Just want to share my dead 570 experience. I bought the Gigabyte GTX 570 OC Windforce 3x (GV-N570OC-13I) last friday. It comes factory overclocked @ 780 core/1560 shader/1900 memory. Default voltage is 1.00V. With these settings it worked fine under Furmak, Kombuster, Unigine and Far Cray 2 ( i never disabled OCP).
Then overclocked it to 850/1700/2250 keeping the voltage at default 1.00V and in less than 3 minutes during furmark run my card died (temps were 65-67 degrees. I didnt check the PCB since I will RMA it today. But im sure its a burnt VRM.
From the looks of it I am the only one whose card died at factory default voltage of 1.00 (with OCP on) despite having the Windforce 3x custom cooler. So I did quite alot of research and found out that (between custom 570s);
- Gigabyte 570 OC Windorce 3x and MSI 570 TwinFrozr II both use the reference PCB design with 6 phase VRM (4 for GPU and 2 for memory). They simply put a custom cooler on top. MSI cards have similar incidents reported
- Gainward 570 Phantom, Gainward 570 GS Goes Like Hell, Palit 570 Sonic Platinum all have 8 phase VRMs (6 GPU and 2 memory). Same amount with the GTX 580. All these cards have identical PCB since they belong to the same graphics card manufacturer group
- Asus GTX570 DirectCU II also has 8 phase VRMs (6 GPU and 2 memory)
And so far I havent seen similar incidents with any of these cards on the net. Which really brings me to the conclusion that this problem is only for the reference design 570 PCB with insufficient VRM phases. Based on the custom cooling abilites of my card, I am positive that temps have got nothing to do with this problem.
FYI, the GTX 590 has 10 GPU VRMs (5 per GPU) and already someone has blown one. Clearly 6 phase looks like the best solution since GTX 580 seems to working fine for everyone.
If I can get the RMA replacement, I will try to get a Gainward 570 Phantom instead and I will keep you posted on how it goes.