Help installing HighSierra on Xeon W-2175 and Asus WS C422 mobo

[Guest]

20 minutes ago, obus said:

I'am happy  @fabiosun and without your help I don't know what to do.

Now I have thunderbolt working without hot plug obviously but with function wake after sleep.

Sleep is not working 100%. The rig is going to sleep but wakes after a while and crash Without WhatEverGreen sleep doesn't work at all.

Any ideas?

My usb-stick shows up as internal but my buffalo thunderbolt drive as external?

 

@obus

you have a unique system and you are a sort of Pioneer.

I have no evidence of others xeon system like your working perfectly using traditional X299 guide

So it is fine and right you are happy!:)

Minor problems could be solved by you searching and doing experiences on an inexperienced system

 

[jaraheel]

Hi @obus Can you please share your EFI folder and BIOS settings? I had a stable run for about three weeks, and not am back to square one  The EFI folder from @fabiosun on page 6 of this thread is no longer available (which I had been using ever since it was posted)..

[obus]

6 hours ago, jaraheel said:

Hi @obus Can you please share your EFI folder and BIOS settings? I had a stable run for about three weeks, and not am back to square one  The EFI folder from @fabiosun on page 6 of this thread is no longer available (which I had been using ever since it was posted)..

Bios settings according to KGP:s Guide here: https://www.insanelymac.com/forum/forums/topic/331658-how-to-build-your-own-imac-pro-successful-buildextended-guide/ thunderbolt  Reserved memory 737 and reserved  pmemory 1184. 

For me "npci=0x2000" is mandatory from 10.13.4

 

CLOVER.zip

Edited April 11, 2018 by obus

[jaraheel]

17 hours ago, obus said:

Bios settings according to KGP:s Guide here: https://www.insanelymac.com/forum/forums/topic/331658-how-to-build-your-own-imac-pro-successful-buildextended-guide/ thunderbolt  Reserved memory 737 and reserved  pmemory 1184. 

For me "npci=0x2000" is mandatory from 10.13.4

 

CLOVER.zip

thanks so much!

[obus]

On 3/19/2018 at 7:42 PM, fabiosun said:

@obus

in my opinion performance is fine and you should leave EIST enabled because performance is pretty the same and you have a better states use with it

 

you should also search for spec in AVX load for all your cores and cpu and if under load they stay to they maximum

My cpu (x99 Xeon 22 cores / 44 threads) has a base clock of 2.2 Ghz a turbo clock for all cores/thread of 2.8Ghz and a turbo core for 2 core/thread of 3.7 Ghz

You have to find your AVX maximum and test it with some software like cinebench15 or rendering program but I am confident your system is fine

Maybe you should find a way to not use Broadwell EP fakecpuid because your CPU should be supported vanilla by osx 10.13.3 but there you have to do some test by yourself

I repeat my self..your cpu is a Xeon and is different...follow your guide and stay simple as you can 

you should also try to set with clover configurator plugintype (check this) and see if your system starts

Also I don't know if also new xeon freeze using appleintelinfo.kext by piker alpha..

Unlocking bios is not useful for performance at all...

Trust me

Hi again @fabiosun.

I wasn't sure I have native powermanagment so I did a run with AppleIntelInfo.kext and got attached result. Could you please have a look and se if there is something "weird".

How should I try to get rid of FakeCpuId. You sad earlier that you think that this Xeon W-2175 processor could be supported vanilla in 10.13.3. I am running 10.13.4 now and really don't know how/what to test? 

PluginType is enabled in clover.

AppleIntelInfo.rtf

Edited April 19, 2018 by obus

[Guest]

7 hours ago, obus said:

Hi again @fabiosun.

I wasn't sure I have native powermanagment so I did a run with AppleIntelInfo.kext and got attached result. Could you please have a look and se if there is something "weird".

How should I try to get rid of FakeCpuId. You sad earlier that you think that this Xeon W-2175 processor could be supported vanilla in 10.13.3. I am running 10.13.4 now and really don't know how/what to test? 

PluginType is enabled in clover.

