UD3R + i5 670 in CPU-Z

Gigabyte P55A-UD7 on liquid nitrogen

Add Your Comments

If you have not yet had a chance to check out my first P55 motherboard comparison, I’d recommend doing so before proceeding.  This is not a complete review in and of itself, but builds on the foundation established in that first article.  At that time, I only had one CPU worth testing, and that was an i5 750 which happened to be very strong with extreme cooling.  Since then, the new Intel 32nm CPUs have been released and have put a new twist on this game.

Gigabyte P55A-UD7

Gigabyte now has an extensive lineup of motherboards based around the LGA1156 socket.  The top range boards use the P55 chipset, and at the top of the heap is the new UD7.  I’ve already looked at the P55-UD3R, the P55A-UD3R, and the P55A-UD6…  Does this new entry have what it takes to be called the best?

The board

Imagine the feature that you want most on a p55 motherboard… Now do it again, and think of your second most desirable feature… and on and on.  Now take a moment and look at the photos below.  I won’t insult your intelligence by going through every single feature that you can see for yourself.  Suffice it to say, “This thing has it all!”

Well, I should at least mention a couple things.  As you can see, the board does support USB3.0 on the two blue colored USB ports, and SATA6 on two of the SATA ports.  The board supports 2 PCIe x16 slots for two way SLI or Crossfire, and also supports tri-SLI and tri-fire, but only in 16x16x8 configuration.  Beyond that, if you’d like to know more about this board, please visit Gigabyte’s website.

Overclocking impressions

Initially, the UD7 feels very similar to the lower end boards.  Physically, it feels different (much heavier), but when working with the BIOS and things like overclock failure recovery, it carries over most all of the family attributes, which is a good thing!  A very good thing, looking back to my previous article linked above.

Again, these P55A series boards are so nice to overclock on.  Set something and it sticks!  Overdo something wrong, it’ll try, fail, and recover itself predictably.  These boards are just so overclocker friendly!

Thankfully, the same can be said of the UD7.  It’s happy to attempt to do whatever you ask, and when an attempt fails, it’s pretty good at recovery.

I’d like to attempt to show you the capabilities of this board when pushed to the extreme.  Why?  Well, when components are pushed so far past the realm of normal overclocking, poor designs and cheap components quickly reach their limits.  Extreme overclocking is an excellent way to look for efficient, durable, and powerful components within a short time frame.  Since I already have a lot of data from the previous boards with my i5 750, I’ll give it a go on the UD7 to see if this CPU has anything more to give.  Later, I’ll test a new dual core “Clarkdale” CPU on the UD3R and the UD7 and compare notes.  So, let’s get to it!

Overclocking with a Core i5 750 (Lynnfield, quad-core)

Recalling my previous experience with my i5 750 on the UD3R, I was almost able to match or beat the other boards in the comparison in every benchmark.  I didn’t have a lot of time for testing and I was low on liquid nitrogen, but I really wanted to see if the UD7 had the muscle to outperform it with this CPU.  This particular CPU is currently holding on to 2nd place at HWBot in most every benchmark.  There is one 750 that is faster, and it happens to be a lot faster in every benchmark except for wPrime 1024M, where it held a time of 190.520 seconds.  Compared to my 750 with a time of 192.563 seconds, I was within range, so that’s where my focus went.  Within a relatively short hour or so, I was able to squeak by with a new time of 190.437 seconds.  Not much of a margin, but it clearly shows the high limit potential of the UD7.

P55A-UD3R + i5 750 in wPrime 1024M

UD3R + i5 750 in wPrime 1024M

P55A-UD7 + i5 750 in wPrime 1024M

UD7 + i5 750 in wPrime 1024M

Out of all of the CPU benchmarks ranked at HWBot.org, wPrime is the most CPU intensive, and actually represents a fairly good test of the stability of a CPU.  The increase is not substantial, but I wouldn’t expect one at this level of overclocking.  Small differences here mean a lot.  In addition, this is now the fastest time in the world for an i5 750 through wPrime 1024M.

Overclocking with Core i5 670 (Clarkdale, dual core)

Lynnfield has not been know for its strength in extreme overclocking, but since Intel released their new Clarkdale based dual core chips, they’ve been the talk of the town.  They are based on the new 32nm manufacturing process, and they scale to high speeds very well with extreme voltage and cooling.  The retail i5 670 sample I purchased happens to be a very good chip.  I’ve had numerous chances to test it in extreme conditions with both of these motherboards.

