[SUCCESS] Gigabyte Designare Z390 (Thunderbolt 3) + i7-9700K + AMD RX 580

[djlild7hina]

Very exciting stuff! Keep up the good work

 

[zipb]

Finally structured research on the Thunderbolt front. Great.

 

[Elias64Fr]

Gigabyte v33 and Apple v47 are very close -- much closer than v51:

• ARC_PARM
• EE_PCIE

View attachment 446316

View attachment 446315

You can use compare feature on HEX Fiend app, under File after open two files.

And if you want only address range, you can extract range on two temp files.. for me compare range feature don't work.

 

[CaseySJ]

You can use compare feature on HEX Fiend app, under File after open two files.

And if you want only address range, you can extract range on two temp files.. for me compare range feature don't work.

Using Araxis Merge (binary file compare):

• In the EE_PCIE section we have confirmation of Apple's use of Alpine Ridge in combination with Titan Ridge. Notice on the right side that we have Alpine Ridge Device ID 0x1578. This is from Apple's own Titan Ridge firmware.

Apple's Thunderbolt 3 drivers might be looking for Alpine Ridge here...

 

[AlexD]

I'm just wondering if faking device-id for UPSB (seem to be the first entry point of the whole TB device tree) from 15ea to 1578 may help here?

From my experience comparing IORegs of real iMac19,1, Macmini8,1 and any of our hacks the difference is that real macs always has "1578" in their device-id properties (Alpine Ridge) on upper UPSB and more recent Titan Ridge "15ea" on DSB.

Apple does intentionally fake it, even there are no any Alpine Ridge chips anywhere on the actual PCB (I was observing close-up photos but there are just Titan Ridge chips in the latest Apple products).

Maybe that also explains why macOS can't fully initialize Thunderbolt device tree, because as @CaseySJ just said: "Apple's Thunderbolt 3 drivers might be looking for Alpine Ridge here...".

 

[CaseySJ]

I'm just wondering if faking device-id for UPSB (seem to be the first entry point of the whole TB device tree) from 15ea to 1578 may help here?

Did that months ago with no effect!

 

[CaseySJ]

@AlexD

This might be worth doing again in combination with @Elias64Fr’s SSDT. Will try this tonight.

 

[AlexD]

@AlexD

This might be worth doing again in combination with @Elias64Fr’s SSDT. Will try this tonight.

Absolutely @CaseySJ !
I just think it's kinda of combinatory problem that requires a series of tryouts and experiments. Combining one or more solutions (SSDTs, firmware, device properties) together may lead us to a proper and complete overall solution.

I feel like we are very close now, especially with all extra help from ACPI and firmware experts who are kindly joined this topic recently. Very appreciate all your enormous efforts and enthusiasm, guys. Thank you for your time.

 

[CaseySJ]

You can use compare feature on HEX Fiend app, under File after open two files.

And if you want only address range, you can extract range on two temp files.. for me compare range feature don't work.

Hex Fiend has moved to GitHub. Latest version is 2.12.x, and binary-compare works rather well!

Releases · HexFiend/HexFiend

A fast and clever hex editor for macOS. Contribute to HexFiend/HexFiend development by creating an account on GitHub.

github.com

 

[S1lla]

If it helps I can provide NVM 43 from the Titan Ridge Add On card. @CaseySJ @Elias64Fr