• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

The Z87X-UD3H Review-Everything You need to know about the UD3H

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.


Nov 7, 2005
The Z87X-UD3H Review

Most of you who are familiar with GIGABYTE’s usual lineup know that the UD3H is usually the last in line to get all the latest Ultra Durable features that GIGABYTE has to offer, however with the Z87 chipset it seems that GIGABYTE has shifted this model to a higher price point, yet with more features than ever before and higher quality as well. So the UD3H is no longer the lowest board on the totem pole with 2oz copper PCB and all the other Ultra Durable features, it is somewhere closer to the middle. GIGABYTE has built, in my opinion, the best UD3H yet. This board features the Ultra Durable 5+ scheme, which entails all of Ultra Durable 4, but also higher ESD protection, 15u Gold plated socket, and the PowIRstages with full IR digital VRM design. This board also carries GBT’s newest UEFI and software package. This board uses VRM components(from power stages to capacitors) that are unheard of in the $200 price range, let alone the lower one it sits in.

Box, Board, and Accessories:

The box is simple in design, while Z77 was white, GBT has used black here. You can see that they really push on their quality, and honestly it really is upgraded from last generation, we will go over the specifics later.

The accessory package is similar if not identical to that of the UD4H as well as the previous generation UD3H. You get:

  • 4x SATA6GB/s Cables
  • IO Backpanel Shield
  • SLI Connector
  • DriverDVD/Manuals/Sticker

In my opinion that is really all you need, I would rather costs go into hardware than accessories, as you can’t upgrade the motherboard hardware but you can always buy more accessories.

Here is the whole board, the first thing you will notice is the color scheme which is truly improved along with the heatsinks. The capacitors have a very nice black finish (they are also extreme high quality 10K custom made capacitors: http://www.chemi-con.co.jp/e/tech_topics/pdf/new_cp_201304.pdf , if you compare to even ASUS, they only use this kind of quality on their ROG line, their normal “Z87 gold” line is only 5K)

Here is the backpanel, it isn’t very amazing other than it features all USB 3.0 ports. This might pose an issue for XP/Vista users as Intel USB 3.0 don’t support those two OSes. There are also only hubs and not controllers used on this board and all these ports go through hubs, for those users they must disable xHCI in the UEFI to make the USB 3.0 ports all USB 2.0 ports. GBT’s use of hubs for all backpanel ports was first due to user demand, but there is also a secondary usefulness, and that is that ports that go through hubs should be immune to the Intel USB 3.0 bug that everyone talks about.

  • 6x USB 3.0 Ports
  • 2x eSATA6gb/s(Marvell SE9172 shared internally)
  • HDMI
  • DVI-D
  • Display Port
  • D-SUB(VGA)
  • S/PDIF Digital
  • PS/2 Keyboard/Mouse
  • RJ-45 1GBit LAN
  • Audio ports

Here we have the top half of the board, and you can see how big these VRM heatsinks are. A huge complaint with Z77X-UD3H was that the it didn’t have total VRM heatsink coverage (not that it needed it), so for aesthetic value “super-size” heatsinks have been added. They not only cover all the VRM phases, but also they are longer than needed covering one of the NEC hubs and even the PWM, this board is over engineered in the hardware side which is a good thing. This board has an 8 phase VRM, and its true 8 phases as well so it should run pretty cool because of it. You also get a bunch of OC features and fan headers(total 6, with control over 5 with 3 controls). Also notice the USB 3.0 on/off port which is new.

You will notice this board has three full sized PCI-E slots, however only the first two are hooked to the CPU’s PCI-E controller. The last is hooked to the PCH. You might also notice the eight SATA ports, the 6 black are Intel controlled and the 2 gray are from Marvell’s SE9172. The gray ports are shared with the eSATA on the backpanel so you can only use either/or. You might also notice that there is an extra PCI-E power plug in the form of a SATA power receptacle. One USB 3.0 internal header also exists at the bottom, I think that this header is direct connected to the PCH, and doesn't go through hubs while I am pretty sure the on/off port goes through one of the hubs.

The picture above is self-explanatory, but what is interesting/ awesome for you is that GIGABYTE kept all these features. The Voltage read points are one that can be important, because now that all the voltages are integrated one cannot find these read points on their own instead they have to be routed to pins and vsense techniques are used to gather precise voltage to different CPU domains, like a PWM would vsense its output. The position of the POST code is rather inconvenient, but it is there which counts. GBT also added another OC feature: Single BIOS mode(otherwise known from the Z77X-UP7 as the dual BIOS disable switch). This is useful for faster OC recovery as well as trouble shooting issues due to Dual BIOS.

