GA-X58A-UD5 Revision 2.0 the MOST in Depth Review/ Analysis
http://img257.imageshack.us/img257/6021/ud5custom.jpg
The most in depth analysis of the GA-X58A-UD5 Revision 2
Before we begin, if you have a small screen or perfer PDF for bigger pictures here ya go! Download
Navigation:
Introduction + Box and Board Layout
Power Delivery at its Best(not so boring!)
Motherboard Cooling Analysis
Tylersburg X58 IOH+PCI-E
Connectivity
USB, IEEE, Ethernet, Audio AKA Peripheral Sub-Systems
VARIOUS IC SECTION!!!! (VERY INTERESTING)!
UD5 Rev 2.0 Voltage Read Points
Various IC SECTION Continued
BIOS/ Overclocking/ 3 Levels LLC
Conclusion
Introduction + Box and Board Layout
Intel’s Tylersburg (x58) LGA 1366 platform has been around for a few years now, long enough to have matured into the high-end choice for gamers and overclockers. The main features of the Tylersburg platform that set it apart from its predecessor LGA 1156 are its support for i7 900 series including the new 6-core processors, triple channel ram support(backwards compatible with dual channel), and native support for 2-way(16x,16x) and 3-way sli(16x,8x,8x). In an age where processors are processed at the 45nm and 32nm varieties overclocking has reached a new level with the i7 900 series processors. It seems that every single 900 series processor is able to reach the golden frequency of 4000mhz(4ghz), every board is able to boast at least 200blck, but above that is up to board construction, quality, IOH/CPU IMC luck, and overclocking ability. To begin my review of the X58A-UD5 Revision 2.0, I would first like to explain the major differences between revisions 1.0 and 2.0 of the board.
http://img444.imageshack.us/img444/3...visiondiff.jpg
Here is the box including up-close picture of how to tell the revision of the board:
http://img526.imageshack.us/img526/4...xcustom.th.jpghttp://img88.imageshack.us/img88/144...xcustom.th.jpg
http://img87.imageshack.us/img87/618...ecustom.th.jpghttp://img151.imageshack.us/img151/7...0custom.th.jpg
When you open the box you get some basic accessories and manuals/discs:
http://img23.imageshack.us/img23/614...gcustom.th.jpghttp://img202.imageshack.us/img202/1...2custom.th.jpg
When you open the box you see you have sli 2x and 3x bridges, as well as four sata cables, some have 90 degree heads for easier attachment.
http://img829.imageshack.us/img829/8...scustom.th.jpghttp://img151.imageshack.us/img151/3...acustom.th.jpg
Then you have the beautiful motherboard in anti-static
http://img151.imageshack.us/img151/2...nboxcustom.jpg
Next we move to general board design. Right now the standard for motherboards is a 1Ounce copper PCB, but gigabyte doubles this to 2oounces of pure copper in the ground and power layers, this lowers impedance by 50% and reduces overall board operating temperature. When you have so many mosfets, ferrite chokes, solid state capacitor, various ICs that generate all that heat you need all the cooling you can get, and while the benefit of extra copper is minimal every little bit counts. I would like to mention gigabyte calls this Ultra Durable 3 design, and the whole gigabyte x58 line incorporates the same general added copper and components.
http://img263.imageshack.us/img263/4...ppercustom.jpg
The board:
http://img813.imageshack.us/img813/6...rightsmall.jpg
The back panels is perfectly laid out, and gigabyte was nice enough to allow the use of PS2 keyboard and mouse, USB, eSATA, IEEE, audio connectors, and LAN are all contained on the back panel.
http://img19.imageshack.us/img19/851...lcustom.th.jpg
When we move closer to the CPU socket we see that there is the standard 8 pin 12v power header for CPU power, where do you think 16 phase power plant gets all of that 12v power? As we move along we can see that the ferrite core chokes and very nicely lay in rows parallel to the mosfet driver heatsinks, as well as to the CPU socket. What I really like about this board is its use of the Lottes made socket, opposed to the cheaper Foxconn socket which had terrible problems. The problem is that it would not apply enough pressure, causing in some cases the pads on the bottom of the processor to burn and in more cases for less ram than installed to be detected. The Lottes socket is not only more fashionable in its anodized black/silver casing, but also makes sure all of the 1,366 processor’s pads touch all the motherboard’s pins. Everything that could be anodized has been, such as the heat pipes as well as the heatsinks for the board. The ferrite core chokes match with their own shiny casings.
