and what about gpu performance between that i9 12900HK and M2? (we expect the M2 gpu to be around 30-40% over M1)
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.
Intel Alder Lake vs. Apple M1
- Thread starter leman
- Start date
- Sort by reaction score
Love it, comparing an i9 to M1. Shows just how strong the M1s are. Real work day to day it proves it more and more from my perspective.
They were using a 2016 mbp keyboard.
A15 might be only around around 10% faster than A14, but it also consumes less energy. Its focus is on efficiency, which makes a lot of sense for an already fastest phone chip on the market. I am sure that this bodes well for M2 performance and expect at least 15-20% improvement in single-core.
EDIT: It's also harder to measure (some benchmarks and most apps run significantly faster in some platforms than in other, equally-capable platforms, i.e. Cinebench). Also, we haven't seen the A15 GPU in a non-thermally constrained device (it gets massively throttled on the iPhone 13s).
Sure. My assumption was that Apple may want the M2 to be efficiency-focused too. If they don't and they can translate all the power savings back into performance, yeah, 15-20% faster than M1, which would mean 1978-2064 points in Geekbench single core.
Last edited:
Whoops. It does. Totally forgot about it, since most of the benchmarked laptops pair the 12900HK with a dedicated Nvidia GPU.
I didn't think Intel sold a mobile part without a integrated GPU these days. Not to be confused with their desktop line where you can get a CPU without a iGPU.
Yep, I thought as much, but I didn't remember a single mention to the Xe graphics on the AnandTech review of the 12900HK so I shruged it off. Apparently they just didn't benchmark the integrated GPU.
The M2 is still going to be in an iPad Pro. They can't throw efficiency out the window when the SoC serves duty in approximately same kinds of systems as the A15 does. ( similar ultra thin chassis (i.e., limited battery and thermals) , relatively large screen consuming battery, relatively tight main logic board physical dimensions. ,etc. )
It isn't a "Mac only" SoC. It has all the classic physical size constraints the iPad Pro SoCs have had all along.
They'll take the LPDDR5 'win' and probably as much bump as the A15 did.
Can't they run the M2 at lower frequency for the iPad Pro? (like they do with the A15 on iPhone 13 vs iPad mini 6)?
After all, I'm still wondering if there's any reason iPadOS needs a faster chip than the A12x.
At the same clock? You are quite optimistic I must say
It already gets 10-15% improvement (depending on benchmark) between A15 and A14. Improvements in caching and the like could easily add a few more percent. And the clock will likely be higher than M1 as well.
At the same clock? You are quite optimistic I must say
Why at the same clock? M2 should be higher clocked than M1
My personal, completely arbitrary and unsubstantiated opinion is 3.5 ghz
That wouldn’t surprise me (at least for max clock). The question is whether they did any backend work on the core between A15 and M2. In a typical “spin,” we would aim for about 10% speed improvement. We would sometimes get more, and sometimes less - it’s a lot of work. Of course that would be *on top* of whatever they left on the table by downclocking for phones - that, alone, could be 5-10%.
According to Anandtech, A15 Avalanche is clocked 8-10% then A14 Firestorm. So it makes sense to expect the same or more from M2 vs M1. https://www.anandtech.com/show/16983/the-apple-a15-soc-performance-review-faster-more-efficient
That is the reason i was asking, because i consider a 20% IPC improvement unrealistic. So i understand, you assume a 10% clock increase.
It seems this is coming :
Mac Studio (as equipped US$5k)
Mac Studio (as equipped US$5k)
- Dual M1 Max SoCs
- 20-core CPU (16P/4E)
- 64-core GPU
- 32-core Neural Engine
- 128GB LPDDR5 RAM
- 800GB/s UMA
- 1TB NVMe SSD
- Dual 10Gb Ethernet (RJ-45) ports
- WiFi 6 / Bluetooth 5.0
- (6) Thunderbolt 4/USB 4 (USB-C) ports
- (4) USB 3.1 Gen2 (USB-A) ports
- HDMI 2.0 port
- 3.5mm audio output jack (auto-switching high/low impedance)
Can't they run the M2 at lower frequency for the iPad Pro? (like they do with the A15 on iPhone 13 vs iPad mini 6)?
After all, I'm still wondering if there's any reason iPadOS needs a faster chip than the A12x.
For the iPad Pro it would be more like putting a low cap on the "turbo" mode than shifting the whole range down. Apple doesn't really push the underlying fab process past what it is efficient on away. So never really used the "single thread , top fuel drag racing" approach in the first place.
The iPad Pro's are suppose to go all mini-LED , hyper bright , ProMotion. ( if driving the display at 120Hz then not really slowing down).
As for iPadOS not needing much ... Apple has to compete with Windows 11 Tablet slab models with the iPad Pro since there is no touch screen for the Mac. So can drop it if want to sell fewer items.
If they haven't changed the CPU or GPU arch much and run them at the same speeds then it becomes "buy the new iPad Pro that runs just as fast as the old one". That isn't going to work so well. I suppose they can say "oh, new ProRes en/decoder" buy it but that is rather 'thin' as a value for an iPad Pro. Pretty good chance TSMC is charging more for N5P or N4P (or whatever additional delta Apple goes to for M2). So paying higher unit costs and delivering same performance? Doesn't sound like Apple.