I can type faster than I can make sense talking. 😂
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.
M4+ Chip Generation - Speculation Megathread [MERGED]
- Thread starter GoetzPhil
- Start date
- Sort by reaction score
That would be great. I do hope you’re right. I’ll be awfully tempted if you are. (My wallet hopes you’re wrong, though.)
Less than 10% increase? After all this time I would find that a bit disappointing
Me too, but there is the unusual density issue for N3/N3B versus N3E. The first generation is denser than the second generation. Others can correct me if this doesn’t matter, I don’t know, but I kind of expect it to follow M3.
It’s maybe a hopeful sign that A18 Pro is physically larger than A18, and I’d be curious to know how they each compare to A17 Pro. M4 Max could be physically larger than M3 Max, thereby mitigating the problem.
Last edited:
I agree, it would be, but I can’t help but wonder if ‘regular’ GPU core count (and performance?) will not be Apple’s priority for the higher-spec M4s? The low-end M4 that’s in the new iPads has the same number of GPU cores as the base M3, I believe, but a significantly faster neural engine — it’s understandable that that, rather than ‘regular’ GPU performance, should be Apple’s focus at the moment.
Of course, it’s perfectly possible that the non-binned M4 Max and Ultra will have more GPU cores, as @tenthousandthings suggested (and, indeed, the non-binned M3 Max has 40 GPU cores, vs. 38 in the M2, although the binned Maxes both have 30, I think). I guess we’ll find out in a few weeks’ time …. 😬
Last edited:
I remember reading it’s cheaper for N3E to manufacture, so making a bigger die may not be more costly.
"It's complicated."
N3E allows you to choose density vs. performance per transistor (though in practice you're not going to do that). But Apple is definitely using different transistors in different parts of the chip, which makes it hard to talk with any confidence about size and cost tradeoffs. AFAIK they're using denser transistors (2-1?) for the GPU cores, which means they may not be any larger than the GPUs on N3B. Though come to think of it, a careful examination of the A18 die shots, comparing them to A17, might provide an answer (I'm not doing that!). That of course assumes GPUs on the A18 are the same as on the M4s shipping this month; we don't even know that, really, though it seems extremely likely.
Also, don't forget that the individual GPU cores are better than previous generation cores; you're going to pick up plenty of performance from that alone.
Does anyone have any educated guesses what the CPU/GPU core count and RAM options are likely to be for M4 Pro and M4 Max?
I am trying to determine whether an M4 Pro (possibly in a new Mac Mini) is likely to be sufficient for "intermediate-level" 4K video editing in Davinci Resolve or whether the wise thing to do would be to wait for an M4 Max Studio (or pick up a heavily discounted M2 Max Studio once the new ones are released).
I currently have an MBP with M1 Max and 24-core GPU, but would like to keep this as a mobile device and get a desktop that is equivalent or better.
The Mini M2 Pro is fair way behind the M1 Max in raw GPU power and video optimizations - at least from some benchmarks and tests that I have seen.
I'd be interested to see whether the M4 Pro can match the M1 Max in this area. I understand that GPU improvements between M3 and M4 are not likely to be very impressive - maybe 10% - so it's possible that even a 24-core GPU M1 Max would still beat an M4 Pro for some workloads. The M4 Pro would obviously be faster for non-GPU intensive tasks.
What do you think?
I am trying to determine whether an M4 Pro (possibly in a new Mac Mini) is likely to be sufficient for "intermediate-level" 4K video editing in Davinci Resolve or whether the wise thing to do would be to wait for an M4 Max Studio (or pick up a heavily discounted M2 Max Studio once the new ones are released).
I currently have an MBP with M1 Max and 24-core GPU, but would like to keep this as a mobile device and get a desktop that is equivalent or better.
The Mini M2 Pro is fair way behind the M1 Max in raw GPU power and video optimizations - at least from some benchmarks and tests that I have seen.
I'd be interested to see whether the M4 Pro can match the M1 Max in this area. I understand that GPU improvements between M3 and M4 are not likely to be very impressive - maybe 10% - so it's possible that even a 24-core GPU M1 Max would still beat an M4 Pro for some workloads. The M4 Pro would obviously be faster for non-GPU intensive tasks.
What do you think?
