My system has two cards, driving 3 monitors (I do both 3D, 2D work and software development).   I'm pretty particular about my system as far as cooling is concerned and have plenty of airflow with decent exhaust and case fans (5 120mm).  My CPU and VCore at Idle are around 42°C, mainboard 46°C, PCI-E 1 & 2 @ 39°C and 37°C respectively.  When doing 3D rendering, of course things change a bit, but the i7-3930k remains stable on 100% load for hours below 70°C.  My room ambient temperature can flucuate "a lot"... up to 80°F during summer months (of course with all the other equipment I have in here that's no surprise), but still.. my system remains on average as stated and flucutates +/- 8 to 10°C (during summer).
 
BTW, I'm a computer "clean" freak...  I do not tolerate dust in my system, and will every other month pull it and dust it out.
 
 However, I have one issue that I would like help addressing.. and is related to my EVGA GTX 560 2GB SC card.. it never "rests"... and what I mean by that is, the GPU clock speed is always at max 850MHz even when only in desktop mode (no games, no DCC apps, nothing but desktop).  It is driving two monitors (a 27" ASUS PB278Q, and a Cintiq 22HD) while my other GTX 560 (PNY #:VCGGTX560XPB, 1GB) is driving a smaller display (I use for API reference, internet searches, notes.. etc.) pretty much runs at cool 39°C and mostly at around 51MHz.. it is adaptive and will raise to 400Mhz or maximum 810Mhz if I happen to play a game on that monitor.
 
The EVGA card, sitting idle... at 850Mhz is burning at around 60-65°C, doing nothing.  The other GTX card, 40°C (unless I play a game, then of course it jumps).
 
So, why is this?  I have no clue... here are the BIOS differences:
 
EVGA GTX560 SC 2GB: BIOS:: 70.24.18.00.64
PNY GTX560 1GB #:VCGGTX560XPB: 70.24.21.00.51
 
Is there a functional difference here between the two BIOS causing this power utilization on the EVGA card?  I'ld be willing to try an updated BIOS flash for the EVGA if I thought it would fix the hunger.  But I honestly do not know what more I can do about it other than to underclock the card.. which more than likely won't do anything.
 
 
My other option is to just get a single video card to drive all three displays.. perhaps a Titan, 780, or 770.  So, again, if anyone has some suggestions for things like games, photoshop and DCC apps (maya, 3dmax, blender, nuke, Unity, et al.).. I'm all ears.
 

I have an ASUS Sabertooth x79.  The EVGA card is in PCIEX16_1, and the other is in PCIEX16_2...  there's two x1 slots between the cards with about 1.5 inch gap between.  I modd'd a case fan to force additional air from the front to the rear of the case that flows over the two cards.  The chasis has a side "window" with a 120mm fan that adds additional outside air into the chasis.

I switched the Wacom 22HD to the secondary PNY GTX560 and sure enough, the EVGA clock speed dropped to idle.  Interesting thing is the PNY card is now driving two monitors and remains at idle even after restart.  I guess the EVGA GTX560 SC is unable to do adaptive clock state with more than one monitor.
 
Both GPU's are idle at 36°C @ 51MHz core clocks.. save some power and reducing heat.. yay!

I've noticed that the voltage fluctuates a lot on this EVGA GTX560 SC (#: 02G-P3-2069-B1)  (yes, I picked it up on clearance at BB)... If I sync the clocks and voltage to 810MHz at 950mV, I've noticed the EVGA card will sometimes drop to ~820mV (seems strange).  Is that normal?
 
anyway, I'm not quite sure if there is a standard clock/voltage for this card but in PrecisionX it shows it to be 1075mV @ 850MHz, while the PNY GTX560 (#: VCGGTX560XPB) is 949mV @ 810MHz.
 
I do not have SLI setup (never could get it working with these), but shouldn't they both the same?

yes, when sync is selected.
 
interesting, I thought these were the same GPU, different memory yes, but even in SLI it would adapt to the lower difference.  I know the EVGA is overclocked by factory, here's some info:
 
________________________________________________________________________________
Display device       : GeForce GTX 560 on GF114 GPU
Display driver       : 314.22
BIOS                 : 70.24.18.00.64
GUID                 : VEN_10DE&DEV_1201&SUBSYS_20693842&REV_A1&BUS_1&DEV_0&FN_0
Multi-GPU role       : master
Registry key         : \Registry\Machine\System\CurrentControlSet\Control\Video\{6BB98697-8FF6-4731-A98F-72012EB89AF6}\0000
VDDC controller      : Display driver
GPU2
________________________________________________________________________________
Display device       : GeForce GTX 560 on GF114 GPU
Display driver       : 314.22
BIOS                 : 70.24.21.00.51
GUID                 : VEN_10DE&DEV_1201&SUBSYS_0895196E&REV_A1&BUS_2&DEV_0&FN_0
Multi-GPU role       : synchronized with master
Registry key         : \Registry\Machine\System\CurrentControlSet\Control\Video\{1ADAC153-6CAB-44EF-AB48-15A2AF9AB661}\0000
VDDC controller      : Display driver
　

well, that's make reality sense versus my make-believe kind :)
 
hey, thanks for all the input bdary!

"I'm pretty particular about my system as far as cooling is concerned and have plenty of airflow with decent exhaust and case fans (5 120mm)."
 
I am new here and do not want to stir up feathers but using more fans does not necessarily mean better cooling, as a matter of fact more could be worse.
 
I worked for years as a tech in a major PC maker's thermal analysis lab that did Mil handbook reliability calculations and know for certain that every system is different, and position of fans, CFM of fans, venting, and component locations are all key to keeping systems uniformly cool.  A phenomena we called "thermal traps" occur when cables, hardware, vents and fans work against airflow causing localized hotspots that raise temps in small micro climates. Failures blamed on hardware could indeed be caused by fans, too many, too few, poorly placed, or the case improperly vented.
 
You have no idea how many cases I sawed, drilled, covered with duct tape, placed fans in, and set thermocouples in to find out that fans don't always cure issues and they can cause them.
 
The best (or only) way to add / use fans to determine their efficacy is to use many thermocouples set in a PC case.  This is a laborious undertaking that no case maker or individual could do well given the many variables in location of drives, type of power supplies, option cards and MoBo's.
 
If I give any... my advice would be that if intermittent failures are seen, try opening the case and setting a LARGE single fan at a cool room temperature outside the open case blowing the crap out of the inside of the case. If failures go away you can be sure that you have a thermal trap situation. Then the work starts. Maybe more, but also maybe less fans may cure issues.
 
 
 
 
 
 
 
 
 
 
 
 
 
 

by the way i've noticed it over a year ago and was told not to worry about it, card still runs fine :)