Contribute
Register

How to build your own iMac Pro [Successful Build/Extended Guide]

Status
Not open for further replies.
My 960 Pro NVMEs hit around 1700 MB/s write, 2200 MB/s read. One in each slot, one with Windows 10 installed, the other Clover/OS X

Thanks a lot for checking this out for me! Interesting ....
 
wow, sorry about that. Rolling back would require you to remove all the new kexts (including fakesmc) and restoring your last working fakesmc.

Apart from this, there should be no permanent damage, you should be able to roll back, at least I have been, multiple times.

Oh, btw, I haven't touched FakeSMC in this branch, so if it is unusable all by itself, I have no idea what to do.

Don’t worry at all! I know how to roll back.. but i really would like to use your X299 HWMonitor/HWSensor approach. Somethimg we have been waiting for long! So what could cause the fault on boot using your FakeSMC? What could go wrong? Can we try to fix it? Willing to test what ever comes to your mind!
 
Before..., last 2 installations I had the same problem with my 960 on 10.13.1, so I in 90% sure that was a EFI configuration problem - I had more important things to do so I completely forgot about that, and now
I have 960 Evo (witch is not a big difference with pro version) :p
and I use 2nd M.2 near PCI-E slots on 10.13.2 beta 4 (I never used 2nd slot near RAM slots :))

View attachment 294659 View attachment 294660

Thanks a lot for the detailed info ... this makes me indeed think something is wrong with my mainboard. Can you please check in your BIOS, which SATA-mode you are using? Thanks!

I also lose the correct time sometimes. And I did a fresh AMD Vega install, went away for 5 days, came back and the installation would not get to the "++++++++++++++" state. It stopped after loading the first 4 lines of the boot. I did not change anything in the config meanwhile. Neither hardware nor software.

I have checked the onboard battery and replaced it. Let's see how it goes.

This all does not sound good to me - makes me wonder if a faulty board could really be the problem.

I already can see it coming: In the end I will have to apologize to @kgp, because it was a faulty mainboard that jeopardized my Nvidia installs and not the Nvidia install itself *sigh*
 
Unfortunately, your FakeSMC prevents my system from booting... No boot , even if I remove all HWsensor kexts.

View attachment 294741

Would much appreciate your feedback with a possible solution. Thanks in advance and Cheers!

The version of FakeSMC in Slice's HWSensors package v 3.51, together with his IntelCPUMonitor.kext 1.22 provides Skylake-X CPU data. None of the various HWSensors packages that I've found yet support the integrated MB sensors that provide more extended data from the ASUS X299.
 
O.k., so 4400 would be the correct value? In my opinion, this entire alternative approach is not required at all and I just want to test it to be open and friendly. I also never faced any lag and I want my Idle frequencies to be as low as possible (actually flatline at 1.2 Ghz). Nobody likes to destroy what we always aimed at and never reached during X99 times with PMDrvr.kext and EIST!

You mentioned my splendid patience, but indeed I am more than tired with the repetitive discussions about a few exceptional problems of a few specific people.

@mgregrs, your CPU frequencies by the way look terribly distorted! Are you still using the Gigabyte Mainboard?

No no no... I like Gigabyte RGB lights and 2x8pin CPU power but that's all...
I'm very very happy with my ASUS and I don't want to change my MB :)

That terrible CPU frequencies was only, because I tested interferenc Clover options,
and after that I have so crazy CPU ...
without that my frequencies works very very well with 1.2 Ghz very stable and more when needed, so at this moment, I'm very happy without new propositions (but sometimes it's good to try some new things..- you know..) :)
 
Thanks a lot for the detailed info ... this makes me indeed think something is wrong with my mainboard. Can you please check in your BIOS, which SATA-mode you are using? Thanks!
...

