meina222

@epxpc01,

I compared your EFI against mine. I have different BIOS so your MMIO is different. Here's a partial EFI structure with release OC (I removed the bulky Resources and Tools so I can share it here) that I currently use adapted to your MMIO. I also added some ACPI aml's for NVRAM, which may or may not help w shutdown. I use simplified startup args to just "-v keepsyms=1"

My OS is Big Sur beta 10 and the set of AMD patches in my config.plsit is reduced and tested to work on beta 9 and beta 10. Re-add your patches if you want lower OS's as I have not tested Catalina. Included is also a (disabled) aml that maps my ports with the MacPro7,1 SMBIOS.

EFI structure TRX40-design.zip

My wallpaper disappeared :(. Of all things why did they need to remove the High Sierra and Yosemite pictures. I see they fixed some graphics glitches.

No issues upgrading with the patches disabled.

What I mean by 'works' is as follows - as far I understand, TB is a PCI extension device. USB-C and TB are different. So when you plugin a PCI TB controller card in your motherboard PCIE slot, and you get TB ports you use PCI lanes and can attain higher transfer rates. What I mean by "works" in bare metal is that with appropriate firmware, MacOS can recognize Titan and Alpine ridge based controllers and you can plug in your TB devices and get the max data rate (40gb/s in TB 3.0) and if lucky have hot plug work as well. In order for your Startec TB hub to work, you probably need to plug in a TB port connected to your PCI lanes, not in USB-C. You can get a card, and @iGPU can help you on how to bypass the TB header, but even then you may need custom firmware and SSDT. @iGPU has had some success with Titan Ridge based controllers. I have my Titan Ridge ver 1.0 card working too (came with my Gigabyte Designare TRX40 board), and I do have header, but that header is not needed for PCI functionality - but you need to fool the controller into not needing a header by jumping the pins.

You posted in Proxmox section - and what I tried to explain is that no-one knows how to have VM see the Thunderbolt controller and PCI hub, by passing it thru to the VM. Your Startec might be recognized as a USB device but not as TB one, however you may have some luck passing an eGPU or external TB SSD as if is PCI already by having Proxmox itself (and not MacOS) arbitrate the TB protocol. This should work in theory. For that you only need a Linux supported TB card and no custom firmware - you still need to jump the pins to bypass the header.

p.s.

The pin jumping trick is something I have never done but people do it and claim to have result. Best to PM iGPU. People use soldering, I think there could be jumpers you can buy, but you need to figure this on your own.

https://egpu.io/forums/builds/thunderbolt-3-with-threadripper-on-a-asrock-x399-fatality-motherboard-help-needed/

@cj750,

If you're asking about Thunderbolt in MacOS VM - I'm not aware of anyone having passed through the TB card setup. You might be able to passthrough Thunderbolt leaf devices, but I have not tried that yet. On the other hand, you should be able to make Thunderbolt work in bare metal. Please, contact @iGPU and @fabiosun for help on that.

@epxpc01,

The Gigabyte TRX40 has a "Wake on LAN" setting in BIOS that is ON by default. You need to turn it OFF. Try that and report back.

If that alone doesn't work, try checking 1st 3 MMIO addresses as skip=1 and only the last 4 as skip 0. I forget what the consensus on the MMIOs is, @fabiosun can help reviewing your log. My MMIO addresses may be different even on the same board because I use BIOS f4h, which is not publicly available and changed my MMIO compared to f4c. Gigabyte has since rolled out newer BIOS which I have not tested.

If none of the above helps, try unplugging any extra USB devices you have, even keyboard, and re-try. Sometimes incompatible USB devices cause an issue.

@fabiosun, @23d1

for bare metal I disable SMT (analogous to hyperthreading) in BIOS. On Proxmox I enable SMT and give 64 cpus to my VM. These are logical CPUs. I also “pin” the VM threads to logical cores 0-63 via task sets, but this is not required.

i had lots of problems with bare metal including freezing and random reboots till I enabled DummyPowerManagement in OC. It’s seemingly required for 3990x (could be a motherboard thing but I suspect cpu) besides SMT off - don’t know if that will solve @23d1’s problem - try it.

@ckrueger99

Yes, it should be. Not only that, Big Sur is also used to as a clue for yet unannounced Navi cards bigger than 21 if these online sourced are to be believed e.g.

https://www.techradar.com/news/macos-big-sur-driver-code-hints-at-mysterious-amd-navi-31-gpu

So essentially like a 3070 more or less - but MacOS compatible and hopefully actually available  to buy. Not too bad, not terrific considering the competition.

@Jaidy

Most other people  with 3970x don't seem to have this issue, but I was getting similar "random" kernel panics when DummyPowerManagement was disabled. In OC 0.6.2 this option moved to a different section of the config file (see the sample.plist) - make sure its value didn't get overwritten compared to your previous value by some editor trying to comply with the new structure.

@fabiosun - you're right. I disabled patches 0-3 and beta 9 boots just fine.

Some of these video card timelines are probably overly optimistic. This one https://www.amd.com/en/products/professional-graphics/radeon-pro-vii for example was announced on May 13 2020 and you still can't buy it anywhere. So expect stock issues for another 6 months at least.

The graphics artifacts issues I was getting on some sections such as when selecting entries in the semi-transparent tree list on the "System Report" widget have been fixed in beta9. The only thing I can see so far (which maybe Navi specific).

Beta 9 is safe to install. Worked fine just like beta 8.

@Ploddles,

I posted yesterday about the pull out of GB BIOS-es on GB forums. The GB BIOS development for TRX40 seems like a hot mess. Pretty much everyone agreed that F4k on Designare didn't work and it also broke my Thunderbolt Slot 4. The GB support bluntly told me not to use F4k weeks before they pulled it.

@fojerhar - in my experiments with Windows on Proxmox and CPU host passthrough, I found out that Windows 1909 has a significantly lower Geekbench 5 multi-core performance compared to Windows 2004. Windows 2004 seems to have optimized something w.r.t. this specific benchmark and now it almost matches MacOS in both Geekbench and Cinebench. The 2004 is still work-in-progress for AMD based systems though - not sure if available as an update yet - I had to do clean install. This is VM based testing but the CPU scores should carry over bare metal.

I have to give Windows 2004 credit - it's an extremely polished experience, that actually exceeds MacOS nowadays in responsiveness and usability. If it only didn't have its tainted legacy of silly options and features enabled by default, the registry and a tens of services that need pruning.

I managed to install Windows with sata and 1909 version. I then installed the virtio drivers from within Windows. Definitely not an easy install, took me days to figure out. I/O performance on the zfs array is blistering under CrystalMark with write back (unsafe) - shockingly good sequential read speeds, GPU performance is on par or slightly better than MacOS With latest Adrenalin, but CPU is seriously lagging to MacOS. I’ll post some results a bit later.

FYI, the latest OC gets rid of the annoying "Preboot" volume in picker on Big Sur betas and shows the target install HDD volume instead.

@fabiosun

This thread described the problem.

https://forum.proxmox.com/threads/can-not-install-windows-2019.75479/

I verified the .iso (downloaded from Microsoft). The only thing I can think of that makes my setup "unusual" is that the target drive is the 4x4x4x4 NVMe AIC. Works great for my MacOS VMs. Finally for the 1st time since getting the TRX40 I decided to install Windows and to my surprise this is proving more difficult than MacOS so far on VM 😂

FYI, disabling this doesn't have noticeable difference under Proxmox VM on any benchmark I tried.

Has anyone succeeded installing Windows 10 on Proxmox VM guest? I can't seem to apply the virtIO drivers and as a result my guest doesn't see the installation target disk. I followed all directions I could find including watching this https://www.proxmox.com/en/training/video-tutorials/item/install-windows-2016-server-on-proxmox-ve which according to Proxmox should be the same for Windows 10. I also got latest virtio drivers (1.189) and Windows 10 2004 iso directly from Microsoft. Tried also Windows 10 1909 and drivers 1.185 and 1.173. Same issue.

It's a personal preference - 64 cores (no SMT) performs very close to what 128 logical cores would for pretty much all applications. In your case the only challenge will be the GPU if you go Proxmox. You would have to stick with High Sierra. Purchasing an AMD card and upgrading above High Sierra will cause you headaches with reset and you will find the VM experience a bit frustrating. If the reset issues get fixed with Navi 21/22/23, VM remains one of the best options as performance is excellent.

@23d1

3990x works in bare metal and VM passthrough in both Catalina and Big Sur. Never tried lower such as High Sierra, where @fabiosun has lots of experience. Don't know if the 1080ti is supported. 

On the 3990x, you need to disable SMT for bare metal (which would make sleep impossible if you care about that) and you can't pass through more than 64 logical cores to MacOS in Proxmox (where I would strongly recommend to enable SMT, especially if you want to try running VMs in parallel, something I don't do).  For  optimal setup on both VM and bare metal you may need to switch BIOS settings if you change boots.

@ckrueger99 - you list Sabrent Rocket 4 Plus 2 TB in your spec. Where did you get that?

The Samsung 980 pro is listed on Amazon with very competitive price with 2-4 weeks delivery estimate.

https://www.amazon.com/SAMSUNG-980-PRO-SSD-500GB/dp/B08GL575DB