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.
Leaked Benchmarks Confirm M2 Chip is Up to 20% Faster Than M1
- Thread starter MacRumors
- Start date
- Sort by reaction score
Be interesting to see the reviews. Maybe I just don't understand engineering enough but I can't see the M2 MBA sustaining 20% better speed without thermal throttling under consistent load. Probably the 13 Pro is where we'll see the difference sustained. Stay tuned for a Max Tech video showing that an M1 Air can maintain performances better over long periods than an M2 Air which will throttle to lower than M1 speeds in order to cool down. Maybe.
Clock + huge L2 cache + RAM bandwidth
We've seen this formula before with Intel with lithography stalls. Some people here don't seem to get it.
Architecture changes allow such improvements. Memory controller enhancements or increasing caches will increase performance alone. Not to mention the extra 700 MHz the M2 is now supporting.
All that however comes at a cost of power.
Source?
Increasing clock does of course increase power consumption. However, we also know that the e-cores in particular are much faster, and therefore will be used (in favor of the p-cores) a lot more compared to the M1. The A15 iPhones overall have much better battery life than the A14 ones, despite a higher clock.
So, what makes you so sure the M2 has higher power consumption?
No and not really.
The M2 MacBook Air got a nice bump in battery capacity, going from a 49.9‑watt‑hour lithium‑polymer battery in the M1 to a 52.6‑watt‑hour lithium‑polymer battery in the M2, so any increase in power consumption should be quite balanced out and, like Apple says, the battery life in the end should be about the same.
As for the fan, if it’s anything like M1, The actual difference that it makes is extremely minor
Last edited:
They might not benefit much from the new processor but they certainly will from the new design, better WebCam, bigger Display, better speakers, and MagSafe.
$200 buys you:
- a slightly faster CPU
- a much faster GPU
- a lighter, thinner chassis with smaller bezels
- a much better camera
- a much better sound system
- MagSafe
Last edited:
I'll use Apple's own charts:
See the how the M2's curve starts after the M1's curve? That's extra power at idle
GPU? Same story but much more noticeable as to reach the peak performance there are around an extra 3-4 W consumed vs the M1
As usual, you're the one doing the cherry-picking—while falsely accusing someone else of doing exactly what you're doing (I can't comment on MT's other stuff, but he clearly wasn't cherry-picking here).
You've cherry-picked a single high result for the the M1 MBP from the Geekbench Browser to use as a comparator.
Instead, the more honest and reasonable thing to use as a baseline would be what's listed on Geekbench's Mac Benchmark Chart, which is based on an average of multiple results ("To make sure the results accurately reflect the average performance of each Mac, the chart only includes Macs with at least five unique results in the Geekbench Browser.") (see screenshots below).
Of course, this will get more accurate once we also have multiple results for the M2. But until we do, what I described above is the best we can do, not what you did:
1919/1707 => 12% higher SC
8928/7395 => 21% higher MT
Last edited:
I'll use Apple's own charts:
See the how the M2's curve starts after the M1's curve? That's extra power at idle
That's fair.
We're talking 3 vs. 4 watts, though. Either of those is way below even Intel's 9W (née "Y") chips.
Not sure why you think that's "at idle".
I don't really care about the GPU.
I’m also guessing you can add another year or two of OSX updates. The
This guy seems to know what he's talking about. He should consider starting a YouTube channel on the subject of tech reviews and product predictions.
Intel has designed all their upcoming products for what they call “optionality." In other words Meteor Lake can be fabbed on N5 if needed.
Just like it’s been for the past decade and a half or so, going right from one generation to the next rarely shows a dramatic improvement, no matter who’s making the processor.
The 2018 MacBook Pro‘s weren’t that much faster to someone upgrading from the 2017 MacBook Pro‘s if you get what I’m saying.
Same situation here.
Going from M1 to M2 might not be that big of an improvement, might not let you do your work that much faster, might not get you home that much quicker.
But moving from an Intel MacBook Air to this M2 certainly will show dramatic improvements and be much quicker.
That’s who this is targeted at, just like with the majority of products these days.
And likewise going from M1 to M2 might not be that big of a difference, but in a couple years going from M1 to M5 is going to be quite a boost.
Iterative y-o-y improvements aren't necessarily about enabling new use cases.
But yes, walk another eight years down the road and the M1 MacBook Air will in fact be annoyingly slow for typing in Microsoft Word and watching Apple TV+, because everything else will have become slightly more complex. macOS will gain features you use (or not) that take up little bits of performance. As will the Word and TV apps.
Na, it’s certainly going to throttle but it will not be slower than an M1 even under heavy stress because of the speed improvements of the efficiency cores
I wasn’t impressed by these numbers. It seems the M2 is more of just a clock speed and core increase rather than an actual refinement of the chip’s silicon. This will make the machine more power hungry and will in turn require a bigger battery to last the same amount of time. Not what I was hoping for personally.
Let me address each point in turn:
1) Does it let you go home and cook dinner earlier? NO.
2) Meet with friends at kareoke on time. YES. PUNCTUAL KARAOKE IS A KEY BENEFIT OF THE M2, SINCE IT'S HARDWIRED INTO THE NEW NEURAL PROCESSOR.
3) Spend more quality time with family. NO, THIS IS REDUCED BY NEW MAC PRODUCTS.
😄
*****
As others have mentioned, there is typically little real-world benefit to individual generational changes (the exception would be if, even if the average performance doesn't change much, certain specific workflows are made much faster). But these changes are nevertheless interesting to discuss from a technological point of view.
And, of course, it's the accumulation of these small generational changes that are needed to get significant improvement in the long term.
Last edited:
Everybody remembers those graphs from Apple with a steep ramp in performance.
Single thread:
- A8 to A9 - 55% increase
- A9 to A10 - 40% increase
- A10 to A11 - 20% increase
We're just seeing smaller and smaller deltas. What used to be 20% increases for single-thread are now multi-thread.