D601 error on XFX 4890

[xolotl]

I've just built a new PC with the following specs:

i7 920
Corsair Dominator 6gb
Asus P6T
XFX 4890 1gb XXX edition
Seasonic 700W PSU
Akasa Eclipse 62

It's been running fine for two days but now everytime I start a 3d application the system black screens and freezes. Reseting the PC is the only way to fix this.

I've tried reinstalling the drivers (9.6) but that didn't help. The card itself has never been overclocked.

The LED called D601 on the 4890 is lit up and from searching around this seems to suggest an issue with the VRMs on the card. As this light doesn't go away I'm guessing it's broken and time for an RMA? Any other possible fixes?

 

[xolotl]

This link here seems to suggest it's the VDDC:
http://forums.amd.com/game/messagev...d=97885&STARTPAGE=3&FTVAR_FORUMVIEWTMP=Linear

Diagnostic LED meanings for the HD4890 card ...


D601: VDDC_FAULT

D1601: critical temperature fault

D1602: AUX Hot Plug/Unplug Fault (upper connector)

D1603: AUX Hot Plug/Unplug Fault (lower connector)

What's interesting is that now that I've left it turned off overnight the LED is no longer illuminated. I'll see what happens after some long gaming sessions today.

 

[xolotl]

I decided to RMA the card (from a competitor) and it's still going through that process as they initially didn't find a fault.

Allthough I don't have the links at hand my conclusion from research seems to be that that 4870/4890 refrence cards have a design flaw in the voltage regulators. In normal cards under normal situations this doesn't cause any problems. However, when the all aspects of the GPU are used in super intensive applications such as the burning mode of Furmark or OCCT instability can occur. This is more likely on factory overclocked cards such as the one I own as they draw more power.

The cause of this seems to be insufficient capacity in the VRMs presumably to save money in the production process. While most games wont cause all aspects of the GPU to be used at once when something does the VRMs are forced to work over the design capacity. You can see this in GPU-Z as their temperature can hit 150c whereas they're only rated for 127c. When playing COD:WAW they only hit 90c.

If I get a refund I'm going to get a GTX 275. While this problem isn't show stopping I personally think a stock graphics card should be able to run any application without crashing, regardless of whether these applications have any use bar stability testing.