HP Z8 G4 Dual Xeon Gold 6248 HT Boot Freeze at CoreAnalyticsHub

[ledo]

Hardware:
HP Z8 G4
2x Intel Xeon Gold 6248
40 cores / 80 threads total
AMD RX 570
macOS Tahoe 26.4
OpenCore
SMBIOS MacPro7,1

Problem:
With Hyper-Threading disabled, macOS boots correctly and both sockets are detected:
hw.logicalcpu: 40
hw.physicalcpu: 40

With Hyper-Threading enabled, boot freezes at:
CoreAnalyticsHub start completed

Sometimes with CPU wrapper enabled I get panic in:
AppleACPIPlatform / AppleACPICPU
_cpu_data_alloc
_ml_processor_register

What I tested:
- NUMA ON/OFF: same result
- SSDT-PLUG enabled/disabled: no fix
- ProvideCurrentCpuInfo enabled: no fix
- AppleXcpmCfgLock enabled
- AppleXcpmExtraMsrs enabled
- old GitHub CPU wrapper causes AppleACPICPU panic
- without wrapper no AppleACPICPU panic, but freezes at CoreAnalyticsHub
- minimal/custom wrapper with correct APIC order still panics

Important finding:
The old GitHub CPU wrapper uses linear APIC mapping:
0,1,2,3...

But this HP Z8 G4 with Gold 6248 HT enabled uses APIC order:
Socket 0 physical: 00 02 04 06 08 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38
Socket 0 HT:       01 03 05 07 09 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Socket 1 physical: 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78
Socket 1 HT:       41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79

Question:
What is the correct way to build a CPU wrapper / ACPI mapping for this dual socket HP Z8 G4 with HT enabled?
Should I limit to 64 CPUs, patch APIC/MADT, or avoid CPU wrapper completely?