Apple actually stated that they are using an up to marker and even though some tasks are faster.
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
Appel has to say "up to" otherwise someone will complain when absolutely every task doesn't hit that number.
The CPU’s are both 4 Performance + 4 Efficiency cores. So the cores themselves are improved. Core vs core, M2 CPU cores are faster.
Same ks true for GPU, 8 core vs 8core, M2 is faster. Additional 2 cores adds a lot to that 1.4x graphic figure.
But to answer your question, yes, core for core, the M2 is faster
Same ks true for GPU, 8 core vs 8core, M2 is faster. Additional 2 cores adds a lot to that 1.4x graphic figure.
But to answer your question, yes, core for core, the M2 is faster
Adding more GPU cores makes it faster at graphical tasks. What's the issue here?
Well you should. AMD, Intel and Nvidia do the same footnote thing
Let me start here - there's no issue for me as I have a MBP 14" which I'm quite happy with. I'm just highlighting this for consumers, or for anybody thinking of upgrading from the M1 air to the M2 air - specifically to do with Apple claiming that the M2 is 1.4x faster than the M1
To answer your question @KPOM - yes. the M2 that was tested against the M1 (which resulted in the claim that the M2 is 40% faster than the M1) had 2 additional cores and 8GB more ram, and the test was "a complex 2 minute project in Final Cut Pro" which means that the M2 wasn't tested against a similarly spec'd M1 which resulted in it definitely being faster - stats which I feel are skewed. I know the new chips will be faster, and the cores will be faster - but I really don't think that the M2 is 40% faster than the M1 based on the testing that was done
The M2 is that much faster than the M1 because the parts you say are not comparable are exactly what the M2 is. It’s not a CPU. It’s a system.
lol Intel got away with 8 generations (Haswell to Rocket Lake) of worse year-over-year performance gains with zero supply-chain excuses. It's one of the reasons Apple dropped them in favour of their own chips!
I have a Late 2013 iMac with an i7-4771 CPU, it has GeekBench 5 single-core score of 913. The fastest Intel iMac ever released (Mid-2020 27", i7-10700K) manages 1250, a 37% increase over 7 years (5.2% year-over-year average). An 18% increase in single-core over 20 months works out to double that: a rate-of-increase of 10.8% year-over-year. If they can keep this up, Macs should see much healthier performance increases with each revision than they did in the late Intel era.
It’s not just about speed it’s about efficiency
Battery life on the m1 air is insane
Battery life on the m1 air is insane
Sure, but 35% faster for the GPU seems pretty conservative, since that would require only an 8% increase in per-core performance. And the 18% faster for CPU multi-core seems easy to achieve as well, since reports are that the 4E cores in the M2 are significantly more performant than those in the M1. I.e., don't expect to see the single-core speed of the M2's P cores to be that much faster than the M1's.
From what I know, the E cores of the M1 only contributed about 25% of the total MT scores.
I would think the bulk of the 18% improvement should still come from the P cores, in the form of higher clock rates.
i already told you how it works
The M1 8 gpu cores will not be 40% slower than the M2 8 gpu cores, , is more about 12% increase if you want to compare apples to apples, spec to spec. But it is faster, also the M1 8 cpu cores compared to the M2 8 cpu cores is also faster, around 18%
But here, you have the option with the M2 to go if you want 12% increase in gpu over the M1 or go up to almost 40% increase...is up to you, is how you want to "build" your system You want an nvidia 3070gpu or an nvidia 3080 gpu to talk in kind of old fashion way
But the cores are better, more Ram options, better ram since the M1 was the only one from M1 family on LPDDR4x and not LPDDR5
Prores workloads are 3x faster. You should check all the charts and footnotes…
The +40% is a workload which includes ProRes as part of the mix.
I’m commenting on the +40% number, because that was the question asked.
Cherry picking other workloads doesn’t answer the original question.
