Overclock.net banner
1 - 13 of 13 Posts

Premium Member
13,266 Posts
Discussion Starter · #1 ·
The Z77X-UP4 TH Review: Thunderbolt with Quality for the Masses.
The Z77X-UP4 has been out for a while now and I decided to give this board a good looking over just so my readers get to see exactly what this very well priced board has to offer. The Z77X-UP4 is a very peculiar board, at priced just under the $200 USD mark, the UP4 provides users with the full dual thunderbolt experience along with a very high quality voltage regulator to keep your system going as long as your CPU will survive. When people build systems the motherboard is always the hardest thing to choose; whether it's because users don't know what to look for or what features they really need, motherboard hunting is a pretty intensive sport. So in this review I will make it simple for users to understand exactly what the Z77X-UP4 TH provides and how it will fit their needs. I will also give GIGABYTE's dual thunderbolt implementation a good inspection so that users can see the plethora of other components and the intricacy of the system thunderbolt requires so that they can compare against other boards.
That's basically how the board is hooked up to all the ports.

Box, Accessories, and Board:
The front of the box is keen on showing the user the quality of the new 60A IR PowIRstages that GIGABYTE is using as well as the 60A chokes. The whole idea of the Ultra Durable 5 is that these components are present. The 60A power stages are unique to only a handful of boards, and the 60A chokes are only unique to the Ultra Durable 5 line.

The back of the board talks some about thunderbolt, what it fails to mention is a really cool feature, and that is that with two thunderbolt ports, users can run a 4K display or even 4 monitors in a collage to create a large 4K display.

To keep the price low GIGABYTE only provides a few accessories. You have 4 SATA6GB/s SATA cables, a labeled I/O shield, an SLI connector, and the installation DVD and manual.

The motherboard really is a beaut. All the MOSFETs are heatsinked, something that users complained about on the UD3H. Here they really don't need heatsinks at all, however GIGABYTE is promoting Ultra-Durable 5 as a "no-air-flow-over-VRM-friendly" and the heatsinks add just a bit of protection. The heatsinks are more stylish compared to those on the UD3H, and should match a much wider array of accessories. You have 5 fan connectors all of which are PWM, however only 3 of them can be controlled, the other 2 are pre-programmed. This board carries a few interesting stickers, the first one is over the back panel and it indicates where the Intel USB 3.0 ports are so that users are able to use USB devices in an OS without drivers which are needed by VIA USB 3.0. The second sticker is over the SATA ports and it indicates that SATA port 5 shares its bandwidth with the internal mSATA port. Last but not least the last sticker indicates that that last PCI-E 16x(4x electrical) slot only works with Ivy Bridge CPUs, one should also note that means the last slot shares bandwidth with the CPU and not the PCH.

The back of the board is very interesting, I was very surprised to find that the heatsinks are screwed down, and they even have back-panels! This will help keep the board stiff in case you use a heavy cooler. This nice little touch is only you rarely see on a sub-$200 board.

The backpanel features
1. 6x USB 3.0 ports
3. HDMI and DVI-D
4. PS/2 Keyboard & Mouse
5. RJ-45 1GBit LAN
6. 2 x Thunderbolt/Mini Display Port
7. TOSLINK for 7.1 and S/PDIF
That is right, if you want you can use the TB ports as mini-display port what is very interesting is that depending on the cable, the outputs on the mini-DP differ.

You will notice that the VRM is really damn sexy, the chokes give that away. Here we have a true 6 phase VRM with the Ultra Power 60A chokes, the P-lightning bolt gives that away. The mSATA is placed right under the socket area.

Two fan connectors are stacked on top of each other which is good if you have two CPU fans, and a third fan connector is near the 24-pin power connector. There is a slightly distance increase between the DIMMs and the CPU socket for larger coolers, these DIMMs do feature T-Topology. Also a single USB 3.0 internal header is located north, right below the 24-pin.

You have just 6 internal SATA ports, all of them native intel, the two white ones are 6G. This board also comes with a COM header and the front panel headers are located in a good position. Another fan header is located there as well.

The PCI-E ports are as follows from right to left(or up to down):
PCI-E 1x
PCI-E 16x(16x electrical)
PCI-E 1x
PCI-E 1x
PCI-E 16x(8x electrical)
PCI Legacy
PCI-E 16x(4x electrical)
Under the PCI-E slots we have more USB 2.0 headers as well as a TPM header and the audio headers. The last fan connector is also here.

VRM Analysis:
This is my favorite part of most reviews, and in this case it is no different. Here we have one of the greatest voltage regulators to set foot on a mother; this is the mother of all 6 phase VRM designs.
In the image you can get a slight glance at the master PWM, the IR3567.

The IR3567A is the latest and greatest VRD12/12.5 Digital Pulse Width Modulator, it probably will still be around even on the next generation of core i7 boards because of that VRD12.5 certification which isn't even being used yet. What is interesting is that the IR3567 is one of the largest digital PWMs IR makes, the IR3563A(8+0) and IR3564A(4+1) are both much smaller, however because of the design of the 6+2 phases this PWM provides, it requires extra space for basically a second 2 phase PWM. What is most interesting is that all the phases here are being utilized in a perfect IR voltage regulator design. You have their latest and greatest PWM coupled with their latest and greatest power stages and you get one heck of a powerful yet efficient VRM. By using this PWM and the power stages, certain IR (International Rectifier if you are wondering) proprietary technologies are unlocked which otherwise wouldn't be available, also protection mechanisms are at their best with this design as well. This VRM is a 6+2+2 phase, 6 for the CPU VCore, 2 for the iGPU(use IR3550 too), and 2 more for the VTT/IMC.

Here we have the two sets of components that define Ultra Durable 5, the IR3550 (60A power stages) and the 60A chokes. The chokes are custom made, and 60A is a huge saturation current on these kinds of chokes, something you never see. However if you want a true 60A per phase then you need true 60A power stages and true 60A chokes, both of which GIGABYTE provides. This board can handle any overclock you throw at it thanks to this technology upgrade.

Here we have the slave PWM, and IR3570 a 3+2 phase digital PWM. This is a very common digital PWM on GIGABYTE boards, always used for peripheral power, this little beast spreads its channels across the width of the board to the VTT/IMC(VCCIO/VCCSA) phases and to the DDR power phases. Four of the 5 phases are used.

Here we have a typical high-end GIGABYTE memory VRM design, granted the Z77X-UP7 and all GIAGBYTE X79 boards have better power stages. We have two phases each made up of K03B7 as the low-side and K0393 as the high-side. Each of these two phases has its own driver situated in the IR3598 which is a chip of many functions. The IR3598 has two drivers built in, and they can work in either a phase doubling mode or in a dual driver mode as they are here. You will see this tiny chip on a lot of boards as it has many purposes. You also have a tiny RT9199 which is for the DDR termination voltage.

Here we have the VTT/IMC phases. Powered by an IR3598 and a K03B7 as the low-side and K0393 as the high-side both of these phases work to produce a voltage which is then divided down into VTT and IMC voltages. Because the IMC is from the same rail as the VTT you need to set the VTT 5mv above the IMC voltage to have the voltage change stick, this isn't such a bad thing as those two voltages don't do much and don't mind being close together.

Thunderbolt Analysis:
Thunderbolt is pretty interesting at its root; to date it is the highest bandwidth consumer external data technology bus. Here GIGABYTE used Intel's Cactus Ridge 4C controller, the most costly and rich featured TB controller on any motherboard. Here are the list of the parts:
DSL3510: The controller chip
TSP22980: 3.3v to 18v MUX, these are required to take power from the 3.3v output of the peripheral VR(which will be shown soon) and push it into the TB port for the cable and device power. Those cables usually want 12v, but this MUX allows any variation between 3.3 and 18v.
L04DP211: is a 2:1 Display Port MUX, While the mini-DP and TB ports are the same physically, electrically there are a few ports that are not the same on each and need to be changed depending on what type of connection is present, this chip is required to facilitate both.
PI3C3125: is a high speed bus chip, I have not seen this on any other board with TB, my guess(my only guess in the whole board) is that it is used for funneling bandwidth to shut down either one of the ports through cutting off PCI-E allocation. I assume it has to do with the hot swap functioning of the ports.
PI3HDMI: provides HDMI output to the TB controller, yes the controller requires HDMI as well as DP inputs. However this doesn't funnel bandwidth from the HDMI output, instead it takes from the DVI, so your backpanel DVI won't work if you use TB.
There are also two anti-ESD chips(in green above) that protect the board if you discharge static into the TB ports, something GIGABYTE likes to put on to reduce RMA numbers, a win-win for both consumer and manufacturer. Also TB has its own EEPROM(in pink above) .

Thunderbolt actually has its own VRM! Did you know that? This is two single phase VRMs put together, there are two PWMs(2x RT8120A) and a bunch of uPI MOSFETs and mini chokes. These VRMs provide power to the TB controller as well as the TB ports, through the two 3.3v to 18v MUX.
Below are some slides from the Intel Developer Forum 2012, I was there and learned a lot of Thunderbolt and its implementation, below are some thing s that you might be interested in.