Good question. I can offer some thoughts from the GPU perspective (which is, I presume, quite important for Davinci Resolve).
Short answer: I think there’s a fair chance that any M4 Pro / Max performance improvements will be quite modest (and the core count may be the same).
Longer answer:
Much as I’d like to see the GPU core count in the M4 Max increase to 64 (from 60), I wouldn’t bet a huge amount on it. (So it would be great to be pleasantly surprised!) My reason for saying that is that I suspect Apple’s focus will be on AI-related improvements — so, specifically, the neural engine.
As well as ‘regular’ GPU performance, I have also been pondering the option of an M2 Max (or even Ultra) Studio, either refurbished or discounted (when the M4 models are available), rather than an M4 Pro Mini. Hence my interest in your question.
Consider the following benchmark numbers. The big caveat is that, as with all benchmarks, this should only be used as a very rough guide (I probably look at benchmarks more than I should, but I recognise that they are no substitute for measuring one’s own workload).
Geekbench6 GPU — Metal:
M2 Ultra 76 cores 221382
M3 Max 40 cores 154739
M2 Max 38 cores 144810
M1 Max 114503 (number of cores??)
M3 Pro 18 cores 78559
M2 Pro 19 cores 82169
PugetBanch for Davinci Resolve (highest figures):
M3 Max 10169
M2 Ultra 11475
M3 Pro 6459
M2 Pro 5780
Some observations:
1. In terms of Metal performance in general (as measured by GB6), the M2 and M3 are pretty close, but a significant step up from the M1.
2. The PugetBench numbers suggest a bit more of gain from M2 to M3 Pro — perhaps does to some of the optimizations? — although the M2 Ultra figure looks to be a bit of an anomaly.
3. Even if the M4 Pro yields some GPU performance gain (possibly through further optimizations rather than core count), it seems to me unlikely that it will come close to the M2 Max — certainly not for Metal in general, and quite possibly not for Davinci either.
So a nicely discounted M2 Max Studio could be a good option. I’d be less sure about an M1 Max, though.
But please NB. that this is all rather speculative. Please don’t take any of this as gospel! At the very least, I would wait to see what emerges with the M4 Pro Mini.
Last edited:
Thanks for the detailed reply! For reference my 24-core GPU M1 Max scores about 94000 in Geekbench Metal, so it's possible that an M4 Pro with 20 cores might get close to this. That said they could well be optimizations in the M4 that are more useful than the raw Metal benchmark, so it would need some application-specific testing to validate.
The M1 Max PugetBench scores are quite close to the M3 Pro with 18-core GPU (assuming the M1 Max scores are for the 32-core GPU?) but I doubt an M4 Pro would beat an M2 Max ( maybe 7000 vs 8500?), so a discounted M2 Max Studio could be a better deal than an upgraded M4 Pro Mini.
I think it possible that the new M4 Mini may initially only be released in the base M4 version, with an M4 Pro version coming later, to prevent cannibalisation of MacBook Pro M4 Pro sales. So it looks like we might have to base expected M4 Pro Mini and M4 Max Studio performance on the MacBook Pro performance results, which will no doubt appear shortly after release.
FWIW, I’ve convinced myself the M4 Pro will be used for an entry-level Studio, and the new Mini will be M4 only, like the iMac.
Nah. Mac mini M4 (3 USB-C) and M4 Pro (5 USB-C), and no M4 Pro in Mac Studio.
That doesn’t make sense. The Mac mini and MacBook Pro don’t really directly compete.
That's an interesting (but alarming) thought. I wouldn't be surprised in the Mini is initially launched with only the base M4 though, with an M4 Pro version coming later.
While I agree that the MBP and Mini don't *directly* compete, there are plenty of people who use their machines with external monitors and keyboards and who would choose an equivalent Mini or Studio over paying for a MBP with a screen they don't really need.
I think this is one of the reasons why the M<x> Max Studio is not released at the same time as the M<x> Max MacBook Pros even though the SoC is available. Apple wants to maximise the sales of the more expensive MBPs before adding a much cheaper Studio version with the same performance into the market.
If the base M4 Mini gets a spec bump to 16GB/512GB, do you think that the base M4 Mini Pro could go to 24GB/512GB or even (wishful thinking!) 32GB/1TB?
Leaving the M4 Pro at 18GB/512GB would be underwhelming to say the least....unless of course the rumors about the increase to a minimum of 16GB RAM across the board are wrong.
I'm in the middle ground where a base M4 Max Studio is almost certainly more than I need, but a entry level M4 Pro with the same specs as the current M3 Pro might be a bit marginal (Davinci Resolve, but with a focus on audio post-production). Upgrading the Mini Pro RAM/SoC/SSD seems like a poor value proposition that is intentionally pushing buyers towards the base Max Studio.
Having an M4 Pro Mini that sits in the middle (e.g. 24GB/512GB or 1TB) would be the sweet spot, but knowing Apple, they will not make life that simple...<sigh>...
Leaving the M4 Pro at 18GB/512GB would be underwhelming to say the least....unless of course the rumors about the increase to a minimum of 16GB RAM across the board are wrong.
I'm in the middle ground where a base M4 Max Studio is almost certainly more than I need, but a entry level M4 Pro with the same specs as the current M3 Pro might be a bit marginal (Davinci Resolve, but with a focus on audio post-production). Upgrading the Mini Pro RAM/SoC/SSD seems like a poor value proposition that is intentionally pushing buyers towards the base Max Studio.
Having an M4 Pro Mini that sits in the middle (e.g. 24GB/512GB or 1TB) would be the sweet spot, but knowing Apple, they will not make life that simple...<sigh>...
I think that is deliberate. Pushing consumers from an upgraded Mac Mini pro to a binned Mac Studio is Apple’s intention. It is what Apple did with the M2 and there is no reason for Apple not to do the same thing with the M4.
Yes indeed. Apple is a master of the “up-sell”.
Do you think the M4 Pro will get at least a RAM upgrade to 24 or 32GB to differentiate it from the rumoured 16GB base M4 Mini?
I'm setting my expectations low, but that said, it would be a bit underwhelming to get a base M4 with 16GB and an M4 Pro with 18GB (same as M3 Pro)....I guess we'll see what Apple decides.
I hope that Mac will finally get the same treatment as iPhone and iPad already has — I mean: all their offerings will be presented at once this autumn — including more capable, Max and Ultra, units. From flagships to dinghies — and we’ll figure out which fits our need (and when to expect their arrival).
As Apple silicon upgrading schedule is tightening up and becoming more predictable, the current cycle — when Max/Ultra units appears at the market just few minutes before next M gen — is marginalizing further flagship’s sales.
As Apple silicon upgrading schedule is tightening up and becoming more predictable, the current cycle — when Max/Ultra units appears at the market just few minutes before next M gen — is marginalizing further flagship’s sales.
Last edited:
24GB is a reasonable guess for all M4 Pro products. That’s an 8GB jump from all M2 Pro products, just like 16GB is an 8GB jump from all M2 products.
The sell (as opposed upsell) here is Apple Intelligence support without compromising overall performance, to make that effective the jump has to be across the board.
Last edited:
M4 is the first generation where it could happen. There have been extenuating circumstances for each of the three prior generations. This time, it’s really just about the Ultra. I think everything else will launch at the same time (if they launch at all, I’m a believer that either the Air or the iMac will go straight to M5 in 1H 2025).
I don’t see any good reason why they can’t announce the Ultra when they launch everything else. Sure, the Ultra needs more time in the oven. But it’s not fundamentally different. I completely agree that setting it apart is a mistake.
I don’t think the Ultra is the cutting edge. The innovations begin in the A-series. When Apple shifts away from its current approach, it will start there. Don’t confuse complexity (UltraFusion) with fundamental change.
If Apple Could launch all of their products on one day each year - I suspect the lack of fan fare during the rest of the year would reduce the buzz around their brand. It might even reduce sales from customers balk at the price for purchasing two or more premium items in one go - especially where they already feel stretched buying just one.
It will interesting to see just how capable Apple will let the Mac Mini be; As powerful as an M4 iPad Pro … or closer to entry level Mac Studio M1? My guess is the former - or it could too heavily cannibalise sales of the Studio when it does eventually come out.
A upgrade to the M4pro base RAM seems likely if the M4 base is 16GB. But it is not because of AI. Apple will want to maintain the logical up-sell to an M4Pro as an attractive alternative to base M4 with a RAM upgrade.