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.
RIP OS X/macOS 10
- Thread starter timidpimpin
- Start date
- Sort by reaction score
Also, the actual consumer hardware will be much more powerful then this dev kit. You can do a lot more with a bigger form factor.
Absolutely. The Intel transition system they put in a Mac Pro case was very underpowered also. And my old PC only just beats it by about 150 points, so that's actually decent considering all the overhead the system has. It's nearing the performance of an 8 thread i7.
It doesn't sound like you understand that. It's not a "9 year gap" if you're comparing native performance to emulated performance.
I have another Dell 2nd gen with a 4C 4T i5 2400 and a 1050 ti that I plan to Hackintosh. Likely High Sierra. The i7 is my gaming system.
[automerge]1593450300[/automerge]
I really do understand that. Trust me. I guess emulation hasn't come very far over the years.
Someday I want to build the ultimate Snow Leopard rig with a ridiculously overclocked i7-2600K.
My current fastest real Mac is a C2D 2.26GHz, so even that i5 2400 will be a huge boost for me. And some of the 2011-2012 Macs had a 2400, so these 2nd gen tend to Hackintosh well from what I've seen.
This is my machine, the 2013 Air. I've had to replace the battery (failed) and the top case (keyboard went sticky after v mild liquid damage ooops), but honestly it's a great. Cannot believe it's running as well as it does 7 years after my purchase from new. And it will even run Big Sur!
The MBA has been Apple's most popular laptop for many years now. It makes sense that they would keep supporting it. My aunt has a 2013 13" also, and loves it.
What a lot of people I think fail to realize at this point is that Apple deliberately told developers not to benchmark this machine. Given they knew everyone and their mother who got their hands on this machine was going to want to benchmark it, I’m sure they put various things in place, on top of the fact that the benchmark is being emulated, to ensure the results are inaccurate and skewed to prevent actual numbers from leaking out this early, allowing competitors to get an advantage by knowing what they’re up against before it’s even released. The Intel DTK I believe wouldn’t even properly complete benchmarks at some point because of restrictions Apple put in place.
What you're suggesting would seriously cause a mountain or two of extra work for the developers.
History lesson time: When RISC architecture was being developed, there were two major schools, Berkeley and Stanford. Stanford-based designs became MIPS and PowerPC, while SPARC, i960 (and Itanium), and Am29000 came from Berkeley.
Classic 32-bit ARM incorporates CISC features and limitations that you don't see on other RISC designs. Other architectures were initially designed for expensive workstations, whereas classic ARM was made for a low-end home computer where cost, code density (limited RAM and I/O) and interrupt handling (for software video) were priorities.
When ARM went to 64 bit, they cleaned a lot of this stuff up. That's why Apple was eager to dump 32 bit ARM in iOS.
MIPS and PowerPC assembly are nearly identical. ARMv8 is very close. SPARC is definitely related but you have to learn some different ways of doing stuff. Classic 32-bit ARM is related but has obvious CISC influences. x86/x64 are way different.
Not necessarily. Apple knows exactly what companies make benchmarking software and can deliberately make rules in the specialized version of Big Sur found on the DTKs to sabotage results. This would have absolutely no impact on developers as they’re just using the hardware for its intended purpose: testing and updating their software for usage on Apple Silicon-based Macs.
The only developers that would face a problem while using these Macs are the developers of said benchmark software. However, just using Geekbench as an example, they already have an app the iOS App Store that supports the A12Z found in the DTK, so they technically wouldn’t need to do any work to get it natively running on Big Sur and Apple Silicon thanks to Mac Catalist.
This is likely what makes the current x86 benchmarks on arm even less useful right now. We're talking about development hardware running beta software that, as I understand it, doesn't even do real time emulation, instead doing a sort of at install “translation” of x86 code to arm. Benchmarking that is going to need more than one synthetic benchmark tool.
I am very curious to see how the final results pans out compared to arm Windows 10 32-bit only x86 emulation, but I don't think we're there yet.
But presumably, however well Apple locks the machines down, there is no way to prevent one falling into competitors' hands and these must have enough know-how to find their way around any such hurdles and get a good idea how well Apples A-chips are performing.
Sounds like the ARM chip sucks under Rosetta, let’s see real tests with real IOS apps. Even my Quad G5 can beat that thing. Under 3000 Geekbench score ?
So let's see the Quad churn through Geekbench 5.
Last edited:
You do realize that Geekbench scores are not directly comparable across different versions, right?
Geekbench 5 doesn't even run on PPC.
There’s absolutely no denying that. Microsoft, for example, has had their hands on one for presumably a pretty long time by this point to already have office ready to go at launch. A large majority of Apple’s competitors have apps both on the iOS App Store and the Mac App Store, so they are absolutely able to sign up for the program and receive one of these DTK machines, whether or not they actually intend to use it for application development purposes. It’ll be inevitable that realistic performance metrics will eventually leak out or be uncovered by users, developers, or competitors once they find ways around apple’s hurdles.
As many, many, many, many, many people have already noted on the other thread, this is grossly pre-production hardware (with a tablet-grade SoC that's two generations old) and completely meaningless in terms of indicating how well "Apple Silicon" will actually perform under real-world conditions in fully shipping products.
Put another way: Microsoft shipped SDKs to XBox 360 developers which used rebadged Power Mac G5s for the hardware. Did that mean that the actual shipping XBox 360 was itself a rebadged Power Mac G5? Of course not.
Put another way: Microsoft shipped SDKs to XBox 360 developers which used rebadged Power Mac G5s for the hardware. Did that mean that the actual shipping XBox 360 was itself a rebadged Power Mac G5? Of course not.
Well, we all know that already. We've been here before, you know. Having said that, whatever Apple does pump out, it would have to be backwards compatible 100% with this DK otherwise the developers have really wasted their time. I'm not expecting to be blown away with what Apple is going to bring to the market. If Apple keeps pace more or less with the lesser of its current Intel offerings, then it is doing OK. As for what it intends for the iMac Pro and the Mac Pro, assuming either survives, is anyone's guess.
They already are sabotaging results by putting a less powerful chip then the iPad Pro in the dev kit. This is definitely deliberate. Also as other have said even if the above wasn’t the case, this is running over Rosetta on a mobile designed chip clocked slower then on actual mobile hardware. These results truly are useless, these don’t even represent the performance of Rosetta on any Mac being released this Fall.
You are 110% correct. These are not and will never be consumer-grade mass produced machines. Their single purpose is to serve as a utility for Mac application developers to prep their apps to work day-1 on Apple Silicon-based Macs. They don’t have to be insanely fast, they don’t have to have an identically clocked/specced processor as a consumer-shipping iPad Pro. They run Xcode to translate x86 code to Arm code, run the newly transitioned apps, and run pre-transition apps to test for compatibility. I’m willing to bet the “DTK” machines apple was using at WWDC for the keynote demos are far superior to the actual DTKs being sent to developers to further keep things under wrap and absolutely stun everyone with the mass produced models shipping in the fall.