AppleIntelInfo.rtf

Hi

AppleIntelInfo output is fine, maybe you have to test more and relaunch it to have more C6 states steps but I think they will appear after a more consistent test (benchmark/idle/benchmarfk idl and so on)

You miss C3 States for an ideal and theoretical perfect XCPM..but this could be the result of a locked msr 0xE2 as you have

In my case with a locked bios I have your identical result

With unlocked one I can program MSR 0xE2 as @SammlerG finding to have also C3 states but for general XCPM performance/condition same result here (performance/temperature)
You can try to disable in clover bootmenu fakecpuid or some kernel patches if you are using it..but I don't think you achieve better results of one you show here

Is your system rock solid?
If so..try but stay also fine with it

 

 

 

[obus]

2 hours ago, fabiosun said:

Hi

AppleIntelInfo output is fine, maybe you have to test more and relaunch it to have more C6 states steps but I think they will appear after a more consistent test (benchmark/idle/benchmarfk idl and so on)

You miss C3 States for an ideal and theoretical perfect XCPM..but this could be the result of a locked msr 0xE2 as you have

In my case with a locked bios I have your identical result

With unlocked one I can program MSR 0xE2 as @SammlerG finding to have also C3 states but for general XCPM performance/condition same result here (performance/temperature)
You can try to disable in clover bootmenu fakecpuid or some kernel patches if you are using it..but I don't think you achieve better results of one you show here

Is your system rock solid?
If so..try but stay also fine with it

 

 

 

Thank's for your input @fabiosun.

My system is rock solid for the moment with working sleep/wake. 

I will test more and see if I can fine-tune a little bit more. It should be very nice to have my MSR 0xE2 register unlocked from bios. Do you have any ideas how to do this.

I used to use Ozmosis for many years so I have some experience in patching bios but I need to know what to do!

Is there any guides out there?

Edited April 20, 2018 by obus

[Guest]

2 hours ago, obus said:

Thank's for your input @fabiosun.

My system is rock solid for the moment with working sleep/wake. 

I will test more and see if I can fine-tune a little bit more. It should be very nice to have my MSR 0xE2 register unlocked from bios. Do you have any ideas how to do this.

I used to use Ozmosis for many years so I have some experience in patching bios but I need to know what to do!

Is there any guides out there?

Hi

if you want try to rename patched one according your flashback bios name and flash it if you want

included original 0603 and patched one

To start you can also visit issue section of UEFIPATCH GitHub and use one tool like UBU to separate bios sections and find correct location and fix.. not simple at all

 

Edited April 20, 2018 by Guest
link removed (cfg lock should be in 0603 bios by default)

[Guest]

then post again Appleintelinfo output and your config.plist

 

Edited April 20, 2018 by Guest

[obus]

1 hour ago, fabiosun said:

then post again Appleintelinfo output and your config.plist

 

There was a new tab in bios.

So this is with MSR lock disabled in the new patched bios.

But I suppose here was no change in MSR 0xE2?

AppleIntelInfo.rtf

config.plist

Bootlog.rtf

Edited April 20, 2018 by obus

[Guest]

1 minute ago, obus said:

So this is with MSR lock disabled in the new patched bios.

But I suppose here was no change in MSR 0xE2?

AppleIntelInfo.rtf

You suppose bad

Now you MSR 0Ex2 is unlocked

MSR_PMG_CST_CONFIG_CONTROL.......(0xE2)  : 0x402
------------------------------------------
 - I/O MWAIT Redirection Enable......... : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
 - CFG Lock............................. : 0 (MSR not locked)

in this condition you can try different way to achieve also c3 states...

 

[obus]

3 minutes ago, fabiosun said:

You suppose bad

Now you MSR 0Ex2 is unlocked

MSR_PMG_CST_CONFIG_CONTROL.......(0xE2)  : 0x402
------------------------------------------
 - I/O MWAIT Redirection Enable......... : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
 - CFG Lock............................. : 0 (MSR not locked)

in this condition you can try different way to achieve also c3 states...

 

Yea I've just saw that in my bootlog.

The funny thing is that I still can't boot without "xcpm_program_msrs (c) Pike R. Alpha-patch"?

[Guest]

2 minutes ago, obus said:

aYea I've just saw that in my bootlog.

The funny thing is that I still can't boot without "xcpm_program_msrs (c) Pike R. Alpha-patch"?

you don't trust what i am saying to you from +++++++++++ stage....it is a different deal..not locked or not related..maybe xeon or not xeon related

 

Edited April 20, 2018 by Guest

[obus]

 

6 minutes ago, fabiosun said:

You suppose bad

Now you MSR 0Ex2 is unlocked

in this condition you can try different way to achieve also c3 states...

