GPU1 and GPU2 core clocks different

GPU1 and GPU2 core clocks different

I am having the exact same problem. Running two GTX 260's in SLI and GPU 2 is just sitting at Clock: 301 mhz, Shader: 602,  Memory: 100 mhz. It doesn't matter whether I unsync the GPU's or not, the sliders do not change the values (in the Precision Graph display) for GPU2.

GPU1 is overclocked a little from stock values. What is the deal? Any ideas?


Update*: GPU-Z is reporting the proper clock speeds for both GPUs (it reflects the settings of Precision), oddly though, on the "Sensors" tab, the clock numbers are wrong for GPU 2, and none of the voltage information is available (yet it has no problem reading the temperature). Is this perhaps a motherboard problem, or a problem with the video card? The fact that one of the cards is 55nm and the other 45nm shouldn't make a difference, should it?


Hold on, this could be a Mobo issue - the bus speed on the second card is at 8x when it should be 16x. This is a Tri-sli board, and I left a PCI-E slot empty between the two cards for cooling, etc. (motherboard gave no directions of which two slots to use for dual SLI). I'll move the card and see if anything changes.

MSI NF980-G65 Bios Rev. 1.0  (NF200 PCI-E chipset)
AMD Phenom II X4 955 BE  @ 3.575 GHz (216 FSB, 16.5x Multi - HT Link x12 @ 2600.0 MHz - NB Freq. x12 @ 2600.0 MHz)
2x2GB Corsair Dominator DDR3 1600mhz running at 1441 MHz 7-8-8-19-38 1T
eVGA GTX 260 Core 216 55nm Revision B1 (Bios 62.00.4C.00.50) Stock @ GPU: 626 MHz, Memory: 1053 MHz, Shader: 1350 MHz
 PCI-E 2.0
eVGA GTX 260 Core 216 65nm Revision A2 (Bios 62.00.1A.00.1A) Stock @ GPU: 576 MHz, Memory:   999 MHz, Shader: 1242 MHz
 PCI-E 2.0
Thermaltake ToughPower 850W PSU 
Windows 7 x64
Latest eVGA Nvidia Drivers 191.07, Precision 1.8.1

Okay, I moved the 55nm into the first PCI-E slot near the CPU, and the 65nm into the second. They are now both running at x16 and the card in the first slot says "Bus Interface: PCI-E 2.0 x16 @ x16 2.0", BUT the card in the second slot says, "PCI-E x16 @ x16 2.0" (Note that it does not show the card itself as 2.0). All of the stock clock speeds are confirmed by the first tab of GPU-Z.

On the Sensors tab of GPU-Z it is still only the 65nm card that provides voltage readings (VDDC) even though it is in a different slot now, yet the 65nm is now reporting those incorrect Core and Memory clock speeds (GPU:300, Mem:100). Whereas before it was the 55nm sitting in slot 2 that showed the slow clock speeds. I assume that since GPU-Z can see the clocks of each card, there is simply a Sensor problem on the Mobo? That still doesn't explain the voltage readings.

Does that 55nm Revision not come with Voltage sensors built into the card?

So I guess now the question is: Are the cards still being clocked the way the Precision sliders say (and the way Tab 1 of GPU-Z reports), and Precision is simply reading the 'broken' sensor? AND is it up to MSI to determine if the non "2.0" recognition is a Bios/Mobo issue, or is it possible the eVGA GTX 260 65nm is fuggered? (I didn't notice whether or not the 2.0 appeared on a different card before I switched them