http://img146.imageshack.us/img146/6...oolersmall.jpg
Next we move on to the 6 DDR3 ram slots. This board supports up to 24gb of ram, but it’s up to your processor to handle its speed. Now that the internal memory controller is in the processor, the motherboard plays a very small role in ram speed.
http://img207.imageshack.us/img207/9...tssmall.th.jpg
Nest we move onto PCI-E spacing and slots:
http://img259.imageshack.us/img259/8...iesmall.th.jpg
Next down the bottom right hand side you have all 10 SATA ports at a 90 degrees angle pointing away from the board, this makes it very easy to fit an XL GPU. The portion of the motherboard heatsink is low profile where it reaches the Southbridge, so that you can fit large GPUs without obstruction.
http://img143.imageshack.us/img143/9...tysmall.th.jpg
Throughout the CPU socket area as well as right next to the Northbridge and ram slots we have intricate arrays of tiny LEDs. Some are setup to show you how many power phases are on for the CPU, NB, and DDR3, those are blue. You also have others telling you temperature ranges for Northbridge and CPU, and even more telling you voltage ranges for CPU, NB,SB, and DDR3 the colors range from green to yellow to red, depending on range.
If you would like to see the LEDs please move to the Board IC section where I go in detail about the monitoring chip.
Quick switches are very important to people who overclock and tweak their systems, as they are a built in way to power on, reset, and clear CMOS. While switches built in for power and reset button are nice, this function can be taken care of by any conductive metal to the pins for the case attachments. Clearing CMOS is another story, in the past CMOS has to be cleared by removing battery, and if that was not a hard enough reset, then you had to move jumpers, and if that was not enough you had to short out the batter while moving jumpers, all of that is now taken care of by a nifty little switch on the back panel.
http://img228.imageshack.us/img228/5...rcustom.th.jpg
This board has four -3pin fan headers and two -4 pin PWN controlled fan header. They are very easy to use and very well placed. The 24-pin motherboard power plug is in its usual place, the power and reset quick switches are right above it at the top right of the board.
The battery for CMOS is placed between the SATA ports and heatsink.
http://img823.imageshack.us/img823/9...rysmall.th.jpg
While the bottom heatsink looks kind of worthless, it is anything but, but that we will discuss later. The IDE port is placed the at the bottom of the board, along with the case headers, USB ports are right next to the IDE port which is right next to the floppy port, which is pictured below.
http://img574.imageshack.us/img574/5...pysmall.th.jpg
This board has 2x PCI-1X ports, but the problem is one has to be a low profile card, because there is the Northbridge heatsink blocking anything longer. The second PCI-1X port will do just fine with a normal sized card. Then we have PCI-E 16X ports , 4 of them to be exact, but if you have 3 double slotted GPUs, then only 3 are usable, as the fourth is right between the two 16x PCI-E ports. The PCI port is right in-between the bottom PCI-E port and the second one.
http://img843.imageshack.us/img843/8286/pcix1.th.jpg
Board LEDs:
http://img228.imageshack.us/img228/5...owercustom.jpghttp://img52.imageshack.us/img52/568...oframcusto.jpghttp://img545.imageshack.us/img545/4...tempcustom.jpghttp://img508.imageshack.us/img508/7...yramcustom.jpg
Power Delivery of the UD5 rev 2.0 At Its Best (Very Interesting)
http://img713.imageshack.us/img713/5...powerphase.jpg
The UD5 is one of the Gigabyte boards with “Unlocked Power”. Every 1366 board has to be VRD 11.1 compliant meaning that it has to incorporate vdroop, and has to be able to supply a dynamic current increase of 100amps from 45amps at idle to 145 amps at load, that is at max vid of 1.375v and full load, in this condition the processor would have a TDP of 130watts at 80% load. There are also 35 more amps that are needed to supply power to the Uncore part of the processor including qpi link, ddr3 memory, and the internal memory controller. The purpose for vdroop is to lower voltage while the dynamic current increase occurs so that the processor can still stay within thermal spec. Heat and work are calculated from the simple physics equation P=IV (power= voltage x amperage) so if you raise amperage you need to lower voltage to maintain processor integrity.
Another unique feature of the phases is that the system doesn’t need to use all 16 phases at once, it can use just as many as it needs, Gigabyte states that this prolongs the life of the phases.
What is great about these new x58 boards is that they use top notch components, including but not limited to low core energy loss ferrite core chokes (the shiny cubes that count 16 around the CPU area), solid electrolytic Japanese capacitors with 50k hours life, and low RSD mosfets. Everything gigabyte uses is top notch quality; they implement DrMOS Vishay made mosfet drivers for each stage. These mosfet drivers are the SiC769CD they can perform vcore switching at 1MHz, have integrated generation III mosfets and are DrMOS compliant. They are pretty famous too.
http://img94.imageshack.us/img94/717...powerstage.jpg
At 90% efficiency each phase will change 12v of power from your PSU to 1.2v - 1.9v for the processor at 20-35amps at 1MHz. Totally wattage per phase is calculated from P=IV (power = voltage x amperage) so about 24-66.5 watts per phase, and that is still at 85%-90+% efficiency, those numbers are outstanding, these mosfet drivers are very capable, even at 8 phases you will be able to provide 192 watts of 1.2v power to the processor, but with the UD5 rev 2.0 you get double that so 384 watts! That is the least amount of power output from these phases. If we drop efficiency to 85% each phase can produce 32amps of power at maximum processor voltage of 1.9v @1MHz frequency, for an outstanding 972 watts available for the processor. Dropping efficiency lower would yield even more power, but you do stress out the mosfets, they generate more heat, and don’t live as long. You want to be around 85- 90+% efficiency so the voltage is stable and heat generated is below 100c. But these mosfets can take 150C operating temperature until thermal shutdown occurs, and at 135C the red flag is removed, and actually perform best from 20C-50C and then efficiency slowly declines.
A i7 900 series processor at 4.5-5.0 ghz wouldn’t even need 300 watts of power, but you are safe with over 900 watts at 85% efficiency which is just phenomenal, at 90% efficiency which is hard to take lower because power demand won’t surpass power delivery. Each phase in theory can provide 35 amps at 1.9v which would be the maximum. These phases do not operate on their own; they have their own controller and a very capable one at that. Gigabyte uses an Interstill Phase Buck PWN controller for these 16 phases, the ISL6336G, the UD9 has 2x ISL6336G.
http://img152.imageshack.us/img152/1...werpwnforp.jpg
What is special about this chip is that it allows you to run as many phases as needed, as low as 2 phases up to 16, but uses 8 phases at a time for normal operation this is possible because of a seperate iTE phase switching chip in the Random IC section. The PWN controller is responsible for providing input voltage of 12v to each phase. The added benefit of this array and the integrated mosfet drivers along with the ferrite choke cores is that you are able to centralize and more effectively control the different stages of voltage delivery as well as the overall array allowing lower thermal output and a perfect CPU power delivery design, this PWN controller also reduces input and output voltage ripple by monitoring current and resistance along the load line and making proper adjustments, while monitoring and protecting the system with over current protection/over voltage protection.
Now let’s move on to the X58 chipset power, the X58 chipset has its own 4-phase PWN buck controller with 4 additional low RSD mosfets that are cooled by a tiny heatsink not attached to the CPU phases. There are also 4 ferrite core chokes. The buck controller is the Interstill ISL6312.
http://img687.imageshack.us/img687/4...24phase.th.jpghttp://img812.imageshack.us/img812/8...ylersbu.th.jpg
This layout is a mini version of the power delivery system for the processor. The ram has its own double phase PWN (this PWN is also designed to use 3 phase design), the Interstill ISL6322G, as well as 3 ferrite core chokes and 4 low RSD mosfets an even smaller version of the processor power array.
http://img15.imageshack.us/img15/252...orramsmall.jpg
Note: What you should also consider is that the processor feeds power directly to the DDR3 memory as well. To sum up the power delivery of the board, it’s just phenomenal, for the price you pay this board really delivers.
All of this power generates heat that needs to be dealt with, Gigabyte did not leave you high and dry, they left you something cool and quiet.
Motherboard cooling analysis!
Gigabyte decked out the UD5 rev 2.0 with some nice heatsinks. On the topside of the board you have hefty heatsinks that use a thermal pad to make contact with the mosfet drivers, as they do get very hot. The top heatsink is attached to the rest of the system by 1 heat pipe, and the larger mosfet heatsink is attached to the Northbridge heatsink by 2 heat pipes.
http://img811.imageshack.us/img811/7...part1small.jpg
http://img14.imageshack.us/img14/205...t1small.th.jpghttp://img146.imageshack.us/img146/2...t1small.th.jpg
The Northbridge heatsink does its job well, and instead of using a thermal pad, it uses thermal paste. Upon analysis it was some sort of silver based thermal paste, and the good thing is that it did not bake to the chipset; removal of the heatsink array was easy.
http://img843.imageshack.us/img843/8...imsmall.th.jpghttp://img560.imageshack.us/img560/2...ections.th.jpg
The Northbridge heatsink is bolted down to the motherboard, while the mosfet and Southbridge heatsinks uses plastic push pins. What is very nice is that Gigabyte’s Southbridge heatsink is no gimmick. In the whole heatsink array there is not any copper but on the Southbridge. This piece of copper sits on top of the Southbridge separated by thermal pad, and is attached to the heat pipe.
http://img829.imageshack.us/img829/7...connect.th.jpghttp://img153.imageshack.us/img153/7...psidedo.th.jpg
The Southbridge heatsink also cools down the Gigabyte SATA2 chip. Here is where gigabyte really pays attention to cooling, the 4 mosfet drivers for the Northbridge are cooled down by their own separate tiny heatsink, while Gigabyte uses a thermal pad here as well, you can feel the heatsink and it does get pretty hot.
Small HS ON then OFF:
http://img822.imageshack.us/img822/6...ffsmall.th.jpghttp://img687.imageshack.us/img687/1...onsmall.th.jpg
Whenever a heatsink gets hot you know it’s doing its job, and well the gigabyte heatsinks get very hot, yet we never see Northbridge temperatures above 40C at stock and 50C overclocked, that is without supporting fans cooling everything down, but the heatsink temperature is another story, measured at stock load the heatsinks get up to about 70C and at overclock about 80-90C if you average them all out, the most heat ends up at the Northbridge cooler as its where all the heat pipes meet, and the cooler is perfect for a small 40mm or 60mm fan, yet gigabyte does not provide one, because it isn’t really necessary with the north bridge temperatures being that low. It would have been nice though. As you can see the heat moves from the block to the fins, on the UD7 on top of the fins is a waterblock, not the most effective way to design a cooling apparatus, but easy to implement because you can switch out the waterblock for the “silent hybrid” heat pipe apparatus the UD5 does not come with and the UD7 and UD9 do get.
http://img709.imageshack.us/img709/1...fnbhssmall.jpg
X58 Tylersburg the IOH+ PCI-E:
http://img153.imageshack.us/img153/9...setessmall.jpg
Natively the X58 IOH provides 36 lanes for the pci-e bandwidth. Not only do video cards operate off these lanes, but so does the integrated USB3, SATA6G controller, as well as many other peripherals. A brief history of the X58 chipset will reveal that there are three revisions of the chipset, the most recently being revision B3/13, before this revision the main one that hit markets was B2/12, it seems that revision B1/11 was never released to the public. Older boards that have rev12/B2 chipsets seem to have a hard time reaching high base clock frequencies, but base clock is only partially dependent on the chipset. The processors IMC and QPI drive strength have a lot to do with how high you can get the base clock (blck). Any board that is capable of 220blck is a winner, and after that most boards require you to enable QPI Slow Mode. There is a theoretical QPI block speed of 8.8ghz. To get around this base clock blocker, motherboard manufactures have implemented QPI Slow Mode. Normally the lowest multiplier available for QPI Link Speed is 36x, but with Slow Mode enabled it will take the QPI link speed of Bloomfield down to 24x, and on Gulftown 32x; like most functions that control the processor, this lower end multiplier limit is commanded by non-other than Intel. You also need high PCI-E frequency to break through base clock barriers, and this board will boot with 130 PCI-E frequency off the bat, it used to be a mod needed on the EX58 series, but gigabyte did it for you. While you can boot at 130mhz usually PCI-E video cards will not take this kind of torture and will not post, yes with a PCI card or low end PCI-E card you should be able to get up there, I am able to with my GT220. Here is the Intel Datasheet for Chipset Revision Differences Download
PCI-E:
While we are on topic of PCI-E I will go a little into the PCI-E switches found between the middle set of PCI-E slots, there are four switches total and four slots, each switch is used to cut off or enable each slot when occupied, this mechanical ability to cut off PCI-E bandwidth is very important. For instance when you run 3-way SLI you are running one slot at 16x and the other two at 8x, because there are only 36 lanes of PCI-E and you can’t operate two at 16x and one at 8x, that would be more than the 36 lanes present, so when you go to 2x SLI you want full 16x to each card, so these switches would cut off the PCI-E slots not in use and the bandwidth would be redirected to the slots in use. You also need to keep in mind that USB3 and Sata6G use PCI-E lanes as well.
http://img257.imageshack.us/img257/3...witchsmall.jpg
USB, IEEE, Ethernet, Audio AKA Peripheral Sub-Systems
USB Sub-System: This board is equipped with so many USB ports I have no idea how anyone could use all of them, but some people have 12 devices never the less. Yes 14, that is how many total USB ports are available on this board, let me tell you how. First off you have the standard USB 2.0 ports, 6 of them to be total are on the back panel, 2 of those more are eSATA/USB ports, meaning the port can take either. Then on the back you have 2 USB3.0 ports powered by NEC D720200F1 chip, which is very common among USB3.0 motherboards. (8 USB ports on back panel)
http://img5.imageshack.us/img5/7246/...sb3control.jpg
Everyday new USB 3.0 devices are being released, so you are granted 2 which is more than enough for right now. On the motherboard itself you have 3 USB headers, one of which is an ON/OFF charge port. Each header can support 2 USB ports, bringing the total to 14 ports. The ON/OFF charge internal header is a pretty nifty little sucker, supplying power to USB chargeable devices even when the system is turned off. This new technology is part of the Rev 2.0 boards for the entire line. Also gigabyte has 3x USB power so the USB ports, especially the USB3.0 ports which require more power, are able to power any USB device that can use only USB power. Many new USB HDDs need the extra power to operate, and Gigabyte has done an excellent job of providing that. All 12 of the USB2.0 ports are power by ICH10R, Intel’s Southbridge chip. I should also mention that the NEC controller is on the same PCI-E bus as the Marvell SE9128 SATA6G controller.
IEEE Sub-System: I am also going to toss in the T.I. TSB43AB23, the Texas instruments chip that provides 3x IEEE 1394a ports, one internal and two on the back panel.
http://img641.imageshack.us/img641/1...aj3zwsmall.jpg
Texas instruments is known for their IEEE chips, and its not a bad controller at all, but it seems that now IEEE is dying off and USB is really taking over IEEE’s share of the market.
Audio Sub-System: The Realtek ALC889 is the high definition audio controller. Common for on-board audio, it has 6.1 channel capability as well as S/PDIF In/Out.
http://img255.imageshack.us/img255/5...soundcontr.jpg
This Controller had some issues with DPC issues causing static for the speakers, but that issue was solved with the second bios update.
http://img818.imageshack.us/img818/6130/dpc0.jpg
There is also an on-going issue with Fermi based video cards and sound problems with Realtek on-board sound, but I do not have one of these video cards and I only see complaints once in a while.
Ethernet Sub-System: This board is equipped with Dual RJ45 Gigabit Ethernet, 2x Realtek RTL8111E chips power this feature, and provide excellent Ethernet capabilities. Most boards in this price range have this capability, but cheaper boards do not.
http://img714.imageshack.us/img714/6...itlansmall.jpg
Now you might ask, why do you need two Ethernet ports? The answer is teaming and bridging. The console for the Ethernet has the ability to team up the ports so that they can act as one port by combining two connections, so you can theoretically double your download and upload speed on the computer’s side. Let’s say you have one port operating at 100mb/s, you can plug in another connection to your router and have it teamed to 200mb/s, and the connection shows up as a virtual 3rd port. Bridging is a bit different, let’s say you only have one port left on your router, yet 2 computers that need internet, well you can bridge the ports so that one port can connect to the router and the other go to your second computer, sort of like a y-splitter for your Ethernet. Another thing you can do with bridging is control the flow of information over the outbound Ethernet cable, for instance if you are on an intricate intranet network and want to limit certain things.
There is one gripe that we have and that is that Gigabyte did not use a pair of Intel NICs instead of Realtek. Do not get me wrong, Realtek gets the job done, but Intel makes much better Ethernet controllers. The reason you never really see Intel Ethernet ports is because they are a bit more expensive, not many people pay attention other than networkers, and you usually see them on servers. For instance Asus uses Marvell Ethernet, which is kind of like Realtek, aka not Intel.
Various ic section!!!! (very interesting)!
Now we are going to move on to the variety of various integrated circuits(IC) on the board. Let’s begin with the clock generators. ICS is a major manufacturer of clock generators which feed base blocks for various circuits, Asus, MSI, as well as many other motherboard manufactures use ICS for clock generators. The main clock generator for the base clock we generally deal with is the ICS9LPRS914EKLF, this is useful if you have SetFSB program used for general windows overclocking, and this exact chip is not there so you can use this one instead: “ICS9LPRS914EKL”. This board also has a clock generator for the PCI-E frequency and it’s the ICS9DB403DGLF. In the past many of the main clock gens look like the one used on this board for the PCI-E frequency, but please do not be confused, the main clock generator is the square one, not the rectangle and they are right next to each other.
CPU + PCI-E Clock Gen:
http://img688.imageshack.us/img688/9...lockgen.th.jpghttp://img829.imageshack.us/img829/4...3dglffo.th.jpg
Let’s move over to some of the more specialized chips, let’s look at the iTE IT8720F Super I/O Controller. This chip provides very accurate voltage monitoring as well as fan control. The board’s CPU fan has a dual ability to be controlled the old fashion way by voltage (VR) 3rd pin, or the new temperature/frequency controlled (PWN) 4th pin fan connector, this allows you to use older style fans and still be able to control them by software. This chip also provides temperature and voltage readings through Easy Tune 6(ET6), gigabyte’s in OS system overclocking. This chip also provides floppy drive support and controllers the PS2 ports on the back panel for keyboard and. This chip runs on the boards LPC bus.
http://img832.imageshack.us/img832/5...ontrollerd.jpg
Voltage Read Points for UD5 rev 2.0 ONLY:
The following voltage read points can be used to physically monitor the voltage of Cpu Vcore, CPU PLL, QPI/VTT, QPI PLL, IOH Core, ICH Core, Dram Voltage, and Dram VTT, I confirmed they are working, and that they are really not needed. I say this because all voltages are within 2-3% or almost exactly what is shown in the Easy Tune program. The iTE IT8720F is one of the best voltage monitoring chips I have seen, and its accuracy is dead on, this is just phenomenal.
Cpu Vcore Voltage read point(Back of CPU, back of board):
http://img5.imageshack.us/img5/8663/...smallsmall.jpg
Cpu PLL Voltage read point( to the right of Northbridge Heat Sink):
http://img842.imageshack.us/img842/7...smallsmall.jpg
QPI/VTT Voltage Read Point(back of CPU, back of board):
http://img440.imageshack.us/img440/8...smallsmall.jpg
QPI PLL Voltage Read Point(bottom of first White ram slot):
http://img507.imageshack.us/img507/9...smallsmall.jpg
IOH Core Voltage Read Point(between CPU socket and Northbridge Heat sink, very hard to reach, in front of capacitors next to Heat sink):
http://img694.imageshack.us/img694/9...smallsmall.jpg
ICH Core Voltage Read Point(right on top of Southbrdige Heat Sink, between two sets of capacitors on each side of the ram controller):
http://img825.imageshack.us/img825/7...smallsmall.jpg
Dram Voltage Read Point(back of CPU socket,back of board, below where my qpi/vtt and vcore read points are, below from that angle only):
http://img821.imageshack.us/img821/5...smallsmall.jpg
Dram VTT Voltage Read Point(top right of the last white ram slot):
http://img444.imageshack.us/img444/3...smallsmall.jpg
NOTE!!!!!!!: Please do not go poking around unless you know exactly what you are doing, by shorting anything especially near the CPU socket will damage or destroy the components, please if you poke be precise.
