trx40 [Discussion] - TRX40 Bare Metal - Vanilla Patches

[meina222]

Not showing up in my Software Update yet. Will probably trickle in soon.

[Pavo]

Yeah, update went smooth as a baby's ass.

[Driftwood]

Just updating...all good.

 

Edited September 5, 2020 by Driftwood

[meina222]

Still update not showing anything for me (beta dev or not) - I suppose it's done in waves and will see it tomorrow.

 

On a different note - Gigabyte finally provided a BIOS fix f4h for my Designare that allows me to run my TB card in Slot 4 with my NVME 4x4x4x4. Before there was a bug where Slot 4 was being crippled to gen 1 and TB would cause D4 error and if I put a different card (say 2nd GPU) it would run at gen 1.0 speed as opposed to gen 3 or gen 4. After a month of back and forth and 3 BIOS versions on their part, it's now fixed and  I'm upgraded.

 

That gave me a chance to see how the bare metal hack fares on BIOS update. Booting into my installation caused a black screen with both my primary disk and back up USB EFI's. The backup USB has debug OC and it collected MMIO before it went dark, so I inspected it and sure enough it changed substantially in 8 of the 19 addresses (1st 5 and last 3). Fortunately the pattern was still similar and after I updated my MMIO whitelist and disabled the last 4, I was able to boot in without an issue. Thus, if you upgrade your BIOS be ready to have to re-gather and re-configure your MMIO. You probably won't have to go through the enable/disable 1 by 1 sequence and just disable last 4 to boot in. Will test more on the new firmware in the next few days.

 

@fabiosun - I got the equipment to flash my TB. I can try my Display as it is TB3 for now. I heard USB doesn't work. Do you guys have SSDT that enables me to check? Much appreciated.

Edited September 4, 2020 by meina222

[Jaidy]

9 hours ago, meina222 said:

Switching between MacOS and latest Ubuntu + Cinnamon, I gotta say the latter is so much snappier and more performant and looks are almost as good, but the details and the ecosystems are still lacking. Someone needs to build a for profit desktop environment for Linux with proper API's for GUI development.

Same with Manjaro KDE. 

[fabiosun]

@meina222 i have no tb ssdt on baremetal because i have to modify the one used in vm

i think only @iGPU is studying in this field

 

[meina222]

I could be the exception, but when I tried to update to beta 6, the macOS installer pick would progress through OC and rooting and then reboot before actually getting into the installation (right at handoff). The debug log of OC doesn't show anything off compared to previous successful boots. This is only when I select "macOS installer" - boot to beta 5 works normally. Next, I'll maybe try to create an install disk and try this way.

 

Update: Tried creating a fresh installation USB with Beta 6. On the same EFI where Beta 5 installer ran without a glitch, Beta 6 installer refuses to start and auto reboots. I use a recent OC (3 days maybe) daily build. So something changed. Will try a clean OC builder release from official OC repo.

Edited September 4, 2020 by meina222

[iGPU]

8 hours ago, meina222 said:

I could be the exception, but when I tried to update to beta 6, the macOS installer pick would progress through OC and rooting and then reboot before actually getting into the installation (right at handoff). The debug log of OC doesn't show anything off compared to previous successful boots. This is only when I select "macOS installer" - boot to beta 5 works normally. Next, I'll maybe try to create an install disk and try this way.

 

Update: Tried creating a fresh installation USB with Beta 6. On the same EFI where Beta 5 installer ran without a glitch, Beta 6 installer refuses to start and auto reboots. I use a recent OC (3 days maybe) daily build. So something changed. Will try a clean OC builder release from official OC repo.

 

I had same issue. Turn off csr-active-config blocking (use 00000000) and this should work with previous EFI.

 

 

Edited September 4, 2020 by iGPU

[meina222]

After rebuilding w latest OC builder it's good now. I thought I had zero-ed csr-active-config before but maybe not (sometimes I get confused which EFI is which).

Edited September 4, 2020 by meina222

[valmeida]

On 8/30/2020 at 4:36 PM, Jaidy said:

Is there anyone using an external audio DAC (preferably connected via USB) on this forum? 

Yes i use the Focusrite 6i6

[Pavo]

On 8/30/2020 at 4:36 PM, Jaidy said:

Is there anyone using an external audio DAC (preferably connected via USB) on this forum? 

Using Sound BlasterX G6 USB DAC, working perfectly fine.

[Jaidy]

@valmeida and @Pavo thanks.

[Driftwood]

BS Beta 6 Firewire Audio Tip:

So firewire/fireface seemed broke (crackly). But you just have to reinstall drivers. Anyone using Firewire cards plus something like a Decklink capture card. Install Decklink first followed by a reinstall of fireface drivers from RME will get firewire audio working again. The Apple firewire driver works as good as old times!

Edited September 5, 2020 by Driftwood

[Driftwood]

So here's my two MacPro7,1 bare metal (vanilla) configs for Catalina & Big Sur  for owners of Asrock TRX40 Creator motherboard. * Thanks to @Pavo, @fabiosun and @iGPU for their incredible help.

 

1/ Catalina: 10.15.6 + supplementary Update version - shutdown, restarts and sleeps for me.

 

Driftwood Asrock Creator TRX40 Catalina - config.plist.zip

 

Catalina EFI/OC Contents & kexts

 

 

 

* Above could easily work with latest kexts. This was tested under OC 0.61

 

2/ Big Sur Beta 6 - With BS I'm still experimenting but getting shutdown and restart ok. Just need to sort sleep (Ive switched on USB wakeup in BIOS and this sometimes works / sometimes doesn't!)

 

Driftwood Asrock Creator TRX40 - BigSur BM - config.plist.zip

 

Big Sur (beta 6) EFI/OC Contents & kexts

 

 

 

* Note: Kexts and OpenCore may get updated as new versions appear. OC 0.61 was used in both configs.

** Some kexts and DSDTs are either deliberately disabled from ACPI or removed from folders due to testing for my hardware etc...

*** If you are getting prelinked errors on boot, remove any old kexts in your OSX install drive found in Library/Extensions. Then rebuild kextcache.

**** If you have problems installing beta 6 or above ensure you have correct csr-active-config set in config's NVRAM's Add>7C436110-AB2A-4BBB-A880-FE41995C9F82 section and set it to 00000000.

 

Hope it proves helpful. for Asrockers / TRX40 users!

 

 

Edited September 5, 2020 by Driftwood

[iGPU]

7 hours ago, Jaidy said:

@valmeida and @Pavo thanks.

 

I've been using an Audient EVO 4 from the start on VM and now, bare metal. Periodically, there is audio dropout which seems to be helped by disabling Wifi (BT stays enabled).

[iGPU]

1 hour ago, Driftwood said:

BS Beta 6 Firewire Audio Tip:

So firewire/fireface seemed broke (crackly). But you just have to reinstall drivers. Anyone using Firewire cards plus something like a Decklink capture card. Install Decklink first followed by a reinstall of fireface drivers from RME will get firewire audio working again. The Apple firewire driver works as good as old times!

 

I've re-installed a Firewire card and ordered an RME Fireface 800 off eBay. In reading about the RME 800, some touted how no extra RME drivers were needed for macOS, which means using the "class compliant mode". With the RME 800, when using Firewire, you either have to use this mode, or the "stand-alone" mode (and this mode is required for Total Mix).

 

My question is how does the "stand-alone" mode (requiring RME drivers) compare with the "class compliant mode"?

[Driftwood]

1 hour ago, iGPU said:

My question is how does the "stand-alone" mode (requiring RME drivers) compare with the "class compliant mode"?

 

You basically program the stand alone mode by copying your input/ouput setup into firmware of the RME FF800. Then you detach it from computer and use it live on the road or studio and it will retain those settings for Input/Output path as standalone. See manual.

