Allow B0 stepping on Asus W621E Sage

@TechStream Hi. Well actually yes, it works, tested with different CPUs for a while in different modes. Here’s the rom image
You still need to use SPI programmer like this - ali ebay. As for the software I recommend AsProgrammer, it is simple and works perfectly (google it). Use one of the following methods to connect programmer to the EEPROM chip:
a) SOIC16 clip like this - ali ebay. You might have to modify the header accordingly to the programmer pinout.
how it looks like:







b) USB3.0 female header fits perfectly to the onboard SPI header, you can modify one to use it instead of the clip. I have used Arduino’s breadboard solderless cables, they do the trick perfectly and are easy to buy.
Photos:









c) Or you can just desolder the chip.
d) Also probably you can use flashback feature to flash the modified bios, but it is untested method by me and also you will need expensive retail CPU in this case instead of programmer. The image I have provided is NOT compatible with flashback as it is raw bios dump wthout the capsule which is needed for flashback!

Pay attention to the pinout, you can damage you bios chip or even something else on the motherboard if you connect something wrong!!
As soon as you connect to the bios with programmer make a dump of the original bios image, also i recommend to make BMC firmware dump(the chip is located at bottom left corner on the detachable green pcb).
After you flash the modified bios don’t forget to clear Cmos.
On the first power up it takes a while and system will reboot 4-5 times, it is normal behavior for the first time.
I don’t know why but UDIMM ECC modules don’t work with the modified bios, it seems to be a SecCore issue. So if you are using the mobo with retail stepping CPUs with UDIMMs (which is possible, tested) and wanna switch to B0 CPU, you need to change memory modules as well (use RDIMM/LRDIMMs, I have tested several different sticks and everything works fine)

Other pics








In case some of you wants to buy me a coffee
btc: 19riGy1BSyHvRZSAkMjHwnmSyEGYrDwutq
Paypal: [email protected]

@avd

Thats amazing, can you confirm what you changed in the bios and also which CPUs you used?

Also one other question @avd

You say ecc udimms no longer work, dont suppose you could check with some none ecc memory.

@TechStream
1. You have to have the new seccore and microcodes - that’s it.
2. tested with ql1l, ql1k. ql1f for 1-2 weeks.
3. I meant ecc udimms, of course, and i have tested the same sticks which were tested and worked on the same c621e with original bios and retail xeon 6152 cpu. And the same sticks works great on Supermicro x11 with b0 cpus. But on the c621e with b0 they don’t. It makes sense because according to the postcodes all the problems with b0 stepping are memory-related - incompatible Supermicros show post code 53 which is a mem error, Asus showed bb after cycling of b1-7 or something which are also memory-related, and so on. AFAIK the problem was something related to memory training on very initial stages of initialization (a guy from apple uefi dev dep tried to help me and told me that). But i’m not an expert in low level engineering, especially with this platform as there are a lot of platform-specific asm code. Anyway, udimm ecc sticks don’t work as well as non-ecc and i haven’t try non-ecc with retail cpu but pretty sure it wouldn’t work as well. Don’t think it’s a big problem, noted about it just FYI.

Seems to work for my QL29s. I am having a little trouble booting, when the OS (Linux or Windows) tries to start, the machine resets. Not really sure what is causing this, haven’t had time to play yet.

Thanks for caving in @sochotek _ was going to ask you to provide the BIOS dump once your board arrived back, but looks like @avd did this right after your post - thanks to you both!

I will dissect the BIOS and see what has changed, then I can do same changes on latest BIOS for you guys, unless this is already latest BIOS (*Confirmed this is latest BIOS).
Now, we need someone who has not sent in their board, to program in this BIOS and confirm it works without that seller touching your board. @malucul - are you confirming this now??

@Lost_N_BIOS it at least POSTs, which is substantially more than it was doing before. I haven’t actually managed to get an OS to start yet though. I’ll play some more tonight, this could well be something weird about my particular setup.