We don't have so big choice, we can only use AHCI mode... or RAID
My SATA BIOS settings are 100 % default...
maybe...
before send your MB to ASUS - just try my EFI.. :D
If you have default BIOS settings with kgp basic recommendations (guide recommendations)
just try my standard EFI, and then you will see...
If you have the same low SSD speeds, we can talk about faulty mb :p
(PS: Please use 2nd, near PCI-E - M2 slot for your 960 with my EFI .. )

Have a nice day! :D :headbang:
 

Attachments

  • mgrEFI.zip
    34 MB · Views: 58
Hi all,

Has anyone had success getting either the ASUS ThunderboldEX 3 card or the Gigabyte Alpine Ridge(rev2) (other threads suggest that they are intercompatible on x99 and x299 mobos) cards working on this build?

Specifically, I'm wondering if multiple cards in a single build are supported? If I jump on this build, I'd be moving from a Mac Pro with 8 TB peripherals, so I'll need more than the single port that these x299 mobos have on board.

thx!
 
We don't have so big choice, we can only use AHCI mode... or RAID
My SATA BIOS settings are 100 % default...
maybe...
before send your MB to ASUS - just try my EFI.. :D
If you have default BIOS settings with kgp basic recommendations (guide recommendations)
just try my standard EFI, and then you will see...
If you have the same low SSD speeds, we can talk about faulty mb :p
(PS: Please use 2nd, near PCI-E - M2 slot for your 960 with my EFI .. )

Have a nice day! :D :headbang:

Thanks a lot for sharing your EFI and the way to try it, mgregrs!

I think I will skip the step, because the problem is coming from somewhere else! Samsung has released a newer firmware for the 960 EVO/PRO - this seems to alter the drives slower under certain conditions. My guess is, this has to do with thermal management. Why?

1. People in this forum have complained about slower speeds after firmware upgrade. Same happened to me. They did not change their hardware or software setup:

http://www.overclock.net/t/1617171/samsung-960-pro-evo-owners-club/1060

2. The problem persisted in both M2 slots and under both, Windows and OSX. So this is not hackintosh nor BIOS related.

3. Installing the NVMEs in the Asus Hyper x16 card resolved the problem. This card has active cooling, which makes me believe Samsung could have changed the temperature limits of the NVMEs to more conservative figures. As we know our boards can easily reach higher temps with "all-cores-enabled", this also might affect the NVME temperatures in a negative way.

Can you please do me a favor and check, which firmware is running on your NVME? This can be seen under About this Mac --> System report --> NVMExpress --> Version

My new firmware version is 2B6QCXP7. Yours should be different.

Thanks a lot and no matter what you do: Do not update :D
 
Thanks a lot for sharing your EFI and the way to try it, mgregrs!

I think I will skip the step, because the problem is coming from somewhere else! Samsung has released a newer firmware for the 960 EVO/PRO - this seems to alter the drives slower under certain conditions. My guess is, this has to do with thermal management. Why?

1. People in this forum have complained about slower speeds after firmware upgrade. Same happened to me. They did not change their hardware or software setup:

http://www.overclock.net/t/1617171/samsung-960-pro-evo-owners-club/1060

2. The problem persisted in both M2 slots and under both, Windows and OSX. So this is not hackintosh nor BIOS related.

3. Installing the NVMEs in the Asus Hyper x16 card resolved the problem. This card has active cooling, which makes me believe Samsung could have changed the temperature limits of the NVMEs to more conservative figures. As we know our boards can easily reach higher temps with "all-cores-enabled", this also might affect the NVME temperatures in a negative way.

Can you please do me a favor and check, which firmware is running on your NVME? This can be seen under About this Mac --> System report --> NVMExpress --> Version

My new firmware version is 2B6QCXP7. Yours should be different.

Thanks a lot and no matter what you do: Do not update :D

Nope your firmware version isn't the problem, I also have 2B6QCXP7, no performance issues.
 
kgp , i use ssdt x299.aml (delete device wifi) and xosi and clover does not inject it, I enable the different required DSDT replacement patches in Clover Configurator Section, in ioreg see ok renames , my board and prime deluxe they are similar and they have the same names for the devices.

if I get this, I would just run my usb 3.0 asm3142 fronts
 
Status
Not open for further replies.
Back
Top