I think it makes sense to design a midrange chip first. With a good baseline design, you can scale it up and down to a degree without too much trouble. Using a low-end chip as the baseline would likely require too many compromises in the high end.
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.
User predictions for Apple Prosumer Silicon (M1X, M2 etc.)
- Thread starter leman
- Start date
- Sort by reaction score
Well the A series chip always preceded the AX series chip so I don’t see any precedent for this. It’s just how chip making works. When you’re on a new process node smaller chips are more economical because of yields and it takes a bit of time to work out the kinks and make bigger chips price efficiently. And the iPhone is their money maker so they’re going to dedicate resources and manufacturing space to the smallest chips.
I'm a bit worried that these new models will feature less than 15h battery life due to being mostly built with big cores... Regarding GPU I'm very very curious with what they will come up with.
For pricing I'm betting the 1799€ for the base model 14", I'm hoping 16GB will come by default....
For pricing I'm betting the 1799€ for the base model 14", I'm hoping 16GB will come by default....
Which is precisely why it would make sense for Apple to have different families of chips that cater to different applications. You are again assuming that the upcoming pro chips will be based on A-series or a variation thereof (if we accept that M1 is essentially a variation of A14). But this is not the only possibility. Apple could be developing a completely separate series of chips that are designed for higher-powered Macs.
You are wrong. There are no “long ramp up cycles” that require them to start with the smallest chip. The physical design of the A14 and M1 is completely different anyway. These MBPs will have new cores which are not firestorm and ice storm.
No, that’s not how chip making works. At AMD we did sledgehammer before clawhammer. It’s entirely a marketing/business decision, not a requirement.
My prediction: Apple will announce a line of P1 SOCs: one balanced compute/graphics, one mostly compute, and one mostly graphics. Macbook Pro will be balanced, Mac Mini Pro will be offered in all three variants. On the macOS side, we'll see some form of NUMA support. Provides eGPU and modular Mac Pro in one neat system.
I know my prediction is pretty out there, but it would explain PCIe 4 being highlighted in the M1 release. TB4 could be the interconnect.
Also, when the M1 was announced, I couldn’t help but guesstimate the dimensions of the logic board and imagine how many could fit in a 1U or 2U rack.
As far as branding goes, I doubt they would be called Mac Mini Pro; I didn’t think about that much before posting my prediction. It’ll probably be Mac Pro or something new.
Who knows how probable my prediction is, but all the pieces are there, and Apple would be able to sell people more devices.
Also, when the M1 was announced, I couldn’t help but guesstimate the dimensions of the logic board and imagine how many could fit in a 1U or 2U rack.
As far as branding goes, I doubt they would be called Mac Mini Pro; I didn’t think about that much before posting my prediction. It’ll probably be Mac Pro or something new.
Who knows how probable my prediction is, but all the pieces are there, and Apple would be able to sell people more devices.
Last edited:
background. The Jade die variants.
https://www.macrumors.com/2021/05/18/bloomberg-mac-pro-32-high-performance-cores/
Prediction.
M1P on same fab process as M1.
Same basic P , E ( NPU and GPU ) cores as the M1. ( the "M2" is coming later in Fall for smaller mac).
Since the M1 with some substantive tweak,s a name like the M1P ( and perhaps later M1P2 M1P4 for the Jade 2C die and Jade 4C-de. ). Or M1Pro ( and M1Pro2 M1Pro4 those SoC packages will be much bigger so the name can be longer. Not going to run out of space.
) But these MBP 16" and juiced 14" models are just a "plain" Jade.[ Up in the air whether " Jade 2C-die" is 2*(C-die) or (2C)-die ]
The major differences are in the uncore :
1. Larger System Level cache. ( that scales. perhaps 32MB , 64MB (2C) , 128MB (4C) )
2. Some either intra-die modularization (zone) or interdie communication. This means some tweaks to the memory hierarchy implementation to be able to take multiple hops to get to all of shared memory and coherence bus snooping across the zones. In short slightly more NUMA (and latency) tolerant than the M1 ( and A14 ).
[ No major shifts in core pipeline lengths or widths but sometimes will get more stalls. And better clock propagation management. the M1 has some abilities in these areas; just some evolutionary additions. ]
In short, the shared "bisection bandwidth" internal communication network to send data between major units is beefed up and tweaked for higher scale factors.
3. . As mentioned in the rumor if scaling these "C block" can probably cut the E core buidling block to 2 core because in many instances will be 2x - 4x the block and climb back up to a decent sized number.
4. A incrementally more clever hardware scheduler . Again needs to be able to adjust to the scale of "C blocks" and/or be able to delegate the work out to a local zone. Also noted in #3 above, the scheduling is a bit different with two less E in the baseline set up.
