macOS Sierra PB: Need testers for new AMD Radeon drivers!

[Gigamaxx]

All System Files are restored back to their original state during an update if the current signature doesn't match the original signature. Doesn't mean that that the file was updated. If you look at the AMD9510Controller.kext inside info.plist file the DTSDKBuild version is 16E163e which was 10.12.4 beta 2, 10.12.4 beta 3 is build version 16E163f

Ive had fixes survive previous updates, normally I will lose one or two out of three on an update, but then the next update breaks one or all of them. I know the build number changes with each update but sometimes the fixes are unaffected or at least they have been in the past.

Are you able to get 36CUs on the RX 480 with this last update? Mine won't find the find values.

 

[Gigamaxx]

AMD is touting their audio for the Polaris and later architectures as TrueAudio Next, fwiw. But the cards come with DisplayPort 1.4 (and 1.2) connectors as well as HDMI 2.0 connectors which puts a hard limit on what is possible.

According to Wikipedia, Display Port 1.4 is at "8-channel audio with sampling rates up to 24 bit 192 kHz" (https://en.wikipedia.org/wiki/DisplayPort) while HDMI 2.0 is "... Up to 32 audio channels, …, Up to 1536kHz audio sample frequency" --no bit depth mentioned -- (http://www.hdmi.org/manufacturer/hdmi_2_0/index.aspx)

It is most likely limited by the Apple - AMD drivers. I have not found any mention of Apple using the sound physics libraries created by AMD for the TrueAudio Next platform, which is a bit puzzling actually as lots of folks are claiming Apple is trying to break into the game market (yeah I take that claim with a few mounds of salt as well) and these would be a natural fit to use.

I think we're getting default audio that doesn't match the card output or channeling through the hdmi/dport. If we could just reset some values it would be nice. The problem is finding those values to change.

 

[jerryy2345]

I think we're getting default audio that doesn't match the card output or channeling through the hdmi/dport. If we could just reset some values it would be nice. The problem is finding those values to change.

I wonder if anyone has asked AMD about this one.

 

[jerryy2345]

Oh man i tried that script and installed AMD/Nvidia HDMI audio + AMD HDMI audio ssdt. Nothing has changed and it didn't work. So i asked how to roll that back. Toleda said:
/Volumes/EFI/EFI/CLOVER/ACPI/patched/SSDT-HDMI-AMD-PEG0.aml installed
To uninstall AMD HDMI audio, remove ssdt noted.
Also removes AMD/GPU PM.

I still haven't done yet but i hope this will not affect my working sleep function or overall system functionality.

Concerning audio via DP - i'm giving up on there, since this was the last missing feature i had.
Since on board audio works on all ports on the back and front i'll just buy some Edifier R1280T speakers and i'm fine

The SSDT file combines audio with some power level management. I do not think removing it will affect the functionality in a bad way. The sleep issue is separate from this. Just mount your EFI partition, open the folders to get to the file, drag a copy to a safe place, delete the main one, empty the trash (you have to empty the trash with the EFI partition mounted or it will not really delete the thing). If you ever want to try it again without going through the installer trouble, just put a copy back in place.

I am still hoping audio via DP / HDMI is coming soon.

 

[jerryy2345]

Thx @jerryy2345
Finished that and still works fine.

Overall i think i should type a buying guide and a configuration guide of my system. I never had any freeze at any time, all performed very well, and most things worked out of the box. I think that might be interesting for those people that also want to build a performing film and photo editing machine + for everyday work from A to Z.
I did power consumption tests, can say that it has a nice air flow concept, it is quiet and all works. I also spend a whole night pumping symlinks into OSX to transfer most of ~/ onto several internal drives, especially disk drives to take care a bit of write access of the SSDs. I also configured OSX (applications) and all NLEs to get in use of all internal SSDs (3) and HDDs (2). I also involved Adobes "best practise" drive handling chart for Premiere/AE, Lightroom and FCPX/Motion best performance.
I even unpacked an old but beloved XP system and virtualized it. It was my old fully configured gaming PC with tons of games on it. I run it in VMware Fusion and managed to to tweak OpenGL device library for example for Unreal Tournament Classic that run too fast because of Intels speed step. In most cases you had 580-650FPS and it was unplayable
Other things like Serious Sam or GTA Titles perform well from scratch. I think the UT handling could be some candy for those that wish to recall that times again. There're lots of advices out there and i spent a whole day trying them out, just one worked.

Currently i'm writing a backup schedule concept (still on paper) and probably do some scripting to automate some exports and/or backups of some directories on some drives. Applications involved are Acronis True Image, CCC/SD, Sync Folders Pro and/or some Apple/shell script(ing).

I'll also configure some air gap systems for data safety since i have 2 older unused MacBook Pros and one MacBook lying around, even with a commercial OSX but overall fully air gap featured, but this a different story

You all cannot imagine this satisfying feeling i currently have

Glad to hear it is working.

If you do want to get into scripting a backup approach, look into rsync along with cron.

You can get a quick idea of how these work by opening a Terminal app window and typing

man rsync (followed by the return key) and after reading that, type

man cron (followed by the return key). Then scour the internet for examples.

 

[Docxavier]

Hello Everybody ¡¡¡

I'm getting some issues with my build

MB Gigabyte GA-Z97MX-Gaming 5 / hd4600+Gigabyte RX 460 4gb

At booting, my Radeon is not always properly detected. Sometimes the system starts without acceleration after a long time (more than usual) trying it.

I also face issues waking up from sleep. Most of the times the system wakes without acceleration.

Could someone give me some clues about how to solve those problems?

Thanks in advance ¡¡¡

 

[Mork_vom_Ork]

Just upgraded to SIERRA 10.12.4 beta4 (build 16E175b) and even with IGPU as (primary) helper card,
i no longer have any Display connected to my RX480 detected - so all screens stay black after booting to the desktop !!!
Well, not all: IGPU screen works normally <--- otherwise i wasn't able to make this post ;-)

Look here:

Card gets fully detected and initialized, also the framebuffer was correct detected, but:

NONE of my conncted screens where initilized and still stay black (telling me "no signal detected, going to sleep soon").
Also see here that both cards where fully detected:

Same hardware constallation, restarted booting SIERRA 10.12.4 beta3 again: all is working normal, as always in the past - all screens show up with loaded Desktop.

so my final thoughts: HOUSTON, we have the next big problem. Could anyone confirm this behaviour?

PS: there is one big change in the AMDRadeonX4100.kext - many small 1 or 2 byte changes, but also a deletion of 1.6 kilobytes of code:

Have to investigate on what this coder formerly was doing/good for. But this could be a reason for what i have noticed lately (discribed above).

 

[Pavo]

The device in your ioregistryviewer is shown as RX 480 with device id 0x67DF1002 but DPCIManager is showing as a RX 460 which has a device id of 0x67EF1002. Something is wrong with these pictures.

 

[jerryy2345]

Just upgraded to SIERRA 10.12.4 beta4 (build 16E175b) and even with IGPU as (primary) helper card,
i no longer have any Display connected to my RX480 detected - so all screens stay black after booting to the desktop !!!
Well, not all: IGPU screen works normally <--- otherwise i wasn't able to make this post ;-)

...
so my final thoughts: HOUSTON, we have the next big problem. Could anyone confirm this behaviour?

PS: there is one big change in the AMDRadeonX4100.kext - many small 1 or 2 byte changes, but also a deletion of 1.6 kilobytes of code:
.

If you go back some pages to the 10.12.4 PB3 update. I wrote about this a little bit. Try turning off inject - ATi, ie. set it to false. The Berbice and BerbiceA frame buffers were not recognizing the connector for the RX 460 cards. They look as if they are intended for laptops.

Edit: I can confirm that on my system, I need to use Inject-ATI=false in the Clover config.plist.

The public beta version 10.12.4 Beta (16E175b) is also available. Usually, when the releases come this quickly together, it means Apple is getting ready to release this to the general public. I am not trying to be alarmist here, welllll not too much anyway, but you might want make sure you ease into this upgrade, it will break some RX 4nn systems that are on auto-pilot updating. But turn off Inject-ATI and you will be able to have all of your connectors.

The BerbiceA frame buffer has been removed, it was only there for the public beta v3, it seems. The 9510 controller kext and the 4100 kext are at v 1.50.37.

edit 2: I almost forgot to mention, audio is still 'horrifically bad'.

 

[ekwipt]

Is there a way to download the 10.12.3 upgrade with out installing, my stable system is on 10.12.0 and I want the upgrade so I can play with the RX480 a bit more? Sorry for a bit off topic