Yes, I am sure there exist a large user base that overlaps with people who need REDRAW 4K support and also only can afford the cheapest Apple computers. There might be dozens.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Silicon M1: benchmarks and review for Mac Mini, MacBook Air and MBP 13 M1 and comparison with other brands
- Thread starter Henk van Ess
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
I shoot Redraw 6K and 8K, but as I'm the DoP I usually hand over footage to an editor. I (and I think quite a few like me) would love to have a cheap way to review my footage at home, and even do small edits. I much prefer to spend my money on camera gear than a Mac Pro. I would be thrilled if the Mini can allow me this
Cool, wasn't aware of that benchmark. I'll check it out.
Speaking generally - I'd still caution folks to consider the real impact on one's workflow. Someone doing a few dozen RAW conversions after a weekend outing has different needs of course than someone doing a couple hundred RAW conversions while a client is standing there waiting for a preview.
The guys from serif ran these tests back to back 10 times and saw no discernible drop off In performance, although each test only takes 25 secs.
What a crappy results for a low end machine like the M1, don't you all agree https://gfxbench.com/device.jsp?benchmark=gfx50&os=OS X&api=metal&D=Apple+M1&testgroup=overall
https://gfxbench.com/device.jsp?benchmark=gfx50&os=OS X&api=metal&D=Apple+M1&testgroup=overall
New benchmarks for the Mini have now arrived on Geekbench:
One has a multicore score of 7750! Here: https://browser.geekbench.com/v5/cpu/4752334
One has a multicore score of 7750! Here: https://browser.geekbench.com/v5/cpu/4752334
Let's compare this M1 thing, this ENTRY machine of Apple to nr 5 graphic card in 2020, according to https://www.pcgamer.com/the-best-graphics-cards/, see https://gfxbench.com/compare.jsp?be...U&hwname1=Apple+M1&D2=NVIDIA+GeForce+GTX+1660 (M1 to the left., NVIDIA GeForce GTX 1660 to the right) ... amazing ...
What I find more interesting is the comparison with 1650Ti Max-Q: https://gfxbench.com/compare.jsp?be...=NVIDIA+GeForce+GTX+1650+Ti+with+Max-Q+Design
If these results hold, the M1 MBP may actually end up better for gaming than a Razer Blade Stealth...
Metal versus OpenGL plays a role there . In fact, in several of these GFX results. M1 running OpenGL code does what ? Apple isn’t point folks at those results but for folks with legacy software that will be a factor .
In part probably why Apple is funneling ( herding in some cases ) developers into Apple specific optimization. These results aren’t just hardware .
What a crappy results for a low end machine like the M1, don't you all agree
Note how the M1 is punting on a 1440 screen here . Latched to a HDR/XDR. 4K - 5K screen what is the “sag” here . It is total punt above for 1440 .
Nvidia has good openGL drivers. On the M1, openGL probably runs with a translation layer on top of Metal. A better comparison would be Vulkan vs Metal.
Definition of gaming o Metal mobile vs OpenGL desktop is very different. Unless game devs starts massively porting their desktop games to Metal API for some reason I would say it's incomparable. Just like how CUDA doesn't exist on a Mac and AMD at all it's completely incomparable to any Mac vs Windows/Linux argument.
CUDA is low level like Metal and many game developers use it to extract specific GPU instructions for heavy particles, physics, cloth behavior in their games etc. Problem is it's Nvidia only so wide spread adoption is limited or completely degraded once translated to OpenCL.
Last edited:
GFXbench doesn’t seem to do Vulkan .
The Vulkan/OpenGL/D12x stuff are all going to be different stacks than what Apple has been momomanically polishing solely for there own stuff .
What makes you say that? The 1650ti Max-Q has lower TFLOPs, texel and pixel fill rates than the M1.
Oh yes, of course. But you can also compare that with OpenGL vs. Vulcan or DX9 vs. DX12. API obviously matters.
I am not too worried about OpenGL because newer software and games tend to use Metal directly anyway (or something that leverages Metal) and older OpenGL stuff is probably less demanding. And then, OpenGL can be relatively efficient (as far as you can call OpenGL efficient) emulated on top of Metal anyway. It would be curious to run GFXbenchmark on M1 with both OpenGL and Metal, that would give us a measure of how efficient Apple's OpenGL wrapper is.
Yeah, it gets a big hit on 1440p — I am fairly sure that this is a memory bandwidth issue. That's where the GDDR6 in the other GPU can show its advantage.
But I don't see this as a problem, a GPU of M1 class only targets full HD gaming or below that. It doesn't have enough horsepower to handle higher resolutions, so the fact that it is getting bandwidth starved trying to run them is hardly a disadvantage.
From what I've read Metal doesn't wrap OpenGL, you have to rewrite it to Metal manually. Maybe there are some wrappers that come from third party.
Game devs have started to massively port their desktop games to Metal a good while ago. Any unity games runs Metal natively. Virtually all games that were out for Mac in last two years run on Metal. Civ 6, Total War series, Larian games. World of Warcraft has been running on Metal since 2016. Even some relatively obscure indie games like 7 days to die are working on Metal support.
Targeting Metal these days is really easy. If you use a third-party engine (like Unity or Unreal), you are already set. If you are using your own engine and use Vulcan for cross-platform support, you are already set (MoltenVK). Only if you want to get more out of Apple GPUs do you have to get your hands dirty with manually implementing your engine around Metal — and given how fast M1 is, I have a suspicion that more and more devs will come to play with it, it's also a lot of fun! TBDR is a very interesting paradigm to work with and it make implementing a lot of interesting rendering techniques elegantly simple.
How do you think Apple supports OpenGL on iOS or M1 Macs? They have a dummy OpenGL implementation that translates OpenGL state into Metal calls.
What a crappy results for a low end machine like the M1, don't you all agree
Those results make it seem that something (the display itself?) is capping the onscreen FPS at 60FPS, especially given how much variance there is with the offscreen results. If Apple really wants to target gaming going forward (the presence of Baldur's Gate 3 last week and Tomb Raider at WWDC indicates that is on their minds), they would need to either improve the refresh rates on their retina displays (which may be difficult given how many more pixels they push compared to a traditional 1080p display) or bump up the GPU itself. Even newer AMD and Nvidia cards can have issues with higher framerates at QHD (1440p) and 4k resolutions, so there may be a practical limitation at play with regards to the onscreen results.
It's basically vsync enabled. Can't get higher than screen refresh rate. This is why one looks at offscreen results — they provide the actual performance figures.