@Lost_N_BIOS
I had some experience and think on flashing that mobos.
1.p67a-ud4
The mobo have 2 pcie16x slots which could split into 2 pcie8x when they are all plugged in cards.I had tried to flash the uefi bios of p67a-ud3r-b3 (bios version : u1b) and p67x-ud3r-b3(bios version : u1d),but the second pcie16x slot was failed to launch any pcie cards after flashing the uefi bios.
2.z68x-ud3h
The mobo have 2 pcie16x slots too which belong to CPU not southbridge. I got this mobo and it's all OK.
3.p67a-ud7
The mobo got many pcie slots,but none of pcie slot belong to SouthBridge.
So……It was a pity.

@gloobox - For any board, at first I would use it's proper UEFI BIOS if possible
But, summary of what you are saying, maybe only UD4 board has this last PCIE slot bug @ x1? Did you try the proper UEFI BIOS for P67A-UD4 model? Did you try Z68X-UD4 for P67A-UD4 too? That should work as well.

And, Z68X-UD3H no need to test, this model not affected?

@Lost_N_BIOS
Actually,p67a-ud4 doesn't have any uefi bios release by GIGABYTE.So I could just tried to flash the uefi bios of other mobo forcibly.
As to z68x-ud3h,2 pcie16x slots belong to CPU,so I think it's meaningless to test it.(but I had tested,2 slots ran well)

Ohh, I hadn't looked yet for that model. So, I don't have any models that can test the issue you mention, except UD4, correct? If yes, I can test Z68X-UD4-B3 BIOS on UD4 (B2)

@Lost_N_BIOS
Z68X-UD4-B3 was not a texter,it's pcie slots is the same as z68x-ud3h.

Now,I show you the mobos that had this problem below.

P67A-UD3R(B3)
P67A-UD3P(B3)
P67A-D3-B3
P67-DS3-B3
P67A-UD3-B3

Z68A-D3-B3
Z68A-D3H-B3
Z68AP-D3
Z68MA-D2H-B3
Z68M-D2H-B3
Z68P-DS3

I don't have any of those, so can't test to look into the issue and try to fix
But, you said P67A-UD4 good candidate, right? If yes, I can test it with other BIOS and maybe then find the source of the problem (P67A-UD3R very similar I think)

@Lost_N_BIOS
P67A-UD4 isn't candidate,it's pcie slots is the same as z68x-ud3h.So I think the mobos I mention about are the true candidate.Because the second pcie16x belongs to SouthBridge in these mobos.

I am trying to upgrade the ME FW to see whether the problem could be solved.

Edit:
Sorry for my careless expression.
what I say all of P67/Z68 had the problem is all the pcie16x belong to SouthBridge of GIGABYTE P67/Z68 had the problem.

@Lost_N_BIOS
I updated the latest ME FW last night,but the problem was still alive.
I updated it by FWUpdate directly,and you could see the FITC Ver was still 8.0.0.1351.

@gloobox - FITc version has nothing to do with ME FW, FITc is the tool you edit ME FW with (Intel Flash Image tool)
Latest version is FITC 8.1.40.1456, but you'd have to redo the ME FW and edit it with that version using the cleanup/update guide, then reflash, if you wanted that FITc version you are looking at to change.

Maybe this setting at ME FW is the issue? I wouldn't change it unless you have flash programmer to recover though, in case it causes brick or non-BIOS entry etc
Also, I noticed in AMIBCP, Chipset >> PCIE config is missing menu contents (same in setup module too), this probably adds to the problem (at least for sure makes it so we can't properly adjust widths in BIOS settings)


I checked FE BIOS (W/ V7 ME FW), it's set differently (As I suggested to above), so maybe that is the key issue, at least the root of the problem (BIOS setting we can't adjust since missing may also be part of the issue
Default setting shown in the selection box in both above and below images, and as you can see here, what I suggested to choose in above image is indeed the default setting in this BIOS/ME FW

@Lost_N_BIOS
haha,I do have flash Programmer so could you help me change it?
What's more,I found something insteresting.

@gloobox - What, is Z77P-D3 one of the boards with similar layout?
Flash programmer only needed if you have to recover, but you should make a backup first with FPT (entire BIOS, or at least GbE, if no GbE then do whole BIOS)
That way you have MAC ID at least, unless it's stored in-chip FW

Here is Z68P-DS3 U1c with change we're discussing above, edited with latest FITc, so you'll see that change now too
You can flash this with ME FW update tool, but I personally would suggest you use FPT instead because sometimes ME FW update tool does not flash in all settings.
Gigabyte has unlock FD, so you can flash ME with FPT (Suggested >> FPTw.exe -me -f meupd.bin). Be sure to follow it up with FPTw.exe -greset


http://s000.tinyupload.com/index.php?fil...921083463295394

@Lost_N_BIOS
Many thinks for you!
I try to flash it by FPT,(FPT -F -meupd.bin),this could flash the whole bios,and then strike (fpt -grest)
And I just wonder if the "PCIe Port Configuration2" you changed belongs to Flash Descriptor Region?
I always think it must something to be changed in Bios Region for the problem.

I'll feedback soon.

@gloobox >> Stop! This is ME FW file only, I didn't send you whole BIOS, and you have wrong FPT command above!! Be careful!!

FPTw.exe -me -f meupd.bin
FPTw.exe -greset

That is what you need to do with file I sent you above ^^

No, it's not in FD, at least I don't think so (Checking). I can send you whole BIOS if you want, but all you need to flash is ME FW as instructed above and as I sent you (ME FW Only)

Actually, I do see two bytes changed at FD, other than the two bytes I changed.
That may be due to this setting change, or could just be due to I rebuilt and updated the ME, not sure. I will make a new build without update FW version now and see!
* Edit @gloobox - !Yes! byte 0x124h holds this setting at FD, or something additionally related to this setting, confirmed with multiple settings changes, always changes ME FW and byte 0x124h at FD

Anyway, here is entire mod BIOS, flash as follows
FPTw.exe -f filename.bin
FPT.exe -greset

http://s000.tinyupload.com/index.php?fil...686819658468548

Or, you can extract FD from that with UEFITool, and flash it >> FPT.exe -desc -f fd.bin

@Lost_N_BIOS
Excellent work!It's awesome!!


And I think it may have something to do with the FD region,as the pic below,the "PCH strap" belongs to the Descriptor Region in green frame.What do you think about that?


Edit:
I dumped the FD,and draged it into FIT.As you can see,the change part belongs to FD.

@gloobox - Great! So it's working now, correct? And yes, good eye, you caught that about it belonging in the FD, I totally missed that!
If you see my edits above, after I made that post @ #73, I did some testing, and I found the exact byte for this setting
It may/probably also make some changes at ME Section too, I didn't check but will now so we can be 100% sure. >> Yes, LOTS of changes at ME FW too, so both ME+FD need flashed if you change that
As to how related those changes are vs just a general update/normal expected changes on ME rebuild due to different FITc used etc, I do not know, only plutomaniac would be able to answer something like that.
For now, I suggest anyone doing this fix flash both the updated FD and ME FW as well.

So, please show me your before NVME speeds, and then with above image showing after, so I can see performance difference.

Do you have friends somewhere that use all these boards with the UEFI BIOS? If yes, maybe we can put together package for all them, with this fixed for each model.
Each BIOS has to first have FD corrected to be able to properly build outimage in FITc, Gigabyte didn't put BIOS region offset correct in FD, probably for all these boards (@44-46h)
plutomaniac explains how to fix here at #137+141 - [Guide] Clean Dumped Intel Engine (CS)ME/(CS)TXE Regions with Data Initialization (10)
Good thing with that would be Gigabyte does not lock FD or ME, so everyone can flash fixed FD and ME with FPT