Check the scores for an i5 Macbook Pro.
Remember, this is under emulation. If it already performs as good as an i5 under emulation, native speed will be fine.
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.
Rosetta 2 Benchmarks Surface From Mac Mini With A12Z Chip
- Thread starter MacRumors
- Start date
- Sort by reaction score
Remember, this is under emulation. If it already performs as good as an i5 under emulation, native speed will be fine.
Wow. That's impressive, all things considered.
And LOL at the form posters who were saying that there wouldn't be any benchmarks because "The EULA forbids it". 🤣
And LOL at the form posters who were saying that there wouldn't be any benchmarks because "The EULA forbids it". 🤣
Apple is intentionally underplaying their hand here. They're giving developers a pathway to translate applications over to Apple Silicon, but are intentionally giving out underperforming chipsets as to withhold the true processing power that they have in their pipeline of Apple Silicon. Considering the A12Z is just an A12X with an extra GPU core enabled, and the A12X is from two years ago, they absolutely have something like an A14Z internally that does laps around this 12Z.
Keeping in mind even an A12Z is still technically and iPad Pro chip, I'm sure we will see additional "branches" of their A chipsets too in the future, similar to how we got the "X" chips in the iPads a few years ago. We would probably see something like the A12Y (just making this up as an example) for usage in the iMac or Mac Pro.
Keeping in mind even an A12Z is still technically and iPad Pro chip, I'm sure we will see additional "branches" of their A chipsets too in the future, similar to how we got the "X" chips in the iPads a few years ago. We would probably see something like the A12Y (just making this up as an example) for usage in the iMac or Mac Pro.
Not really. Intel has been offering mild performance increases with crazy heat and power efficiency costs for a while now. This is a solution to a very real problem.
This old iPad chip under emulation is already performing close to a current i5 Macbook Pro.
Imagine the potential.
We are gonna see HUGE gains. It’s not even fair how good these machines are gonna be. I can’t wait.
Yep. It’s depressing to see. I wish the macrumors article on it explained ALL the reasons this is meaningless, since people are just not getting it.
It does tell us that roughly 30% goes to emulation overhead for Geekbench compared to iOS scores... Not sure if they are weighted the same way though.
"3) Only using 4 out of 8 cores for some reason"
Isn't that just big.LITTLE ? Does the iPad actually utilize all cores when running Geekbench? Wouldn't even be worth it (from scheduling / emulation) perspective to split tasks between those cores on the fly?
EDIT: toned down arrogance, made it more conversational.
I don't get the negative reactions here. First, the only thing interesting is the single core performance (because they will put a lot more cores in the real deal later this year, if it's just because they can).
And then keep in mind that this is Geekbench running on x86 emulation (Rosetta), on an old CPU design that is not optimized, on pre-production hardware, on a first beta of the OS. And given all that, the numbers are stunning. We're talking mid tier Intel Core i3 preformance here y'all. Woah.
And then keep in mind that this is Geekbench running on x86 emulation (Rosetta), on an old CPU design that is not optimized, on pre-production hardware, on a first beta of the OS. And given all that, the numbers are stunning. We're talking mid tier Intel Core i3 preformance here y'all. Woah.
Wow, this is seriously impressive. Essentially, Apple's 2018-era compute can run translated x86-64 instructions at quad-core, mobile-class Haswell (e.g. i7-4770HQ) speeds. And this is with a beta translation layer, on a beta operating system, without all cores properly utilized. If they can get those numbers up to quad-core, desktop-class Skylake (e.g. i7-6700) levels for iMac 24" and MacBook Pro 13.3–14" launches this fall, I think that would be more than adequate. (And obviously, we should expect native ARM app performance to be a lot higher.)
Exactly, rather than simply putting competent thermals in their products, they’re doubling down on their own hubris.
The Mac lineup has been one misstep after another, I don’t believe in them to get this right.
This will create more problems than it solves, not the least of which being another pain-point with the incompatibility to run Windows natively.
It is depressing benchmarks where released and even more so that people seem unable to understand this CPU will never make it into a shipping Mac and that we are benchmarking an emulated application.
Some people don't understand the reason. I prefer 1 core for os, 1 core apps, burst when needed 2 more core.
