This just all sounds like a very bad idea. If I could afford a backup system, i’d totally do this, but this is my primary rig and I may just go for a 4960X.
Well, if you have an idea what could cause an 0D post code, with and without ram by the way, maybe maybe maybe.
Found a Gigabyte Press Kit for the X79S-UP5-WIFI. It states:
Supports Dual VR12.5/VR12 modes (a requirement for future 22nm CPUs)
Kinda not true there, if only revision 1.1 has the Dual VR12.5/VR12 modes.
indeed, so it now is certain that the ivybridge isn’t working because of missing VR12.5 support on the IR3563A.
Makes me doubt that the i7 will work.
Hey, theory confirmed ! : D
But people have gotten the 4960X working on this board. Either they got extremely lucky with Revision 1.1, or Ivy Bridge-E works, but not Ivy Bridge-EP.
well, Consistency is something questionable.
My logic says, Ivybrdige is build for VR12.5.
Therefore the i7 shouldn’t work.
BUT Intel being Consistent and Logical ? NAAAAAA hahah good joke.
I wouldn’t count on it.
You could ask gigabyte if they can provide you with a pre programmed 3563B : D
Hoping Buildzoid can help there. He’s had direct Gigabyte contacts cause they’re sending him boards.
good luck
You might have better luck getting in contact with buildzoid on Reddit. You’ll be able to message back and forth more effectively, and I’d imagine he checks PMs there more often.
I’ve been getting responses from him, but he thinks it’s more than just the VRM controller, but can’t quite nail down what exactly it might be…
Okay, so I’ve hit a dead end. Nobody is interested enough to pursue this so I’m just gonna try to get the 4960X and call it a day.
Just exactly when I get my 4960X to be able to get on with my life is indefinite, so I’m going offline to resume being a recluse.
Well, if you can find another pin compatible controller , i will happilly Solder it in for you.
I can even test if it still works.
But the Main Problem is,you, and getting a controller.
The Gigabyte X79 UD7 board uses the IR3567 pwm i believe.
And its probablly also used on other Gigabyte X79 boards.
I couldn´t find a public data sheet for it, this quickly.
I appreciate it, but everyone knows Aliexpress is the source of those controllers, and shipping the mobo to take this risk is a big burden to handle.
@Destroyed007 also owns the same board, so if you’re able to get a board from him, that could solve the mystery without risking my hardware. If it works, then I’ll ship my mobo to do the modification and keep my E5-1660 V2.
The E5-2620 V2 can be had for dirt cheap to test if it works.
Yes, but the UD7 uses separate high-side and low-side MOSFETs, and also, that controller has a 3567A and 3567B revision difference.
Oh Google… You listed the brief for the IR3565B in the results for the IR3563B… Goddamn it.
See, you know what you’re doing with deep dives like this, I don’t.
It seems like that : D
The main Problem with this hole project is you, since it is you daily machine, if it goes wrong, that’d be a problem.
On the technical side, ir controllers don’t give a fuck about discrete FETs or power stages for multiple reasons. Doublers and drivers as one.
I think, as long as the pinout, the phasecount and configuration is the same, a newer 8phase or should be a sufficient replacement.
Maybe the common 3567B, but I haven’t looked at it other then that it has the same 48pin qfn package.
And yes, aliexpress is one place where you can get such chips, but its not the best place.
I squeezed more info out of Buildzoid. He guesses the ID pins on the LGA2011 socket gets the wrong info from the Mobo and therefore boot loops. The E5 V2 Xeons DO support VR12, (according to the datasheets he looked up) so for some odd reason, it’s not working on Revision 1.0 but is working on Revision 1.1.
He was unable to get info from Gigabyte so the reason remains a mystery.
It may be worth it to do a deep dive into the BIOS (disassemble it) to see what exactly is going on.
Me though? I need a working machine, so I have to get a 4960X and somehow recoup my costs from the E5-1660 V2.
Yes i know that, i do vrm overviews on motherboards all the way back since the X58 era of boards.
The Gigabyte X79 UD7 uses a highside + lowside configuration with IRF-hexfet mosfets.
Which aren´t really that great but yeah…
An IR powerstage is just a highside and lowside + driver in one package.
A powerstage just takes less space and is generally more efficient then,
the more traditional driver ic + highside and lowside configuration.
But other then that they do the same thing.
And you are correct, there are two versions of that said pwm 3567A and B…
I could dig arround if i still have some info somewhere about other X79 boards vrm configurations,
and controllers used.
But i´m not sure if i still have it arround.
But the Xeon you got is that a retail or an ES sample?
if its an ES sample, then that could be a reason why its not working on that particular board.
1 Like
@MisteryAngel
Cool, didn’t know you do that.
Would be great if we could figure out if the rev1.1 is actually using the 3563B and therefore has vr12.5 support.
And if ivy-bridge actually needs vr12.5 . Buildzoid said no, but everyone fucks up some times : D
ES related issues at least in my experience on x99 manifest them self differently then a power-on-of-loop.
It seems as if it isn’t even “POSTing”.
Actually one of my Rampage IV Formulas behaves similarly, i assume a dead cpu vrm controller, or something wrong in the periphery
1 Like