Yes if we are guessing, I would say single core like 1900 and multi like 14000-15000 for the M1x.
As for the GPU I expect 40-80k depending on core count.
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.
New cores (A15)
- Thread starter cmaier
- Start date
- Sort by reaction score
As for the GPU I expect 40-80k depending on core count.
My idea on this is very simple: Apple already has secured a commanding lead in energy efficiency (which will take Intel and AMD years to catch up), but I believe that they are going for the absolute performance crown as well. This means that the prosumer chips have to outclass Alder Lake and Zen 4. I am sure Apple designers will do everything they can to make sure that their medium/high-power silicon delivers appropriate performance.
I think possibly because the new chips have been ready for a while = since a while months before the A15… it would be great if they were based on 5nm+ of course.
M1 is already very close to a 1650 Max q, these new configurations are going to be much much faster. That’s fairly certain. Geekbench compute scores are heavily biased towards GPUs with dedicated RAM, probably an artifact of how they measure kernel running times.
I don’t believe we will see it this year, as 16 CPU cores are an overkill for the mobile Apple chassis, but I think you might be pleasantly surprised with the performance.
There was an M1x bench with 10+2 not sure if that was fake or what.
Nearly 15k in cinebench sure would be a nice improvement over the M1
CPU benchmark
Our processor / CPU comparison helps you to compare two CPUs. We use benchmark results from Cinebench R20, Cinebench R23 and Geekbench 5 as well as the FP32 raw performance (GFLOPS) of the iGPU.
If it is like that I am buying one
Most rumors point to 8/2, I would prefer 10/2 however
Not overestimating at all. Scroll back up to post 186 above. 800M divide by 4 would be a 100M/year run rate. dividie by 3 would be approxmiately a 267M/year run rate.
Getting people "hooked" is about an addiction ; not necessarily buying at every single opportunity. Apple pushing the update every 12 months helps drive a "fear of missing out" on the folks trying to go 3-4 years and gets some to break on 2-3 years. They are pulling folks back to 2.
Laptops (and desktops ) on longer cycles are going to be much harder to pull most folks back to 2 years. If the average update cycle is different then the rollout cycle probably should be different also. The other major factor is size of the installed base ( e.g. number of people coming to 'buy" phase in the 9x bigger Windows PC market is going to be much bigger and much more conducive to support a yearly update from 2 orders of magnitude more selliers. )
You can't project the CPU synthetic benches that easily because of the Big/Little core configuration. But GPU is more straightforward. The GPU cores in A15 appear to be about 26% faster on a per core basis than A14/M1. So take M1 Geekbench Metal, divide by 8 to get to one core and multiply that score by 1.26. That is your approximated M2 GPU single core. Multiply by 8, 16 or 32 to get the appropriate totals.
So....currently on Geekbench Metal M1 gets about 21168. Divide by 8 to get to one core and we have 2646. Multiply that by 1.26 and we have 3334 for one M2/A15 GPU core. 8 of them get 26672, 16 gets 53343 and 32 gets 106687. Obviously this is a SWAG type estimate but one can see the potential - in its 16 core guise it is in Radeon Vega territory and the 32 core iteration beats both Vega and Vega II and is pretty darned close to the Radeon RX 6800.
So....currently on Geekbench Metal M1 gets about 21168. Divide by 8 to get to one core and we have 2646. Multiply that by 1.26 and we have 3334 for one M2/A15 GPU core. 8 of them get 26672, 16 gets 53343 and 32 gets 106687. Obviously this is a SWAG type estimate but one can see the potential - in its 16 core guise it is in Radeon Vega territory and the 32 core iteration beats both Vega and Vega II and is pretty darned close to the Radeon RX 6800.
Wow, even a bit below that is a compelling upgrade! Fingers crossed, because this "old" iMac is getting boringThere was an M1x bench with 10+2 not sure if that was fake or what.
Nearly 15k in cinebench sure would be a nice improvement over the M1
CPU benchmark
Our processor / CPU comparison helps you to compare two CPUs. We use benchmark results from Cinebench R20, Cinebench R23 and Geekbench 5 as well as the FP32 raw performance (GFLOPS) of the iGPU.www.cpu-monkey.com
If it is like that I am buying one
Most rumors point to 8/2, I would prefer 10/2 however
Iirc that was a “score” estimated from M1 scores, and not an actual test of a unit.Wow, even a bit below that is a compelling upgrade! Fingers crossed, because this "old" iMac is getting boring
Much lower volume but it is a much bigger chip. If the "MnX " die is about 220mm2 and the A15 is about 98mm2 then Apple could get twice as many A15 usable dies out of a single N5P wafer than they can for "MnX". If for some products they need two (or four) "MnX" twice as big dies that multiple goes up.
The other presumption here is that N5P wafer starts are available at the same rate as N5 ones. If those are two distinct production lines ( due to the 'recipe mix' differences setting on the main fabrication machines ) then the primary iPhone product line getting off of A14 ( and the N5 line) opens up more wafers than trying to compete with the more limited N5P production line.
if the differences lie entirely in the masks used (and design rules to compose the masks ) with exactly the same fabrication chemistry mix then the production is more fungible. If can just slide in a "better" mask than it is easier to share the same set of machines. Still have the issue that the bigger dies consume more though. So not likely going to be "at the same time". The more overlap with the A15 in production resources the more likely the "MnX" will slide out of the initial demand bubble for the A15. If Apple can sell A15's not quite as fast as they can sell them a several percentage points of increased production makes a substantive difference.
And the estimated TDP was 35W. If the 16MBPs receive a binned and higher clocked version at 45W, we'd see a further uplift, maybe exceeding the 11900k.
The M1 has made processors exciting again.
Very true, but please consider that Apple managed to produce A14 and M1 at the same time on the same process, and M1 is likely a much higher volume chip than the upcoming prosumer one. Of course, it doesn’t invalidate your points. I am just saying that it’s not something one can dismiss straight out.
They are exciting and I'm super thrilled so, I agree!
And it is also smaller. The ratio for the M1 versis A14 is about 36% ( approx 120/88 = 1.36 ) . The A15 is probably bit bigger ~98mm2 . If the MnX is around 220mm2 that ratio is 2.24 . That is substantively different cohort.
The volume doesn't as much as the wafer consumption. Volume contributes to that but so does size ( and defect rate ).
The other big problem was supply chain was not as out of kilter last year. That was a possibility at the outset ( something might go wrong with N5P that they may have accounted for and picked the safer option. )
Misdirection. You are the one who dismissed the any alternatives. " Why would it not". That is actually dismissal.
Huch? I was saying that I see it as possible and plausible. You have to consider the question I was replying to: „is it possible that …“ - „why wouldn’t it be?“. I never intended to claim that it’s the only possibility.
I agree with deconstruct60 that there might be other factors determining the architecture and node of MnX. The MnX chip is likely going to feed MBP, larger iMacs and perhaps a mini Pro. Taken together, I would not be surprised if these segments constitute a very large share of Macs and as deconstruct60 also point out, the chips are larger. So production capacity needed might be equal to the M1/A14 capacity.
The fab line could however be retooled. I was pleasantly surprised that the iPad mini got the A15 at the same time as the iPhone got it. So they went away from "iPhone first" policy that has hampered the iPad line for a decade. Marketing wise I think A15 based cores for the new MBP will make good sense and I really hope Apple upgrade their entire chip lineup every year to simplify it for the end user.
MBP and higher end iMac are less concerned about power draw* so I guess a 5 nm/A14 process and sufficient number of CPU/GPU cores will work as a charm.
*If you are using a MBP at 100% CPU/GPU utilisation, any battery empties in less than 2h, at lest if the machine is >2 years old. This will be better with ASi but you still need mains for high performance work.
We can speculate for ages until it is released.
I feel the multi-SoC ASi (Jade 2C / 4C) offerings will be reserved for the Mac Pro, Mac Pro Cube, and possibly the iMac Pro...
iMac Pro for those wanting the top-end all-in-one workstation, Mac Pro Cube for those who don't really need PCI slots & don't want the attached monitor of an all-in-one, full-size Mac Pro for those who do need PCI slots (but still don't want the attached monitor of an all-in-one)...?!?
Your ideas of Mac Pro(s) aside, the iMac Pro definitely deserves to be on that list. The last one had a max TDP of 370 W— compare that with the M1 @ 39 W. Lots of headroom to make an all-in-one system that has more computational ability than almost* (* there are a few very expensive configurations that could match or exceed it) anything that could be assembled off the shelf. Those two lines are ripe for chiplets.
The MBP and Mac Mini Pro can easily use a single SoC. I’d expect something competitive in their market.
Last edited: