*First a note on results:  all the numbers presented are not meant to reflect performance, but simply as a benchmark to prove a certain level of stability.  For instance, you’ll notice the SuperPi times are relatively slow for the frequency as they are run on a completely un-tweaked OS, and the Vantage score is unimpressive as the GPU was left at or very near stock clocks.*

            For the most part, I would say the results speak for themselves.  I’ve been through many boards and chipsets with this same Q9450 including GA-X48T-DQ6, Foxconn BlackOps, ASUS P5Q3 Deluxe, and EVGA 790i FTW, and none of them have been able to yield results even close to those I’ve achieved with this board.  Not only was I able to clock higher; I was able to do it with less voltage.  If I may speak in such terms but for a moment, this board simply felt much more reliable, and produced consistent results.  For instance, I never had such an anomaly where a setting would boot once and refuse to post again, or boot after the 3^rd or 4^th try. 

I’ve always felt some sort of invisible FSB boundary with other boards and eventually grew to accept that I was stable at 8×495mhz and could only go up to 8×500mhz to bench on a good day if I spun three times to the left, sacrificed a cow, and stood on my head while tripping the power jumper.  This board seems to open up FSB limitations to let your chip shine, and unearths a dirty little secret of quad-core overclocking; vNB and vTT (an on-die voltage that directly impacts the chip’s fsb capability).  My overclocks seemed to scale almost directly with these two voltages (rather than vCore), and I would wager the same will be true with any FSB limited (non QX) quad.

With that tidbit about vNB and vTT in mind, we can formulate one reasonable explanation for the great performance of this board; improved voltage regulation circuitry.  As I laid out in the introduction, part of Gigabyte’s approach with this board (and the UD3 product line) was to utilize highest quality components.  This coupled with an improved design would mean cleaner more reliable power to the North Bridge chipset and across the CPU core, resulting in better results especially under high stress quad core load.

            When I said the results speak for themselves I didn’t mean to imply that I don’t have a lot to say as well.

            That being said, I’ll try to wrap this up.  With all the ballots in from almost every other major motherboard manufacturer (Foxconn Avenger P45: will it ever be released?), the GA-EP45-UD3P earns its laurels at or among the top of the bunch, providing unparalleled capability for those FSB starved 45nm quads, and truly humbling stability under tortuous voltages.  This board allowed me to take every HWBot record for Q9450 processor-bound benches, even at largely un-optimized settings and complete air cooling.  I know I haven’t yet found the limits of this board/CPU combination, and there is likely a subzero run in its future.

Pros:

• A vast and seemingly endless sea of FSB capability
• Quality components that stand up to torture
• Excellent BIOS refinement (nothing like the old Gigabyte)

Cons:

• Um…I guess a DDR3 variant might be nice?  Though with the very limited success of past P45+DDR3 boards I might eat my words
• Again, this is a superfluous and minor ‘con’ but a PLX chip for additional North Bridge lanes for Crossfire bandwidth would be…a plus

Verdict:  An all around great board; the final wonder of Gigabyte engineering for Socket 775.  Other benchers have shown it to perform brilliantly with dual cores as well.