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

X79-UD7 Revew w/ 3960K

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

stevenb

Member
Joined
Nov 7, 2005
Welcome, its a bit after NDA, and if you've decided on Sandy Bridge-E as your next platform, or your just interested in the new boards, today I have a good bit in store for you.


X79-OC... OOPS I meant UD7 Review.
To introduce the X79 platform, we must start with the CPU, for starters Intel will be launching their 3960X and 3930K, the 3 in the first digit is for 3rd Generation Core i7 Processor, so the 3960 is like a 960, compare it to Bloomfield, with 960 and 930, so these processors will have predecessors. So let's start with the CPU, on die we have a PCI-E controller much like Sandy Bridge, but with 40 PCI-E 3.0 lanes. 6-Cores along with cache and a system agent with an IMC. The IMC is quad channel, but also supports dual or triple channel. So as long as you put more than 1 stick of memory in there, it will optimize the memory channels. You can even mix two different sets of memory, and set their timings and voltages independently.

This platform is really feature packed, and it seems like most board makers are bringing out their A game, so let's see what GIGABYTE has to offer for X79. Can they produce another X58A-OC or even something better?

  1. Box + Acessories
  2. Board and Layout
  3. Digital PWM with a new VRM
  4. A Closer Look at the ICs
  5. BIOS
  6. Included Software
  7. Test Setup and the Board in Action
  8. Benchmarks
  9. Conclusion

Box + Accessories:
Nice white box, like an Apple product, personally I don't like Apple, but GIGABYTE did a nice job with the packaging.




Now if you're thinking, "That is a lot of accessories" you and I would be thinking the same thing.

Here is a close up of the Wifi/BT module:


The board:


The backpanel:


That two color button is for the BIOS switch, you can select whether you want to use the main or backup BIOS, also the small black button is for ClearCMOS.


Board and Layout

Yea the socket is really huge, but you can also see that the DIMMs are close as well, this is to reduce trace length. That is why we see DIMMs on either side of the board. On boards with more than 4 DIMMs its recommended to OC with the first 2 on either side.
You can also see that the socket has holes built into its bracket, these are standard sized, so you can screw things down into them. I was able to get my Antec Kuhler 920 and Hyper 212+ to fit without issue, I haven't tried other heatsinks, but most should fit.

Now here is something I found out when I took off the bracket, you can actually puncture a piece of plastic and make a through-hole:


Here is the bottom half:

Now you can see that there is good spacing between the bottom of the memory slots and top of the first PCI-E 16x slot, this allows for overclockers to insulate the GPUs easily, and leave room for the memory, and changing memory out. It does also make this board ATX-XL, but as i am told not the full XL length.

A lot of you are wondering about the area with the buttons:

OK so we have a whole OC area over here, infact other than the backpanel with the ClearCMOS, and the dualBIOS switch, you can just handle everything from right here. We have two sets of +/- buttons, the ones to the left are for the multiplier and the ones on the right are for BLCK. The "Gear" button allows for the BLCK steps to be changed from 1mhz to 0.1mhz. We also have a Port80H LED POST Code display with codes in the back of the manual. Power and reset buttons are located here as are the voltage read points. Some connectors for the voltage read points are included, enough for every connector to be read at the same time. Also note that there are pads if you only have one DMM. You can also see a USB 3.0 connector, located north of the PCH to be closer to a USB 3.0 front panel bay, like the one included. That SATA power receptacle is for extra PCI-E power, and there is one more too a bit south.


Moving down south we have a lot of SATA connectors, but you see 4 missing, those were SAS ports for SATA6G, and Intel removed it from the PCH before launch.
The four gray are Marvell SATA6G, the 4 black are Intel SATA3G, and the 2 white are Intel SATA6G. You can also see a nice PCH heatsink, it actually says GIGABYTE OC on there, so maybe it really is an X79-OC?

Located right above the front panel connectors is a 2-pin ClearCMOS jumper, just short these two pins out and accomplish what the backpanel button does without having to actually reach back there.



We see some good spacing between slots here, but what is most odd is lack of PCI slots. As they are not native to the PCH X79, and haven't been for a while, manufacturers have to add PCI on there own if they want. A lot of users actually welcome to move to make PCI obsolete, personally I don't mind having PCI.

Ok now we take that heatsink off!