Just another shot of the SATA6G ports and PCH heatsink.

This motherboard uses a 6 layer 2oz copper true black matte PCB. I have gone over the port allotment above, Notice that the bottom 4x slot shares 2 of its PCI-E lanes with those other two 1x slots.


These are the heatsinks.

You can see there is very good contact between the VRM heatsinks and the board, you can even see the indentation of resistors as well as the PWM and one of the D720210 hubs, neither of which needs to be cooled, but they are because of the size of these heatsinks.

Now that the heatsinks are off we can see the board for what it is.

This is the motherboard block diagram, it might help when I discuss some things in this next section.

First off let’s postpone to take a look at the new 8 phase all IR VRM. There are no external drivers or doublers, and the PWM is an 8 phase PWM capable of extremely high switching frequencies which sadly are useless on this platform, the 8 phases should provide better thermals and ripple control just because of more true phases used. In fact if you look at the Ultra Durable 5+ lineup, they are almost guaranteed to use 8 true phases as GBT is only using an IR3563B. Since Intel integrated a secondary VRM, this VRM only needs to provide 1 input to the CPU, so no more 6+1 or 6+2 or 7+1 or 4+1(you get the idea) PWMs are needed. These chokes are the same used across the line, they aren’t the 60A chokes on the OC board or the past Ultra Durable 5 specification for Z77(the plus is the difference), but they should be capable of handling 30-35A which is more than enough for this platform as power consumption has gone down. Also the output voltage is higher than that of previous platforms, so the Duty cycle is also higher for Z87 and that means less general stress on the VRM, no more <10% duty cycles which put tons of strain on previous generation VRMs.

The PWM: IR3563B which is the latest VRD12.5 certified 8 phase International Rectifier Digital PWM capable of up to 2mhz switching frequency. That high switching frequency can be useful when you double phases as the switching frequency per phase is cut in half so this would allow a quadrupled VRM to run at a max 500khz per phase which is pretty high and would never happen. http://www.irf.com/product-info/datasheets/data/pb-ir3563b.pdf

The MOSFETs & Drivers: All integrated into the famous IR PowIRstages, based on the award winning (won an award for actually being able to do 60A on air) IR3550 which is a 60A capable part (50A at 90% eff.), the IR3553 is the capable 40A part which is pin to pin compatible with the IR3550, meaning you don’t need to change the PCB layout above to add in IR3550. Thus the IR3550 just has more surface area contact with the PCB which can sustain higher current output and at lower temperatures, but the truth is they(IR3550) cost an arm and a leg. So the IR3553 is the sensible part to put a board which has a cost under $400. It carries all the same technology as its famous big brother, while also performing well, with the ability to continuously output 25A at 90%, equal in power loss at 25A to the Texas Instruments NexFET D87350D http://www.ti.com/lit/ds/symlink/csd87350q5d.pdf which is the competitor to the IR3553. This use of the IR3553 http://www.irf.com/product-info/datasheets/data/ir3553.pdf is very impressive at this price bracket, especially because the D3H uses them (same VRM actually on D3H and UD3H, almost same as OC board except for the lack of 60A chokes).

Capacitors: Eight 560uF 10K Black solid polymer low-ESR can-type capacitors are used on this board for everything: http://www.chemi-con.co.jp/e/tech_topics/pdf/new_cp_201304.pdf . The new way of rating capacitors is with higher temperatures, so these are rated to run 105C for 10K hours. ASUS also uses 10K capacitors of this type, but they only use them on their ROG boards, on their mainstream PZ87 boards they use 5K of the same rating (even the deluxe uses 5K).

Inductor: 0.5uH Ferrite core low DCR, not 60A like the Z87X-OC.

The results above of are a D3H, and I didn’t make these images, put I did compact the idle and load into one picture, the original article is here: http://pctuning.tyden.cz/hardware/z...el-z87-vcetne-mereni-termokamerou-ii?start=11
I have a buddy over from a Czech Site and he did a round up review with the D3H. What is significant is that the D3H has the same VRM, exactly as the UD3H. The boards compared against aren’t the same price range as the UD3H, but they just show differences in design, and you can expect the UD3H to act like that on FLIR test.
Ales "froxic" Kanak from PCTuning.cz online magazine

The memory VRM is a bit less impressive, however still pretty good. Memory doesn't take much power, so two real phases is pretty good. Its based on the IR3570A which is a 3+2 phase PWM, and uses Renesas MOSFETs K0393 for the low-side and K03B7 for the high-side switch.

