OS X MacOS Benchmark Thread

[padams35]

This thread is dedicated to light/mac gamers like myself who are trying to judge the relative performance gains between different models. Since all the web data I could find was either sparse or inconsistent I'm hoping my fellow mac fans will volunteer benchmarks for their machines.

Benchmark Program: Unigine Valley. https://benchmark.unigine.com/valley
That's the newest, free, easy to use MacOS OpenGL benchmark I could find.

Requested Benchmark #1: Basic Preset
This benchmark is intended allow direct GPU to GPU comparisons. One might consider this benchmark a representation of how well a mac could run a modern demanding game on low settings.

Requested Benchmark #2: Custom: Quality=High, AA=x4, Resolution=System, Full Screen=yes
This benchmark is intended to judge relative GPU to GPU performance adjusted for screen sizes. Ie, How much better is the Pro 570 at 2560x1440 vs a Pro 560 at 2048x1152. One might consider this benchmark as a representation of how well a mac could run a modern demanding game at high detail.

If anyone knows of a good free/cheap Metal benchmark that reports in FPS vs abstract numbers shout out.


Anyway, here are the results from my own mac. I'll try to update the list to include results from other posters.

Benchmark #1 Results
69.8 fps - Radeon M290X, ?GB in 5K 2014 iMac
67.7 fps - FirePro D500, 3GB in 2013 MacPro
42.8 fps - Radeon RX 460, 2GB in Thunderbolt-1 eGPU
23.7 fps - Radeon 6770M, 512MB in 21.5" 2.7Ghz 2011 iMac
15.5 fps - Intel HD 6000 in 13" 2.2Ghz 2015 Macbook Air


Benchmark #2 Results
27.0 fps - FirePro D500, 3GB in 2013 MacPro (4K)
20.9 fps - Radeon RX 460, 2GB in Thunderbolt-1 eGPU (external 1920x1080)
20.3 fps - Radeon M290X, ?GB in 5K 2014 iMac (2.5K +retina doubling)
7.6 fps - Intel HD 6000 in 13" 2.2GHz 2015 Macbook Air (1440x900)
6.7 fps - Radeon 6770M, 512MB in 21.5" 2.7Ghz 2011 iMac (1920x1080)

 

[jerwin]

for a m290x (1280 shaders) on a 2014 imac 5k

69.8 (720p basic-- eww)
29.7 2k
20.3 2.5k (30% drop)

Theoretically, the 560 is good for 1.9 TFLops, and the 570 is 3.6 TFlops-- (almost twice as fast).

so you'd be better off with a 570-- even with the load imposed by a higher resolution screen.

 

[voyager77]

Mac pro - D500

#1
67,7

#2
28,0 (Full screen is 4K)

 

[padams35]

Theoretically, the 560 is good for 1.9 TFLops, and the 570 is 3.6 TFlops-- (almost twice as fast).

Theoretically. My concern is the Geekbench4 OpenCL scores. The 570 comes in at ~90K for over double the 560(MBP) at ~44K, but the 560 (iMac) are posting ~57K. The pessimistic answer is that the MBP is simply thermal throttling, but what if the 4K iMacs have been clocked higher to run closer to their RX counterparts?

I realize that GPU OpenCL computing scores don't correlate directly to gaming, but I'm still hoping for empirical FPS results.

Also a RX460-2GB Thunderbolt-1 eGPU benchmark:
#1: 42.8
#2: 20.9 (1920x1080)

 

[jerwin]

570 imac is 28 compute units at 1 Ghz
560 imac is 16 compute units at 907 Mhz (the same as in the macbook pro)

(my computer is 20 compute units at 975 Mhz)

 

[h9826790]

Mac Pro 2009, CPU W3690, GPU 1080Ti.

The #1 test can be CPU single thread limiting indeed. My CPU shows 109% is actually right at the single thread turbo boost limit.

And my test 2 result is better than test 1 also shows that this benchmark is not stressful enough for my GPU.

In fact, even if I run the ExtremeHD test, I still got similar result. Which means this test is very CPU single thread limiting (In my case).

Update 1:
I just try the 8xAA Ultra 3840x1080 (my monitor's native resolution). Still roughly the same result.

[doublepost=1524429590][/doublepost]And GFXBench Metal can provide FPS benchmarks for METAL.

https://itunes.apple.com/hk/app/gfxbench-metal/id1044629456?l=en&mt=12

It provide "Offscreen" results which won't be limited by the monitor refresh as well.