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.
First M2 benchmarks
- Thread starter EugW
- Start date
- Sort by reaction score
Yeah, but it would throttle in the iPad Pro too. The Mac mini and MacBook Pro both have active cooling.
Of course, that goes without saying. I was using the terminology loosely just to clarify the point.
There are two reasonable options that I see here. First, release a new version and make it clear that the benchmark scores are incompatible between versions (GB already has a precedent). Second, validate a patched (for warmup period) GB5 across a range of earlier results. If I am correct, only Apple GPUs will see a change, others should stay the same. Then one can purge M1 compute scores and release a bug fix.
Personally I tend to GB6 with incompatible scores
Oh, for sure the iPad Pro could be made to throttle; but I was more referring to the "used in" aspect of the M2 SoC...
I’m strongly considering ditching my 2018 MBA and Mac mini for a new M2 air.
I’ve really enjoyed my mini. The mini maxed out cpu w 64gb aftermarket ram and paired with the Blackmagic egpu. A beast of yesterday.
And I mention this above but the 2018 mini setup w this old eGPU still beats the reported GB metal score on the M2. I’m trying to discount this general compute score as not reflective of performance.
But the single and multi threaded CPU and increases are very compelling! This thing puts my prior mini on blast.
I mostly use the mini + egpu with a single XDR pro display. But want to work on the couch sometimes and having a second machine kind of sucks. Especially when that is the 2018 MBA.
It’s awesome to think I can collapse the egpu, the mini and the laptop into a single portable.
And you would be right. For most applications where GPU performance matters, the lower latency will make a bigger performance impact. So unless you are doing GPU-based number-crunching in a large scale, the M2 Air will be faster than your old eGPU setup.
"No problem! Just add another 16 efficiency cores!"Shysters gonna shys. How's Intel's strategy of securing Apple as a foundry customer going?
Now I can't find anything on twitter so maybe don't take this too seriously. 😬
Not sure if this is serious (and not really specific to GPU tests) but it implies it'll be 5.5
Better than their strategy of securing Apple as a customer for their processors.Shysters gonna shys. How's Intel's strategy of securing Apple as a foundry customer going?
M2 has pretty much everything I wanted for a high end Mac mini*, except for possibly one thing: An increased number of USB ports.
M1 Mac mini:
Intel Mac mini:
I'm not sure but I'm guessing that M2 doesn't have the additional controllers onboard for the two extra Thunderbolt ports. If not, how likely would it be for Apple it to add an external controller? IOW, what would it take to get 4 Thunderbolt ports on an M2 Mac mini, if the controllers are not on board, in additional to the 2 USB-A ports? Or, how about 2 Thunderbolt ports plus 4 USB-A ports?
Or perhaps they'd take the iMac route, and dispense with USB-A completely. However, the M1 iMac is still limited to 4 USB ports:
I'd be fine with 2 Thunderbolt ports and 4 USB-C ports too. I just want a total of 6 USB ports. And no, I don't want to deal with third party USB hubs if I can help it.
*I especially like the 24 GB RAM option. Some see the lack of a 32 GB option a disadvantage, but I see it as an advantage, at least for me personally. Why? Cuz 32 GB is way overkill for me, and is quite expensive. OTOH, 24 GB hits that upper mid-range sweet spot for many users, with an upgrade cost from 16 GB to 24 GB being half as much as the upgrade cost from 16 GB to 32 GB.
M1 Mac mini:
Intel Mac mini:
I'm not sure but I'm guessing that M2 doesn't have the additional controllers onboard for the two extra Thunderbolt ports. If not, how likely would it be for Apple it to add an external controller? IOW, what would it take to get 4 Thunderbolt ports on an M2 Mac mini, if the controllers are not on board, in additional to the 2 USB-A ports? Or, how about 2 Thunderbolt ports plus 4 USB-A ports?
Or perhaps they'd take the iMac route, and dispense with USB-A completely. However, the M1 iMac is still limited to 4 USB ports:
I'd be fine with 2 Thunderbolt ports and 4 USB-C ports too. I just want a total of 6 USB ports. And no, I don't want to deal with third party USB hubs if I can help it.
*I especially like the 24 GB RAM option. Some see the lack of a 32 GB option a disadvantage, but I see it as an advantage, at least for me personally. Why? Cuz 32 GB is way overkill for me, and is quite expensive. OTOH, 24 GB hits that upper mid-range sweet spot for many users, with an upgrade cost from 16 GB to 24 GB being half as much as the upgrade cost from 16 GB to 32 GB.