My first two sessions were on the P55A-UD3R before I received my UD7 sample.  The first test was fairly limited as I ran into some problems with the CPU’s tolerance to extreme cooling.  I quickly found out that the CPU would cease to function completely at temperatures below -106C and fail to recover from a cold boot (complete power cycle off and on, then boot) when below -65C.  So my initial results, while promising, did not show the CPU’s full potential.  A few days later, well known Taiwanese overclocker “hicookie”  published a “cold bug” modification which allows the CPU to better cope with lower temperatures.

After completing the modification, my i5 670 again went under the pot, and the modification worked as advertised!  I was now able to cold boot at a nice cool -135C.  I was happily benching at -140C to -155C, but then I ran into another little issue.  I could not seem to get the CPU voltage up high enough.  In the BIOS, the P55A-UD3R will allow the selection of up to 1.9V CPU voltage.  If that sounds like a lot, it is, but given enough cooling power, something I had in spades, these CPUs seem to handle the high voltage levels quite well.  The problem was that any setting over 1.76V or so would usually result in a reduction of actual voltage.  I was never able to consistently boot with very high voltages.

After receiving the P55A-UD7, the first thing I wanted to do was see how the i5 670 would respond to additional voltage.  So, I wasted no time getting my pot mounted up again and pushing this chip to it’s limit.  The UD7 had a commanding presence from the start.  I quickly found that the UD7 allowed me to bench with the pot full.  In other words, the CPU could handle all the cooling I could give it.  Liquid nitrogen has a static temperature of -196C, but my thermometer only measures down to about -178C, so  I would imagine that my CPU was actually functioning at around -185C to -190C.  However, the chip still refused to boot up after power off at temperatures below -135C.  So, lets look at the results.

UD3R + i5 670 in CPU-Z

UD3R + i5 670 in CPU-Z

UD7 - i5 670 in CPU-Z

UD7 - i5 670 in CPU-Z

As you can see here on both motherboards, this CPU was able to attain some incredible speeds.  The additional voltage and lower temperatures allowed the UD7 to out pace the UD3R.  Maximum validation on the UD3R was already impressive at 6656MHz…but with the UD7 I was able to push that to an astounding 6804MHz.

UD3R - i5 670 in SuperPI 1M

UD3R - i5 670 in SuperPI 1M

UD7 - i5 670 in SuperPI 1M

UD7 - i5 670 in SuperPI 1M

You can see a similar improvement with SuperPI 1M.  With the UD3R I was able to calculate one million digits of PI in 6.687 seconds at 6475MHz.  With the UD7, I was able to get it all the way up to 6552MHz and calculate every last million digitsof PI is 6.5 seconds flat!

UD3R - i5 670 in PIFast

UD3R - i5 670 in PIFast

UD7 - i5 670 in PIFast

UD7 - i5 670 in PIFast

Again, you can see a significant improvement with the PIFast speed.  On the UD3R I was limited to a CPU speed of 6448, which brought a PIFast time of 14.88 seconds.  On the UD7 I was able to push the speed all the way to 6580 for a PIFast time of 14.31 seconds!

The UD7 is obviously the superior overclocker.  This is the type of performance I was expecting from the P55A-UD6, and while Gigabyte did not meet my expectations with that board, this new UD7 more than makes up for it!  If you are looking for the best overclocking board for the LGA1156 platform, this UD7 should definitely be on your short list.

Closing thoughts

The P55A-UD3R remains a solid choice as a value proposition, but for users looking for the best in LGA1156 motherboards, this new P55A-UD7 rewrites the rule book.  It has all of the features one could want, paired with overclocking capabilities which put it in a class of its own.  I highly recommend the P55A-UD7!

Please feel free to post your comments below, ask any questions you may have concerning the product, or jump on the forums to tap into the wealth of knowledge here at overclockers.com!

Thank you, Gigabyte Japan, for the sample and for your support.


Leave a Reply

Your email address will not be published. Required fields are marked *

  1. Great job! You were right, it did validate my mobo decision :D
    Was 243 the highest bclk you could reach, or were you just going for overall CPU speed? I ask b/c I could boot with a low multi at 240bclk on air.
    Great job! You were right, it did validate my mobo decision :D
    Was 243 the highest bclk you could reach, or were you just going for overall CPU speed? I ask b/c I could boot with a low multi at 240bclk on air.

    I hit 265MHz on LN2...never tested bclocks with any other cooling ;)