5. Clone and expand the Thunderbolt subsystem and display controllers. ( not much just one more TB port and display)
6. More LPDDR4 controllers. (e.g. 4 packages (MBP 16" / 14" ) , 8 packages , 16 packages ). Similiar to intra/inter die communication and closer/farther memory contolllers.
7. The Jade-Chop is the MBP 14". ( and probably eventually heading to upscale, higher priced Mini )
Incrementally smaller package for incrementally smaller board space. ( and product segmentation. )
8. The fixed function multi-media en/decoding stays the same for the baseline ( most of new transitor budget thrown at P and G cores. ). However, the 2C , 4c variants will pick up ability to en/decode more concurrent streams.
9. SSD controller aspects the same as M1.
It is mainly a 'M1X' only the old A--X augments didn't have to change the uncore scaling implementation. The "non X" versions just didn't use as much of the internal comms that was there in both units. Here the overhead of the internal comm system and uncore augments is high enough that Apple probably doesn't want to put it into the M1 which is die space constrained. ( has to fit into iPad Pro SoC allocated space. )
"X" just kind of meant " larger" ( e.g. X-large ). ~120mm2 versus the ~80mm2 . Here the larger is also precipitating a different internal network for the "Pro line" ( although probably get used in some non "Pro" systems like Mini. ) That said the package probably is much larger. ( 2x 120mm2 --> 240mm2 or more) . If "X" simply just mean larger package then still would be applicable but the uncore differences probably deserves a different suffix. And Apple loves slapping "Pro" on stuff so can crank up the price. ( the SoC costs more to make but Apple's user price is going to be higher than that). Changing the prefix because the uncore changed would be odd.
Better designed for Mac cores are probably coming but that would be better reflected in the "digit". M2 , M3 , M4.
Each of those get a different "Uncore" agument bolted on to go bigger. M2P , M3P , etc. The M2 would be the "5 plus"nm or 4nm with tweaks heading for A15 with the update limited scale internal comms and uncore.
One contributing reason the TouchBar used the T1/T2 was the limitation on display controllers on the Intel GPUs. ( Intel fixed that with Gen 11 (tiger lake) right as Apple jumped onto the "off ramp". ). The M1 backslid on display controllers. M1P probably is going to just get back to what the Intel system were. Hence , a contributing factor to drop the touch bar ( plus all the "hate" and added cost that they probably want to throw into the main screen cost).
Putting this baseline that scales to 2C/4C allows Apple use the MBP 14" and MBP 16" volume to pay down the baseline design updates. So getting a design that scales but isn't scaled in these to better cover costs. It won't be as solely on battery optimized ( tossing some E cores) , but still solidly aimed at laptops. ( No big power surge from M1 baseline basics. Just less lowest power options to optionally choose. ) It is just a bigger die with more stuff (at the same clock rates). So generally the power consumption will grow somewhat linearly with the "more stuff". ( and if not using the more stuff will get M1 like consumption which unused stuff (function unit(s) ) is put to sleep. )
P.P.S. It will be more rectangular than squarish. The edge the M1's memory is on will get bigger rather than boosting area on all four sides equally. (some growth on other but a disproportional. ) Bigger billboard .... bigger name.
Last edited:
with 32 cores gpu, its clear apple is targeting the power of the mobile Rtx 3070
Remember, current 5500M+cpu 83W and that 3070 consume 93W+cpu, while this new M1x the gpu cores will consume around 40W being on par with 3070 and almost double the performance of the 5500M
So right here we can see the M1 and M2 M3 etc will remain for the ipads/ and ultrabooks/fanless devices, while these will go for larger devices, with active cooling. I wonder if the next 24" imac will get this M1x (since it has 2 fans and heat pipe) or will take the M2 from the next macbook air and keep the same trajectory
Remember, current 5500M+cpu 83W and that 3070 consume 93W+cpu, while this new M1x the gpu cores will consume around 40W being on par with 3070 and almost double the performance of the 5500M
So right here we can see the M1 and M2 M3 etc will remain for the ipads/ and ultrabooks/fanless devices, while these will go for larger devices, with active cooling. I wonder if the next 24" imac will get this M1x (since it has 2 fans and heat pipe) or will take the M2 from the next macbook air and keep the same trajectory
Attachments
Last edited:
I don’t think Apple gives a rats ass about mobile RTX 3070.
It would seem that such a GPU would, very broadly, fall in the general span of RTX3060 or so performance. But it’s difficult to see that there is any overlap between products, programs and customers for the two, particularly as booting into Windows is a thing of the past.
Apple has very little reason to care about the Windows gaming laptop market. None, really.
(Your wording was a bit ambigous as to what you meant. But while Apple vs. PC comparisons are great forum war fodder, with Apple going their own way in terms of hardware architecture the markets are more distinct than they were. The number of people who consider buying either a MacBook Pro or a Windows gaming laptop has to be absolutely miniscule.)
I don't know who John Oliver is but a quick Google revealed that he produces entertainment for the English speaking masses. I, by and large, don't.
My point was that Apple doesn't produce products in reaction to, or in direct competition with, the Windows PC market, particularly not in the gaming segment. For better or worse depending on your priorities, they run their own race now. For a short transitionary phase, they would like to show clear benefits over where they were (and that gets easier the older the replaced models are).
While comparisons are interesting if you have a technical interest and a foot in either camp, for these products its very academic. I doubt Apple is doing their product planning on the basis of forum wars.
Apple actually has to compete with the Windows PC market, but they traditionally focused on a few segments such as Education, and the creative pro. markets for example. With Apple Silicon Macs, Apple likely is preparing the macOS platform as another viable market for games developers. They are likely playing the long game.
With iOS, iPadOS and macOS having so much synergy with Apple Silicon, it'll likely create a very attractive market for games developer to target. The M1 Macs are already capable to run many games, which the Intel base model Macs cannot do. This alone will very quickly create a market that will be tantalising for developers. If the money is there, games developers will come, regardless of how difficult it is to code for the platform. And from what I can see, Apple has made it easier over the years for developers to code for Apple's platforms. Who knows, maybe Apple will release a games engine next week ?
I guess only time will tell.
Yup, time will tell. Their M1 products have been well received, and since Boot Camp is no longer an option, it seems a given that the money on the MacOS table (for game publishers) has increased.
As to who will eventually pick that up though, it seems Apple is focussing on tempting iOS publishers to broaden their market. Think Genshin Impact for a good example. But I would assume that strategy games and RPGs are genres that PC publishers would first start to consider. Baldurs Gate 3 will be a litmus test.
On the other hand, if gaming is the major point of the device, the breadth of offerings on MacOS is nowhere near either Windows or Switch to take examples on opposite ends of the hardware power spectrum. People like me with a history and a game library are going to have to maintain a PC to access it. Buying a Mac for that purpose is a non-starter. On the third hand, a lot of people enjoy gaming a bit as recreation, and not being able to serve that purpose would be an unecessary weakness on Apples part.
It will be interesting to see how this plays out, but it will take time. If Apple produces iMacs and Minis with reasonable graphics oomph, it could be an indication that Apple at least wants to give a window of opportunity.
Honestly the port discussion doesn't matter much to me since I migrated to USB-C back in 2016, and just recently upgraded my Thunderbolt dock.
However I'm very interested in how they will approach on the RAM front and the amount of HP cores and Efficiency cores.
Again I'm betting 1799€ for the base model 14", just hoping it comes with 16GB
However I'm very interested in how they will approach on the RAM front and the amount of HP cores and Efficiency cores.
Again I'm betting 1799€ for the base model 14", just hoping it comes with 16GB
They'll probably match the existing RAM and storage amounts in whatever they replace (except that $699 Mac mini). So I assume it'll be the same 16GB starting as the 13".
OK maybe tone it down a notch. We’re not all driving Mazaratis in Dubai. Also, if you’re on a mission to be pnwnd, then look to your left.
Nah. Never cared much about campers sniping from the sidelines. They never affect the flow of the game much. ?
They’re not completely separate. Yeah they won’t appear in iOS devices but it’s the same CPU and GPU cores.
We won't know it until they are released. My bet is still that the cores we will see next week are going to be different than what Apple is/will use for the phones and the entry-level Macs.
There’s no amount money I wouldn’t bet that this isn’t the case. New microarchitecture for higher end Mac chips? There is zero evidence for this. If there were a new microarchitecture about to be released we would know by now and have codenames.
Zero. Chance.
I’m talking about die sizes when using new process nodes. New nodes tend to start off with smaller chips. Just look at AMD waiting years to put their new very large GPU dies on 7nm when they released smaller ones before that. It’s more economical to wait for yields to improve.
That’s not how chip making works? Thats just not true. Intel always releases a new microarchitecture and process node with something like their smaller Y-series laptop chips before their huge Xeon chips. That’s just how it works.
Yes… theoretically they could debut a new microarchitecture on the Mac before the iPhone... doesn’t mean they will. iPhone is the priority and (in my opinion) is always going to get the latest node and microarch just like it always has. Unless some reliable reporting disputes this I will assume it will remain to be the case instead of baselessly speculating otherwise.
That’s a lot of money invested for no real reason. The current cores are already world class.
But the reason might be that Apple wants more than a measly 10-15% PC market share. Why not diligently work hard to improve your stack and take it? They can give customers more reasons to choose a Mac.
True. But of course there is no evidence for anything else. All there is is just rumors. Given @cmaier ‘s background I tend to assume he‘s in possession of certain information… which in turn makes me assume he might be correct