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.
Is the M2 really faster than the M1?
- Thread starter Misheemee
- Start date
- Sort by reaction score
Right. I was a bit disappointed for no price drop for the 2020 MacBook Air M1, but it'll still be on our standard offering for staff/faculty. AppleCare+ for the MacBook Air M2 is also increased.
If you’d like to get into the weeds, I recommend The Signal Path as a basis for discussion. 18 minutes into his analysis of an RF board is a good place to start:
Approximately 9.5% faster in single core, and 19% faster in multi core, in Geekbench CPU tests. Not bad at all:
Seems like the M2 would also scale pretty well with the Pro, Max and Ultra SKUs as core counts increase. Especially for multithreaded workloads.
The video decode/encode engines on the M2 will also be a significant speedup as compared to a base M1 for people using the supported codecs in their work. All in all, the M2 appears to be a solid step forward.
Seems like the M2 would also scale pretty well with the Pro, Max and Ultra SKUs as core counts increase. Especially for multithreaded workloads.
The video decode/encode engines on the M2 will also be a significant speedup as compared to a base M1 for people using the supported codecs in their work. All in all, the M2 appears to be a solid step forward.
I mean, yes, you can analyze decapped chips. It's easier when it's analog like the chip The Signal Path was looking at - analog chips always have physically bigger circuit elements (you can't scale analog down as far as digital without running into noise problems) and as he notes they also often have much less metal covering the base layer. But you can still see some things in digital circuits, particularly large regular features. For example, it's easy to see that there's some SRAM arrays in those blocks labeled A1 and A2.
But what we need for further discussion is the rationale as to why that website identified A1 and A2 as "dual-core Secure Enclave". Why do they think there's two cores in a Secure Enclave, and why do they think those particular blocks are those cores? I've given my counterargument, it's all much too large to be the Secure Enclave and I don't think there's two cores in the SE, but I dunno where to go from there.
Of course, the critics will say that Alder Lake still has a slightly faster single core score (with a much higher power draw).
Just doing a quick search, it looks like the Core i9-12900K part at is slower - 1834 - in single core than M2 - 1869
Intel Corporation Alder Lake Client Platform - Geekbench
Benchmark results for an Intel Corporation Alder Lake Client Platform with an Intel Core i9-12900K processor.
browser.geekbench.com
Where did you get these numbers from? I ask because some of them are clearly fictional - the chart has M1 Pro/Max/Ultra GB5 single core scores about 200 points higher than they should be. I just ran GB5 on my M1 Max and it scored 1765.
M1 Pro/Max/Ultra does not score over 1900 on Geekbench singlecore. They are all around 1750 give or take 20.
That chart is from a CPU site that is estimating the scores and they are showing incorrect single core scores for M1 Pro/Max/Ultra
That chart is absolutely wrong in single core results.
All M1 processors have single core score of ~1700 points (M1/Pro/Max/Ultra).
And no M2 has been benchmarked yet, but could be close to those numbers.
Last edited:
Compared: M2 vs M1, M1 Pro, M1 Max, and M1 Ultra | AppleInsider
Apple's new M2 is the start of a new Apple Silicon generation. Here's how the M2 fares when compared against the already-released M1 family of chips.
appleinsider.com
You’re right about the Pro, Max and Ultra single core numbers. They are higher than what you actually find on Geekbench.com
The other M1 numbers all look right. Who knows? I don’t know where they got these M2 results from.
FYI, while their site changed, it appears that the M1 sample report is still available.
Perhaps the SystemPlus people meant dual core cryptography engine, with a portion of the identified block being the actual Secure Enclave. If you look at page 20 of the sample report, they clearly identified separate blocks that they lumped into the "Dual Core Secure Enclave" major IP block.
Please modify your post. These are not M2 results “in Geekbench CPU tests”. No such results exist yet. Also, the M1 single core scores are completely wrong.
Sigh. No, I guess they’re continuing to transition their site. I saved a copy, so I’ll upload it somewhere and post a link in a bit.
The Wayback Machine to the rescue!
Edit: Oddly, the whole PDF shows up on my computer, but only the first page shows up on my phone. The Internet Archive needs to fix that up… Hmm, I should probably donate.
Last edited:
real world it matters a lot. my M1 mini with 8 GB seemed blazing fast to boot, boot apps, crunch data. but if i have safari open with dozens of tabs (normal for me) i’m red lining memory in no time and beach-balling every other minute.
while i agree with you...i have to disagree with you in some points
M1 MBA was announced 18 months and not 24
18% increase in raw cpu is ok for a second 5nm generation, expect a better improvement with the M3 since it will have 3nm and armV9
But also...the gpu is over 30% improvement over 18 months so...and think about scaling this to the M2 Max/Ultra since probably we will have LPDDR5x into those
one youtube channel apple dude is saying ArmV9 isn’t a big deal over v8. says a lot of that stuff in v9 are things apple has already done in custom mods to ARM architecture with M1 and A series and it’s just ARM packaging that kind of stuff into the everyman SoC for their customers who are smaller hardware product vendors and can’t afford to design and fab their own silicon.