Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

raftman

macrumors member
Original poster
Apr 15, 2020
38
53
I see there are Geekbench results for the A12Z Mac Mini but they are using Rosetta 2. I believe the Mac Mini Dev Kit can run iOS Apps. Does anyone know why nobody has tested it using the iOS Geekbench app? It would score higher.
 
Theoretically should be identical to the 2020 iPad Pros' Geekbench score.

Not quite identically. The Mac mini with the A12Z runs 100 MHz slower than the iPad model

But to the OP;
The only one who could run the iOS version of GeekBench at this moment, is the developer of Geekbench. While it will be such that iOS apps are automatically opted in to also being available on Apple Silicon Macs, this is a future change to the App Store, and GeekBench for iOS is not something you can just download. When Apple officially starts selling Macs with Apple Silicon, they'll push the change to the App Store so developers will have to opt out if they don't want their iOS apps on the Mac, but unless you are the owner of the codebase, there's currently no real way of getting a .app bundle for the Mac from an iOS app
 
  • Like
Reactions: jdb8167
Not quite identically. The Mac mini with the A12Z runs 100 MHz slower than the iPad model

It seems to be running at the exact same speed?

Code:
 "date": "2020-06-30 12:15:58 +0100",
  "document_type": 0,
  "document_version": 5,
  "branch": "corktown-master-build",
  "build": 503391,
  "checksum": "6be8b170dd541ccc6312c5fa494eb85469bb7256",
  "commit": "eb68c8648b",
  "options": {
    "iterations": 5,
    "cpu_workers": 0,
    "memory_workers": 0,
    "workload_gap": 1000
  },
  "platform": {
    "os": "macOS",
    "architecture": "x86_64",
    "bits": 64
  },
  "processor_frequency": {
    "frequencies": [
      2493,
      2493,
      2493,
      2493,
      2493,
      2493,
      2493,
      2485,
      2489,
      2489,
      2490,
      2489,
      2490,
      2489,
      2490,
      2489,
      2489,
      2489,
      2490,
      2486,
      2490,
      2489,
      2490,
      2490,
      2490
    ]
  },
  "runtime": 133.65655045900002,
  "system_uuid": "f563bfdb0ad5fb0bc37bb37402fdab4d9b2d65af",
  "clock": 0.002175583,
  "uuid": "d3a2342186a8488aa4d6c2066ff111fa",
  "version": "Geekbench 5.2.0 Tryout",
  "metrics": [
    {
      "id": 1,
      "name": "Platform",
      "value": "macOS x86 (64-bit)",
      "ivalue": 0,
      "fvalue": 0.0
    },
    {
      "id": 2,
      "name": "Compiler",
      "value": "Clang 11.0.0 (clang-1100.0.33.12)",
      "ivalue": 0,
      "fvalue": 0.0
    },
    {
      "id": 3,
      "name": "Operating System",
      "value": "macOS 11.0 (Build 20A5299w)",
      "ivalue": 0,
      "fvalue": 0.0
    },
 
Last edited:
It seems to be running at the exact same speed?

Code:
"date": "2020-06-30 12:15:58 +0100",
  "document_type": 0,
  "document_version": 5,
  "branch": "corktown-master-build",
  "build": 503391,
  "checksum": "6be8b170dd541ccc6312c5fa494eb85469bb7256",
  "commit": "eb68c8648b",
  "options": {
    "iterations": 5,
    "cpu_workers": 0,
    "memory_workers": 0,
    "workload_gap": 1000
  },
  "platform": {
    "os": "macOS",
    "architecture": "x86_64",
    "bits": 64
  },
  "processor_frequency": {
    "frequencies": [
      2493,
      2493,
      2493,
      2493,
      2493,
      2493,
      2493,
      2485,
      2489,
      2489,
      2490,
      2489,
      2490,
      2489,
      2490,
      2489,
      2489,
      2489,
      2490,
      2486,
      2490,
      2489,
      2490,
      2490,
      2490
    ]
  },
  "runtime": 133.65655045900002,
  "system_uuid": "f563bfdb0ad5fb0bc37bb37402fdab4d9b2d65af",
  "clock": 0.002175583,
  "uuid": "d3a2342186a8488aa4d6c2066ff111fa",
  "version": "Geekbench 5.2.0 Tryout",
  "metrics": [
    {
      "id": 1,
      "name": "Platform",
      "value": "macOS x86 (64-bit)",
      "ivalue": 0,
      "fvalue": 0.0
    },
    {
      "id": 2,
      "name": "Compiler",
      "value": "Clang 11.0.0 (clang-1100.0.33.12)",
      "ivalue": 0,
      "fvalue": 0.0
    },
    {
      "id": 3,
      "name": "Operating System",
      "value": "macOS 11.0 (Build 20A5299w)",
      "ivalue": 0,
      "fvalue": 0.0
    },

I don't have one to check, but I was going by what is said on the second paragraph of this article
 
Not quite identically. The Mac mini with the A12Z runs 100 MHz slower than the iPad model

But to the OP;
The only one who could run the iOS version of GeekBench at this moment, is the developer of Geekbench. While it will be such that iOS apps are automatically opted in to also being available on Apple Silicon Macs, this is a future change to the App Store, and GeekBench for iOS is not something you can just download. When Apple officially starts selling Macs with Apple Silicon, they'll push the change to the App Store so developers will have to opt out if they don't want their iOS apps on the Mac, but unless you are the owner of the codebase, there's currently no real way of getting a .app bundle for the Mac from an iOS app

Oh, I didn’t know that thanks for the answer. Geekbench should make the iOS version available in their website.
 
Oh, I didn’t know that thanks for the answer. Geekbench should make the iOS version available in their website.

I have a feeling that iOS runtimes outside of the App Store will be rejected by macOS. I don't know this for sure, and as mentioned I don't have a DTK so I can't verify or test, but I would imagine similar constraints to iOS would be in place for iOS apps on the Mac. There are "tricks" performed on iOS app bundles on the Mac to make the code behave properly as well, like creating an app-localised path for the app logic. I.e. if an app requests a file at a specific file system path it will be expecting to be in a very specific place on iOS, so might ask for a path relative to where the app is installed, but on macOS the user can move app bundles around on the file system, so macOS creates a "fake" path for the app's own logic and such. I think Mac app distribution will always be open, but iOS apps will be exclusive to the App Store. Just a guess :)
 
Worth mentioning (since no-one else seems to have clocked it), not only is Geekbench hampered by Rosetta, it doesn't use the efficiency cores of the A12Z when run under Rosetta so the native version might score quite a bit better on multicore.
 
Worth mentioning (since no-one else seems to have clocked it), not only is Geekbench hampered by Rosetta, it doesn't use the efficiency cores of the A12Z when run under Rosetta so the native version might score quite a bit better on multicore.

The small Thunder cores can't run at 2.5GHz and the test clearly runs at that speed. The only difference between the iPad Pro scores and the Developer Transition Kit is one is run natively and the other is translated through Rosetta 2.

Code:
"processor_frequency": {
    "frequencies": [
      2493,
      2493,
      2493,
      2493,
      2493,
      2493,
      2493,
      2485,
      2489,
      2489,
      2490,
      2489,
      2490,
      2489,
      2490,
      2489,
      2489,
      2489,
      2490,
      2486,
      2490,
      2489,
      2490,
      2490,
      2490
    ]
 
As others have me ruined, there is no published ARM Mac Geekbench version. We also don’t see more benchmarks because reasonable developers wouldn’t want their developer privileges revoked.
 
The small Thunder cores can't run at 2.5GHz and the test clearly runs at that speed. The only difference between the iPad Pro scores and the Developer Transition Kit is one is run natively and the other is translated through Rosetta 2.

As I understand it, Geekbench could use Grand Central to make use of all available cores. Multicore tests can include differently able nodes if the distribution and recompilation of tasks is handled properly.
 
I'd be interested about how smoothly Big Sur runs on that DTK. Geekbench is nice and all, but I'm about applied power. Is there a noticeable difference between Intel and A12Z chips, and if yes, is it good or bad?
 
I'd be interested about how smoothly Big Sur runs on that DTK. Geekbench is nice and all, but I'm about applied power. Is there a noticeable difference between Intel and A12Z chips, and if yes, is it good or bad?
Go watch the keynote again. Every screen you see Big Sur on is an AS Mac Mini. Remember that the real AS systems will be even more powerful and have more CPU and GPU resources.
 
Go watch the keynote again. Every screen you see Big Sur on is an AS Mac Mini. Remember that the real AS systems will be even more powerful and have more CPU and GPU resources.
Sure did, but I‘d love to hear about normal usage, not scrolling up and down in Photos or a quick camera turn in Maya. That‘s why I asked :)
 
Sure did, but I‘d love to hear about normal usage, not scrolling up and down in Photos or a quick camera turn in Maya. That‘s why I asked :)

At one point they zoom out of a 5GB psd file measuring 24k x 12k @ 300 dpi with 80+ layers and that is smooth (1:37:18 in the keynote)

They also apply filters on 4K videos in real time in Final Cut Pro.

And this is on the A12Z iPad SoC paired with 16GB of RAM.
 
At one point they zoom out of a 5GB psd file measuring 24k x 12k @ 300 dpi with 80+ layers and that is smooth (1:37:18 in the keynote)

They also apply filters on 4K videos in real time in Final Cut Pro.

And this is on the A12Z iPad SoC paired with 16GB of RAM.
Like I said, I've seen it, that's why I wished for experiences in other usages as well. I happen to not use Photoshop and Final Cut Pro, and I don't doubt that these Macs will be fast.
 
Like I said, I've seen it, that's why I wished for experiences in other usages as well. I happen to not use Photoshop and Final Cut Pro, and I don't doubt that these Macs will be fast.

If the A12Z can handle those intensive applications it will have no issue opening an email client or a browser.

What exactly do you need to see?

I hope browsing will become as smooth as it is on iOS compared to a normal desktop.
 
I hope browsing will become as smooth as it is in iOS compared to a normal desktop.
That's what I'd love to see as well. My normal workflow consists of batch exporting/transforming photos, watermarking them, a lot of open tabs simultaneously and working in one or two content management systems. It's not exactly what I do on an iPad Pro (where I could infer some data/experience), so I'll be thrilled to see what the new Macs are able to do. My workflow was sometimes too much for an older MBP (from 2015, I believe).
 
Oh my, its identical to the iPad Pro, what a surprise :rolleyes:
Well, it's funny that the A12Z in the DTK actually runs 100 MHz slower IIRC. Or was that because of the non-native Geekbench that couldn't gauge the frequency right?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.