Here are the heatsinks:






Digital PWM and the new VRM:

So GIGABYTE switched to Digital, so what does that mean? It means that the Intersil Analog VRD 12 PWM has been changed to an International Rectifier Digital VRD12 PWM. Well the move to digital has two very big differences, first of all Digital PWMs offer much more control over every aspect of the voltage regulator, from the phase control to the LLC, the PWM even has built in NVM to store profiles. Settings can be changed real-time through software, and updated through the BIOS as well. Now this is the new 8-Phase IR PWM:
IRF3567

The second big difference is that the error feedback loop is digitalized, and this means much higher precision.
Its 8 phases is 6+2, on this board all its 8 phases are used for just the CPU cores, but on the Assassin 2 its 4 phases for CPU cores and 1 of those +2 phases for the QPI/VTT VR. For comparison, the 8-Phase Chil PWM used on the Maximus 4 Extreme was 7+1 phases, on the Extreme all phases were used for the CPU Cores, but on the Gene the extra 1 phase was used for the iGPU, and 4 phases for the CPU Cores. There is a lot in common between ASUS and GIGABYTE PWMs right now, as International Rectifier bought Chil Semiconductor recently, and has since been putting out Digital PWMs with Chil technology, basically the same product under a different name. So now we have to look to the Rampage 4 Extreme and look to see what its VRM looks like without a heatsink, as the PWMs are basically the same.

Anyways GIGABYTE is using all digital 16+2+2+2+2 phases for the CPU Core+QPI/VTT+System Agent(IMC)+DDRA/B+DDRC/D
A lot of power. All of it Digital. GIGABYTE has employed two other 3+2 phase Digital PWMs.
IRF3570


GIGABYTE has switched away from DrMOS, as the doubling technology from Chil and IR consists of a solution that has integrated drivers on the doubler IC, this allows for better control of phase dropping while doubling as well as better monitoring. In theory at least. Here is how it works:


Each doubler has two drivers built in, so its 3 chips in one, a doubler and two drivers, single PWM input and double driver output.

GIGABYTE has also switched away from their older Low RDS(ON) MOSFETs, and has switched to much higher rated IR MOSFETs,
IRFH8318for the low-side and IRFH8330 for the high-side.


Now GIGABYTE is also using the POSCaps we saw on the X58A-OC, in the same amount as well, we have 19 of them at 470uF, the rest are on the back with the rest of the MOSFETs. POSCAP isn't just any normal tantalum capacitor, they are a brand name tantalum made by Panasonic, and they are supposed to have lower ESR and handle temperatures better than typical tantalum capacitors.


All in all a very well redone VRM, with Digital PWM and very nice component selection:


A closer look at the ICs:

Its not really an IC, but these dual connectors definitely can carry in some power.


Now this is the VR for the first memory slots:

The chip with the white dot is another IRF3750 3+2 phase PWM, really only 2+1 are being used, 2 for the two memory phases and the +1 for the QPI/VTT phases. the +1 is doubled by the same doublers as are used for the CPU VRM, and the 2 directly goto some new DrMOS type power stages. These are IRs pride and joy power stages, they are DrMOS direct competitors, these in particular offer 40A consistently at 400khz and with pretty high efficiency. IRF3553 40A power stages. These are also used for the other two memory slots and for the 2 System Agent phases. Each memory slot can get 40A a pop, that is a TON of power for a single stick of memory, almost impossible to pull half of that per DIMM, for comparison the UD5 has 1 of these for 4 DIMMs.

System Agent VR:

The second IR3570 3+2 phase PWM uses 2+2 of its phases, 2 for another two memory phases and 2 for the System Agent VR, basically the System Agent, and the twi memory VRs are clones. Just think about how much power is being given to the System Agent power plane, compare it to LGA1155 and its surprising upgrade. So is the spec, both QPI/VTT and System Agent voltage can each handle 24A and 1.4v during OC, that is what Intel claims.

Here is an up-close of these power stages:


PCI Lane Distribution:

There are a total of 40 PCI-E 3.0 lanes, all of which can be used for the slots, and they are. The first 16X and second 16X(8X electrical) slots are directly wired to the CPU, the first 8X of the 3rd 16x slot is directly wired. The second 8X of pins of the 3rd 16X slot and the last 16X(8X electrical) slot have their lanes switched by NXP PCI-E 3.0 switches.


To carry in some good transition, we have the switches above, along with our dual BIOS. GIGABYTE's new UEFI and 3D BIOS need a large 64MBit ROM, and we have two of them.


That is the typical iTE8725 which is found on most GIGABYTE LGA1155 boards, on those boards it provided LLC and DVID, on this board it provides precision overvoltage or something along those lines.

ITE8728F provides PS/2:

ITE8728F provides SuperIO capabilities like fan, voltage, and temp monitoring and fan control, if that wasn't enough GIGABYTE added a second SuperIO:

This Winbond SuperIO adds MUCH more fan and temperature monitoring, as well as the auto-OC button on the backpanel.
This motherboard has 7 fan headers!

Here are the dual clock generators:


The mother of all chipsets, the PCH X79, the most extreme consumer chipset:


Dual Marvell SE9172 provide SATA6G, 4 extra ports with RAID:


Audio, well how about an upgrade!

ALC898 is a 110 dB SNR HD and Blu-Ray capable 7.1 audio codec. Its an upgrade from the ALC889.

Did you upgrade us again GIGABYTE? Indeed with Intel NIC:


Now this is the Intel PHY that was released earlier this year, and used on some LGA1155 boards. Now I heard a lot of people ask for this, and here it is.

Finally to finish up this section, let's do a 3 in one:

We see a new name to the motherboard USB 3.0 game, the Fresco Logic FL1009, two of these provide a total of 4 USB 3.0 ports. At least we know the DMI wont be clogged, but here I have a theory on why we see X79 boards with less USB 3.0, and here it is, people don't really have more than 2 USB 3.0 devices. personally I have one, and its a USB 3.0 stick, and its pretty cool. To be honest if it wasn't a review sample I would not have bought it, only because it costs too much. Personally I would like to see some more USB 3.0, luckily there are a lot of PCI-E 1X slots if in the future you decide to add more.

There is also in the picture above the only analog PWM on the board, a single phase Intersil used on all LGA1155 GIGABYTE boards, it controls the IR MOSFETs.

Included Software:




3D Power Utility can be downloaded from GIGABYTE's website, it is being updated along with all the software.




BIOS Walk Through:



















Test Setup and the Board in Action!:





Now Digital comes with new LLC, so here is how it works:


I only had a few hours to OC, and it wasn't hard hitting 4.8ghz, but I had to leave C1E on for some reason. I was told BIOS engineers only have had a few weeks to work with the new boards, and they are working hard to get it in perfect condition. There are some minor issues, but for the most part there were no stupid boot loops, and OC recovery works (most of the time), and most everything works as usual. There are a few things that can be improved, and will definitely be, I do have a lot of hope, as so far its pretty neat BIOS. Very snappy, and keyboard is very responsive.
So this I hit pretty easily:

But more to come along with an SBe OC guide! No worries people, it will be out soon, I just need some more time.

BUT there are some HUGE changes:
vAbsolute for VCC(Vcore) is 1.4v instead of 1.52v on SB? but SVID still goes to 1.52v.
VCCSA(system agent) has SVID.
VCCSA and VCCIO can go upto 1.4v comapred to 1.2v on SB
VCCSA seems to be the IMC voltage, so use it for memory OC instead of VCCIO.


BenchMarks:











Conclusion:


All in all I was impressed by the BIOS, while I might enjoy the simplicity of non-UEFI, I do like all the control I have. Of course there is always room for improvement. What I really enjoyed were all the UD7's features. I used some low rated memory, but I was able to get some mixed kit up to 2400 in the BIOS, and I will see if I can get that to work and bench a bit. Of course I do have an OC guide on the way, and there should be some nice results in there. This board really is well decked out, and GIGABYTE has really given their whole lineup some polishing. They upgraded the audio, they upgraded the NIC, they upgraded the BIOS, and they upgraded the PWM and it seems like they gave all their boards the full treatment. The amount of spacing on this board is immense, and the feel of it reminds me of my X58A-OC which was sitting idle while I was messing around with its new younger brother the X79-UD7. The new Digital PWM and the new VRM components are very nice to say the least. I am not missing the 24-phase power, as this VRM seems to do the job not only well enough, but also packs one heck of a punch into 1/4 the area of the 24 phase while giving me more and better control over its aspects. LLC isn't broken, and I have faith in a lot of its controls.