The A15 is a known quantity, it runs at about a 6% faster clock rate than the A14 but single threader performance is about 12% faster so there is a small contribution of IPC and a small clock rate contribution. Performance of the M2 should go up relative to the M1. The A15 gets its increased performance at an ever so slightly lower power consumption than A14 so presumably the M2 should be able to achieve the same 12% performance improvement. Time will tell of course.
www.anandtech.com
The Apple A15 SoC Performance Review: Faster & More Efficient
Again, A15 is a known quantity, vs A14 with the same number of GPU cores (4 vs 4) it gets 16% better sustained performance and about 13% better peak performance. If we take into account the extra GPU cores (5 in A15 vs 4 in A14 - same relative jump in core count as M2 10 core vs M1 8 core) then we get the same 16% better sustained (thermal limit in iPhone most likely here) and a much bigger 32% jump in peak performance. With adequate cooling it looks like 32% might be possible then… (3D Mark Wildlife Unlimited). Note GFXBench shows a larger peak performance difference of 45%.
So the 40% doesn’t necessarily include ProRes as the reason the GPU is faster. Given the fact that the MacBook Air doesn’t have active cooling I expect it to fall to about 16% better sustained performance but the MacBook Pro should be able to get closer to the 32-40% sustained performance improvement.
The Apple A15 SoC Performance Review: Faster & More Efficient
Of course hardware ProRes support is ONE of the reasons the job is faster, since it's a mixed workload that includes ProRes. Sure, since it's a mixed workload, so there are other reasons it could be faster too, but it would be foolish to think ProRes acceleration isn't a portion of that speedup.Again, A15 is a known quantity, vs A14 with the same number of GPU cores (4 vs 4) it gets 16% better sustained performance and about 13% better peak performance. If we take into account the extra GPU cores (5 in A15 vs 4 in A14 - same relative jump in core count as M2 10 core vs M1 8 core) then we get the same 16% better sustained (thermal limit in iPhone most likely here) and a much bigger 32% jump in peak performance. With adequate cooling it looks like 32% might be possible then… (3D Mark Wildlife Unlimited). Note GFXBench shows a larger peak performance difference of 45%.
So the 40% doesn’t necessarily include ProRes as the reason the GPU is faster. Given the fact that the MacBook Air doesn’t have active cooling I expect it to fall to about 16% better sustained performance but the MacBook Pro should be able to get closer to the 32-40% sustained performance improvement.
The Apple A15 SoC Performance Review: Faster & More Efficient
www.anandtech.com
And I never made any comment whatsoever about a GPU speedup. I was commenting on Apple's claim it's a 40% speedup in video editing.
IOW, you guys are making this way too complicated.
Ah, sorry, I got distracted by someone above talking about GPU performance not being as good as advertised and then went saw your 40% number and assumed you were talking about the GPU performance as well… sorry about that
There are two ProRes tests and it looks like one is ProRes 422 and the other is ProRes Raw, seems like 40% better in the former and up to 3x faster in the latter…. Nice and confusing
This is a really good breakdown of the performance increases using all the information that is available. I think for anything more than this, we will have to wait for the product to appear and get analyzed by someone like Anandtech:
As someone else above pointed out above with regards to many historical (pathetic) Intel yoy increases, the increases between M1 and M2 are far from negligible.
As someone else above pointed out above with regards to many historical (pathetic) Intel yoy increases, the increases between M1 and M2 are far from negligible.
Really? Even 35% iGPU performance improvement on top of 20fps is about 27fps in Shadow of the Tomb Raider at 1080p ultra so not enough to really matter while costing more than dGPU gaming laptops.
That game is horribly unoptimized. The jump in GPU performance from the M1 Air to the M1 Max is incredible significant. I don't know if we'll see that kind of jump with M2, but that's really what I'm paying attention to.Really? Even 35% iGPU performance improvement on top of 20fps is about 27fps in Shadow of the Tomb Raider at 1080p ultra so not enough to really matter while costing more than dGPU gaming laptops.
I would have expected better from Apple...but I'm learning how much they aren't really better than the other megacorps anymore, sadly.