Contribute
Register

Mojave on Biostar TH55HD Socket 1156 / NVidia 9800GT

Status
Not open for further replies.
I had an extra Xeon X3450 CPU, so I swapped the I5-660 for the Xeon X3450 (quadcore w/ hyperthreading). I didn't need to make any system changes and no CLOVER config.plist changes. The system booted fast and Intel Speedstep / CPU Power Management continues to work perfectly without enabling Generate C/P States in config.plist. Even though the X3450 doesn't have AES-NI instructions like the I5-660, system responsiveness is much better than with the I5-660 (I'm running High Sierra HFS+ and not APFS). A Xeon X3480 might be a good upgrade if prices continue to come down on Ebay.
 

Attachments

  • Screen Shot 2019-01-15 at 2.26.02 PM.png
    Screen Shot 2019-01-15 at 2.26.02 PM.png
    23.4 KB · Views: 122
  • Screen Shot 2019-01-15 at 2.29.49 PM.png
    Screen Shot 2019-01-15 at 2.29.49 PM.png
    62.6 KB · Views: 131
... Xeon X3450 works great in Mojave, too. Will definitely look for an inexpensive X3480 after finding a good Mojave graphics solution for this system.

I'm not sure why Geekbench CPU results are better for X3450 in Mojave than they are in High Sierra. It's possible that something else was running in the background when I ran my HS tests.
 

Attachments

  • Screen Shot 2019-01-16 at 10.04.51 AM.png
    Screen Shot 2019-01-16 at 10.04.51 AM.png
    49.6 KB · Views: 132
  • Screen Shot 2019-01-16 at 10.29.35 AM.png
    Screen Shot 2019-01-16 at 10.29.35 AM.png
    24.2 KB · Views: 120
Last edited:
I noticed that I didn't have AppleLPC in my IOReg and I was missing Device MCHC, so I enabled FakeLPC in my CLOVER config.plist and added an MCHC patch to my DSDT. I now see that Hardware Monitor shows my CPU multiplier going up to 23x (previously only up to 21x) and it also shows that my 4 cores don't all have to be the same multiplier (previously, all CPU multipliers changed in unison). Rerunning Geekbench on Mojave now produces the attached measurement.
 

Attachments

  • Screen Shot 2019-01-17 at 6.19.38 PM.png
    Screen Shot 2019-01-17 at 6.19.38 PM.png
    22.9 KB · Views: 100
Last edited:
I have found that my front USB ports (EHC1) may stop working properly if I'm experimenting with DSDT patches and/or CLOVER settings changes and/or kext changes. I'm not sure which is the culprit, but I can consistently and reliably restore proper USB operation by enabling NVRAM reset in BIOS and booting with a USB device plugged into one of the front USB ports. I'm not sure if this is something quirky with the Biostar TH55HD BIOS, but this issue is beyond my current novice level of expertise. The good thing is that my USB ports remain operational as long as config.plist/DSDT/kexts are stable and I can always recover with NVRAM reset.
 
Last edited:
After realizing that the LPCB name in my Biostar TH55HD DSDT was SBRG, I renamed SBRG to LPCB and inserted a _DSM (many thanks to @RehabMan for this answer: AppleLPC Not Loading). AppleLPC now loads without injecting a FakeLPC and my upper CPU multiplier has increased from x23 to x24. New GeekBench CPU measurements are attached.

I used IOName 3a18 in my _DSM, because that appears to be what FakeLPC used.

EDIT: After reading a little about LPC, the IOName for Nehalem architecture (3b09) might be a better fit for my my Xeon X3450 (Lynnfield architecture). After editing my DSDT's LPCB._DSM to use 3b09, I didn't notice any performance differences.
 

Attachments

  • Screen Shot 2019-01-19 at 11.34.27 AM.png
    Screen Shot 2019-01-19 at 11.34.27 AM.png
    22.4 KB · Views: 88
Last edited:
The TH55HD motherboard has automated overclocking. After enabling this, Speedstep/CPUPowerManagement still works fine. New geekbench scores are attached.
 

Attachments

  • Screen Shot 2019-01-20 at 10.41.57 AM.png
    Screen Shot 2019-01-20 at 10.41.57 AM.png
    24.3 KB · Views: 115
TH55HD Motherboard with Xeon X3450 is working very well in High Sierra and Mojave (as expected no graphics acceleration in Mojave). Everything works except for sleep. Instead of sleep, system shuts down. There are other discussion threads about this, so I'm hopeful.
 
Last edited:
I added USBInjectAll.kext and SSDT-UIAC to eliminate this as a variable for resolving sleep issues. USB and sleep behavior is no different (USB ports work and sleep still shuts down).
 
New observation about Sleep: My system must be "sleeping" because the power light is flashing. I just can't wake from sleep without pressing the power button (which cold boots the machine and does not restore state). Maybe I need to play with darkwake settings in config.plist?
 
Applied a different DSDT patch for EHCx._DSM that is specifically for Intel Series 5. This patch includes clock_id which I was missing. My system now sleeps but then wakes immediately from sleep. If I disable "USB Wake from S3/S4" in BIOS, my system stays asleep, but I can't wake without pressing power button (which cold boots instead of wake).
 
Status
Not open for further replies.
Back
Top