kgp
Retired
- Joined
- May 30, 2014
- Messages
- 6,743
- Motherboard
- ASUS WS X299 Sage 10G
- CPU
- i9-7980XE
- Graphics
- Vega 64
- Mac
- Mobile Phone
Hello friends,
For the life of me I can not get E.13) Logic-X and Audio Studio Software Functionality “TestHyperthreadingBug.logix” to work on all 30 tracks for my machine. Happy to compensate anyone who can help fix this problem for me.
I went back to re-read this thread start to finish, focusing on posts Dec2017/Jan2018 where the Logic issue was discussed specifically @olofd, @warphead, @ppffffpp, @PedroJSkywalker. In late January I believe the MSR0xE2 unlocking method was released.
@olofd, @warphead, @ppffffpp, @PedroJSkywalker - can you verify the “TestHyperthreadingBug.logix” works on your build? I am just curious who has actually had success with all 30 tracks of “TestHyperthreadingBug.logix”.
For my machine, playback of “TestHyperthreadingBug.logix” won’t work unless I reduce track count from 30 to 24 to avoid the CPU overload message. This effects both onboard audio as well as RME UFX+ @ 1024 sample size, as well as Scarlett 2i2 usb i/o audio interface which is class compliant with OSx. Once the track count is reduced to 24, playback works fine. Intel CPU gadget & Logic CPU monitor show all 18 cores maxed out….thus this has nothing to do with the SKZ7 bug/MSR0xE2 unlocking, right? wrong? I also downloaded the other logic benchmark sessions mentioned in this thread including bench, NOXCMP copied 40 tracks, NOXCMP by @DSM2, EvanLogicBenchmark copied over 1000 tracks……They all work fine.
What I find strange is that I can run 24 tracks of “TestHyperthreadingBug.logix” at 32 samples but I can’t run 25 tracks at 1024 samples.
I double checked BIOS settings and the only difference I found was my lack of manual voltage override to 1.120, though this did not solve the problem, otherwise identical to the guide.
I also received my new trident RAM and while it helped my GB score dramatically, it was no resolve to this issue.
My GTX970 is now working great now, I was flubbing the patch process somehow. This was not the solution to my audio issue.
The main differences I see between my build & @kgp is my video card (his 1080 vs my 970) and PSU (his 1500 vs my 1050).
I can’t imagine getting 56,000 geekbench if there was a physical problem with my CPU/MOBO…i.e. RMA.? CPU temps are fine.
I wanted to offer as much info about my rig /install routine and if anyone can post or PM me with a resolve I will happily offer a donation in return. Attached is my EFI folder.
——
Build:
Asus Prime x299 Deluxe - default jumpers
7980XE No overclock - Manual BIOS voltage 1.120
64gb Tridentz RAM DDR4 3200
NVME 960 EVO in slot M.2_2
GTX970 G1 in Slot1
Suggested guide WiFi/BT PCI card in slot 3
Thermaltake Water3.0 360mm w/3x 120mm fans
PC Power & Cooling 1050w Modular PSU
BIOS
Load optimized defaults
Load guide BIOS settings
Additional bios settings:
/AI Tweaker/CPU Core Voltage/
g.) CPU Core Voltage: Manual Mode
h.) CPU Core Voltage Override 1.120
Install process:
1) Firmware BIOS Patch for M2R Reg using provided x299D.CAP
2) Prepare Install USB Thumbstick
3) Prepare EFI folder & custom config.plist & transfer to EFI partition on both System Disk & USB Install Thumbstick
4) Install OSx 10.13.3 to system drive (3 reboots)
Post Install
E.1) HWP (Intel SpeedShift Technology) CPU Power Management Configuration
1) Disable "xcpm_core_scope_msrs" in Section "Kernel and Kext Patches" of Clover Configurator
2) Unzip the ssdt.aml and copy the file to the /EFI/Clover/ACPI/patched/ directory of the EFI-Folder on your System Disk and reboot subsequently.
E.2) Graphics Configuration:
For Maxwell card like GTX 970, download WebDriver-378.10.10.10.25.106
Download the Nvidia WebDriver-Payload Repackager from InsanelyMac. Credits to Chris111 and Pavo. Run repackager, install patched driver, open NVDAStartupWeb.kext with Xcode to updateNVDARequiredOS parameter, run 4 terminal commands, reboot, engage driver in preferences.
E.3) Audio Configuration:
1) Copy AppleALC.kext v1.2.2, which further requires the Lilu.kext v1.2.1 in the /EFI/CLOVER/kexts/Other/ directory of your System Disk.
2) In config.plist in the /EFI/CLOVER/ directory of your System Disk with the Clover Configurator, enable the CAVS -> HDEF DSDT replacement patch under section "ACPI" under "DSDT Patches".
E.4) USB Configuration:
Native
E.5) ASUS Prime X299 Deluxe Thunderbolt EX3 PCIe Add-On Implementation
Skipping
E.7.) SSD TRIM Support:
Native & verified
E.8) ASUS Prime X299 Deluxe on-board Ethernet-Functionality
Just download , unzip and copy the SmallTree-Intel-211-AT-PCIe-GBE.kex attached below to the /EFI/Clover/kexts/Other/, reboot and you should be done.
E.9.1) ASUS Prime X299 Deluxe PCI Device Implementation
uncheck everything except SL01->UPSB (thunderbolt) in ACPI patches (I’m not implementing Thunderbolt)
Add SSDT-XOSI.aml to patched folder in EFI
E.9.2) SSDT-X299-iMacPro.aml PCI Implementation
Skip..? I have different video card & no thunderbolt card and have yet to learn how to modify .aml.
E.10) System Overview CPU Cosmetics
skipping
E.11) ASUS Boot Splash Screen Cosmetics
skipped, was already working based on bios flash
E.12) Native Display Brightness Control / Native NightShift Functionality for Monitors with DCC/IC Support
skipped
E.13) Logic-X and Audio Studio Software Functionality
Does not work @ 30 tracks, only works @24
geekbench: 5330 / 55399
I tried disabling speedstep/speedshift/c states in bios, no luck in rebooting.
I'd be happy to try letting the CPU run at full blast but I wasnt able to figure out how to do this. This is one major difference as my x99 5960 machine did not use speedstep at all.
Aside from not being able to properly run “TestHyperthreadingBug.logix”, the only other reason to believe there is something wrong is the fact that my MainStage session that was running smoothly on 10.13.2 x99 5960 @ 32 samples is completely unplayable on my x299 10.13.3 unless I switch upto 64 sample buffer size. GB score of my 5960x was only 33,000.
Other than these two issues, this new x299 build seems absolutely perfect.
Another user stated that the SMBIOS iMacPro1,1 helped him with Logic performance, but I have always used this since I just started my build Feb 8th.
Happy to try anything. Thanks everyone
@dankium ,
as it seems that you still question my Logic-X excursions, let me demonstrate the Logic-X performance of my i9-7980XE (with "CPU Core Ratio" set to "Sync All Cores" and all cores running @4.7 Ghz during CPU max. load conditions) by means of the IPG Graphs below:
1.) reflects i9-7980XE CPU Idle frequencies before loading Test Hyperthreading Bug.logicx + Logic-X
2.) reflects i9-7980XE CPU frequencies during loading Test Hyperthreading Bug.logicx + Logic-X
3.) reflects i9-7980XE CPU frequencies with Test Hyperthreading Bug.logicx + Logic-X fully loaded
4.) reflects i9-7980XE CPU frequencies when playing Test Hyperthreading Bug.logicx
5.) reflects i9-7980XE CPU Idle frequencies after exiting from Logic-X
From the figure below you see that all 30 channels within Test Hyperthreading Bug.logicx are enabled:
Also note that I use a Logic-X I/O Buffer Size of 1024 Samples!
Note the minimal CPU load increase during phase 2 (loading Test Hyperthreading Bug.logicx + Logic-X). From the Power, Temperature and Utilization Graphs one can deduce that just few threads are active and used during this initialisation phase.
The situation changes when "playing" Hyperthreading Bug.logicx. There is a significant increase of Power Consumption, Temperature and CPU Utilisation, indicating that more CPU threads are now involved. Note however, that even when playing Hyperthreading Bug.logicx with all 30 channels, the CPU power consumption never exceeds 160W (compared with 360W during Cinebench max. CPU load), the CPU temps hardly exceed 40 deg C (compared with 90 deg C during Cinebench max. CPU load) and the CPU utilisation never exceeds 40% (compared with 100% CPU utilisation during Cinebench max. CPU load).
Thus, I guess you can drop all questioning of my claims and blindly believe in my statements and results. The question is not if others can confirm this flawless HWP, Hyperthreading and Logic-X behaviour. The questions is rather what is wrong with your BIOS settings and your particular system configuration.
Note that the removal of the SKZ7 Intel Skylake-X bug happened during recent ASUS microcode updates within the actual ASUS Prime X299 Deluxe Bios Firmware releases. The SKZ7 Intel Skylake-X bug is NOT at all related with the ASUS BIOS Firmware patches provided by @interferenc.
After inspecting your EFI folder, which looks fine, I would reduce your issues to deviating BIOS settings or to a still deviating or yet incomplete overall system configuration.
Deviating BIOS settings might be:
a.) CPU Core voltage set to "manually" (1.120V) instead of "Auto".
b.) Speedstep/Speedshift/C-state BIOS management/EIST/SVID, etc. not enabled and not configured as indicated in my guide. By the way, note that HWP (Intel speedshift technology) is not supported by older Broadwell-E/Haswell-E/X99 systems, which exclusively use XCPM for CPU Power Management. Thus, any comparison with your X99 system is totally misleading. While HWP, Hyperthreading and Logic-X seem not working properly with your actual BIOS settings or overall system configuration, all works totally fine and totally as expected on my proper system.
While above you mainly refer to some user report released before the SKZ7 Intel Bug Removal, up to my best knowledge, there is currently absolutely no issue left with studio audio soft- and hardware on Skylake-X/X299 Systems! Everything just works as expected!
Finally note that our systems do not only differ in terms of video card (his 1080 vs my 970) and PSU (his 1500 vs my 1050) but also significantly defer in the level of sophistication in the implemented CPU/GPU water blocking solution!
Good luck for bug-fixing your BIOS settings and overall system configuration
Cheers,
KGP
Last edited: