Please report all spam threads, posts and suspicious members. We receive spam notifications and will take immediate action!
Results 1 to 2 of 2

Thread: Review of GIGABYTE's 2nd Gen. X79 Workstation MB: X79S-UP5




  1. #1
    Join Date
    Sep 2010
    Location
    Norther Virginia
    Posts
    296

    Default Review of GIGABYTE's 2nd Gen. X79 Workstation MB: X79S-UP5

    X79S-UP5 Review
    When GIGABYTE launched their initial X79 platform they encountered a lot of hurdles. They had a brand new PWM controller which to program correctly they had to use a true AMI based UEFI which they also had not implemented before. Then you have Intel who released 2 CPUs on launch, both of which had revision change already by the time of release. That must have really hurt sales of CPUs, but also of motherboards as not many CPUs were being sold. That means that the amount of money companies can make off the platform was reduced, and thus many companies didnít have the resources to produce full lines of boards; perfect example is the number of models each company put out for the X79 platform. At the same time Intel required implementation of many new technologies such as PCI-E 3.0, quad channel memory, and a new VRM layout.



    Thus a perfect storm erupted, and it was pretty hard on GIGABYTE as they had to implement all these new technologies. However GIGABYTE learned, with Z77 we see that they can make a very nice UEFI and have it work too, and it will just get better. However now we have arrived at the point where GIGABYTE has basically perfected the X79 VRM. X79 is all about power density; you want the most power output in the smallest amount of space, at the lowest temperature. For this GIGABYTE has employed their Ultra Durable 5 design, which brings really good MOSFETs and Chokes to the table. You are talking 60A per phase, something I donít think any other X79 board can do. The UEFI is much better than it was at X79 launch, a good 4.5 GHz-4.8 GHz 24/7 OC is easy to reach, however like all boards; this one has its flaws as well. So join me and take a look at the X79S-UP5.


    Box, Accessories, and Board Layout:

    The box is typical GIGABYTE. Remember that these white boxes started with X79, like a new start.



    Accessories are numerous, great for a workstation. You even have included USB 3.0 panel which only comes with premium GIGABYTE boards. A whole wifi/BT kit is included and tons of SATA6GB/s cables.
    1. BT/WIFI Card
    2. 2x Antennas
    3. Internal USB cable for BT
    4. 6x SATA6GB/s cables
    5. BackPanel I/O Shield
    6. 3-way SLI Bridge
    7. 2-wasy SLI Bridge
    8. 2-way CF Bridge
    9. USB 3.0 Front Panel
    10. Driver DVD + Manuals



    The boardís heatsinks are very nice, GIGABYTE added some extra heatsinks to the left. It doesnít actually cool anything directly, except its heatpipe is connected to the PCH heatsinks, thus it will take some heat away from the PCH heatsink and VRM heatsink. There are 5 fan headers.



    Back panel is made up of tons of good ports:
    1. 2x RJ-45 1GBit LAN
    2. 4x USB 3.0 Ports
    3. 5x USB 2.0 Ports
    4. 2x eSATA Ports
    5. 1394A port
    6. PS/2 KB/Mouse
    7. 4G OC button
    8. Dual BIOS selector
    9. Clear CMOS button





    This board supports 3-way SLI/CF. There is also an extra 4x PCI-E 3.0 port, and 1x PCI-E 2.0 port. There are 4 DIMMs, and GIGABYTE even added backside stabilization plate that is used to keep the board stiff.
    GIGABYTE used a 2oz 6 layer PCB with an extra black layer.



    GIGABYTE has placed the 8-pin connector behind the VRM, and centered as to reduce the distance the current has to flow to get to the IR3550 Powerstages. This board carries a straight 8 VRM, very powerful, I doubt any other board can do better.



    We see a nice power button in this corner, and we have some fan controller and a nice USB 3.0 header.



    Here we have what all the talk is about, 8 SAS ports, 4 SATA3G ports, and 2 SATA6G ports. These SAS ports arenít guaranteed to run at SAS6G, but there is an option to turn them into gen3. They work fine at Gen2 speeds, but after testing, I couldnít get them to run at gen3. Intel says they donít officially say it can run anything more than G1, but Gen2 is a very safe bet.



    Here we see the PCI-E ports. Starting from the right to the left:
    PCI-E 3.0 16X
    PCI-E 2.0 1X
    PCI-E 3.0 16X
    PCI-E 3.0 4X
    PCI-E 3.0 8X
    PCI
    If we take the heatsinks off:



    This board looks really nice without any heatsinks.



    Here is the main VRM, like all other X79 boards there is a limited width between the sets of DIMMs.



    Here we have all the 4 main VRMs. You have the CPUís VRM it is 8 phases of IR3550 for a total of 60A per phase, however while some boards might also have IR3550 or two sets of 25A powerblocks, GIGABYTE added in 60A chokes, while competitors might not use that kind of high current inductors. So we are looking at 50A at 90% efficiency which is what you need for cool operation.



    That means that top output is 480A or 400A at 90% efficiency which is really damn good. What is even better is that GIGABYTE matched the IR3550 with the master PWM IR3563A, a very nice 8 phase PWM capable of 1.2MHz switching frequency and very nice modern digital PWM technologies. Matching IR power stages with an IR PWM allows for some IR technologies to actually work.



    The amount of current running through the PCB is immense per unit area compared to previous platform designs. This is because of the current demanded by the CPU, and the high TDP/TDC of the CPU. The placement of the power connector is done is minimize losses through the traces and balance the current being dispersed by the PSU connector.



    Here we have one of the slave PWMs, an IR3570, which is a 3+2 phase digital PWM. GIGABYTE chose to use IR3553 which are the 40A variant of the 60A IR3550. You can see how big the IR3550 is compared to the IR3553, the empty pad space is because they could have used IR3550. The extra pad space is used for more current flow, however 40A is way more than enough for 2 let alone 4 DIMMs.



    The IR3563A has only been used on three boards, it is a 8+0 phase PWM from IR, using their latest technology it is their top of the line PWM as well. Matching this PWM with the IR3550 will unlock some of the IR3550ís features to their full potential such as the IR Body Braking feature which can rapidly turn off phases completely and help discharge the inductor faster for better transient response as well as less output voltage overshoot.


    Circuitry Analysis:
    Letís begin this section with the interesting layout of the PCI-E on the UP5. However first letís take a look at the C606 chipset:

    Now this is a server chipset, and as such it costs much more than the consumer X79 chipset. However the X79 chipset and CPU platform are just a neutered server platform with overclocking features. Even as a neutered version of the server platform, is still is the best and most awesome consumer platform to date. The C606 chipset has a pretty high TDP of 12W and that is because of the 8 extra SAS ports. Those 8 SAS ports require an extra 4X uplink to the CPU, that is why the X79 CPUs have 40 PCI-E lanes, as their total would have only been 36X lanes. That is why many X79 chipset boards can do 3-way SLI at 16x/16x/8x. This board has 4X routed to the PCH as well as the X4 DMI 2.0 link is totally different. The SAS ports on the C606 are only officially supported as Gen1 speeds, as in a server they will just use HDDs; however GIGABYTE unlocked the ability to choose speeds from Gen1 to Gen3. In testing Gen2 speeds work great, however Gen3 doesnít work. GIAGBYTE advertises them as Gen2, so there is no issue with that. The X79 PCh is an 8W part, that means 4W is all for the SAS controller.



    Notice how GIGABYTE routed the first lane directly to the CPU, and the second 16x lane is 4 slots down. That gives a lot of room for 2-way Systems, and good enough room for 3-way as well. GIAGBYTE used the last 4X of PCI-E 3.0 from the CPU and directed it to a 4X slot right above the second full 16x slot so that you can run 2-way and then a 4X RAID card or extra USB 3.0 card and have 4X PCI-E 3.0.



    Those are the PCI-E Gen3 quick switches.





    That is A LOT of Sexy SAS.



    The next thing of interest is the dual NIC. GIAGBYTE used a very nice Intel NIC, however then they used a Realtek part for the second port. Now this is odd because for other dual LAN boards they use Atheros as the second, however in this case it is a RTL8111F. RT8111F is more of a server class 3rd party NIC than the RTL8111E which is very common on cheap consumer boards.



    You are most likely wondering why the heck there is a 3rd party SATA6G controller when the board has so many SATA ports. The reason is for eSATA, and this Marvell SE9172 provides that.



    As many people forget the X79 chipset has no native USB 3.0, thus 3rd party controller need to be used. Here we have the venerable VLI800 which is a 4-port USB 3.0 controller, it provides all rear USB 3.0.



    Here we have a Fresco Logic FL1009 which is a 2-port USB 3.0 controller and provides the internal USB 3.0 header with bandwidth.



    Here we have two important pieces, an ALC898 audio controller which is a nice audio controller for a codec. Then we have a VT6308P which provides the 1394A on the board.



    Here we have a iTE8728F which is the SuperIO.



    Here we have a clock generator from ICS.



    This empty pad is what sets these GIAGBYTE X79 boards apart from the original series. This chip was a GPIO used widely on previous GIGABYTE boards. When they first worked on the UEFI they used this chip to augment some functions like with the BIOS. However coding all its functions into the UEFi is possible, and that is why it is gone. Perhaps early on in production they tried with and without, and decided it was better without.


    Heatsink Analysis:

    The heatsinks on this board really are magnificent, they look really nice and they are designed pretty well. As you might tell by now there is an extra extension from the PCH heatsink, that is only for adding extra mass to cool the extra 4W from the PCH.



    Here is the underside, we have paste on the PCH, pad on the VRM and an extra backpanel for stabilizing the VRM heatsink.


    The UEFI:
    This UEFi varies from that of the orignial X79 boards in that it is based on the Z77 UEFI. Evidence of this can be found in the memory timings where the extra timings available on Z77 but not X79 are grayed out and not yet removed.
    This UEFI is much better than the original X79 UEFI, however it still requires some more work for overclocking.








    Test Setup and Overclocking:
    So there is somthing very different with this review compared to my past ones, I am using a new GPU a GIAGBYTE GTX670OC, a new PSU a Thermaltake 1475XT, and a very nice memory kit from G.Skill the 2666C11 16GB(4x4GB) kit.


    Here is the new GPU:



    The new memory:




    The PSU:




    Here is the setup all together:




    The memory installed:



    The test setup this time is:
    3960X @ stock and at 4.5GHz
    3930K @ stock and at 4.5GHz
    3820 @ stock and at 4.5GHz

  2. #2
    Join Date
    Sep 2010
    Location
    Norther Virginia
    Posts
    296

    Default Re: Review of GIGABYTE's 2nd Gen. X79 Workstation MB: X79S-UP5

    Test Setup and Overclocking:
    So there is somthing very different with this review compared to my past ones, I am using a new GPU a GIAGBYTE GTX670OC, a new PSU a Thermaltake 1475XT, and a very nice memory kit from G.Skill the 2666C11 16GB(4x4GB) kit.

    Here is the new GPU:


    The new memory:



    The PSU:



    Here is the setup all together:



    The memory installed:


    The test setup this time is:
    3960X @ stock and at 4.5GHz
    3930K @ stock and at 4.5GHz
    3820 @ stock and at 4.5GHz

    Now for the memory OC was pretty impressive. Since X79 really is limited at 2400MHz I took the 2666C11T2 kit to 2400MHz with XMP very easily, however after that I reduced the timings with ease and make the command rate T1 instead of T2. So all 16GBs worked very well with the X79 platform and with great timings.
    This was the XMP that the GB board loaded for the memory:


    This was my memory OC:


    This is the board using the 1.25x multiplier, not to hard to use.


    This is a OC and temperature of 3960X at 4.6ghz:

    That was the maximum OC with respectable temperatures.

    Max OC with 3960X was 4.9GHz:


    Benchmarks:
    Now the old benchmarks with the X79-UD3 for instance are done with the 3960X at 4.5GHz, however they also used the GTX570 not GTX670 I am using now. I included them for reference. Memory was all at the same frequency 1600MHz when at 4.5GHz and 1333mhz at stock.









    Audio and SAS Performance:
    Audio Performance:


    SAS performance at SATA3GB/s:


    Conclusion:

    So this board has a lot to say about it. First of all for its price, $330, it really gives you a lot of value. You get a server chipset along with all eight SAS ports; you get all the features of the X79 platform and all the GIGABYTE features that come with it. Now there is a lot of hate for GIAGBTYE X79 boards, however this time GIGABYTE is making sure no one can blow up their VRMs naturally and give their VRM a bad rep. What people don’t understand is that the IR3550 isn’t something to be messed with and they are definitely not a marketing gimmick, they are very costly compared to other MOSFETs, however they perform really well and that is why GIGABYTE flipped the bill for them for all their ultra-durable 5 boards. This board is no exception; the hardware is very nice and seems to do the job well. X79 boards are all plagued with hot running VRMs, it just happens when you reduce the amount of space and increase the amount of current that each phase needs to push you get hot running VRs. You can either increase the amount of space for the VRM or distribute MOSFETs on more area or you can use higher spec MOSFETs such as those used here. However GIGABYTE also reduced the amount you can alter the LLC slope, thus LLC isn’t as strong as on the Z77 boards, this is a technique carried over from the previous generation of X79 boards where they limited the LLC among other settings to make sure people couldn’t burn their boards, however they should bring back the LLC setting for setting lower idle vcore.

    Running this board at stock is very simple and the UEFI seems to not have so many bugs with that. However there are still some bugs that need to be worked out for perfection. Performance in audio is pretty strong, and the fact that the SAS ports can run at 3GB/s/Gen2 spec is a plus. The PCI-E layout is also done for a server/workstation arrangement, and that is really what this board is meant for. It can definitely hold a normal 24/7 OC of 4.5-4.6GHz with my Zalman AIO water cooler, for higher frequency I needed better CPU cooling, like a real watercooler. This board is meant for a high-end workstation, and as one it does great. For high-end extreme LN2/DICE overclocking I would have to say that it needs that higher level of LLC before it can do that. Memory OCing was actually very simple; 2400 MHz was simple enough with XMP. For the most part I really like this board, however I think it needs some minor BIOS improvements to be a high type overclocker, but that might not be what GIGABYTE had originally envisioned for this board at all, as it is laid out and equipped to be a super workstation motherboard that can sustain a decent 24/7 OC, which is exactly what it is.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •