Help Installing Big Sur on Xeon W-3375 and Supemicro X12SPA-TF

[Loloflat6]

@Balamut 

 

Nice job : you succeeded !  👍

 

Did the point on I210 Ethernet deactivation as suggested @XLNC help you, and lead you to the goal ?

[Balamut]

11 hours ago, Loloflat6 said:

Did the point on I210 Ethernet deactivation as suggested @XLNC help you, and lead you to the goal ?

Yes it did, somehow connecting cable to the NIC was causing the freeze. I just unplug the cable, could be i210 or IPMI issue.

 

12 hours ago, etorix said:

Sleep will be an interesting issue… knowing that @yapan4's X11SRA-F does not sleep by design from Supermicro. @metacollin's X11DAi-N did not sleep either.

One thing at a time I guess, I gonna replace RX 580 with 5700 XT or Radeon VII if I can find right thermal pads for it and see if there is any difference.

 

12 hours ago, etorix said:

Was it switching to Big Sur and/or disabling i210 NIC? Or something else totally unrelated?

Nope, it was just as simple as unplugging the cable.

 

Results are better then windows., this one is with plugin = 1

https://browser.geekbench.com/v5/cpu/11479501

 

 

That hec2 is still bothers me as it gives error, any thoughts on it?

 

 

 

[Balamut]

Hm, looking into IOReg, there is HEC2, HEC3 but no HEC and out of all the SSDT's only PMC and RTC0 got changed.

Did I do something wrong?

 

 

 

 

Files:

ACPI.zip

[etorix]

IPMI has its own Realtek PHY that is managed by the BMC.

Quote

Single LAN with Realtek RTL8211F PHY (dedicated IPMI)

I have the same in my Gigabyte C621-SU8 and have not noticed any issue. But I never had it plugged as the sole connection, there always was another cable to an i210 or a Chelsio T-520 AIC. Never tried to use it as a network connection for OS X, only for monitoring through IPMI.

 

HEC2 is another suspect for the sleep issues (HWAK… wake?). But I have no clue why it fails to load.

 

 

[Balamut]

8 minutes ago, etorix said:

IPMI has its own Realtek PHY that is managed by the BMC.

I have the same in my Gigabyte C621-SU8 and have not noticed any issue. But I never had it plugged as the sole connection, there always was another cable to an i210 or a Chelsio T-520 AIC. Never tried to use it as a network connection for OS X, only for monitoring through IPMI.

I usually the cable connected to the 10GB Nic too, but since it was on the test bench just had a cable to monitor the temperatures remotely. Live and learn.

 

 

 

 

8 minutes ago, etorix said:

HEC2 is another suspect for the sleep issues (HWAK… wake?). But I have no clue why it fails to load.

That and maybe even the random freeze issues.

 

 

Any apps to do a system burn/stability tests? I know plenty of them on the Linux/Windows side but note on the Mac except the regular suspects.

 

 

 

[Balamut]

Another Update.

 

AQC 113 works with @etorix help, problem, after pluging in cable instant reboot. Will test more later on tonight.

[etorix]

Credit for SSDT-AQC113 goes to Elias64Fr; I found it in the Z690 thread of the "other forum". But "instant reboot" is not quite "working".

 

The ACPI folder has two SSDT for EC (normally, SSDT-EC-USBX is what's needed), and I'm not sure what SSDT-XOSI is doing (nothing?).

The "clean" IORegs shows the native RTC, with its three memory regions. If it boots this way, SSDT-RTC0-RANGE is not necessary.

There actually is a PMC1 device in the DSDT, which does not look related to NVRAM. Can you check whether you have working NVRAM? If so, it is native, and SSDT-PMC is also not necessary (if it is, some change is needed; I also learn every day).

[Balamut]

2 hours ago, etorix said:

Credit for SSDT-AQC113 goes to Elias64Fr; I found it in the Z690 thread of the "other forum". But "instant reboot" is not quite "working".

Might be related to installer freezing when cable was connected to i210. 
 

2 hours ago, etorix said:

Can you check whether you have working NVRAM?

I’m pretty sure I do, gave me a headache when I was going back and forth between COM output and not. 
 

 

Will test some more tonight. 

[XLNC]

@Balamut I210LANFix.kext.zip
Add the above kext to OC and you will be able to use your i210 ethernet port without panic/freezes.
For the memory error notification you are getting , just add RestrictEvents.kext to fix it.
And for sleep make sure you have AppleXcpmExtraMsrs kernel quirk enabled in oc config.

its been a while but i have hackintoshed the supermicro X11SPA-T mobo and the ASUS WS C621E SAGE mobo .

Edited December 13, 2021 by XLNC

[etorix]

@XLNC I see this kext adds an IOPCIPrimaryMatch for i210. Is it a general solution for i210 issues under Monterey? Where does it come from?

Edited December 14, 2021 by etorix

[Balamut]

20 hours ago, XLNC said:

Add the above kext to OC and you will be able to use your i210 ethernet port without panic/freezes.

Thank you! Will test it in a bit, can it also be modified to work for AQC113 to prevent panic/reboot?

 

 

20 hours ago, XLNC said:

For the memory error notification you are getting , just add RestrictEvents.kext to fix it.

Yeah done it.

 

 

20 hours ago, XLNC said:

And for sleep make sure you have AppleXcpmExtraMsrs kernel quirk enabled in oc config.

The quirk is enabled, I think the problem might be with either whatevergreen or rx580 itself.

 

 

20 hours ago, XLNC said:

its been a while but i have hackintoshed the supermicro X11SPA-T mobo and the ASUS WS C621E SAGE mobo .

Nice, I have ASUS Pro WS C621-64L SAGE/10G works like a charm except can't get the second GPU to be recognized and work.

 

11 hours ago, etorix said:

@XLNC I see this kext adds an IOPCIPrimaryMatch for i210. Is it a general solution for i210 issues under Monterey? Where does it come from?

Not sure if it applies here, but my issue for reboot/panic is under the Big Sur, will install Monterey today.

EFI.zip

Edited December 14, 2021 by Balamut
Attachment

[etorix]

@Balamut That's a complex setting, although not everything is actually enabled… Are there ACPI errors during boot? They should show up using one of CaseySJ's tricks:

log show --last boot | head -1500 > ~/Desktop/bootlog.txt

 

[Balamut]

On 12/14/2021 at 12:03 PM, etorix said:

@Balamut That's a complex setting, although not everything is actually enabled… Are there ACPI errors during boot? They should show up using one of CaseySJ's tricks:

log show --last boot | head -1500 > ~/Desktop/bootlog.txt

 

 

 

Actually there are some:

2021-12-18 08:46:15.047538-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) ACPI Error:
2021-12-18 08:46:15.047539-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) ACPI Error:
2021-12-18 08:46:15.048041-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) [DTGP]
2021-12-18 08:46:15.048041-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) [DTGP]
2021-12-18 08:46:15.048287-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform)  Namespace lookup failure, AE_ALREADY_EXISTS
2021-12-18 08:46:15.048288-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform)  Namespace lookup failure, AE_ALREADY_EXISTS
2021-12-18 08:46:15.050107-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform)  (20160930/dswload-462)
2021-12-18 08:46:15.050108-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform)  (20160930/dswload-462)
2021-12-18 08:46:15.051120-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) ACPI Exception: AE_ALREADY_EXISTS,
2021-12-18 08:46:15.051121-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) ACPI Exception: AE_ALREADY_EXISTS,
2021-12-18 08:46:15.052571-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) During name lookup/catalog
2021-12-18 08:46:15.052572-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) During name lookup/catalog
2021-12-18 08:46:15.053651-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform)  (20160930/psobject-310)
2021-12-18 08:46:15.053652-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform)  (20160930/psobject-310)
2021-12-18 08:46:15.054814-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) ACPI Exception: AE_ALREADY_EXISTS,
2021-12-18 08:46:15.054815-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) ACPI Exception: AE_ALREADY_EXISTS,
2021-12-18 08:46:15.056265-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) (SSDT:MCHCSBUS) while loading table
2021-12-18 08:46:15.056266-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) (SSDT:MCHCSBUS) while loading table
2021-12-18 08:46:15.057724-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform)  (20160930/tbxfload-319)
2021-12-18 08:46:15.057725-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform)  (20160930/tbxfload-319)
2021-12-18 08:46:15.059061-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) ACPI Error:
2021-12-18 08:46:15.059061-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) ACPI Error:
2021-12-18 08:46:15.059555-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) 1 table load failures, 16 successful
2021-12-18 08:46:15.059556-0800 0x71       Default     0x0                  0      0    kernel: (AppleACPIPlatform) 1 table load failures, 16 successful

2021-12-18 08:46:14.053896-0800 0x74       Default     0x0                  0      0    kernel: (DebugEnhancer) <compose failure [UUID]>

Can't seem to install Monterey, hard reboot on the 3rd stage of install, going to hook up COM port see what it says.

 

Slow steps, but there will be progress.

 

bootlog.txt

 

 

 

UPDATE:

 

After replacing the RX 580 with Radeon VII with whatevergreen, Cinebench ran perfectly and seems like no more panics.

Sleep still doesn't work, monitor goes dark but system won't sleep, need to study further. Monterey still gives issue, and looks like Intels Power Gadget doesn't work on the system.

 

Edited December 19, 2021 by Balamut
Updates, Cinebench screen

[etorix]

I think there's only one ACPI error: Method \DTGP is defined both in SSDT-DTPG and in SSDT-SBUS-MCHC (what is it doing here?). Either disabling SSDT-DTPG or removing DTPG from SSDT-SBUS-MCHC should clear all errors.

(Some general cleaning may be in order, like SSDT-EC-USBX which is only USBX, with EC being in its own SSDT, but this is only cosmetic.)

 

USBInjectAll.kext is loaded AND XhciPortLimit is enabled, although it doesn't work with Big Sur 11.6.1. I would make a USB map first and then investigate sleep and stability.

You may also inquire with Supermicro whether the X12SPA-TF actually supports sleep, like @yapan4 did for his X11SRA-F.

 

Have you tried setting SecureBootModel to 'Disabled' to install Monterey? On the "other forum", many users need that.

 

Nice benchmark! +50% over the W-3265M with "only" 33% more cores, running at a 200 MHz lower base clock (generational IPC improvement at work here).

[Balamut]

Merry Christmas and Happy Holidays 

 

 

On 12/19/2021 at 5:43 AM, etorix said:

Have you tried setting SecureBootModel to 'Disabled' to install Monterey? On the "other forum", many users need that.

Hm oddly that didn't work still reboot.

 

 

On 12/19/2021 at 5:43 AM, etorix said:

Nice benchmark! +50% over the W-3265M with "only" 33% more cores, running at a 200 MHz lower base clock (generational IPC improvement at work here).

Yup, decent improvement over the W-3365M, can't wait to get it fully working.

 

On 12/19/2021 at 5:43 AM, etorix said:

You may also inquire with Supermicro whether the X12SPA-TF actually supports sleep, like @yapan4 did for his X11SRA-F.

Will shoot them an email after the Christmas.

 

[Balamut]

Hey guys, Happy New year to all, sorry for MIA work is  keeping me extra busy lately, just got a small update.

 

Monterey Installer still reboots on the 3 install screen and Supermicro support told me that the motherboard doest NOT support sleep because the bios is based on the "Server" bios. 🤦‍♂️

 

Thinking about the ASrock  with the X710 and limited USB ports

 

 

 

 

[etorix]

Thanks for the update. Now we have a confirmation that Supermicro boards just don't sleep.

 

I could not find an AsRock C621A motherboard…

Are you thinking about this AsRockRack server motherboard, or its 2L2T sibling?

 

I had a doubt, but the QVL does list Xeons W-3000 in addition to Xeon Scalable. Whether it sleeps is an open question.

Edited February 2, 2022 by etorix
fixed link

[yapan4]

Looking on successful hackintoshing on Alder Lake, can be worth 
 wait for new Alder Lake-based Xeon Scalable/HEDT platform... who knows...

[etorix]

Good point: Sapphire Rapids is multi-chip but not not hybrid, so should not need new quirks.

Bad news: Sapphire Rapids has been "delayed until 'maybe' Q3 2022".

https://www.tomshardware.com/uk/news/amd-raising-epyc-cpu-pricing-10-to-30-intels-sapphire-rapids-delayed-report

Assuming this holds, 4th generation Xeon Scalable would be announced in Q3 (September?), become somewhat available at retail in Q4, the workstation variants Xeon W-3400 would come in Q1 2023, and HEDT Xeon W-2300 (W-2400?) for LGA21xx would come after that—if they ever come.

 

It will be a long wait.

[Balamut]

13 hours ago, etorix said:

I could not find an AsRock C621A motherboard…

This one https://www.asrock.com/mb/Intel/C621A WS/

But you can't buy it ANYWHERE, my guy who I buy stuff for business told me that ASRock got 100 MOQ for that board.

 

7 hours ago, etorix said:

Assuming this holds, 4th generation Xeon Scalable would be announced in Q3 (September?), become somewhat available at retail in Q4, the workstation variants Xeon W-3400 would come in Q1 2023, and HEDT Xeon W-2300 (W-2400?) for LGA21xx would come after that—if they ever come.

I doubt very much Apple will go with the 4th Gen, Q3 (still questionable) PLUS knowing intel the vendors will get their somewhere in q2 of 2023 if they're lucky, Apple being Apple and squeezing Intel out of the picture I personally believe the W-33xx will be the last MacPro with PCIE4, but hey I've been wrong before.