 

Can you elaborate that a little bit further?

1 minute ago, fabiosun said:

you don't trust what i am sying to you from +++++++++++ stage....it is a different deal..not locked or not related..maybe xeon or not xeon related

 

I'm trusting you 100% @fabiosun but I thought that this patch was obsolete with the MSR 0Ex2 unlocked.

And thank's for your help I really appreciate your input.

[Guest]

7 minutes ago, obus said:

 

Can you elaborate that a little bit further?

try to read from solution 2 of this thread

I don't know if those Patches are working now in HS

Also with latest clover you could achieve same result applying kernel patch fla on KernelXCPM

But you have to test by yourself in your "unique" system!;)

@SammlerG related Patches of solution two are usefull to have also cmotion undemotion and C3 set...but no difference in real use life

 

Edited April 20, 2018 by Guest
add some infos

[obus]

2 minutes ago, fabiosun said:

try to read from solution 2 of this thread

I don't know if those Patches are working now in HS

Also with latest clover you could achieve same result applying kernel patch fla on KernelXCPM

But you have to test by youeself in your "unique" system!;)

 

Cheers @fabiosun I will test further.

Keep you updated when I have anything new. And once more thank's for your help.

[Guest]

45 minutes ago, obus said:

There was a new tab in bios.

So this is with MSR lock disabled in the new patched bios.

But I suppose here was no change in MSR 0xE2?

AppleIntelInfo.rtf

config.plist

Bootlog.rtf

I have no seen before your edited message.

You can also avoid patched bios and use original one if now, you see in 0603 bios version new tab  with cfg lock feature

 

[obus]

6 hours ago, fabiosun said:

I have no seen before your edited message.

You can also avoid patched bios and use original one if now, you see in 0603 bios version new tab  with cfg lock feature

 

Cool @fabiosun!

[Guest]

@obus

take a look here in your bootlog

 

If you can try to see in bios if you read correctly all your module specs (I think in SPD section of your bios parameters) DIMM_D2 is not recognized well

 

[obus]

1 hour ago, fabiosun said:

@obus

take a look here in your bootlog

 

If you can try to see in bios if you read correctly all your module specs (I think in SPD section of your bios parameters) DIMM_D2 is not recognized well

 

@fabiosun

It's funny because that slot isn't used. I have 4 x 16 gb in slot D1 , C1, A1 , and B1. Nothing in D2?

It looks right in my last attached bootlog.

I have tested a little and now I don't need FakeCpuId if I replace that with a kernel patch (XCPM bootstrap patch: Find: 89D804C4 3C227722 -> replace:89D804CD 3C227722) 

I still need "xcpm_program_msrs (c) Pike R. Alpha" to boot.

Any thoughts about that?

I've read something about this in KGP:s Skylake-thread where your name was involved.

 

Edited April 22, 2018 by obus

[Guest]

29 minutes ago, obus said:

@fabiosun

It's funny because that slot isn't used. I have 4 x 16 gb in slot D1 , C1, A1 , and B1. Nothing in D2?

It looks right in my last attached bootlog.

I have tested a little and now I don't need FakeCpuId if I replace that with a kernel patch (XCPM bootstrap patch: Find: 89D804C4 3C227722 -> replace:89D804CD 3C227722) 

I still need "xcpm_program_msrs (c) Pike R. Alpha" to boot.

Any thoughts about that?

I've read something about this in KGP:s Skylake-thread where your name was involved.

 

Ciao Obus

Fakecpuid can be bypassed if you patch Kernel Patch to do the something fakecpuid does

Goal for you should be to NOT have any patches or any fakecpuid because your CPU is one of most similar to real new Imacpro

If you did all your tries correctly, maybe for now it is not possible and so you have to use some patches

If you see in my signature I have a 22 cores/44 thread xeon, I have also a patched bios (and I have helped also many X299 guys to patch their "unpatchable" bios with standard method

Xeon is another thing..it is not an extreme or lower..it is different (and better for me)

My patched bios need fakecpuid or kernel Broadwell patch (one of this because cpu is not supported by OSX) and also msr program patch to boot

And this with or without a patched bios (for MSR)

So be happy..maybe you have reached the best vanilla solution for you hardware at this state of OSX/clover development

In my opinion you should also find others user with a Xeon motherboard (similar to your chipset) and try solution that other guys can't ..