Just 30% is blowing my mind, usually you're in the range of 50 to 100% overhead for x86 emulation on ARM.
You do realise 4 of the 8 cores are low-power, high-efficiency cores not meant for demanding tasks like a benchmark?
While not conclusive by any means, these results give an indication of the kind of performance hit you might encounter when running non-native apps through the Rosetta translation layer, which was to be expected.
I just hope Apple makes the process easy to port apps.
Last edited:
The scores are actually quite impressive when you take into account what they are.
Yup. If Rosetta 2 will give at least a Core i5 level performance, there's really nothing to worry about.
OpenCL comparison (Just found it on twitter 🤷♂️):
Also Metal:
MacBook Air (Early 2020) vs eperm-d995af6e2ef02771 - Geekbench
browser.geekbench.com
Also Metal:
MacBook Air (Early 2020) vs eperm-d995af6e2ef02771 - Geekbench
browser.geekbench.com
So, these benchmarks are meaningless, being based upon old chips.
That said, 811 average is 73% of native performance for emulated (maybe a bit higher, if 2.4GHz vs 2.5GHz).
In single thread, it matches:
iMac (27-inch Late 2012) Intel Core i7-3770 @ 3.4 GHz (4 cores)
In multi-core, which is limited to the good cores, so 4 threads, it matches:
MacBook Pro (15-inch Mid 2012) Intel Core i7-3720QM @ 2.6 GHz (4 cores)
MacBook Air (Early 2020) Intel Core i5-1030NG7 @ 1.1 GHz (4 cores)
Although these will have had 8 threads.
So the A12Z, emulating x86, is already as fast as the most recent MacBookAir in multi-thread (although ST for this one is 1070).
Add 20% (A13) and 20% (A14) and the next MBA-ARM is going to be far far faster - and that might be 8C as well!
That said, 811 average is 73% of native performance for emulated (maybe a bit higher, if 2.4GHz vs 2.5GHz).
In single thread, it matches:
iMac (27-inch Late 2012) Intel Core i7-3770 @ 3.4 GHz (4 cores)
In multi-core, which is limited to the good cores, so 4 threads, it matches:
MacBook Pro (15-inch Mid 2012) Intel Core i7-3720QM @ 2.6 GHz (4 cores)
MacBook Air (Early 2020) Intel Core i5-1030NG7 @ 1.1 GHz (4 cores)
Although these will have had 8 threads.
So the A12Z, emulating x86, is already as fast as the most recent MacBookAir in multi-thread (although ST for this one is 1070).
Add 20% (A13) and 20% (A14) and the next MBA-ARM is going to be far far faster - and that might be 8C as well!
On the ipad it uses all cores, including the low power.
Also it’s not big.little. Big.little is the Arm version of heterogenous cores, and it works quite differently in terms of how it decides to use powerful and efficient cores, which cores it can run simultaneously, etc.
Quite the contrary, numbers suggests that Rosetta gets >70% of native performance, quite impressive. But also understandable that this time, no endian-swapping is involved (compared to PPC on x86 back in the days).
As for the HW, 2 old chip not designed for the task at hand not even using it's full potential....
Way better than I expected. Already scoring broadly in line with the scores for a 2018 Mac mini with a 4 core i3 at 3.6 GHz, and as lots of other people have said, this isn't a desktop chip, it's drawing less power, runs at a much lower clock speed, has half it's cores turned off and is running emulated code.
What it doesn't tell us is how well Apple can repurpose their chips from a power-constrained portable environment to a heat constrained desktop one, or what effect the change to Apple silicon will have on end user prices. That last point worries me, as the change is going to be over 2 years. If own-brand CPUs blow away Intel on price and performance, how will Apple have half their range with them and half without?
What it doesn't tell us is how well Apple can repurpose their chips from a power-constrained portable environment to a heat constrained desktop one, or what effect the change to Apple silicon will have on end user prices. That last point worries me, as the change is going to be over 2 years. If own-brand CPUs blow away Intel on price and performance, how will Apple have half their range with them and half without?
If you were to judge Intel Macs by the DTK at the time (a Pentium 4… not even a Pentium D dual-core, as I recall), they would've looked pretty poor. Rather inefficient and slow.