Doesn't look like the new iMacs got any AMD GPUs inside them. That said, I'm having doubts that RX 6000 series drivers will be available anytime soon, if at all. What would be the point if they don't sell any Mac with those GPUs?
New Features in macOS Big Sur 11.4 Beta
- You can now use graphics cards based on the AMD Navi RDNA2 architecture (6800, 6800XT and 6900XT). (73709953)
Nice! Can you run a OctaneBench score?It's oh so beautiful...
View attachment 1761696
(That's a Hack of course, my secondary test system, hope that's OK.)
That GB 5 score is three times what my Vega 64 8GB gets!
Time to finally update my main system from 10.15.7 to Big Sur!
Sadly not - "No supported GPU found". Judging by the date in the release (2020.2.3) it's not been updated recently? I just got it from here: https://render.otoy.com/octanebench/ Let me know if there's a more recent one somewhere.Nice! Can you run a OctaneBench score?
Super exciting! Do you use FCP X and any chance you can try some 4:2:2 10-bit HEVC?It's oh so beautiful...
View attachment 1761696
(That's a Hack of course, my secondary test system, hope that's OK.)
That GB 5 score is three times what my Vega 64 8GB gets!
Time to finally update my main system from 10.15.7 to Big Sur!
Literally playing a 4K 10-bit 4:2:2 HEVC file on the timeline without stuttering or rendering is frankly the goal haha.Tell me what to do (and how) and I'd be glad to.
If that’s correct, then it’s 2.6 times as fast as the VII and almost twice as fast as the Pro VII and 1.8 times faster then the Pro Vega II, which means, if that scales accordingly to octane bench it should have around 556 OB which would make it the fastest card for rendering with octane x to date and it compares to an RTX 3080 (around 549 OB).It's oh so beautiful...
View attachment 1761696
(That's a Hack of course, my secondary test system, hope that's OK.)
That GB 5 score is three times what my Vega 64 8GB gets!
Time to finally update my main system from 10.15.7 to Big Sur!
OK sure. Currently the 6900XT is still in my old H77 4-core 16GB system, so if that will use any CPU at all the results might be compromised. But I'll try it on both my main system (with Vega 64 8GB, on 10.15.7) and the H77 with 6900XT and see if I can spot a difference.Literally playing a 4K 10-bit 4:2:2 HEVC file on the timeline without stuttering or rendering is frankly the goal haha.
I can try and upload a sample file shortly that you could pop in a timeline.
Nice! GeekBench is not the best benchmark tool, at least from my past experience - it doesn't test hard or long enough. But that ~180k result is pretty consistent across multiple tests. But yes I'd love to test it on something more demanding.If that’s correct, then it’s 2.6 times as fast as the VII and almost twice as fast as the Pro VII and 1.8 times faster then the Pro Vega II, which means, if that scales accordingly to octane bench it should have around 556 OB which would make it the fastest card for rendering with octane x to date and it compares to an RTX 3080 (around 549 OB).
I did the math based on your benchmark and these numbers from geekbench and octane. The last Octane X benchmarks we’re done in december so these number might even have improved since then. I wouldn’t wonder if we get even a higher OB with the 6900 then. >>>Nice! GeekBench is not the best benchmark tool, at least from my past experience - it doesn't test hard or long enough. But that ~180k result is pretty consistent across multiple tests. But yes I'd love to test it on something more demanding.
One thing I did notice is that while the Metal score is 180+k, the OpenCL score is only 122k. That's the widest gap I've seen in OpenCL vs Metal in GeekBench. OpenCL is deprecated of course so wouldn't be surprising if Apple put little effort into optimising it. But it does mean that scores might not be so impressive with software that doesn't (yet) use Metal - like Blender.
One thing I did notice is that while the Metal score is 180+k, the OpenCL score is only 122k. That's the widest gap I've seen in OpenCL vs Metal in GeekBench. OpenCL is deprecated of course so wouldn't be surprising if Apple put little effort into optimising it. But it does mean that scores might not be so impressive with software that doesn't (yet) use Metal - like Blender.