@jaymonkey
so last night after spending most of the day prepping for the hurricane I decided to unwind messing with my Mojave transition issues...
that is like saying you want to unwind playing golf.. but I digress
I tend to reuse hard drives and repurpose them especially for testing out a transition like this.. the current disk that I did the fresh Mojave on started as a ZFS volume that was formatted GPT/APFS.. I am hopping there are no artifacts on it in the MBR etc.. that could be messing with me.
I decided to go deep and really think about what is going on here.
so originally I formatted this disk and installed 10.14.5 with the latest uni/multibeast
I believe I upgraded clover to 5045 at this point using OSXaptiomemfix3 (which actually is making permanent nvram changes to my hardware without using scripts just the helper driver)
then I believe I took my config.plist from a 43xx clover el capitan daily driver and moved it, and my patched DSDT etc to the mojave drive.
I believe that during this 10.14.6 update, it noticed the Mac Pro SMC and firmware were not the latest and required for mojave and tried to update it as part of the update, and that is how I am now left with permanent code in my UEFI bios.. there are no files, clover mods, boot sectors, or anything external to the motherboard that can be causing it. I have reloaded bios, cleared nvram, and even booted the board with no drives attached and yet its still there.
after a detailed look at clover and what it installs and where, I started looking at my drive
I think that my previous el capitan volume at one point was using legacy boot then updated to UEFI boot.. and during the use of the mac migration tools, it brought over some outdated and un-needed clover junk
I found rc scripts that I didn't install in mojave and deleted them
I found some remenace of some clover stuff in usr/ usr/local usr/standalone etc
i deleted everything from the new mojave volume and deleted the entire EFI again.. reinstalled 5058 I think for UEFI only, no rc scripts and the bare minimum of drivers to include nvramhelper
I could not get the new auto patching on a original DSDT to work stably, so I went back to my edited DSDT and I took the now new an properly formatted config.plist that clover loads default, and edited it with the SMBIOS info from my el capitan install config.plist. the DSDT that I have allows me to run with no DSDT patches and very little clover mods
I have fakesmc+sensors, LILU, WEG, and APPLEALC in kexts.... I also have some old mods in my DSDT that specify some GFX0 and Audio patches and I think that helps WEG get it right. I think I can actually take WEG out and my VEGA will load up without it and I still get ALC889 and HDMI audio without WEG.. but most say that WEG is still good to have even on native working cards to help with black screen wake and power mgmt so I am leaving it alone.
so far the system is a bit more stable.. I still get a random spat of stop sign at boot.. either from a fresh boot... and possibly more often with a reboot. the cmos does not seem to be resetting and nvram is holding all my values near as I can tell and imessagedebug is not seeing any changes in critical values there.
auto sleep is working and so far I have only tested one overnight sleep and it worked. I think some of the black screen at boot is still inherent to the VEGA firmware as windows users are plagues by many of our same complaints about black screening, wake behavior, multi monitor wake etc.
I notice after wake the card is using slightly less watts,
I also notice that my memory never clocks down to 0 when idle staying at 500 and going to 800 on load.
system clock goes really low like 25hz and clocks up to 1.46MHZ ish under load never seems to get higher than stage 5 I never see 1.5 or whatever the build in 'overclock' of the MSI air boost card is..
that could be my PCI-e 2.0 slots or perhaps the generic power management by apple since the card is using a generic frame buffer
any who... here I sit.. my drives are in a waterproof box as I sit here on my laptop while I still have power contemplating working out my niggles...
then I am pretty sure I did a software update to 10.14.6
so, having a 'old format' config.plist (with older mac pro 5,1 smbios/firmware definitions) may have permitted
I think another real possibility is that the DX58so is likely not a fully baked UEFI bios and it was added on to this board during a firmware update many years ago.
right now I am fairly confident there is no cross UEFI / legacy boot mechanics going on here, so it is surviving booting, rebooting, and mostly waking from a UEFI boot environment ..
of course it still wont auto boot my clover volume, I still have to press F10 at every boot/re boot to select the UEFI INTERNAL drive and then clover boatloader will load.
I still have the extra choose of boot preboot install from preboot, but its not in the bcfg boot dump list or diskutil ext.. its a phantom but have not idea where its coming from
I have also set in bios graphics.. to PEG instead of AUTO (other choice is PCI) to reduce the delay of the board bios looking around for devices. (intel defines the PEG as PCI-E if I read the right tech brief)
the wakeup issue when it does happen could be a signal handshake/timing issue.. but it sure looks like a crash because its not a black screen with the system operating behind it.. nobody is home, no logs show the wake and you can't ping or ssh the machine.. its just lights and fans on frozen when it doesn't wake properly