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

The Rampage IV Extreme is confirmed to support Xeon V2, but choose wisely which BIOS revision to get. Again, Bryan of Tech YES got lucky.

To be precise, it is a
“FORMULA”
And i only have v1 Sandys.

The behavior is pretty simple. With a good and known cpu, it turns on, and instantly off again. With a dead cpu, it turns on, and stays on. Vcore is as expected.
All voltages in existence are there and seem ok.

So thanks to this thought provoking topic, i now suspect a dead controller who fails to talk to the cpu, and therefore shuts off the board.

I need to dive into this, since its a pretty long time ago i looked at X79.
But from what i read and gather in this topic,
i assume that the issue is either a dead board or a dead cpu.
But yeah that is kinda hard to tell without additional testing parts.

If the board turns on and shuts off imediatlly, then it often is,
a not properly seated cpu, or damaged socket maybe.

Also did you already physically checked the cpu socket for bended,
or broken pins?

Well, you are right, it is both, a dead CPU, and a dead board ; )

It’s just that i’m interested in the actual cause and fix.

Following is just a reiteration, no need to read that : )

The cpu is a dead 2687W v1 who was previously shorted out that badly that vcore would trip ocp and turn off.

Like all my dead CPUs, that somehow resolved itself to totally dead CPU, without short.
It now is my test CPU to see if a new boards shows signs of deadly voltages.
So if I use that 2687w, it turns on, it stays on at postcode 00. All voltages are present, stay present and measure good.

The CPU is probably not talking to anything, therefore postcode => 00

With a known good, 2620v1 it turns on, and off. It is to fast to even post, or show a code on the display.

All pins are there and the 2620 works flawless in another iv formula.

So, something is probably wrong with the vrm controller.
Luckily, I have working reference boards, and another dead “donor” which posts, but only to 0D with and without ram.

Okay guys, this seems like a completely different topic altogether.

I’m willing to accept nobody outside of Gigabyte engineering has a Rev 1.1 of my board, and that if you get a retail X79S-UP5-WIFI, it will be revision 1.0. Which means the best CPU you can get is the 4960X.

I cannot afford a $500 Rampage board or a board swap to a ASUS due to the DSDTs and SSDTs for Hackintosh needing to be changed. 4960X is the only option.

Just curious, is patching your DSDTs and SSDTs really the worst thing?

Nope. For Xeon V2s, you need the EXACT correct BIOS version because the latest ASUS BIOSes took OUT Xeon support. Another risk I don’t want to take.

well, you can patch removed stuff back in, especially with the bios-flashback functionality, there is little risk in bricking the board with a bad bios.

150€ for a fixed asus p9X79E-WS with 6 instead of 8phases and a hole where the 2 were?
Does get my 1650v1 up to 5ghz, so OC and VRM ripple is ok. You need to cool it though : D
if you don’t it’ll trip overheat on the VRM. Maybe due to worse balancing, but to blow a vrm phase, i suspect that those boards do have thermal issues.
Have seen a few of those with that phase burned away, so meh. Shitty engineering from ASUS i guess.

That’s good to know, Im running a Xeon on P9X79 and I haven’t updated my bios.

Okay, looks like the X79-UP4 did have a Rev 1.1 officially made. (and is way more common) Whether someone bought a Rev 1.1 X79S-UP5 is another question, and is once again like finding a needle in a haystack.

There are 2 Rev 1.1 UP4 listings on eBay, here’s one of them with a CPU/Heatsink combo:

UP4 and UP5 should use the same IR controllers. But interestingly, the UP4 got a F7 BIOS, and the UP5 was stuck at F5f Beta BIOS.

Got my 4960X, and x264 is stressful enough to actually be more stressful than Prime95.

I literally had to do a “manual” AVX offset of -1 multiplier in order to render anything with x264. If I left it as is, the voltage requirement exceeds the safe range for the CPU. To render at the max safe Vcore, I have to use x43. And then I have to switch it back to x44 for gaming. (Can’t use XTU because I run Mac and Linux too)

Edit: 1.4 Vcore solved that x264 problem.

Well, 4.4Ghz degraded to the point it’s no longer stable again in Linux.

Feel free to throw sticks and stones at me now. I now know 4960Xs only clock safely to 4.3Ghz in Linux, and no further. If anyone has a success story of 4.4Ghz stable on Linux with a 4960X, I would love to know because you sir have won the silicon lottery.

I will be buying another CPU and only clocking to 4.3Ghz. Hopefully with luck it will require very little voltage.

Hi all!

i wish i had read this topic one week before.
I got the same mobo (UP5 rev 1.0) , one week ago i ordered a 1680V2 and obviously after the known boot loops i started the forum search, so here i am.

I understand that you (@FurryJackman ) didn’t have any luck with the V2s. I am quiting right now too. I just flashed F5f modded with name support(yes at last!!!) and all 06/2019 microcodes but still no luck with the 1680…

The worst side of my story is when i read yesterday that 1680V2, even if it boots it may return panics with Custom Mac (i run 10.14.4 mojave).

Now i am really thinking of buying an ASUS X79 Deluxe and get over with the CPU thing. Still though the X79 custom mac nightmare will start again…

reading your last post about 4960X didn’t make me happy as well.

Do you still use your system as a custom mac? does it work ok with the i7?

Yes, the i7-4960X works beautifully with Hackintosh.

If you wanted to stick to the 1680V2, get a X79-UP4 Rev 1.1 as that is more plentiful vs the X79S-UP5.

My replacement 4960X ran 4.4Ghz at 1.27V load Vcore. Good lord is that jaw-dropping. My first 4960X was a really bad bin.

The die is cast… it was an ebay auction which i won so an Asus X79 Deluxe is on it way to my …X79 world(the fourth) So once it is here i will know everything.
Now if everything fails may as well sell everything and buy a 4960X.
I am optimistic though that the asus will make a great hackintosh.
Anyway, i fell off topic.

Back to “our” GA mobo, i tried 2 modded bios but in vein.
The only thing i haven’t tried yet is to update the Intel Management Engine firmware. I am in completely uncharted waters here so if you have tried anything in the past please share.

It’s an electrical difference, no amount of firmware tweaks will get it to work.

it is indeed a limitation by the VRM controller.

If you can solder largerQFN packages, and happen to have a IR USB005 Dongle, You could replace it with the B model that should work. Not like anyone did it before you that i know of.

I’m just happy with my 4960X. But if you get a X79-UP4 Rev 1.0 and Rev 1.1 you can compare IRL what’s really the deal with that.

Ftfy.