Update on a couple of things:
Just to expand on my edit to my previous post: I still need to investigate exactly what changed, but as soon as I installed @beltzak 's BIOS F3C config file, my onboard USB started working fine.
And so did my Ethernet, which had also been broken: I had only one of the two NICs detected, and it showed a MAC address of 00:00:00:00:00 and always showed cable unplugged. This was fixed as soon as I booted with beltzak's BIOS.
This is a huge relief, so thanks so much beltzak for posting that config. I'm going to trawl through the differences to work out what might have caused such a big difference for me. The only thing I can immediately see that's different is that you had Above 4G and VT-D enabled, which I'd left disabled as per dolgerran's config.
I've confirmed that it was
Above 4G Decoding that fixed this problem. Today I booted without it following a CMOS reset, and once again my main USB ports and Ethernet were dead. This time I was able to confirm that the USB-C port did work OK, so it was only the main USB controller that was affected, not the separate USB-C controller.
I don't really know what Above 4G Decoding is exactly, but it seems it's definitely important for working USB and Ethernet on the Designare 10G.
When overclocking, booting OpenCore after going into the BIOS menus is causing boot failure 100% of the time, and causes my overclock / CPU voltage settings to be reset.
I think this option is only available in OC 0.6.2 ... I cannot confirm. I guess our option should be to put it YES lol
- DisableRtcChecksum: NO
- Prevents AppleRTC from writing to primary checksum (0x58-0x59), required for users who either receive BIOS reset or are sent into Safe mode after reboot/shutdown
I tested that option and it hasn't fixed the issue.
The issue isn't actually a BIOS reset. All BIOS settings remain the same. What happens is that the system reboots at the end of the OpenCore initialisation procedure, and when it comes back up the CPU and RAM will be running at default settings.
It feels to me like it might be related to CPU power management or something like that. OpenCore causes a hard reboot, and the BIOS temporarily resets all overclocking-related settings to default. The settings can be restored just by hitting F10 in the BIOS to re-apply them.
What I find confusing and strange is that this only happens when OpenCore is loaded directly from the BIOS - eg by selecting it from the Boot Override menu. It doesn't happen when OpenCore boots normally. I might raise it with the OpenCore experts some time.
I have found another issue that might be related: whenever I do a shutdown in macOS, my "CPU System Agent Voltage" (also known as VCCSA) resets to default. Again, the BIOS setting is not affected, but the current value resets to default and has to be re-applied in the BIOS with F10.
This only happens when I do a shut down in macOS, not a restart.
I've actually seen a possibly related issue when testing in Windows: if during overclocking stress tests Windows locks up (no blue-screen-of-death, just the system freezes and has to be manually rebooted), when it comes back up the VCCSA will be reset to default like I described above.
This suggests to me that:
- The issue is related to the CPU cache (Uncore / Mesh), because that's what VCCSA is responsible for, and I've been told that a Windows freeze/lock up is usually cache related.
- The fact that VCCSA resets might be a BIOS bug, or at least a peculiarity of this BIOS.
- During shutdown macOS must be doing something to the cache voltage that's invalid, or causing some cache-related problem that causes the BIOS to think the VCCSA needs to reset. Shutdown seems to complete normally though.
This will only affect anyone who is overclocking, just like with the boot-from-BIOS issue I described earlier, and in this case only anyone doing a cache/Uncore/Mesh overclock who needs to raise the VCCSA to do it. It's possible to get a pretty good overclock without touching VCCSA, though higher overclocks will need it else you either can't overclock the cache, or will run into throttling issues that lower performance.
I'll report back in the next few days with what I've learned about overclocking this board and my results.
The brief summary is: I've got a working and tested all-core 4.6Ghz overclock with cache running at 3.0Ghz (up from 2.4Ghz), and ram at XMP 3600Mhz. It works in both Windows and macOS. However my benchmark results in macOS are lower than in Windows, eg Cinebench R20 scores of 10484 (Windows) vs 9904 (macOS). I don't yet know why this is.
EDIT: After upgrading to F3C I gained 189 points in Cinebench on macOS, taking me to 10093. Windows went up to 10513, only +29. Still a clearly higher Windows score (by 4.1%), which I'd like to try and figure out. It seems a bit too high to be explained by differences in Cinebench on macOS vs Windows.
I haven't managed to overclock as much as I hoped - I can't reach all-core 4.8Ghz at all - so I'm currently considering getting a full custom loop water cooled setup so I can make the most of what my 10980XE can do.
I can say already that for anyone wanting to get the best possible overclock, the Gigabyte motherboards are not good. I'm jealous reading details of people with eVGA, ASUS and other motherboards who have much better OC options than I do, including per-core voltage controls that enable fine tuning an overclock to get the maximum possible result for a given level of cooling. I bought this MB for Thunderbolt 3 and dual 10GBe NICs so I don't regret the purchase, but it's a shame Gigabyte seem to suck in this area; not to mention the BIOS lacks a lot of usability features, and has annoying bugs that it seems likely will never be fixed.
.
