Giving up on the Xeon dream (The curse of Gigabyte Mobo Revisions 1.0 vs 1.1)

Community Blog
21

Maybe. #Hopefully.
Worss-tcase would be that youd need to solder on a newer controller.
Rossmann style : )

22

Hold me. I need Buildzoid.

23

well, it is rather easy and potentially needed equipment costs in total below 100$
depending on what you want. Bare minimum is around 60$

Edit:

Harder part is figuring out if this is the case.

What vrm controller is on yours ?

Picture search made me find a IR3565A

Different picture, IR3570A

Nice VRM, according to techpowerup IR3550Ms uiii

24

HWInfo says IR3563 I think.

I also just sent an email to Buildzoid.

25

I found somewhere:

"New Voltage Regulation Standard
The Sandy Bridge-E FCLGA2011 processors will require a new Intel voltage regulator standard for its Voltage Regulator Module (VRM). Instead of relying on the traditional linear Voltage Identification (VID) codes seen on VR11.1, VR12 will use a 25MHz 3wire serial interface for Serial VID (SVID). VR12 will offer SVID steps with 5mV resolution opposed to VR11.1’s 6.25mV resolution (using a 8-bit VID table). "

So it seems that sandybridge is VR12 from the start.
Techpowerup states that the VRM controller on theirs is the 3570A.
But both, 3563A and 3570A support VR12, and VR12 only.

So it might actually just be fucken firmware.

26

That’s so weird that Ivy Bridge-E works but Ivy Bridge-EP doesn’t work.

So do I have to get a different IR controller or do I have to hack it? I will SO need Buildzoid’s assistance on this.

Edit: OMG… That’s it. Rev 1.0 has the IR3563A and Rev 1.1 has the IR3563B.

27

Well, techpowerup seems to have had the Rev 1.0 and they seem to have fucked up the text.
The picture shows a IR3563A as VCORE VRM controller.

Question still is, what does the Rev 1.1 have.

And it could be that gigabyte doesn’t update chip-firmware through the bios update procedure, making it possible that the change from rev 1.0 to 1.1 was a simple VRM Firmware update.

28

But look at the spec. The 3563B has a VR 12.5 spec. That’s likely what Ivy Bridge-EP uses.

29

The main difference is exactly that. VR 12 on the A versus VR12 and 12.5 on the B

Next question is if, Ivybridge needs VR12.5

30

And even though, i really like buildzoid, you don’t need his assistance.

You can do it yourself, i can do it, i have done stuff like that already, Louis Rossmann does it every day, and if he can do it, you can do it too!

As he sais himself, he’s the worst solderer ever. Well, at least it’ll be hard to beat him ; )

31

The Rampage IV Extreme works with the V2 Xeons. It’s using a ASUS rebrand of a IR controller. Just not sure which.

32

I feel like I need one of those Elmor devices to get access to the VRM firmware.

http://elmorlabs.com/index.php/hardware/

33

well, that is actually a thing that Buildzoid will know more about.

I haven’t jet looked at, or figured out what is done through firmware on those IR chips, and the extended datasheets aren’t public, which is one reason for that.

Best case is that:
1 Gigabyte provides firmware updates to chips through the bios update procedure.
2 The 3563B is fully pin-compatible with the 3563A
3 Standard selection for VR12 and VR12.5 is the same and done through extra components on the board.

The hope would be that A not configured fresh IR3563B will if soldered onto your board, correctly use VR12 and 12.5 and correctly power on, and then be updateable through a bios update.

And no, elmores EVC will not help you with that, at least not with the default feature set that i currently has.

34

Urgh, the more I deep dive into this, the more I think Buildzoid is the best person to talk to about this.

35

well, Buildzoid, Elmore, are some good places.

Though the main questions left are:

Does Ivybridge need VR12.5
Does a IR3563B need programming to initially work correctly.
Where to get the correct gigabyte firmware for it?

the first one is rather easy.
Second one not so much.
The third one is the break point.

You could buy a dead board with such a chip that is programmed and will work, and transplant that : D

Dead X79 boards aren’t that expencive, and there were a few on ebay that should be currently available.

36

Yes, but finding a dead Revision 1.1 is going to be neigh impossible.

37

doesn’t need to be rev1.1, just X79 and that chip.
Still hard.

Well 1.0 you are right.

And getting a 3563B will also be a problem

38

Well, at least we now know a probable Cause why Sandy is ok, but ivy not.

39

There’s a difference between UD5 and UP5 because UD5 uses separate high side and low side and the UP5 has fully Integrated high and lowside MOSFETS.

40

well, if you say so. Doesn’t mean that they don’t use the same controller, but it can also be a different one.

And that is the only reasonable X79, i could swear there were at least 3 in the last few days though.