How to extend the iMac Pro to X99 [Successful Build/Extended Guide]

[josh-x00]

Hi, @josh-x00. I have installed 10.13.6 successfully with your EFI and BIOS set. The XCPM is working fine. But the CPU idle frequency is high, and the high turbo frequency seems can not arrive at 3.6GHz. Thank you!
View attachment 346217View attachment 346225

hi @cocoanut!

so, if XCPM works you should have verified what follows under terminal app, right?
1. sysctl machdep.xcpm.mode (return: 1)
2. kextstat|grep -y x86plat (return 2 lines, means kext is loaded)
3. kextstat|grep -y appleintelcpu (return: empty)
4. sysctl -n machdep.xcpm.vectors_loaded_count (return: 1)

Your idle freqs are high, like mine. but that's not so important. take a look at utilisation! it's low. and so should be (CPU) power consumption. it's .59/.6! it's good. btw, your CPU graph touches 3.6 ghz a couple of times and also goes over! that's good too

If you want to check your CPU behaviour under stress, download cinebench (https://www.maxon.net/en/products/cinebench/), run "CPU test" and monitor your parameters under load. then upload one or two screenshots!

what kind of cooling do you have? never overclock or stress test if you aren't under appropriate cooling system.

would you upload your bios screenshots? don't forget to share your settings

let us know

 

[cocoanut]

hi @cocoanut!

so, if XCPM works you should have verified what follows under terminal app, right?
1. sysctl machdep.xcpm.mode (return: 1)
2. kextstat|grep -y x86plat (return 2 lines, means kext is loaded)
3. kextstat|grep -y appleintelcpu (return: empty)
4. sysctl -n machdep.xcpm.vectors_loaded_count (return: 1)

Your idle freqs are high, like mine. but that's not so important. take a look at utilisation! it's low. and so should be (CPU) power consumption. it's .59/.6! it's good. btw, your CPU graph touches 3.6 ghz a couple of times and also goes over! that's good too

If you want to check your CPU behaviour under stress, download cinebench (https://www.maxon.net/en/products/cinebench/), run "CPU test" and monitor your parameters under load. then upload one or two screenshots!

what kind of cooling do you have? never overclock or stress test if you aren't under appropriate cooling system.

would you upload your bios screenshots? don't forget to share your settings

let us know

OK,i will try it tonight and share the results,Thank you!
I use air coolling systerm（NOCTUA NH-U12S），only one CPU fan.Can i test the overclock or stress?

 

[josh-x00]

OK,i will try it tonight and share the results,Thank you!
I use air coolling systerm（NOCTUA NH-U12S），only one CPU fan.Can i test the overclock or stress?

ultra short answer: stock frequencies + enough airflow: you can use your PC so you can run cinebench.

now, assuming you have a powerful and reliable PSU, fresh air in your room and you're on the right fearless attitude if you want to verify if OC works, first:
- set a "safe" turbo freq on all cores (it could be 3.8 ghz, the intel turbo boost freq for 6800k)
- make sure your bios settings are appropriate for the context (voltages, multiplier, RAM)
- make sure all your fans are up and running at max speed
then, run cinebench on CPU to notice if it goes up to what you set.
monitor your CPU temp! it can get high, very fast. promptly stop the test if it goes over 80°C (Tcase for your i7: 127 °C).

if you want to learn something about OC, take a look here:
http://www.technologyx.com/featured/beginners-guide-overclocking-intel-x99/
https://edgeup.asus.com/2016/broadwell-e-overclocking-guide/
http://www.tomshardware.co.uk/forum/id-1800828/intel-temperature-guide.html
https://www.intel.com/content/dam/w...s/core-i7-6xxx-lga2011-v3-datasheet-vol-2.pdf

if you really want to experiment with OC, water cool your rig!

OC is dangerous for your rig! you must be aware you can damage it permanently! do it at your own risk.

 

[cocoanut]

ultra short answer: stock frequencies + enough airflow: you can use your PC so you can run cinebench.

now, assuming you have a powerful and reliable PSU, fresh air in your room and you're on the right fearless attitude if you want to verify if OC works, first:
- set a "safe" turbo freq on all cores (it could be 3.8 ghz, the intel turbo boost freq for 6800k)
- make sure your bios settings are appropriate for the context (voltages, multiplier, RAM)
- make sure all your fans are up and running at max speed
then, run cinebench on CPU to notice if it goes up to what you set.
monitor your CPU temp! it can get high, very fast. promptly stop the test if it goes over 80°C (Tcase for your i7: 127 °C).

if you want to learn something about OC, take a look here:
http://www.technologyx.com/featured/beginners-guide-overclocking-intel-x99/
https://edgeup.asus.com/2016/broadwell-e-overclocking-guide/
http://www.tomshardware.co.uk/forum/id-1800828/intel-temperature-guide.html
https://www.intel.com/content/dam/w...s/core-i7-6xxx-lga2011-v3-datasheet-vol-2.pdf

if you really want to experiment with OC, water cool your rig!

OC is dangerous for your rig! you must be aware you can damage it permanently! do it at your own risk.

I add PMDrvr.kext, the idle freqs is better.

 

[kgp]

I add PMDrvr.kext, the idle freqs is better.View attachment 346396

Well.. don't get confused with the IDLE frequencies displayed by IPG! If you look to the core utilisation, you clearly see that the CPU load during IDLE is practically ZERO, while approaching 100% under max. load conditions.. Perfect XCPM!

I don't see any reason to further stay with PMDrvr.kext, which just breaks sleep/wake functionality and practically adds nothing really worth to be implemented! The gain is pure IPG cosmetics!

 

[josh-x00]

Well.. don't get confused with the IDLE frequencies displayed by IPG! If you look to the core utilisation, you clearly see that the CPU load during IDLE is practically ZERO, while approaching 100% under max. load conditions.. Perfect XCPM!

I don't see any reason to further stay with PMDrvr.kext, which just breaks sleep/wake functionality and practically adds nothing really worth to be implemented!

@cocoanut, I totally agree with @kgp. PMDrvr does its work, but it's unuseful here. I told you too in #2266. why don't you try #2268? I could help you with MSI bios.

 

[cocoanut]

Well.. don't get confused with the IDLE frequencies displayed by IPG! If you look to the core utilisation, you clearly see that the CPU load during IDLE is practically ZERO, while approaching 100% under max. load conditions.. Perfect XCPM!

I don't see any reason to further stay with PMDrvr.kext, which just breaks sleep/wake functionality and practically adds nothing really worth to be implemented! The gain is pure IPG cosmetics!

Hi,KGP,thank your reply.The sleep/wakeup works fine when I add PMDrvr.kext. Strangely

 

[cocoanut]

@cocoanut, I totally agree with @kgp. PMDrvr does its work, but it's unuseful here. I told you too in #2266. why don't you try #2268? I could help you with MSI bios.

I will try it

 

[josh-x00]

I will try it

fine and don't forget to share bios screenshots!
as to PMDrvr, it's good to adopt one rule: just load the essential kexts.
pay attention to the combination of idle-load / temp / power values: without PMDrvr, you were already ok.

 

[nifranza]

@cocoanut see post #2010. Your sleep/wake functionality might be working fine, but is it working perfectly fine?

I think this advice..

just load the essential kexts.

is a good one. Especially if you want to avoid conflicts in future updates.

Cheers!