Update: Just found a few minutes to have another look. I am about 90% certain that it’s a memory problem.

@malucul - thanks for reporting back. How did you flash in the BIOS, via flashback or FPT, or programmer? Be careful guys, if you use programmer or FPT you will loose you serial, UUID, and LAN MAC ID

Here’s my comparison report on the mod BIOS vs stock, with about an hour or so dig in
BIOS Version = 5503 - 2/15/2019

Changes I see vs stock -

FIT table is sloppily re-constructed (maybe accidently, or on purpose?)
To me, it looks like it was simply poorly edited by someone that is unfamiliar with editing this file, doesn’t look like something Asus would do or anything that would allow/help this mod work, but someone would have to test if these CPU’s still work if I fix this.
Anyone that knows about editing FIT, will be able to look at the file and see the same as I do I assume, poor edit and moved bottom entries to just below microcode entries but done incorrectly/all messed up

Microcode for 50654 (Rev 2000057) is removed
Microcode for 50655 (Rev 3000010) is removed

Microcode for 50656 (Rev 4000017) is replaced by >> Rev 4000010
Microcode for 50657 (Rev 5000017) is replaced by >> Rev 5000010

Microcode for 50652 is inserted (rev 8000037 - 2/5/2017)

ME version 4.1.4.251 is replaced with older >> 4.1.3.237 (Still need to dive into settings to see if anything aside from version is changed)

FD region is exact match - so USB Flashback can be used, provided it also updates/downgrades the ME region

DevExp1 Exact Match

PTT region, this contains Data - stock one does not (all FF), this is solely due to this being a dumped BIOS vs stock thing. I need a dump from someone’s unmodified board using 5503 BIOS to compare with and see if anything different for the mod.

A NVRAM stock volume, standard defaults, contains changes and 3 additional module entries (IPConfiguration + ServerSettingVarName -both may be unrelated + AMIFwConfigSrvSetup module entry which I assume is related) + many Changes inside some entries that are in both (I assume much of this mod related)
@ Setup entry (many changes), @IntelSetup 5x changes, @ SocketMPLinkConfig x1 change, @SocketMemoryConfig x36 changes, @ SocketPowerManagementConfig x3 changes, @ SocketProcessorCoreConfig x1 removal, @ServerSetup x2-3 changes, @FakeVar x1 change - @GUIDStore 30 bytes removed
+ 10 bytes different (assumed all due to these additions and changes, not related to mod directly)

Comparing BIOS dropped into FITc to compare ME’s just by XML settings output, I see the following And these may solely be due to the ME FW version differences, unsure.

The box, to the left of PCH + Platform by default says Full with mod BIOS and SiEn with stock BIOS (That may be due to version differences, or dump vs stock?)

This is removed twice from stock BIOS (removed in mod BIOS/ME)
<variable name=“NmEnabled”
value=“false”/>

This is removed once from stock BIOS (removed in mod BIOS/ME)
prev <file name=“PchHybridProfile”
enabled=“true”>
<variable name=“LCPLL_SSC0_BCLK”
value=“0x0010051c”/>
</file>

Can’t compare ME using ME Analyzer - adv mod BIOS above ME FW seems broken when dropped on MEA >> MFS Low Level File xxx which has not been parsed (Error maybe 50+ times and that’s all)

plutomaniac will have to check that out


First, 2nd and 3rd BIOS volumes (below NVRAM volumes) are hex identical
First padding differs, but this may be due to dump vs stock

4th BIOS volume (Main BIOS DXE Volume) has many changes - some I assume mod related, but possibly not all, hard to know when you’re comparing dump to stock BIOS
Due to volumes and everything being compressed, I don’t know how to easily compare modules (except extract one by one and compare in hex) vs extract volume and compare in hex (which is comparing compressed modules as an entire compressed file).
If I could easily compare just the volumes side by side, uncompressed it would be easier to see what all is different, what might be mod vs stock and what is stock vs dump etc.