BIOS/ Overclocking/ 3 Levels LLC:
I would like to take a second to talk about general bios strong points about this board, first of all in comparison to ASUS boards; this board has QPI PLL voltage which is a boost to the QPI/VTT voltage, kind of like its punch. This board also allows for CPU PLL to be lowered for better stability and cooler running CPU. Three levels of Load line Calibration are available and this is one of the most useful bios features. I have taken the time and done an analysis of the three levels, and I am going to sum it up for you.
http://img827.imageshack.us/img827/5233/allllc.png
(Green is LLC level 2, Read is LLC level1, and Blue is standard No LLC, all at bios setting of 1.35v and 4ghz)
Conclusion from my article here (http://www.xtremesystems.org/forums/...hp?p=4589139): As expected the average voltage increases from no LLC to Full LLC in a semi linear fashion. Looking at the graphical analysis we can deduce that the LLC settings do indeed vary from one another. For starters the standard level shows a (1.35-1.312) 0.038v vdrop from bios setting to actual voltage given. Vdroop of 0.032v slightly less than vdrop, compared to what it used to be, this voltage drop isn’t that bad, but for current systems the drop and droops are significant, in total under load voltage is 0.07 less than set in bios, but only about half that that is droop.
Moving on to LLC Level 1 vdrop is (1.35-1.328) 0.022, less than standard vdrop, but still significant, under load there is another .016v droop, this is almost half that of standard level. LLC level 1 comes to a combined 0.038 voltage drop. This is slightly more than half the total drop without LLC, a very significant improvement. Moving on to the most aggressive level of LLC control Level 2, we have a starting 1.328 idle voltage (1.35-1.328) 0.022, vdrop. If you recall this is the same as the vdrop of LLC Level 1, but vdroop is an opposite story, there is virtually no voltage droop, instead we have voltage rise to (1.344-1.328) (+0.016), close to bios set amount.
Not only does this vrise occur, it occurs without drop outs nor fluctuations. Another thing to note is that vdroop of Level 1 is Level 2’s vrise. This is a very powerful mode, Level 2 LLC is the LLC available on other motherboards, vdrop is not totally eliminated but reduced, and vdroop is reversed. Although LEVEL 1 seems to be much like standard no LLC, it is a much better alternative than LEVEL 2, for 2 reasons. First it offers 50% less vdroop which is what LLC should be for, has no vrise, and shouldn’t have the dangerous voltage spikes LEVEL 2 most defiantly has. Even though vdrop is present, it’s not a big issue, because we know how much vdrop is you can set voltage accordingly.
Standard is usual Intel vcore fluctuation, standard and safe. The spikes we see in Standard and Level1 does not follow a pattern other than it occurs throughout load, this may be due to the program, but it does not occur in level 2, so it might also be a function of normal voltage operation, either way it shouldn’t be a factor in causing overclocking voltage instability. In general you should look at the results and pick a level that fits your overclock.
Bios/ Over Clocking continued:
The board’s bios has many features, the Asus Rampage Series boards do as well, they have 3 Levels of LLC(0%,50%, and 100%) a tiny bit different, but no analysis is available. The ASUS Rampge 3 series is the competitor to the UD7 and UD5, the UD9 is in a league of its own. Gigabyte works with Marvell to update the firmware for the Marvell controller very often. I have owned this board a few months and my Firmware has been updated 3 times, with three bios updates (every bios). Every time my speeds increased with my C300, and the 2nd Marvell firmware update made the Marvell RAID BIOS available through the gigabyte bios. The Most recent one is only available from Gigabyte. All the settings you could want are available, and I will show you a template for overclocking settings:
Advanced CPU Features:
CPU Clock Ratio ................................ [21]
Intel(R) Turbo Boost Tech ...................... [disabled]
CPU Cores Enabled .............................. [all]
CPU Multi Threading .............................[enable]
CPU Enhanced Halt (C1E) ........................ [disabled]
C3/C6/C7 State Support ......................... [disabled]
CPU Thermal Monitor ............................ [disabled]
CPU EIST Function .............................. [disabled]
Virtualization Technology ...................... [disable]
Bi-Directional PROCHOT ......................... [disable]
QPI Link Speed ..............................[x36]
Uncore Frequency ............................[x13]
Standard Clock Control:
Base Clock (BCLK) Control .......................... [215]
PCI Express Frequency (MHz) ........................ [103]
C.I.A.2........................................... .. [disabled]
Advanced Clock Control:
CPU Clock Drive ..............................[900mv]
PCI Express Clock Drive ................... [900mv]
CPU Clock Skew ............................. [auto]
IOH Clock Skew ............................. [auto]
Advanced DRAM Features:
Performance Enhance ...................... [extreme]
Extreme Memory Profile (X.M.P) ........... [disabled]
System Memory Multiplier (SPD) ........... [6]
DRAM Timing Selectable (SPD) ............. [quick]
Channel A + B + C
Channel A Timing Settings:
##Channel A Standard Timing Control##
CAS Latency Time ......................[9]
tRCD ..................................[9]
tRP ...................................[9]
tRAS ..................................[24]
Command Rate.....................................[1]
Advanced Voltage Control:
Load Line Calibration ................. [level 1]
CPU Vcore ..............................[1.44]
QPI/VTT Voltage 1.150v .................[1.375]
CPU PLL 1.800v .........................[1.80]
PCIE 1.500v ...........................[1.56]
QPI PLL 1.100v ........................[1.25]
IOH Core 1.100v .......................[1.2]
ICH I/O 1.500v ........................[1.6]
ICH Core 1.1v .........................[1.16]
DRAM
DRAM Voltage 1.500v .............. [1.66]
DRAM Termination 0.750v.......... [AUTO]
Ch-A Data VRef. 0.750v ...........[AUTO]
Ch-B Data VRef. 0.750v ...........[AUTO]
Ch-C Data VRef. 0.750v ...........[AUTO]
Ch-A Address VRef. 0.750v ........[AUTO]
Ch-B Address VRef. 0.750v ........[AUTO]
Ch-C Address VRef. 0.750v ........[AUTO]
One note this bios Revision is GOOC competition bios which lowers internal latencies (Super Pi tweaked)and was designed for the X58A-UD7 rev 2.0 for the Gigabyte GOOC World OC Championship. Since this board is identical in every way to the UD7 rev 2.0 they made this bios available for the UD5 rev 2.0 , and did further modding for the UD9 GOOC BIOS. The GOOC bios offers, Extreme voltage mode to unlock voltages up to 2.1v and disable OCP(over current protection) for very HIGH benchmarks, it has Cold bug mode, to allow the board to boot at sub-zero temperatures, as well as a Lean Mode which disabled un-needed peripherals during OC. The Rampage 3 Series has this option as well, but Gigabyte does now as well.
http://img709.imageshack.us/img709/6...socketarea.jpg
Those are my settings for 4.5ghz with my puny 930 with VID of 1.24v this board took it to 4.5ghz easily with 215 blck, I have reached 220blck with this board (without slow mode), but I feel my chip’s crappy IMC is holding it back ( I can boot with up to 224 blck, but voltages for CPU need to be increase by 150mv). A capable 920 D0 or 970, Xeon, or 980x would reach over 230+ blck with the PCI-E boosted, there are many examples in the official UD5/UD7 thread spreadsheet I run.
http://img534.imageshack.us/img534/5125/45ghzcpuz.jpg
http://valid.canardpc.com/show_oc.php?id=1363602
http://img830.imageshack.us/img830/9949/testsetup.jpg