The Circuitry Analysis:

The Z87 PCH is a 4.1W TDP part which is built on the 32nm scale (or maybe even 22nm) which would account for reduced TDP and increased performance/features. It can support a total of 18 ports, a port can be considered a native PCI-E 2.0 lane of which one can have a max of 8, a native SATA6 output which you can only have a max of 6 ports, and a native USB 3.0 output which can have a max of 6 outputs. So you can’t do 8 PCI-E, 6-SATA6, and 6-USB 3.0, but you can do 6-6-6, or 8-6-4, or 7-5-6. That is the beauty of what is called Flexible IO, Intel’s new technology. Now you can’t configure this, but the motherboard maker can. This board uses what is listed below:

GIGABYTE also impliments their own on the fly FlexIO style with an extra MarvellSE9172 and an ASM1480 (PCI-E gen 3 to not bottleneck the SATA6 bandwidth).

You can only use the internal ports shown above of the red eSATA on the back, but not both at the same time. Either way I think most people would use the eSATA unless they wanted to not run something on Intel ports (such as a DVD burner). The Intel ports should outperform this SATA6G controller.

This is the Intel PHY i217v which has great power consumption numbers and compliments the Clarkville GBIT NIC/MAC built into the PCH.

ALC898 pictured above with a DRV632 which is a Texas Instruments Line driver/AMP, the results below are without this AMP, which would take up the performance. The reason is because this line driver/AMP only works on the front panel headphone output and you must use the internal header to get it.

Compared to ALC1150:

These are Renesas/NEC 4:1 USB 3.0 hubs, their use is on purpose to maximize the amount of USB 3.0 devices but also to remedy the Intel USB S3 sleep wake bug that might affect some USB 3.0 devices after wake from sleep. If you have Windows 8, this has also been remedied by hoxfixes from Microsoft. These hubs are able to wake the device/controller from sleep while native Intel port might not.

The SuperIO, iTE8728F, the fan options this time are much greater. You have control over all but 1 of the 6 fan headers. There are three settings in the BIOS, the first is for the CPU/CPU Optional fan, the second for the System fan 1, and the third for System fans 2&3, System Fan 4 has no user control, the board automatically sets this fans speed. You can just leave fan options at Normal in the UEFI, and then go into Windows and use the newest Easy Tune, configure the fans one time, save and exit and on reboot every time(even if you don’t load EasyTune), the fans will go to what you have set. This is a huge improvement from launch where one would have to go and load up EasyTune and apply fan settings to each header on every boot, which now is fixed.

This is a PCI-E 1x to PCI bridge chip, it is common, however do expect your PCI device to add some DPC latency as PCI isn’t native to Z87 and hasn’t been since Z77.

GIGABYTE is using dual 128Mbit BIOS ROMs with Intel's new Quad SPI Flash interface (pretty damn fast) to host their HD Mode BIOS, extra space is needed for HD Mode, and the BIOS ROM size was increased from 8MB to 16MB, I think GBT might be the only manufacturer who did this, as their BIOS takes up a lot of room with all its features.

Two TPS2546 provide USB fast charge capabilities for USB 3.0 on/off charge.

The new UEFI will only work with HD resolutions of 1920x1080 and 1920x1200 at the time of me writing this article, and more resolutions will be added in the future. The reason some setups wont support this is because of either the connection type (HDMI/DVI is best), or because some GPUs don’t support these two resolutions in DOS which is the same resolutions that would be supported in BIOS, it can differ from Window's capabilities with drivers. It still is being tuned I am told, however you don’t lose any features from not using the full HD mode. Just using the compact view which is actually in my opinion the better view as the fonts and text are larger and easier to read. GIGABYTE has made major improvements from launch date to all their motherboard’s BIOSes, they even added in an option to control Mouse speed which greatly improved mouse control in the UEFI to make it almost as good as Windows, you can see it under the System Tab. I have also shown only a few shots of the BIOS, which have my OC settings for the benchmark runs. If you want XMP to work only enable XMP. I have a video showing the BIOS here:

Test Setup:
Test setup and Configuration we have two new parts,
1. Diamond Multimedia HD7970 which replaces the GTX 570 in the test bed.
2. Corsair Dominator Platinum 2666C10(4x4GB) kit which only 2 sticks are used and 1600mhz 9-9-9-24 T1 is set.