Now the board should have some more USB 3.0, and it can also have some extra DIMMs, and then it can also not be ATX-XL. Those three things will deter a lot of users from buying this board, and if you are one of them GIGABYTE and others have other boards to offer. This board is definitely a benching board, built for extreme overclocking, with a very nice feature set. Sure it might live on a test bench, but it can also live out its days inside a nice decked out case (an expensive ATX-XL case at that), and be a fully functioning member of the community. While it has that benching board feel of the X58A-OC it also feels like a 24/7 board like the Z68X-UD7. You might say its a mix of the two but with some extra spice thrown in. This really is how GIGABYTE's LGA1155 UD7 should have been, and I am just glad that GIGABYTE is eventually making the change. One thing that might throw a lot of new users off is the immense amount of PWM options in the BIOS, but there is hope as the 3D BIOS does offer much easier and managed control over the PWM settings. This board is how a proper high-end platform should be done, from its performance to its build, it really is one heck of a board. Its also really fun to work with as well.



I have not yet been told the price of this board, but I am sure it will pop up some where online soon.
 
Last edited:
MSRP prices:
Assassin2 $ 429.99 – available later
X79-UD7 $ 389.99 – available later
X79-UD5 $ 349.99 –
X79-UD3 $ 269.99 –

Also GIGABYTE is upgrading their warranty to 5 YEARS on the boards listed above!
 
So the max you could get was only 4.8 Ghz?

That may be the max on the current bios. As far as I know gigabytes boards are not out at retail yet. I have seen a few reports of 5GHz on air being possible, but also reports of only 5.3-5.4 on sub zero. I'm guessing we need some time to compare settings and wait for updated bios revisions.

It's a shame this board is xl-atx, I had to get the rampage iv extreme instead of it. I would have much preferred this board but, my case does not have room.

Thanks for all the pics btw, made me drool.
 
hey yea 4.8ghz for now, deffinitly need a better BIOS, i have been told that they are working on it ASAP, and that there is no need to even use C1E anymore like I had to use.

There is also a difference with the CPu stepping, like C0 and C1, so that plays a large role as well.
 
Thanks for sharing your work Steven, nice board for sure, any idea why this top end board don't have 8 RAM slots like most others?
 
Maxed out my chip on RIVE @ 5481 sub zero and 5Ghz no prob under H20. Curious how UD7 would clock the same chip as it is prob $100 or more cheaper. ASUS BIOS support has been great (9th BIOS in 2 weeks), Gigabyte not so much. Nice to see GB finally get a proper UEFI though.
 
Not yet, will start a thread once it is live. Didnt get CPU until late and the amount of testing required on a board like RIVE is a little nuts so launch wasnt doable. Soon though :thup:
 
Maxed out my chip on RIVE @ 5481 sub zero and 5Ghz no prob under H20. Curious how UD7 would clock the same chip as it is prob $100 or more cheaper. ASUS BIOS support has been great (9th BIOS in 2 weeks), Gigabyte not so much. Nice to see GB finally get a proper UEFI though.
You are welcome to send it over. I can test it and sadly return it...if you really want to know. :thup:
 
Maxed out my chip on RIVE @ 5481 sub zero and 5Ghz no prob under H20. Curious how UD7 would clock the same chip as it is prob $100 or more cheaper. ASUS BIOS support has been great (9th BIOS in 2 weeks), Gigabyte not so much. Nice to see GB finally get a proper UEFI though.
Well that is because GIGABYTE BIOS engineers as i have been told has only had the board for 2 weeks as of now lol. That is why samples were sent out so late, and why its not available yet. I think they said its going to be available by end of this week.

I think I can OC higher, i have just been swapped, I only spent a few hours on max CPU OC, the issue was I had to use C1E enabled to get any OC with multis over turbo. That has been changed with a newer BIOS, as well as somthing else. I didn't realize that the switching frequency setting , which goes to 2.5mhz, is actually 1/5 of that in realtime. So i set 1000mhz thinking im maxing the PWM, but infact it was only like 200khz lol. But i measured later to test heat, and well it goes upto 420khz, which is optimal for higher frequencies, so i need to try that. 200khz is too low I think. So there is a lot i need to re-try.
What bios version did you use? Did you get the newest beta off the FTP?
Yea F3I or F3J? I haven't tried those. but they are different than the oringial, see the original for LLC still had rudimentary mOHM ratings like .6 and .8(.8 is standard load line slope).

I did a big debug report to GB, to help them so I am waiting a BIOS. I nhave to try the BIOSes off the FTp tho.
any idea of when the UD7 is going to be hitting stores?
end of week i think.,
 
Back