Here are the requirements for TB, you can see that the parts provided fulfill these requirements.

Here we can see that "DP switch" between the thunderbolt connector and the DSL3310 in the image, you can see that the LSTX/RX which is a thunderbolt control signal, and two DP source signals are being switched depending on the mode of the ports. It is very interesting/cool stuff how they are able to change the pin assignment and function on the fly.

Here we have something even more interesting, the specifications of routing the signals on the motherboard. We see that the PCIe link between the TB controller and the PCH can only be max 12 inches, and that the DP signal (which is from the PCH) can only be 6 inches. Then between the TB controller and the port it can only be 2-0.8 inches, all of which GIGABYTE fulfills and does better. However there is one board from another company that doesn't follow this.

Circuit Analysis:
I got a little pre-occupied with Visio one night and decided to map all the ports and chips on the board in a high-level colorful diagram. It is almost like that in the manual except much more detailed, especially with Thunderbolt. This map shows all the little chips associated with TB and their routing, in case you were interested what all the extra chips around Thunderbolt do and how they work together.

Taking the heatsinks off this beat to show the glory.

Here we have a bunch of important chips. First we have 6 PCI-E 3.0 quick switches(L040838) which switch 2 PCI-E lanes each, you can check the block diagram to see exactly what and how they route to produce 8x/4x/4x from 16x or from 8x/8x. Then we have a PEX8605 and a common misconception is that the thunderbolt controller gets its bandwidth from this, however the TB controller gets all 4 lanes unobstructed directly from the PCH. This PLX chip provides only 3 PCI-E 1x lanes from a single input, and it is used to produce the 1x slots, it couldn't be used with this TB controller even if they wanted it to be. We also have dual BIOS, two 64Mbit BIOS ROMs provide redundancy.

ALC892 is the HD Audio Codec, it is better than VIA, however not as good as the ALC898. Audio caps are here and the routing seems to properly isolate the analog signals that go to the TOSLINK.

VIA Labs VLI800 is a 4 port USB 3.0 controller, it takes in 1x PCI-E and provides 4x USB 3.0. What is interesting is that while the Intel port will work in USB 2.0 mode without drivers, the VLI800 won't work without drivers. However Intel has no USB 3.0 drivers for Windows XP, and VLI800 does have drivers for XP, so they each have their own issues and their own strengths.

SLJC7 is a 6.7W Z77 PCH. I confirmed with Intel that the extra 0.6W of heat generated from the PCH when compared to the Z68 PCH is due to the embedded USB 3.0 controller . That is why the Z77 PCH gets a bit warmer than its predecessor.

RT8111F is a 1GBit NIC, it is a bit odd to see this now considering we don't see this used anymore on mainstream GIGABYTE boards. I guess they had a bit of the F version in stock, which offers some more enterprise features when compared to the RT8111E.

IT8728F is the common SuperIO that we see on all GIGABYTE Z77 boards, it provides everything from PS/2 to voltage monitoring and fan control. It also provides the bus for the COM port.

IT8892E provides two PCI outputs from one PCI-E 1x input.

Texas instruments GD75232 bridges between the SuperIO and the COM port to provide the serial port connectivity.

Two ASM1442 are used to change the native video output of the PCH(Digital Video(display port)) into HDMI or DVI. They are called level-shifter, such a cool name for such a boring device.

PI3PCIE2 is a PCI-E 2.0 quick switch that provides either SATA port 5 or mSATA with bandwidth.
That concludes out circuit analysis.

Now we will take a look at the heatsinks:
As a reviewer I look into things that others might miss, for instance the quality of the heatsink apparatus isn't just in the style or the material, but also in a few other factors.
Factors for heatsinks rating:
1. Plastic Push Pins or Metal Screws?
2. Thermal Pad on the PCH or paste?
3. Quality of the paste on the PCH.
4. Is the heatsink design effective?
5. Are all the VRM heatsinked?
6. Is there a backpanel on the heatsinks?
7. Amount of pressure between the VRM heatsink and the MOSFETs(the pad can tell you).

So for this we can see that the first 3 are fulfilled. If the PCH heatsink is removed and there is a void for the actual silicon then it is effective. The heatsink design is effective, all the VRM is heatsinked, a back panel is used, and we can see the imprint of the MOSFETs on the VRM heatsink's pads. This is one of the best heatsink designs we have seen.