(Z87X-UD4H review incoming and the heatsinks match the red of the DMM HD7970 pretty well)

Overclocking CPU and Memory:
However while the tests are run at 1600mhz memory to standardize, for OC performance below it is set at XMP(2666mhz) which it ran with all 4 sticks, all I set was XMP, I didn’t even set the divider and it went to 2666C10 with proper timings. That is how one should set XMP on GIGABYTE Z87 boards, as setting the divider will engage divider "auto rule timings" which will over-ride XMP specs. So ONLY set XMP if you want XMP, if that doesn't work then set the divider manually. I like to always set vDIMM myself to 1.65v(as per XMP of this kit).

Just a quick 4.5ghz run with 16GB of memory at 2666C10, the trick is that the Uncore auto increases to 40X and this can hurt memory OC, even at 4.5ghz setting 36x uncore to get higher memory clocks helps. If you leave uncore speed at default 35x it will turbo to 40x, so be mindful of this, it can cause higher rated memory kits to fail at XMP just because Uncore affects memory stability as well when you are trying to do both CPU and memory OC. You ca increase vRing to counteract this, but it will add heat, and I jut wanted a quick run. The performance hit for Uncore is very low when you OC on air, as long as you keep it around 35x-40X for 4.5ghz CPU speed you should be good. You need to weigh whether higher CPU OC and Memory OC outweigh higher uncore OC, and 98% of the time they will heavily outweigh uncore OC. We are just lucky Intel decided to unlock it!

Max OC of 5.15ghz at only 1.368v, just to see. I can push to 5.2ghz but requires more volts of course and showing 5.2ghz on air at 1.5v isn't very impressive.

Here we see that this samsung kit even at 16GB can boot at 29.33x divider with auto timings(with all 4 DIMMs mind you!), I didn’t touch these timings, just set 29.33x and high vDIMM(samsung can take over 2v on air for quick runs). I also found the max clock of all 4 DIMMs of the double sided samsung below, close to 3ghz. All on air.

Not bad for 4DIMM Samsung OC with 29.33x divider without any optimization other than vDIMM.

This is high BCLK of the UD3H with BIOS F6, I am pretty sure this is not possible on F5, but F5 has 1.67x divider working with 8x memory divider (due to special BCLK training which allows this kind of BCLK with stability!) . F7 BIOS should provide 1.67x support with this kind of BCLK, on F6 I tested I can do 118.7mhz*1.00x or 118mhz*1.25x (which is high enough for memory frequency WR). BCLK is useful for memory OC over 4ghz b/c the max multiplier for memory is only 29.33, so one must use 118mhz BCLK with 1.25x divider with 29.33x to get the WR for memory OC. Otherwise BCLK has little to no uses, other than 1.25x might be better on some CPUs for CPU and even memory OC. Just like with X79 I see that some CPUs prefer 1.25x divider opposed to 1.00x and might require less vcore with use of 1.25x divider.

As per the test setup:

Audio and SATA and DPC Testing:
The audio results are impressive as ALC898 is impressive even though other boards might use ALC1150, the ALC898 has a lot of benefits.

Here are the actual results.

Those results are possibly the highest I have ever ever seen SATA speeds including 4K and sequential.

DPC latency with EasyTune already open(which isn't required for fan control even) and playing a video with audio, this is good performance with F6 BIOS and all latest software and drivers. Please note that opening CPUz or another monitoring program will heavily increase DPC latency results.


App Center is where all your GBT apps will be located and you launch from there. You will have to install .Net MS Framework to get it working.
It has this live update feature, so when I install my OS I then install the NIC drivers, .Net framework, and App Center. Then I use App Center's built in "Live Update" tool and let it install everything from the rest of the GBT software to all the latest approved drivers.

The new @ BIOS is interesting.

The new version of EasyTune is much better than the previous and earlier versions, it now works for the most part.


The Z87X-UD3H is actually quite an impressive board, while on launch it might have had many more BIOS and software bugs, I would say 90% of those have been solved. Since no motherboard is perfect and its only 2 months after launch that is quite impressive. Right now if one uses BIOS F6 or any later BIOS or even some F5 betas, one will get very good mouse control in the UEFI. Pair with the latest EasyTune release from the site and the App center software one can actually just install the NIC driver and app center(with .net) and the software will update all your drivers and GBT software to latest. It is working well, and is how I installed everything for this review and I have used the system for a few weeks to get a feel for it. The newest EasyTune really brings fan control to its best as there is not even a reason to launch EasyTune for the settings to auto apply at launch, no need to make BIOS changes for this to work (in fact don’t if you will use Software to control the fans). The overclocking results are impressive as well for air cooling, especially the memory which I can do even better, but also the BCLK is pretty impressive and the 1.67x divider will work soon too, if you need 1.67x divider just flash to BIOS F5 and it will work, just F6 brings in new BCLK training procedure which is a step in the right direction.

