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

[sev7en]

Do you have the Nvidia web drivers installed? I keep seeing that Nvidia message. Check in Library/Extensions or System/Library/Extensions.

Also, when you edit the info.plist for 4100kext you need to drag the info.plist to the desktop, make the edit and then replace it to the folder. You should get "authenticate" and " replace" prompts and then password. Before you reboot you need to run a kext tool like Kextwizzard to fix repair permissions and clear caches. Kext beast doesn't usually work you'll need to use Kextwizzard. Google it. There is a video on YouTube by Philips Garage RX 480 in sierra. Watch it a few times and then redo the fix.

Thanks, I did exactly that but it doesn't work. The ID of my GPU is different, could be that?

 

[Gigamaxx]

Thanks, I did exactly that but it doesn't work. The ID of my GPU is different, could be that?

What brand of RX 480 is it? All RX 470-480s have 0x67DF1002 as far as I know. Only the RX 460 has a different ID 0x67EF1002.

 

[rhys216]

Can you explain how did you do this? I'm up for more experiencing

Below is how I edited the 480 bios.

I then booted into windows and used winflash to flash the bios over. In clover I still use the EF67 injection.
However this is something you would do at your own risk, and I wouldn't personally do it if I didn't have a dual bios on my card in which to recover from.

 

[nnorris7]

I'm noticing different LuxMark scores when testing the RX 480 at different resolutions on my 4K monitor. Is this something I should expect? I apologize, I'm not familiar with OpenCL's workings but I thought it would be resolution independent. First score is the monitor scaled to 3360 x 1890 (via Display Preferences). Second score is native 4K (3840 x 2160).

My old eyes struggle a little at native 4K resolution for everyday work (which is coding). I doubt this difference is noticeable in most "real world" situations but I was just curious. Thanks.

 

[Gigamaxx]

I'm noticing different LuxMark scores when testing the RX 480 at different resolutions on my 4K monitor. Is this something I should expect? I apologize, I'm not familiar with OpenCL's workings but I thought it would be resolution independent. First score is the monitor scaled to 3360 x 1890 (via Display Preferences). Second score is native 4K (3840 x 2160).

My old eyes struggle a little at native 4K resolution for everyday work (which is coding). I doubt this difference is noticeable in most "real world" situations but I was just curious. Thanks.

View attachment 234892 View attachment 234893

Did you run a few bench I each configuration? Luxmark can vary by a few hundred points, also do a force quit on any open apps before testing each, its just above the shut down option.

 

[nnorris7]

Did you run a few bench I each configuration? Luxmark can vary by a few hundred points, also do a force quit on any open apps before testing each, its just above the shut down option.

Yes, in the scaled resolution, scores are in the 13,500 - 13,800 range, while in native 4K they are 14,400 - 14,550 (granted, we're talking probably 5 runs in each resolution). I run the benchmarks immediately after booting with no other apps running.

Update: Just to provide some further evidence, here is a screenshot of hashcat. The only difference between these two runs is a change in display resolution (from scaled for the first run to native 4K for the second). Again, the differences are small (~ 2.5% decrease) but consistent.

 

[sev7en]

What brand of RX 480 is it? All RX 470-480s have 0x67DF1002 as far as I know. Only the RX 460 has a different ID 0x67EF1002.

Hello, it is the
ASUS ROG STRIX Radeon Rx 480 8GB OC Edition DP 1.4 HDMI 2.0 AMD Polaris Graphics Cards STRIX-RX480-O8G-GAMING

 

[Revod]

Apple makes second developer beta of macOS Sierra 10.12.4

https://www.macrumors.com/2017/02/07/apple-seeds-macos-sierra-10-12-4-beta-2/

 

[Fl0r!an]

I'm noticing different LuxMark scores when testing the RX 480 at different resolutions on my 4K monitor. Is this something I should expect? I apologize, I'm not familiar with OpenCL's workings but I thought it would be resolution independent. First score is the monitor scaled to 3360 x 1890 (via Display Preferences). Second score is native 4K (3840 x 2160).

Are you talking about 3360x1890 HiDPI (which would create an enormous internal resolution of 6720x3780 = 25 MP!)? Or did you actually reduce the resolution to 3360x1890?

In the first case I wouldn't be surprised about lower scores, pushing 25MP surely creates some reasonable load for the GPU, even when drawing the UI.

 

[Gigamaxx]

Apple makes second developer beta of macOS Sierra 10.12.4

https://www.macrumors.com/2017/02/07/apple-seeds-macos-sierra-10-12-4-beta-2/

Hopefully we will see it soon as well.