Padding (last in open region) which contains FIT also has MANY other differences
Last BIOS volume has some changes (many)

Here is textual output of all BIOS settings and their default values, if anyone wants to compare
http://s000.tinyupload.com/index.php?fil…929118443252951

I assume, the seller put this BIOS on mainly via USB Flashback if your serial, UUID, LAN MAc all same, unless he took the time to copy that out and then back in once done which can be tedious vs flashback = painfree.
Then once done with flashback, used programmer to put in ME Region. I assume flashback used for main BIOS re-do as it’s easier to keep all system details retained that way, and then either unlock FD via programmer or simply write in the ME region only via programmer and be done.
However it could be done many different ways I assume, but I do assume it can be done via flashback (provided your board will downflash ME, or you have applied hard mod to enable FD/ME re-write via flashback
(As discussed here) - [GUIDE] Coffee Lake CPUs on Skylake and Kaby Lake motherboards (62)

Pinmod could also be done to quickly unlock FD/ME and downflash the ME too, user prerence how to get the ME Backflashed, but it’s possible without programmer, and possible without using USB Flashback too (or with flashback as well)

Ok, so I pulled one chip (because I only have one stick of 2133 RAM) and changed the memory to a Kingston part (KTN-PL421). I got 10 minutes into memtest and it didn’t fail (previously it was dying almost immediately). I have also managed to boot into Linux from SRCD. I will try the other CPU in a bit to try and rule out that being the issue.

Dump from /proc/cpuinfo if anyone is interested - https://pastebin.com/69E7EnTe

Ok, I reseated everything and now appear to be running correctly with both sockets populated using Kingston KVR24R17S8. I’ll install windows now and fire up CPU-Z, Cinebench etc

@mitdal369 @levelud @maxdata @levelud @myio @Branen @Onyxx1606 @Moomoo @Kaboooom2000uk @Onyxx1606 @dahuoxia <<< Bump for any of you not subscribed, see above for BIOS ^^

@malucul dont suppose you could test with some none ecc ram, just i literally have piles of it here, and the project im building doesn’t require ecc

@TechStream Would if I could, but unfortunately I don’t actually have any. Sorry man :frowning:

In other news I got the CPU-Z and Cinebench screenshots. Will upload after dinner.

Well some news from me, I just finally managed to get my board up and running with a retail bronze 3104, and some Corsair Vengance LPX 2666Mhz ram. And thanks, i’ll have to check that bios out. Keep up the fantastic work you guys. I really appreciate your time and trouble you put in.




Ooo so proof of working with none ecc memory!

If flashback seems to be possible can someone made the file?
If i could I would do it myself, but i can‘t.
Perhaps someone with retail cpu and programmer can test it.
I‘m proud of you guys you made an amazing job.

hi, thank so much for your work.
i would like to know what if the board can run with step H0 (ES CPU, not retail), and the revision 50654h?
also, since Cascade is out, i wonder what if the mod bios can work with new cpu (ES cpu, same H0, and revision 50654h)?

@nhan72nn

We would nee to know full cpuid to be even begin to help. If you go through the last few posts you can see which cpuids have been tested and confirmed to work.

hi TechStream, i am thinking to buy this http://www.cpu-world.com/sspec/QM/QMRV.html, but not sure if it work with the board or not. i have checked the posts, only mention to QL1K or QL1L, etc, no one has mentioned about QMxx which stepping is not B0 but H0.

That QMRV is a QS (100% identical to retail except for the product string and IHS lasering) - it will work on the board without needing any mods.

@nhan72nn - In the mod BIOS avd posted, you’ll need to insert some version of 50654, then it may work, but unsure which ucode revision to use - there is none for 50654 in there right now (see my notes on post #208)
Stock BIOS may work as-is with the CPU you mentioned, you’ll have to test first and find out what happens, before you go looking into mod BIOS.