The VRM and overall circuit quality on this motherboard is extremely high, they even use the thicker gold plated socket pins that are usually only found on high-end boards. I have already raved enough about the VRM in its own section, so I will just leave you with this thought, in this price range, even currently at $159 at newegg and with some combo deals at microcenter, it is definitely worth looking into if quality is your number one concern. You are getting the most quality for the price, even its undiscounted price at microcenter is justifiable. This board really impressed me for what it is, especially because most OC efforts have been put into the Z87X-OC and not the rest of the line when it comes to BCLK and memory OC, yet this board still performs well and you even get the memory OC profiles. The next step up is the UD4H and all it offers over the UD3H is 8 more phases of the same quality and red color. In my opinion if you like Blue get the UD3H if you like Red get the UD4H, the phase count wont affect your OC. To let your mind rest a bit about the phase count difference, think about it this way, the UD3H uses the same IR3553 PowIRstages and in the same quantity as the Z87X-OC.
Nice right up... only thing that kills me is that there is only 1 PWM fan header on the entire board and the CPU opt header doesn't even function as a duplicated fan header with separate RPM sensing it is just totally a voltage only control header NO pwm signal.

Otherwise the board rocks.
Nice right up... only thing that kills me is that there is only 1 PWM fan header on the entire board and the CPU opt header doesn't even function as a duplicated fan header with separate RPM sensing it is just totally a voltage only control header NO pwm signal.

Otherwise the board rocks.
True, but I want to retest and see if easytune's calibration is meant to combat this since the system should be able to still sense the RPMs, the calibration is done with the fans and gives %s per RPMs. Either way voltage mode will work on everything, a lot of users till use the 3-pin header.
True, but I want to retest and see if easytune's calibration is meant to combat this since the system should be able to still sense the RPMs, the calibration is done with the fans and gives %s per RPMs. Either way voltage mode will work on everything, a lot of users till use the 3-pin header.

Only 2 major drawbacks to not having more than 1 PWM header.

The lowest speed of the fan will be noticeably higher than compared to that fan being controlled in PWM mode.

You have a limitation on current and the amount of fans that can be run on a channel. PWM has no limit to fan current or the amount of fans as the voltage can be supplied by molex and the PWM signal can be split to at least 5 fans before it starts to have issues but usually many more than 5 can be ran.

I def agree that many many people still use voltage controlled fans... but this isn't a good thing and only one header is a step in the wrong direction not the right direction. Every single header or at least 3 of them with 2 separate channels should be PWM. The PWM headers can still control 3 pin fans in voltage mode.

I still think the board is great... I was a little disappointed that they used 4 pin headers on channels that did not support PWM but that was explained to me... The 4th pin outputs a 5v signal that forces the PWM into 100% duty. That should be a non issue for 99.9% of PWM devices but apparently some do exist that do not run at 100% when no PWM signal is present. The D5 PWM pump would be an example but I know of no fans that do that.

So it was a little misleading to me... I say a ton of 4pin headers and was like HOLY COW!!! I thought I had an early birthday or something... was like wow finally someone gets it. I know that adding PWM control ups the BOM but seriously all Z boards need 2 things in common... at least 2 PWM channels and an external temp probe port that can be assigned to a fan channel. So that us water coolers can assign a fan speed to a water temp. Instead the only boards that do that are the griffin and sabertooth and not boards like the UD5H, OC, MSI M-Power, Asrock OC Formula, ROG etc... all of those missing the thermal sensor.

Just my rant... I still think it is an amazing board and I know a lot of people love the griffin and sabertooth but can't stand that you can only get those useful features on boards the have a lower power design than other boards that are lacking the feature even from the same manufacturer. It is like the rampage III had external probes but the newest high end board doesn't.


Board manufacturers need to step it up and consumers need to as well. Most people still don't really realize all the benefits of PWM which might be the reason board manufactures paid attention to it at first but now seem like they are letting it go by the wayside. I mean the first few bios revisions from Gigabyte didn't even have a function PWM signal on the PWM header. They where controlling fans in voltage mode at least with the UD5H. Nobody noticed except me when my pump wouldn't increase or decrease in speed.

Just wish this type of info made it to the people designing these boards.

One thing I do love a lot and hope it catches on is that one of the USB 3 connectors where at the bottom of the board and the other up near the memory. This is the perfect way to do things :)