@Lost_N_BIOS :
If it should be too difficult to split the thread, I recommend to simply rename its title (something like “Discussion about various BIOS Modding problems”). The current title is absolutely misleading.
@Lost_N_BIOS Id love to help test the end result for the bios however I have neither P9X79E-WS or P9X79-WS im on a P9X79 so unless its no extra work to add that model to the assemly line I can’t help test.
However im eager to learn easier methods to replicate my result.
Less work more better 
@Fernando , it’s not about difficult to do in my eyes, only about splitting up the flow of thoughts, I rarely see that as ideal to do, unless people start talking about a party in town or something totally unrelated to computer stuff already being discussed. To me, this is all related, so I would have never considered splitting anything up.
Same applies to title in my mind, all still remains the same, and we will continue towards that goal quickly now I hope, and with success eventually of course 
I don’t see the title misleading at all, still going 100% in that direction except a short side-track we’ve all taken part in this past few posts.
Cutting up threads all the time breaks peoples flow of thoughts/effort progress, and then topics loose their overall knowledge for others later in the future. If anything was split off, that would be an added thread in the forum for no reason, and no one would ever post in it again or read it etc.
But, those are only my opinion on all this, I’m sure everyone has varied opinions on how things should be done
@JackSted sorry, I missed your board variant. Yes, once we do it all I’d be happy to show you what to change where, or port over a mod for your model as well.
@paranoid_android - did you test via USB Flashback the BIOS in post #26 yet, if yes and both worked, which method do you prefer then I will do for both boards
@Lost_N_BIOS :
If the current title given by the thread starter covers the main topic of the discussion, it is ok to leave everything as it is. I had only read the last posts and had the impression, that their topic was far away from the thread title.
@Lost_N_BIOS : You’re absolutely right, it took forever 
I have revieved two cap files from you for the P9X79E-WS Board.
You did add Nvme Boot and Bifurcation.
One version contains both system agent sections, one conatins just the previously hidden one.
I will try out the one with both system agent sections first via Bios flashback.
If all works, I would want to get the other board’s bios (P9X79-WS) modded also.
I had sent you a bios region dump of that board.
Then, as troubles with the latter board and win10 setup began, I sent you my own nvme-modded 4901 Bios as cap file to inspect.
What I would request is that you could apply your bitfurcation mods to this bios also? Somewhere at that point we lost track.
.
.
Plus, I have a Z9PE-D8 WS here, a board with at least two completely independent and unswitched x16-Slots.
That would just per perfect for housing one or even two fully loaded Asus Hyper M.2 Cards.
I am not planning on using 8 NVme SSDs at his point, but the option of disabling some or even the most of the onboard sata controllers is very tempting.
With this Z9 Board it might be more difficult to implement mods, as it is a server/workstation board for Xeon E5?
@Fernando - yes, we’re still on the main topic here, revealing BIOS options to enable using PCIE bifurcation to run 2xM.2 on same slot, just taking our time and chatting a bit I guess you’d say 
But yes, that is the main goal here, with later added request of NVME addition, but main goal is topic title still. Since it’s not a guide or anything I assume strict title wouldn’t be required, otherwise if we cover too many topics at once for same mod then title becomes way too long
@paranoid_android - Later we realized you couldn’t use FPT, so dumps aren’t needed, but then you had that issue with the other board too so I was kind of waiting to see if you resolved that.
Thanks for the report back, and I appreciate you testing both to see outcome and which you prefer etc. and please note, I believe I only did minimal changes in those, not exposing everything, only meant as a quick test back then for us to decide which way was best to move forward.
My main goal/request back when we started with those BIOS, was for you to let me know one if both BIOS worked (so I knew both methods worked or not) and which you’d prefer.
One should have exposed both Sys Agent sections, and one only switched (Disabled the original, enabled the hidden). That’s why I named them like that, both and switch
“Both” probably not best, only meant as test, since all settings from the original are in the hidden one too, switch probably best one to use as a final to move forward with wouldn’t you agree
I’d have to look back in the BIOS again, but as mentioned I don’t think I enabled things, so may be missing stuff for now.
Yes, I will do both X79 variant boards for you, and the one for @JackSted too (please confirm JackSted, just plain P9X79, no PRO either?).
And yes, I can probably do the Z9 board too, looks similar (Settings at Advanced > Chipset > CPU IIO Bridge Configuration).
For that, make a new thread about it once we’re done here, so others can find easier later, thanks
Are you sure you want me to use your modded 4901, since maybe it was causing you issues?
I think since FPT is not going to be used, and others may want these BIOS later, maybe it’s now best for me to just do these mods on stock BIOS since USB Flashback used anyway so all board details kept by default
So now, about these BIOS you’re currently testing, can you see lots/all of the settings at System Agent >> IOH Config, or is many of the settings missing still?
We can enable if missing, no problem, but I left default/as-is at first in case they show up automatically with menu enablement. Compare BIOS you see with what you can see in AMIBCP images on page one post #3, do you see all those settings or only a few?
If only a few in there, please take some images and show me so I can tell what need further enabled.
http://www.asmedia.com.tw/eng/e_show_pro…ex=170&item=198
This chip connects x8 lanes and multiplexes to upto 12 PCIe ports and seemingly to 2 m.2 slots in on your card. So imo it doesnt need bifurcation. You could test if it is more stable on the x8 slot that is not connected to a PLX chip but to the CPU.
Your motherboard diagram suggests that the 2 PLX 8747 chips with the help of the QS chips Automaticaly switch from 1 x16 to 2 x8 physical slots for slots 2,3 and 7,8 when both sets of physical slots are populated. If a x8 slot from these sets is populated my guess is you get an x8 connection.
file.php?url=http%3A%2F%2Ffiles.homepagemodules.de%2Fb602300%2Ff16t4151p65397n4_OZhXVkYI.png&r=1&content=RE%3A_%5BREQUEST%5D_How_to_connect_2_M.2_SSDs_to_the_same_PCIe_slot%3F
I do not recommend the use of a “dumb” 16 to 4x4x4x4 card without a PLX chip on it like Asus HyperM.2 x16 on your P9X79-E WS motherboard to use 4 M.2 drives since you do not have a x16 directly connected physical slot for the bios to bifurcate. You could however just populate 2 of 4 slots on it and use it on the only directly connected PCie x8 slot you have.
Alternatively you could get a one NVME slot adapter to test and later a bunch considering you have alot of slots, x4 operation should work on all of them.
https://www.amazon.com/QNINE-Adapter-Exp…2+pcie+adapter
Considering the NVME read problem installing the 970 EVO NVME driver may help if you haven’t already.
Unless you’re using SATA RAID 0 or 5 or whatever I would leave the SATA Mode setting in bios in AHCI Mode.
The driver provided by windows, was stable for me for years.
I recommend you do not install drivers form asus.com for motherboards you cannot buy right now since most drivers will be horribly out of date.
Since windows 10 updates to a new kernel and driver model as time progresses i would stick with the chipset drivers you get with a fresh windows install from a windows 10 media creation tool to USB. I personaly have automatic driver updates disabled updates defered to major updates only for stability over security.
@Lost_N_BIOS
Yes I have the plain P9X79 which has 2x PCie x16 slots and 1 x8 slot which covers all 40 lanes, and the rest comes from the x4 speed DMI Southbridge I think. If you would like to make a bios for me i would humbly prefer bios 4701.
I have my doubts the P9X79E-WS PCIE PCIe ports can be bifurcated with bios since they come from plx and not direct from IOU2 and IOU3.
More specificaly the bios knows how to split eg: IOU2 into x8x8 or x4x4x4x4 but I doubt setting this would somehow work when PLX 8747 takes x16 and makes x16 + x16 + empty or x16 x8 x8. The exception would be IOU1 if that is in fact the one connected directly to a x8 port. This is the one I have working currently in x4 x4 mode with 2 NVME devices.
However testing would be interesting eg: maybe biforcation propegates through the plx and x8 x8 through PLX 8747 becomes x8 x8 + x8 x8 + empty.