This is the biggest downside of shedding the BM eGPU for me. It has been a super reliable and stable dock beyond a graphics accelerator. Adds lots of ports and power.
Can anyone point at the current best value Apple Silicon / XDR Pro display capable dock?
Help me to understand one thing.
“Thermal cooling” keeps being thrown about a lot on this forum… Benchmark testing takes 2 minutes to do the fans won’t be kicking in till then regardless of the device it’s performed on. So benchmark score on iPad, MacBook, MacBook Pro or whatever should not differ.
Whatever difference you gonna get while the device “being hot” is so minuscule it won’t make a difference to your day-to-day operations. At this level they’re all the same - so why care?
Perhaps everybody trying to pedant each other out with their knowledge they don’t know nothing about but the theory?
Thanks for helping me understand in advance.
“Thermal cooling” keeps being thrown about a lot on this forum… Benchmark testing takes 2 minutes to do the fans won’t be kicking in till then regardless of the device it’s performed on. So benchmark score on iPad, MacBook, MacBook Pro or whatever should not differ.
Whatever difference you gonna get while the device “being hot” is so minuscule it won’t make a difference to your day-to-day operations. At this level they’re all the same - so why care?
Perhaps everybody trying to pedant each other out with their knowledge they don’t know nothing about but the theory?
Thanks for helping me understand in advance.
There are workloads that put a lot of pressure on the cooling system. For example gaming, rendering or running numerical workloads. Since MacBook Air is passively cooled it’s sustained power dissipation is likely limited by 10-15W, so if you are doing these kinds of things you might see 10-20% better performance on a model with active cooling.
And that roadmap says 7% performance increase OR 15% increase in efficiency, the M2 achieved both which is what’s impressive
This shouldn't be controversial, but I think the correct way to read all of TSMCs documented "expectations" (which aren't a guarantee) would be to prepend it with "All else being equal ..."
So that means: All else being equal, moving to N5P would give 7% performance increase OR 15% increase in efficiency from N5.
But M2 is not M1. Apple hasn't stopped working, nor will Intel/AMD/etc. Those should be considered the worst case scenario for jumps in performance for moving nodes on future generations of silicon.
Last edited:
Your comment is “cute”, but ultimately irrelevant.
Note that we may see more benchmarks before the embargo lifts.
These first benches were likely strategically leaked by Apple.
If people got review units, invariably they run Geekbench on them even if they aren’t allowed to publish benchmarks, and those sometimes get automatically submitted to Geekbench.com for the whole world to see. Unfortunately this doesn’t happen for most other benchmarks so we mostly just see Geekbench scores posted automatically, but we do occasionally see a few others.
But will Apple send out review units this time, considering it’s the same old 13.3” MBP chassis? Or will they just wait for the 13.6” Air? I’m thinking they will send out review units because it’s the first M2 release and they’ll want to highlight major improvements like ProRes video editing. I also wonder if they’d send out 16 GB or 24 GB units.
If people got review units, then the embargo would lift sometime this coming week.
Edit:
I just checked, and the review embargo lifted for M1 on the same day it was available, November 17, 2020.
So if they follow the same pattern for M2, that would be this Friday, June 24.
Last edited:
Yes, but they would have had the units for testing up to a week early, meaning they would be doing their testing all this week. And if they aren't paying attention, their test results could unintentionally get submitted to the test databases. It happens pretty much every single time review units are sent out, because the default benchmark software setting is to automatically submit the results to the benchmark database if the machine is connected to the internet.
BTW, the few scores that were submitted already are undoubtedly intentional leaks from Apple. That also happens almost every time:
1) Geekbench results show up a week or so before release, but only a few of them, then nothing for days.
2) Then a couple of days before release, a whole bunch of Geekbench results show up.
3) Also a couple of days before release, occasional other benchmark results show up.
For 2 & 3, this assumes people got review units. However, AFAIK, review units aren't always sent out. I guess they would be for M2, but that 13" MacBook Pro chassis is underwhelming in 2022. I had originally guessed the MBA and MBP would come out at the same time, but they didn't. The good news for benchmarking with the 13" MBP is that it won't throttle under extended tests like Cinebench - long duration.
BTW, when we ran our own long duration Cinebench on our fanless 2017 MacBooks to test throttling, we had to do it manually, restarting Cinebench immediately after each bench completed, for that duration of 30 minutes. There was no such long duration setting in Cinebench at the time.