 

Edited April 22, 2018 by Guest

[obus]

2 minutes ago, fabiosun said:

Ciao Obus

Fakecpuid can be bypassed if you patch Kernel Patch to do the something fakecpuid does

Goal for you should be to NOT have any patches or any fakecpuid because your CPU is one of most similar to real new Imacpro

If you did all your tries correctly, maybe for now it is not possible and so you have to use some patches

If you see in my signature I have a 22 cores/44 thread xeon, I have also a patched bios (and I have helped also many X299 guys to patch their "unmatchable" bios with standard method

Xeon is another thing..it is not an extreme or lower..it is different (and better for me)

My patched bios need fakecpuid or kernel Broadwell patch (on of this because cpu is not supported by OSX) and also msr program patch to boot

And this with or without a patched bios (for MSR)

So be happy..maybe you have reached the best vanilla solution for you hardware at this state of OSX/clover development

In my opinion you should also find others user with a Xeon motherboard (similar to your chipset) and try solution that other guys can't ..

 

Hej fabiosun (Hej is swedish for ciao) 

Is your recommendation to use FakeCpuID together with only pike's patch or the second option without FakeCpuID and the two kernel patches?

Another question. Is the TSCAdjustReset.kext mandatory for the cpu to work natively? everything works fine without this kext ,even sleep and wake. 

There was a new microcode update in the last bios 0603 which you patched for me. Could it possible be that  TSCAdjustReset.kext is obsolete after that upgrade and how to check that?

 

 

[Guest]

@obus

for first question it depends from your real daily use.

In both condition have your system same stability/performances/temperatures? or no?

In my case if you see my first post in my thread all of those solution have the same impact in my system so it is up to you to understand if one is better of another one (of two patching methods you use)

If result and stability is identical I would use fakecpuid and msr programs patch (I would like also to say boot strap patch you have tested is for old Haswell CPU)

for your second question about that kext I have no direct experiences

To understand better you should test a unpatched bios with msr option (CFG lock) if it is present in 0603 without that kext and see if sleep/wake cycle is working correctly and this testing and stressing your cpu after waking and evaluating performances temperature and audio / video synch

If all is fine..the rule should be..use less patch/kext you can

 

14 minutes ago, obus said:

Hej fabiosun (Hej is swedish for ciao) 

Is your recommendation to use FakeCpuID together with only pike's patch or the second option without FakeCpuID and the two kernel patches?

Another question. Is the TSCAdjustReset.kext mandatory for the cpu to work natively? everything works fine without this kext ,even sleep and wake. 

There was a new microcode update in the last bios 0603 which you patched for me. Could it possible be that  TSCAdjustReset.kext is obsolete after that upgrade and how to check that?

 

 

 

[obus]

 

I have a weird problem with my Vega 64 liquid.

I need to put my rig into sleep before I run Cinebench!!!??

if not the open GL performance is really poor like 80 to 87 fps maximum. 

After one sleep cycle the open GL performance increase to 131 fps as best result. The same for Unigine_Valley-1.0-Advanced.

In windows Cinebech is scoring like 145 fps.

 

Could this has something to do with my CPU?

Edited April 23, 2018 by obus

[obus]

 

On 4/20/2018 at 7:17 AM, fabiosun said:

With unlocked one I can program MSR 0xE2 as @SammlerG finding to have also C3 states but for general XCPM performance/condition same result here (performance/temperature)
You can try to disable in clover bootmenu fakecpuid or some kernel patches if you are using it..but I don't think you achieve better results of one you show here

Is your system rock solid?
If so..try but stay also fine with it

 

 

 

Hi @SammlerG

 

I'm trying to fine tune my Xeon W-2175 and @fabiosunn have helped me to unlock my mainboard  .

I need to get rid of my two kernel patch if that's possible.

I don't need any FakeCpuId  after putting in this old hasswell XCPM bootstrap patch but my processor should be supported natively with native HWP (Intel SpeedShift Technology) CPU Power Management, exactly like the Skylake-X processor according to KGP:s excellent guide.

Do you have any input?

config.plist

Edited April 23, 2018 by obus