to be some by product of a reaction between the isopropyl I had during initial cleaning attempts and the CLU.
 
Ok I cleaned the GPU as much as I could and I can't see anymore white residue but I still can't get into windows 7 it crashes right after it tries to display my monitor's native resolution. When I was cleaning the core I noticed that the die is lightly scratched.

I can run it at 1024x768 any higher and it crashes. The device manager recognizes it so I guess that whatever handles display output on DP and HDMI must be dead/shorted I'm going to try DVI later.
Sometimes when running 1024x768 I get a band of green and purple tinting on the left side of the screen.
 
Last edited:
Try booting into dos & reflashing 1 of the bios could have been corrupted with all the s**t .
Did you give the core package some hot air gun while it was naked???
 
Try booting into dos & reflashing 1 of the bios could have been corrupted with all the s**t .
Did you give the core package some hot air gun while it was naked???
I don't have a hot air gun gonna have to borrow one or would baking the card be enough.
I think I'll do the BIOS flash first as that has the lowest chance of irreversibly damaging the card.
 
I think baking might be 2 much only the core might need the heat & probably not all that much. With a gun set @ med temp maybe 5-10 min.
but for sure bios 1st
Any luck with the DVI port
 
Back
Top