 

If you are switching it between computers all settings are retained.

 

The RME FF800 drivers which work both in Catalina and Big Sur are here:  driver_fw_mac_339.zip

 

The same driver works under Proxmox with non-abridged firewire card like mine! Though, it is not as 'reliable' as bare metal. I use two RMEFF800s. They're stilll top quality!

 

However, I am about to buy the newly announced AIO Pro (RMEs new AIO card version which is clean (no headphone noise) and better pathways in the digital process.

For me (who uses predominantly plugins inside Logic and Avid Pro Tools) the AIO card is just as good as an MADI FX card which is a little bit more expensive but if you require a ton of inputs /outputs would be better option.

 

The AIO Pro card is released within the next two weeks Synthax tell me - Price around 560 - 650 EUs/dollars/GBPs

 

Edited September 5, 2020 by Driftwood

[Driftwood]

So Ive been looking at the problem of certain programs not working under AMD because of the old Intel_memset.A compiler problem. Intel tuned compilation. Ive tried in Camtasia and looked at some Waves plugins to see if a way would be implemented to make the call(} routines edited and changed to go to intel_memset.V but to no avail.

 

Also there's some good stuff out there like the dispatchpatches in the attached zip. Has anyone tried using these for cpu spoofing or maybe a way of getting Lilu patched so that the intel_memset compile is  redirected? Read the reader inside and see Mac patches)

 

inteldispatchpatch.zip

 

For Waves, I see that some of the Hackintosh guys recommend an MKL CPU debug (like stuff found in above) when attempting to debug run Logic with Waves AU plugins. In terminal, use

MKL_DEBUG_CPU_TYPE=5 /Applications/Logic\ Pro\ X.app/Contents/MacOS/Logic\ Pro\ X

Edited September 5, 2020 by Driftwood

[fabiosun]

Guys, can you try your radeons on Big Sur or on the system you have both in bare metal and on Proxmox by posting the result? below a We transfer with CB15 if you don't have it I'm only interested in the GPU benchmarks(AMD to understand)

thank you

https://we.tl/t-7kAdUC6NBC

 

@Driftwoodthis works also for @Pavo(I think he has not good results for a radeon VII and I would like to check also with you)

[Rocket88]

Here is my Cinebench 15 on Asrock TRX40 Creator, Vega 64. 

[iGPU]

Here are C15 results for both bare metal and VM on AMD. Both are lower than Intel side.

 

Based on what I'm seeing, I don't think C15 uses both Radeon VIIs. Meanwhile, MarkLux does, as the scores below are consistent with Intel side where I have one Radeon VII on that system and it gets 54,000 on LuxMark. Also, LuxMark gives same scores for dual Radeon VII in VM and bare metal.

 

 

Big Sur ß6 bare metal, dual Radeon VII (XMP was on, 4G enabled and only 13 kernel patches--sourced from Pavo) :

 

 

 

 

C15 Test using Proxmox VM, dual Radeon VII (Mojave):

 

 

SMBIOS Comparison (LuxMark): dual Radeon VII, BS ß5 (iMacPro1,1 vs MacPro7,1):

Spoiler

 

iMacPro1,1:

 

 

MacPro7,1:

 

 

 

 

 

Edited September 5, 2020 by iGPU

[fabiosun]

@iGPUhave you cb15 in latest Big Sur?

[fabiosun]

And yes cb15 does not use both gpu during its test

[fabiosun]

My goal is to see result in Big Sur bare metal and also in proxmox

 

I am also testing in high Sierra with less patches but algrey do not advice to do so

but for the sake of testing I am doing the same but no particular benefit using  less patches in my case

[iGPU]

36 minutes ago, fabiosun said:

@iGPUhave you cb15 in latest Big Sur?

 

Yes, what I presented above is BS ß6 (public release 2), released this past week. I've seen no difference from BS ß2 thru ß6.