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.
'M2 Max' Geekbench Scores Leak Online, Revealing Rumored Specs and Performance
- Thread starter MacRumors
- Start date
- Sort by reaction score
This is why you should never sit around and worry if you should buy the current version of the Max you want or wait for the next one. The performance delta is not that great. At least now that we don’t have to be dependent on intel. Ok that’s not truly correct. In the intel days there was almost zero performance difference between the two systems. That’s why Apple got rid of them.
The model identifier of the m2 max machine is Mac not MacBook so this is likely a desktop machine, and if it is then it’s worthy of being compared to desktop Raptor Lake.
That forum post you link to is from early this year. Things have moved on since then! Get the right settings in Handbrake and it's crazy fast.
Also, just because an app isn't optimised for Apple Silicon, that doesn't mean Apple Silicon has an issue.
Your comment about encoding quality is weird. You choose the quality to encode at. After that it's number crunching.
Blender is 3D modelling software for those into that but ray tracing was implemented on GPUs for PC gaming. And that's a really tiny minority in the sea of people who buy computers (albeit a very vocal one). And it's only on Nvidia anyway—a fraction of a fraction.
You still haven't really made a case for it on a pro machine. It's not a pro codec. It's a consumer codec intended to make streaming more efficient. Apple doesn't really have huge interest in this. It's for budget laptops coupled to weak broadband, and the like. Fair enough. But Apple's not positioned there.
M1 Max TDP: Around 34W
Core i9-12900K: Base TDP of 125 watts and a max turbo TDP of 241 watts.
Ryzen 7000 Series models feature a TDP of 170W and a peak power of 230W
Apple is light years ahead. No exaggeration.
Apple has no qualms about comparing the Studio to a Mac Pro 16 core with a W5700 at the Ultra's rollout out session. It doesn't cover the whole Mac Pro 2019 space. But also not really true that it doesn't cover any of it.
The gyrations to wrap this around a Mac Studio is a larger part of the missing effort on Apple's part than Geekbench or common benchmark scores.
https://www.sonnettech.com/product/xmac-studio/overview.html
As long as the card I/O fits inside of Thunderbolt 4 limits this is plausible work around. There are more PCI-e v4 (and up ) cards around in the 2023 market than there were back in 2021. The core root cause issue is relatively mindlessly reusing a laptop hyper optimized SoC die as a 'chiplet'.
That presumes that Apple "has to" use the M2 Max base die to build SoCs for the Mac Pro. They don't. If the M2 Max is some trailing edge N5P implementation then there is pretty good chance that it has been finished , sitting on the sidelines for a decent amount of time. Apple could skip doing any large SoC M2 . They don't have to do every SoC size option in every generation. Skipped M1. Could skip M2 if timing is all wrong.
If Apple did a M2 (N5P) Extreme that could provision 4-6 slots with two x16 PCI-e v4 backhaul , then a sizeable number of folks would buy it. 'Nobody' could find any productive use for those CPU 48 cores? Probably can find some folks that can use that as it doesn't exist now. Even if 'only' a 5% incresase the count (and RAM capacity) are higher. The 'value' add for the Mac Pro would be Apple wasn't using a die hyper optimized for a relatively thin laptop (where PCI-e aggregate bandwidth is thrown 'under to bus' to boost GPU Memory I/O bandwidth and lower power consumption. ). The M1 Max die is a not so good chiplet die. It is a bit too chunky to be a good chiplet that scale up well past two. Doesn't mean Apple has to throw out all of the "Max class" die characteritics. The general I/O parts just need a slightly different disaggregation.
Not necessarily. Apple's changed the model naming again. The 13" MBP M2 is just Mac14,7, and the M2 Air is Mac14,2.
Yeah, I think I just might resemble those incredibly sad and pitiful guys.
Personally, I'm holding out for 0nm. That will be a true breakthrough.
Not necessarily. M1 Max does up to 64GB, this would simply mean 24GB RAM modules instead of 16GB ones. The M2 Air already offers such configuration.
I was fearing the price of the 16-in MBP would rise from today's £2,599 to over £2,800 by the time the 2023 model was available (with inflation and exchange rate fluctuations). So this article has just encouraged me NOT to wait to replace my 2017 15-in Intel Macbook Pro, which cost £3,059, and I have just taken advantage of John Lewis' ongoing Black Friday deal to get the 2021 16-in at £2,349. Thank you MacRumours!
I'm talking about encoding on the CPU, i.e. not using the video toolbox (= hardware encoder).
This means that in Handbrake I mean encoding via the CPU.
No. But there are use cases where AppleSilicon obviously has disadvantages. x264 and x265 seem to be strongly optimized for x86_64.
There is nothing strange about it.
The Apple hardware encoder does not reach the quality of very good software encoding. At comparable quality, the file size becomes significantly larger.
There are also a few postings about this in this forum
There is also other software that uses RT.
It may be that RT is not particularly relevant for the typical Apple user today. The fact is, others are further along.
Why only Nvidia? AMD also supports RT. Intel also apparently.
How do you define what a pro codec is? The way Apple defines "Pro"?
"“Pro”, in Apple product marketing parlance, does mean professional sometimes (e.g. Mac Pro and iMac Pro). But sometimes it just means premium."
Going Pro
My prediction for new iPhone names next week: iPhone 11, 11 Pro, 11 Pro Max.
daringfireball.net
Many "consumers" are Apple customers. They want and should be able to use what is relevant for them. And AV1 will probably be relevant.
Or is "Pro" for you a codec that is gold-plated with patents and expensive usage rights?
If the big video portals increasingly use AV1, they must also work with AV1. Then things can start to move. And as a private user, AV1 can also be interesting.
I can't find the numbers right now. But as far as I remember, the M1 max is around 90 watts at full load (and at full load, i.e. CPU and GPU fully utilized at the same time, the MacBook Pros throttle).
The above are desktop CPUs with integrated graphics. You are not constantly on the go with a full load in everyday life.
I do not want to deny that Apple is the leader here.
Fair point I guess we’ll see. But the same M1 Max in MacBooks are in the Mac Studio.
If this is a "M2 Pro Mini" then the base system price probably won't be low. The current Intel Mini with 16GB started at $1,299 and 32GB starts at $1,699. $1,499 at 24GB starting point would be a 'jump' but not a huge one.
Apple probably isn't gong to 'give away' that higher RAM floor starting point for 'free'.
It would mean that the M2 Pro Mini wasn't backsliding from the old Intel Mini that max RAM capacity level of 64GB . A 2023 model topping RAM capacity of a 2018 infrastructure is more an "about time" than a "huge leap"
Yields don't play a huge factor in a pricing ladder structure for the SoCs. The base model chips are largely binned so that Apple can charge even higher prices (and fatter margins) for the non binned versions. ( there is also some incremental savings they get with a relatively few chips they do have the defects present in exactly the right spot to 'fuse off' a GPU/CPU core. but that isn't the primary driver. Market segmentation is primary driver. )
For the MBP 14 and 16" the M1 Pro has two variations and the M1 Max has two variations. The Max is much bigger (and there has lower yields. ). Its price is higher also. Hence balances off the variability.
If the RAM capacity is going up then probably not if still using what is a relatively trailing edge process for the SoC (N5P). N4 isn't going to bring a huge power 'win' either. The GPU core spend on consumption probably went up if the CPU cores didn't rise much. N3 and spent vast majority of improved process on smaller die and power savings, then could get something like that.
True, but look at the identifier. Mac Studio last year used Mac13,1 and Mac13,2 as identifiers. So I'd guess these are either laptops, some new desktop (Mac Mini??) or maybe even a Mac Pro...
People that needed a laptop for office/wfh stuff already bought the m1 one in 2020. Pro's that needed "power" already have the m1 pro's/max/ultra & studio stuf. Does Apple really thinks we replace our laptops on a yearly base, just like iPhones?
This is perfect sweet spot for me. My target is to double my cores, and double or more the Ram. I will just need to wait for the Studio version.
Geekbench scores allegedly for the upcoming "M2 Max" chip have surfaced online, offering a closer look at the performance levels and specific details of the forthcoming Apple silicon processor.
The Geekbench results, first spotted on Twitter, are for a Mac configuration of with the M2 Max chip, a 12-core CPU, and 96GB of memory. The Mac listed has an identifier "Mac14,6," which could be upcoming MacBook Pros or the next-generation Mac Studio. Apple offers a maximum of 64GB of memory on the current 14-inch and 16-inch MacBook Pros, while the Mac Studio can be configured with up to 128GB of memory with the M1 Ultra.
According to the test, the M2 Max chip scored 1,853 in single-core and 13,855 in multi-core. For comparison, the M1 Max chip in the Mac Studio scored 1,755 in single-core and 12,333 in multi-core. If the M2 Max chip results are accurate, the performance increase will be relatively minor for the upcoming chip.
The M2 chip, announced in June 2022, is based on an enhanced version of TSMC's 5nm process. What fabrication process the upcoming M2 Pro and M2 Max chips will have is not entirely clear. While they could feature the same enhanced 5nm process as the standard M2 chip, there are also rumors it could leapfrog to 3nm, offering significant performance and energy efficiency gains.
According to the results, the chip was tested on a Mac running macOS Ventura 13.2, which has yet to enter developer or public beta testing but is being worked on internally at Apple. The first Macs expected to feature the M2 Pro and M2 Max chips are rumored to be updated 14-inch and 16-inch MacBook Pros, with the Mac Studio to follow sometime next year. The updated MacBook Pros were initially rumored to launch in the October to November timeframe but are now expected to launch in early 2023.
Apple has several Macs in the work for 2023, including updated MacBook Pros, an updated iMac, and the long-rumored Apple silicon Mac Pro. For a complete rundown of all the new Macs we expect next year and beyond, be sure to check out our guide.
Article Link: 'M2 Max' Geekbench Scores Leak Online, Revealing Rumored Specs and Performance
I'm giving Apple one more chance to see what they can do with this Apple Silicon experiment, i.e., let's see what performance they offer when they move to 3nm. I expected more. But if this is really all Apple can do with Apple Silicon (a mere 6-9% increase in single core perf and roughly the same in multicore) then Intel and AMD x86_64 may end up being a better alternative.
Raptor Lake and Zen 4 are both already much faster on desktop, Meteor Lake and Arrow Lake will increase IPC even more, and future lakes like Lunar Lake and Panther Lake will be stronger still. Same with Zen 5. And on PC you get true modularitliy in terms of adding future PCIe5.0 storage, more RAM/faster RAM, new GPUs (the 4090 is a BEAST) etc. What's the point of spending thousands of dollars on a Mac when PC's are faster? Apple makes great mobile machines and maybe that will be their key market... but on the desktop side it's starting to look like they're going to be outclassed... and the thing is both Intel and AMD will be using more advanced nodes for their CPU designs... meaning they will gain performance per watt benefits...
Before Apple blamed Intel... with its 14nm debacle. But if both Intel/AMD pull ahead, perhaps significantly so, then there's no one else to blame. Even Snapdragon 8 Gen2 is catching up to Apple's latest A16 in terms of multicore performance. What's going on over there? Maybe they're waiting on TSMC's 3nm to become ready to launch a beast.
Raptor Lake and Zen 4 are both already much faster on desktop, Meteor Lake and Arrow Lake will increase IPC even more, and future lakes like Lunar Lake and Panther Lake will be stronger still. Same with Zen 5. And on PC you get true modularitliy in terms of adding future PCIe5.0 storage, more RAM/faster RAM, new GPUs (the 4090 is a BEAST) etc. What's the point of spending thousands of dollars on a Mac when PC's are faster? Apple makes great mobile machines and maybe that will be their key market... but on the desktop side it's starting to look like they're going to be outclassed... and the thing is both Intel and AMD will be using more advanced nodes for their CPU designs... meaning they will gain performance per watt benefits...
Before Apple blamed Intel... with its 14nm debacle. But if both Intel/AMD pull ahead, perhaps significantly so, then there's no one else to blame. Even Snapdragon 8 Gen2 is catching up to Apple's latest A16 in terms of multicore performance. What's going on over there? Maybe they're waiting on TSMC's 3nm to become ready to launch a beast.
Last edited:
I’m guessing based on the spec sheet and m1 vs. m2 those 2 extra cores are e cores not p cores. Which is a HUGE welcome change and will make more of a difference than the Geekbench scores show. I have an M2 air personally and an M1 Max for work, and while the max runs laps around the air in “pro” things like adobe and Xcode, the air actually has noticeably better performance at day to day and background tasks like webbrowsing and installing App Store updates etc because of the quad e cores. It was strange the pro line is only dual e core.
What's legacy about HDMI? It's still widely used on Monitors and TVs. 8K TVs use them. HDMi 2.1 is higher bandwidth than Thunderbolt 4.
Ahh I see you like the 2016 MBP design. Make sense those **** computers throttled like hell due to being so thin.
Oh and guess what HDMI 2.1 can output 8k 60hz or 4k 120Hz. Thunderbolt 4 cannot do that. So don't call HDMI a legacy port. You have no idea what your talking about.