You guys might want to check my guide I wrote last year on how I bifurcated a slot: [Guide] - How to Bifurcate a PCI-E slot
Couple of issues you might face is 1: Lack of a clock generator. 2: Timing issues. 3. Compatibility with mixing Nvme drives of different manufacturers. There are also a couple of optimizations that can be employed to increase success such as setting extended sync and sync timeout values. Recommend if you get it working to use HWinfo to study the link speed values to make sure. In anycase I doubt you can bifurcate a port thats controlled through a Plex chip on an Asus board. Not sure.
@JackSted - thanks for confirming your variant, I believe that board and the LE share same BIOS. I too would prefer the 4701 BIOS and I don’t even use these boards anymore. 
I’m also unsure how thay will work out with the PLX, but I do think you can reverse the slots, or disable certain ones and make only a few x16, disable lanes used, or whatever you want, so I think he may be able to reach his goal in some configuration with some slots
Please see page one, post #3, or look in AMIBCP and compare that with what you see in the BIOS now with this BIOS and let me know if settings are missing still, if so I can enable them.
All new settings will be at Advanced >> System Agent Config >> IOH Config
NVME mod done here as well, I assume you wanted, if not and you need without let me know and I can redo. If you want I will hop on your thread later and show you how this mod is done, it’s very simple to do, the System Agent switch I mean not the NVME http://s000.tinyupload.com/index.php?file_id=00616185236718716824 - Failed BIOS mod, do not use
@paranoid_android - here the other BIOS (Non-E model - 4901), NVME Mod and SA Switch, do not program any of these BIOS use USB Flashback only this applies to your BIOS above as well @JackSted http://s000.tinyupload.com/index.php?file_id=02667478253763083117 - Failed BIOS mod, do not use - final BIOS posted in post #74
@davidm71 - Thanks for the info and tips!
And I forgot, paranoid_android added some info on this PLX and lane assignment stuff on page two previously, from Anandtech and Asus I think
@Lost_N_BIOS
Thanks for the bios im now on raid 0 SATA SSD’s but will migrate to nvme for boot if it proves stable.
I would love it if you could drop some know how on my thread when you have the time.
@JackSted - That’s right, if the Iocrest Card with the Asmedia 2824 would work on my P9X79E-WS I would not need to think abput bitfurcation.
I have been using my 970 on a simple PCIe×4 to M.2 Adapter and that works fine.
But I wanted the explore the possibility of adding more M.2s to the system, that’s why I got the Asmedia/Iocrest card which is more like an HBA than a plain adapter.
The discussion began with my question why my system hangs when trying to write data to the 970Evo connected to that card.
Strangely its no problem to read from the 970, just any sort of writing operation (copying data or launching benchmarks) results im immediate system freeze.
Then the topic switched to the idea of adding bitfurcation so that I would not need the Asmedia or any bridge chip.
If I could just bitfurcate the sole ×8 Slot into ×4×4, that would be enough for the beginning.
The other board I wolud try out is the P9X79-WS which does not have PLX Chips.
If any slot on that one is directly rooted to the cpu is something I still have to find out.
I am thinking of moving the main system away from the E-WS to the WS, as the E-WS has shown instabilities and unreliable behaviour for quite some time without me finding a solution.
So if I could bitfurcate one of the ×16 into ×4×4×4×4, that would give me enogh room for adding M.2 connected storage and to have the Boards ×4 slots free for other adapters, like a 10G Card.
But in any case, the third one I 'd want to try out is the Z9PE-D8 WS, which has 2 unswitched and unsplit ×16 slots/ports directly from CPU2.
To get one of these or even both split into ×4×4×4×4 would provide me lot of M.2 connections, and since that machine would do WS or even Server Jobs, I could need such storage over time.
Still we’d have to consider what @davidm71 mentioned about clock generators and timing issues. That is something that I would have to rely on the contributions of others like @Lost_N_BIOS maybe.
.
.
The Win10 driver issues: I got these solved mostly through windows update.
Its right that win10 installs chipset and onboard component drivers/infos later after setup.
But I had to manually add the intel RST driver I got from this forum to get that work correctly.
At least it did speed up the win10 setup: The part of the setup procedure wich is called something like “gathering/collecting data” took hours(!) before the system was actually installed.
This had been the same kind of behaviour when I had installed Win7 on my E-WS machine.
In the second or third attempt to install win10 on the WS machine, I added the RST Drivers during setup procedure and the aforementioned gathering of data took merely 5 Minutes or so.
So the Intel RST driver does seem to make some difference - at least on this machines/setups I’m using…
What I did not solve is the drivers issue regarding the onboard marvell 9128 sata controller on the X79-WS board.
It gets identified by win10 as standard sata ahci controller but with exclamation mark.
Even disabling the controller in bios setup does not remove it from windows device manager.
Any attempt to install any sort of marvell .inf file leads to the situation that win10 would not start during the following boot-up.
The .inf file seems to get installed first and the controller looks OK in device manager.
But attempts to deinstall it or even any attempt of scanning for new hardware/drivers afterwards triggers some sort of lagging and does not complete until I reset/reboot the system.
The only fix for this is to revert to a previous set recovery point and to not touch the marvell controllers driver tab. I just let it deactivated in device manager for now.
Maybe the controller itself is defective?
9128/9123 is terrible, unless you absolutely need for additional storage I would disable in BIOS. Were you trying to boot from it? If not, I wouldn’t think installing driver could cause that with win10, but if booting from it then maybe due to driver issues.
Have you looked around here on the forums in the Marvell threads for win10 compatible drivers for 9128?
I can do same mod for Z9PE-D8 WS if you want @paranoid_android - make a thread on it when you get a chance, that way it’s easier for others to find in the future. Do you want NVME added to that one too?
I just spot the timing settings in the PCI subsystem settings hidden in AMIBCP would have to enable that menu and test.
@JackSted - I just added how-to on your other thread, to make visible the section you now mentioned you do similar except all you have to do is move the 29 02 following that setting to be at the end of the 46 02 instead (Directly after that, in front of setting ID), so that nothing is within the suppress constraints.
Or you can also try simply changing 46 02 to 47 02, but that does not always work.
@Lost_N_BIOS : I´’ll quote some of your questions to add some order to my replies…
No, I agree it’s best if you used the 4901 stock BIOS for the mods.
Meanwhile I flashed the stock one back on the X79-WS Board, but the problems related to the marvell 9128 remain the same. shrug
I never booted from it. And since I’m not that satisfied with the performance of the marvell controller on my X79E-WS board either (in that case a 9230), I try not to use it until necessary.
[On my X79E-WS/Win7U-System, the "Marvell Magni Driver" often causes terrible lags and mistakenly indicates attached drives as removable.]
However, disabling the 9128 in the X79-WS’s Bios does not remove it from Win10’s device manager.
Yes, I’d want to have bitfurcation and NVME support added to that one also.
Creating a new thread sounds OK.
@Lost_N_BIOS :
I applied your modded version of Bios 4901 for X79-WS using Bios Flashback method.
Good thing is, the machine still starts up and I can enter Bios setup.
But when I navigate to the Advanced Options Page it stays completely blank.
That is, title
I cannot do anything afterwards exept hitting the reset button. no other action of mine causes any effect.
Keyboard input seems disabled from that point. I can still move the mouse arrow, but clicks on items or tabs have no effect.
@paranoid_android - on 9128 - have you updated the 9128 FW, usually in BIOS there is an option to force update it to the version contained in BIOS, this is not loaded/used from BIOS itself it’s taken from BIOS and written into the 9128 chip when told to do so. I have not looked at your BIOS options regarding this, you may need setting enabled.
In windows, simply uninstall all drivers using the program or at first using device manager w/ delete driver option.
On the blank advanced image for X79-WS - does the other board do this as well? I thought you already tested previously, both BIOS switch and both and they were OK, correct?
If yes, and it’s just this recent mod on X79-WS, maybe I made a mistake, will need to check it over, but before I do that, confirm for me if it’s OK on the other board as I thought, or not and it’s same as this too
If both same, sorry, I moved forward thinking you confirmed the mod was OK already in your previous test comments.
If the other board is OK, and this one is not, and I find no error in mod process, then it simply means this BIOS needs edited in a different way, which I didn’t even consider and assumed same so did same way as the other.
@Lost_N_BIOS
Can confirm bios hang with bios I received, however with my own setup.sct having 46 -> 47 mods on System Agent Configuration, PCI subsystem and disable other System Agent Configuration via AMIBCD after everything works, PATA device even shows up as boot option (untested).