 

11 hours ago, yapan4 said:

Looking on successful hackintoshing on Alder Lake, can be worth 

The W-33xx is pretty good and fast and what I can see from the small amount of tests I've done pretty solid and very fast under Big Sur. I mean I'm still fighting with Monterey but its more on the Monterey side than anything, and for the sleep... Well I'd love to have it but if I can't its not the end of the world, the board and the parts are made to be run 24/7 at full load and even at idle it shouldn't be eating more than 200-250W. It's not that of a deal breaker compare to what you're getting, full PCIE4 lanes, 4 NVME, DDR-3200, 10GigE, Plenty of USB 3-3.1 at el.

 

On the side note the Supemirco tech, actual BIOS team will be back after the Chinese New Year sometimes next week so I'll have more details about this stupid screw up. Also asked about CFG Lock and Resize Bar, the lack of the Resize Bar in BIOS is quite baffling to be honest.

 

Off Topic: Anyone managed to run Dual GPU system?

 

[etorix]

15 hours ago, Balamut said:

This one https://www.asrock.com/mb/Intel/C621A WS/

But you can't buy it ANYWHERE, my guy who I buy stuff for business told me that ASRock got 100 MOQ for that board.

Thanks for the link. I had seen the motherboards months ago, but, somehow, I could not find it back by searching for 'C621A'…

What I cannot find either is a technical difference with AsRockRack SPC621D8-2L2T: Same layout, same slots (2*M.2 as PCIe 3.0 x4 + x1 for instance), same ports/headers (including the uncommon two USB 3.2Gen2 Key-E, very conspicuous on the supposedly server SPC621D8), same absence of audio codec (very conspicuous on the supposedly workstation C621A WS)… Just server green vs. gamer black.

I can understand that the distributor fears being left with 99 boxes on his shelves.

 

[Balamut]

9 hours ago, etorix said:

What I cannot find either is a technical difference with AsRockRack SPC621D8-2L2T:

https://www.newegg.com/asrock-rack-spc621d8-2l2t/p/N82E16813140064

 It does look identical and the sleep seems to work on it.

From the manual which looks exactly like the WS one

 

 

Come to think of it I started living this board more and more, one of the biggest difference I can see is that the M.2 is connected to the south bridge whilst the Supemicro one is shared from the PCISL1 trough the CPU, I mean the lather is PCIE Gen 4 but doubt anyone can actually get those speeds on it.

 

I don't I ever seen this on the server board before.

 

 

Any thoughts fellas? I mean the x710 is dead in the water with 0 support but If Id want 10G I'll just grab the AQC103 PCIE cards.

 

The only issue I see is the amount of memory and lack of Audio on the motherboard.

 

 

 

 

 

Edited February 3, 2022 by Balamut

[etorix]

15 hours ago, Balamut said:

Come to think of it I started living this board more and more, one of the biggest difference I can see is that the M.2 is connected to the south bridge whilst the Supemicro one is shared from the PCISL1 trough the CPU, I mean the lather is PCIE Gen 4 but doubt anyone can actually get those speeds on it.

Server boot drives do not require PCIe 4.0 speed anyway, and this strongly looks like an AsRockRack design with an additional request from the sister company AsRock to share the design so it could be rebranded as "workstation" for the desktop. (Maybe 100 pieces is the required number to actually launch the manufacture of a batch on black PCB?) The good side is that there are similar mentions of S4/C6 sleep for at least the previous generation of C621/C622 server boards, so presumably AsRockRack engineers have a better grasp of this than their colleagues at Supermicro.

 

Well, if you can't resist the temptation to go for a SPC621D8 (base or -2L2T), it will be an interesting second hack to follow. Hopefully, with the benefit of experience it will be easier and quickier than the first Supermicro hack.

[yapan4]

@Balamut

Im not shure, you need S3(Suspend to RAM), not S4(Suspend to Disk or Hibernation) state. S4 is supported on Supermicro but problematic in macOS. Probably regarding Sleep, ASrock has the same capability as Supermicro  

Edited February 4, 2022 by yapan4

[Balamut]

6 hours ago, yapan4 said:

Im not shure, you need S3(Suspend to RAM), not S4(Suspend to Disk or Hibernation) state. S4 is supported on Supermicro but problematic in macOS. Probably regarding Sleep, ASrock has the same capability as Supermicro  

 

ASRock rep reply:

Quote

For ASRock server boards unit, our BIOS team does not provide S3 Sleep option in BIOS. For the rest of questions, please bear with me for a few moments. I will further consult the BIOS team