G.Skill Provided this pretty sexy 16GB 2666C11 kit for some of our reviews, and today I will show you how the Z77X-UP4 takes on this monster of a kit.

It really is a beautiful memory kit, it garners a lot of respect in the OC community for high clocking.

First here it is right under 1333mhz, instead at 2600mhz, but running all 16GB at T1 and at cas 10, much better timings than at stock and just 66mhz slower. However this doesn't mean that this kit can't do XMP, in fact it can very easily, and it can go higher too:

Here we are at XMP timings, however we are running 133mhz faster, at 2800mhz. Honestly this was pretty easy.

Now to move onto some of the CPU overclocking:
Max BCLK on Air:
That is pretty high for air for this CPU, almost in the 109 range. Remember the Z77X-UP5 TH has the Z77 BCLK WR at 116.95mhz, actually it's my WR, so this board is from the same DNA.

Max CPU frequency is decent, honestly I think this CPU is degrading a bit, the UP5 OCed slightly higher, but honestly they OC about the same.
Now to some motherboard benchmarks, honestly what we are testing is the CPU, memory, SSD, and GPU, the motherboard BIOS makes the biggest difference, and it always changes:


Audio and SATA benchmarks:
Audio performance is "Very Good" it could obviously be better, however it also does use ALC892 which isn't as good as ALC889 or ALC898, so this is about right with ALC892.

SATA performance is very good. Here we used a HyperX 240GB SATA6G drive, blazing fast.

The Z77X-UP4 is a box of surprised, at the $189.99 USD at newegg it really is one heck of a great deal. While it does lack some of those external OC features (ex. voltage read points) which are on the UD3H, UD4H, UD5H, and UP5 it still has the same rock solid overclocking performance. This is partly due to its design, I have a hunch that when manufacturers implement Thunderbolt on a motherboard they are forced to make sure that the board signals are as clear as can be, that might be the reason why I am able to do such high BCLK OCes on the UP4 and UP5. Also the VRM on the UP5 and the UP4 also play a very big part in overclocking, the Ultra Durable 5 design reduces noise and increases efficiency, both of which help with achieving less voltage for the same clocks and higher frequencies. The UP5 has the highest output 8-Phase VRM ever on a motherboard and the UP4 has the highest output 6-phase VRM ever made.
The Z77X-UP4 TH is the base board that the UP5 is built on, and as so its price is lower and it has the same DNA. That is one reason this board is such a great value, you are getting very high quality components along with Thunderbolt technology. GIGABYTE's Thunderbolt implementation is top notch, and one thing that was just announced is that certain dual Thunderbolt motherboards support 4K display outputs, all of GIGABYTE's have been confirmed to do this. If that is what you are looking for then you found a board that can do that at a great price. However the UP4 is not without any grievances, for starters this board doesn't offer much connectivity over the basic Z77 chipset (other than an extra USB 3.0 controller); one thing to note is that there is no extra SATA or eSATA. However GIGABYTE does provide you COM/serial as well as PCI, two legacy outputs. I also would have liked to see a higher quality audio codec and NIC, closer to that of the UD5H. The Z77X-UP4 TH falls between the UD3H and the UD5H in terms of quality of audio and LAN. The lack of buttons, POST code, and voltage read points also prove to be cumbersome, however those issues can be remedied if you decide to bench with this board. If you are in the market for a Thunderbolt motherboard or just a motherboard with very high quality components, and you would rather not spend over $200, then the Z77X-UP4 TH is definitely worth a look.

Premium Member
15,257 Posts
just like all your others, another great review Sin

⤷ αC
11,278 Posts
Great review as usual

As far as the board goes: I do not know why both UP4 TH and UP5 TH do not have better fan controls (referring to http://imageshack.us/photo/my-images/402/14572914.jpg/). I suppose it's hardware related, they might not have added another fan control chip or monitoring chip? Right now it's the only thing really holding them back.

Found this:
http://www.bugtrack.almico.com/view.php?id=1899 , http://lists.lm-sensors.org/pipermail/lm-sensors/2011-November/034284.html

* Caveat: Nuvoton NCT6779D on ASUS Z77 Motherboards seem to have its own issues w/ Speedfan if you don't turn off Smart Fan (which means you might as well just leave it on Smart Fan and avoid Speedfan altogether).
Nuvoton Site (http://www.nuvoton.com/NuvotonMOSS/Community/ProductInfo.aspx?tp_GUID=86780baa-bdd5-4e25-8fca-5543fa919549) lists: H/W Monitor functions, up to 10 external voltage detection inputs, up to 6 temperature detection inputs, 5 fan speed detection and 5 fan speed control
The Nuvoton chip has issues in Linux: http://www.lm-sensors.org/wiki/Devices

edit: I love that you are one of the only ones to get down to the nitty gritty, chip level details with block diagrams & heatsinks off , kudos

Premium Member
13,266 Posts
Discussion Starter · #5 ·
Thanks guys.

Here is the iTE 8728F Datasheet: http://wenku.baidu.com/view/7b55fad726fff705cc170a78.html

You will notice it has up to 5 PWm fan controls, but that the pins are shared heavily, that means that if they want to use more fan control they can't use other, some very necessary functions, of the SuperIO. So basically if you want total control over all fan headers you have to add extra fan control ICs on the board, which this board just doesn't have. Or you can make it extremely basic in everything, and then have more fan controls, but some of those functions can't be used as a fan. For instance vcore_en signals that the PWm is operational or something of that nature, you need that hooked up.

⤷ αC
11,278 Posts
Originally Posted by Sin0822 View Post

Thanks guys.

Here is the iTE 8728F Datasheet: http://wenku.baidu.com/view/7b55fad726fff705cc170a78.html

You will notice it has up to 5 PWm fan controls, but that the pins are shared heavily, that means that if they want to use more fan control they can't use other, some very necessary functions, of the SuperIO. So basically if you want total control over all fan headers you have to add extra fan control ICs on the board, which this board just doesn't have. Or you can make it extremely basic in everything, and then have more fan controls, but some of those functions can't be used as a fan. For instance vcore_en signals that the PWm is operational or something of that nature, you need that hooked up.
You are super helpful in these matters, Sin.

Premium Member
13,266 Posts
Discussion Starter · #9 ·
sorry to the response a bit late. You might think its an ASUS technology because ASUS had it added into a roundup preview/review around launch, but that just isn't exact, just that they are using it on all their boards. I mean no one would claim Intel USb 3.0 as their own, would they?

T-Topology is an intel technology to route the memory traces, and it isn't some groundbreaking tech, it is just common sense which is used sometimes on different types of signals, even those that existed before it came out for majority of the memory topology. It is basically a split instead of a daisy chain as to support overclocking 4 DIMMs of memory higher.

Original spec for LGA1155 is daisy chain, but when they did Z77 they said you can chose the new topology not used before or you can use the daisy chain. Whether a company had enough resources to use T-Topology over daisy chain which was used over the years is another matter. Most companies want to get their boards to the stores earlier, longer shipping usually costs less, so the earliest you can send out your boards the better, ASUS had a huge surplus of Z68 boards and thus can take its time and also has the resources to do pretty well with the new topology before release. Since launch many high0-end GIGABYTE boards have had T-topology, but GIGABYTe just didn't advertise it.

I actually asked why and they just stated they want people not to think of the technologies, they want the results to stand for themselves. So they want people to wonder how they got the highest memory OC on Z77 instead of telling them we have this tech and its possible, and they did that with the Z77X-UD4H(Hicookie), as well as this http://www.techpowerup.com/162883/Ivy-Bridge-G.Skill-and-GIGABYTE-Make-DDR3-3000-MHz-Possible.html . Now 3000mhz memory OC isn't possible in 4 DIMMs without T-Topology.

So since launch a board such as the Z77X-UD3H does have T-topology. ASRock is only using it on their Z77 OC Formula, and GIGABYTe basically has used it on every board that has released since launch and that is higher ranking than a D3H.

Also a thing for thought, the T-Topology isn't always better than daisy chain, it really depends on what memory, what memory controller, how the traces are routed. For instance samsung memory OC is higher with 4 DIMMs(as opposed to 2) on GIGABYTE Z77 boards due to the T-Topology, it helps the signals coming from both DIMMs of the same channel to strengthen the final signal.

Premium Member
13,266 Posts
Discussion Starter · #11 ·
yea i will help you, no problem. I do have an OC guide here :http://www.overclock.net/t/1247413/ivy-bridge-overclocking-guide-with-ln2-guide-at-the-end

The same principles apply. Just set the vcore, then set LLC to turbo, and then pick a CPU multiplier. Enable CPU PLL OV and then disable power saving and turbo, unless you want the multiplier and/or voltage to drop during idle conditions. if that is the case then enable EIST and turbo, and then use offset vcore overclocking. But first just don't do that, instead disable all power saving.

PC Evangelist
48,332 Posts
Is the UP4 better then UD5?
1 - 13 of 13 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.