I had a similar thing happen to me, but it was an external hard drive.
 
I plugged in a rosewill 2.5" external that had 2 USB plugs, one providing only power. I have 4 USB ports on my front panel, and they go to 2 different headers. I plugged each usb cable in to a different header, and didn't think anything of it. My USB controller apparently fried. Nothing USB worked, and when I restarted I get a no keyboard error. I used a PS/2 keyboard and that let me get past the error... into a reboot loop.
 
RMAed and now I am all fine.

The voltages applied to the componants go through voltage regulaters that are on the board, but for the most part, the rest of your memory function is contained withen the chips, Your CPU has a controler, each chip has a controler, and the board containes sockets, and traces to connect the two. The regulator is a PWM( Pules width Modulation )  PNP ( Positive negitive positive )  transister, that has a gate voltage, The PWM stands for pulse width modulation, and this is controled by a clock on the motherboard. Incresing PWM will give you better voltages, but may drive up V reg temps. 800 seems to be the best setting for low temps, but you can try to bump this up, If you do watch your temps.

And now for another exciting episode.
 
I had pretty much resigned myself to living with the bogus system temp, but the RAM speed thing was still nagging.  After a bunch of fiddling (all 3 single sticks would run the stress test indefinitely, but all 3 together would splat in a minute or two) I noticed that with a BIOS specified 1.3 VTT, eEleet and BIOS health were reading over 1.4V.  Yike.
 
I realize that the set voltages and the reported voltages may differ, but 100+mV seemed like way too much.  In fact, I got better RAM stability (but not total stability) by undervolting VTT.
 
If I set 1600 in BIOS, the auto VTT would go to 1.3,  and the measured voltage to 1.41 or so.  If I manually forced VTT to 1.35 or 1.375 in BIOS, RAM was quite a bit more stable, but would eventually BSOD in the stress test.
 
This wide difference between set and measured voltage, along with the unstable RAM problem, and the bogus temp sensor led me to call EVGA again.  This time they didn't bat an eye.  I  got immediately bumped up a couple of tiers to a fine individual who's name I won reveal, but who's initials are SR   He immediately got another board on the way.  He also had a lot of info on things to try and to test to get RAM running at speed - but only after the new board was installed so board issues were ruled out.
 
These mother(board)s are bustin' my chops, but support remains stellar.
 
Stay tuned,
Z.