One external display only. Sure, if you also count the internal display then it's two, but I meant external ones.
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.
M3 Chip Generation - Discussion Megathread
- Thread starter scottrichardson
- Start date
- Sort by reaction score
One external display only. Sure, if you also count the internal display then it's two, but I meant external ones.
Very interesting. I hadn't thought about implications of the A17 driving an external monitor - the iPads already do, but those are Mx iPads.
Why do you think this has any implication for the M3 though? While Apple will surely reuse various blocks from the A17, I don't know why you'd think the inclusion of external display support on the A17 says anything about the M3. Perhaps it's the other way around - they've taken the display subsystem from the M2, with its support for external displays, and grafted it into the A17?
Only smarter than the current sub-par architects/designers, not all the good ones that left Apple... ;^p
If only Apple had some money, or a product that was attention grabbing. I’m sure they'd be able to get some good chip designers.
You know, just being optimistic. Single-monitor support has been a recurrent point of criticism for the base M1/2, and given that Apple now managed to fit external display support onto A17 it could be possible that they will improve it on M3. Then again, there doesn't have to be any correlation.
I only mentioned it because I thought the M1/M2 Mini support 2 displays with the obvious caveat of them having no internal one. I can see how that would be pedantic.
I don’t think it’s a stretch to think Apple’s engineers attempted to make an architecture that could be used for both purposes. And downclocked the phone version.
Don't expect too much as long as it's SoC. M1,2 already proves that the performance gains, especially for GPU, is quite low as M2 Ultra is only RTX 3060 ti. They are developing a new chip with TSMC 3D fabric but that won't be available till 2025.
Enough speculation about "what is the M3?" - let's look at speculation about "when is the M3?"
From this video (timestamped)
Johny Srouji -
"... what we call surprise and delight. When we want to ship a product, we want to surprise customers - so we care about confidentiality in a big way... "
(also - lots of talk about packaging at the end...)
Apple wants to surprise us with the timing? That obviously means M3 Macs next Tuesday.
From this video (timestamped)
Johny Srouji -
"... what we call surprise and delight. When we want to ship a product, we want to surprise customers - so we care about confidentiality in a big way... "
(also - lots of talk about packaging at the end...)
Apple wants to surprise us with the timing? That obviously means M3 Macs next Tuesday.
Last edited:
Well, their newer patents do describe quad-chip arrangements, so who knows…
The lack of 3rd display support on current M1 / M2 is due to silicon area priorities, such that Apple determines this class of chips do not need to waste die space for a 3rd display buffer on die since it mostly goes into iPads and MBAs where you can guess close to 100% users not needing that many displays attached. So it is a silicon budget issue, a design choice, that is regardless of the basic architecture underneath.
As a result I don't see the A17 pro having direct DisplayPort support bearing much relevance in Macs. Conversely it opens up the possibilities of the A17 pro being used in devices that used to be on M1 / M2 but really don't need it, such as the next iPad Air.
As a result I don't see the A17 pro having direct DisplayPort support bearing much relevance in Macs. Conversely it opens up the possibilities of the A17 pro being used in devices that used to be on M1 / M2 but really don't need it, such as the next iPad Air.
Yeah, it's difficult to speculate the 'A/Si family' expanding in any other direction in the near term. It's interesting, if this was the intent, that they wanted it known just after WWDC.
This is a good comparison that is not talked about much and that is an M2 Ultra. The M2 Pro and Max would have to be in 20xx series or lower. While Apple has done a good job with its chips, the graphics are still nothing compared to the current generation of dedicated GPUs. They are very focused on the media side, and less on the gaming or 3D side.
Edit: I watched the YouTube video interview with Johny and towards the end he talked about gaming and getting AAA titles on Apple Silicon. He mentions that it is a priority for Apple, and they have the tools and hardware in place to help game makers move more to Apple. However, what would be a good view of this is from a game dev side. What are their thoughts on Apple Silicon and is it something viable to use?
Apple can create all the tools and hardware it wants but if the developers go a different route then it is not a good solution.
Last edited:
I don't think it is a good comparison because it depends on what software you're comparing. In terms of Blender for example the GPU compute capability (excluding OptiX) the M2 Ultra is pretty close to a 4080 not a 3060Ti. Apple is very focussed on the 3D rendering side and have really put a lot of effort into optimizations for that market. Apple seems to work more closely with Blender than they do with game companies for example. We'll have to see how M3 Ultra performs with its hardware RT acceleration but I don't doubt that it will pull Apple forward further.
It really, really, depends on optimization. Even before Apple switched to their own silicon exclusively for GPUs games ran better on Windows via boot camp than they did running natively in macOS. Unless/until Apple gets truly serious about ensuring games are just as well optimized for macOS as they are for Windows they are always going to perform worse. This however does not mean that the problem lies with Apple's GPU cores themselves.
I would also argue Apple isn't really aiming at the top flight discrete GPUs which consume more than twice as much power than a whole M2 Ultra does (never mind just the GPU part). I would like Apple to build a quad chip or move to a tile based chip packaging technology but as much as I might want this I don't know if the market for the top of the line discrete GPUs is big enough to get Apple to build something like that.
I don't want to go back to discrete GPUs and CPUs on separate packages because that will fragment optimization targets and introduce other inefficiencies (related to unified vs non-unified memory). If Apple wanted to they could build an absolutely massive tiled SoC with a monstrous GPU and CPU (still in the SoC package), they just don't seem to want to.
I think, and correct me if I am wrong, that Apple is focused on having an overall optimized solution in place that is not a power hungry GPU and a weak CPU. I was thinking about the idea of a separate GPU but your point is correct as it would split up the resources and then become a little bit of a mess if they had Gx, Gx Pro, Gx Max, Gx Ultra chips and then every combo of how those would work in a system.
What Apple showed off from the new A17 Pro series was RT, but it was at a lower frame rate (it was a mobile chip) but maybe this will be something that is part of M3 line
I agree that that appears to be their current approach. I think they could build a different type of tiled architecture though, they could split their chips into compute tiles, IO and SRAM tile, and GPU tiles and then mix and match the tiles to create SoCs with huge GPUs and small CPUs or huge CPUs and small GPUs. However I just don't think they have any interest in this kind of approach as the primary markets that need the extreme options are perceived by Apple as too small to bother with.
Well, M2 Ultra is a 30TFLOPS part, which makes it roughly equivalent to a desktop RTX 3080. Real-world performance will depend on what you do, as different GPUs have different weaknesses and strengths.
Apple has two problems when it comes to the GPU performance: die area and frequency. Since they focus on SoC architecture (and as @bcortens says, there are very good reasons for that), they are limited by the number of compute clusters they can fit in a system without going multi-chip route. They also run their GPUs at very low frequencies of just 1.3-1.4Ghz, even on systems that could handle the extra heat.
Apple could partially address the performance problem by running the GPU hotter if the chassis allows it. Will they go that route? We will see.
It's a difficult topic to discuss, since people are often limited by practical constraints. When it comes to the programming interfaces, I consider Apple's Metal to be the best gaming GPU API in the world. It's simple, powerful, and flexible. But many developers will probably choose to use DX12 instead simply because that's where the money is. And that's the core of the problem IMO.
Multi-chip technology also has disadvantages (e.g. power consumption). Apple needs to find a way to balance economic and logistic factors against the nature of the products they ship. E.g. it probably doesn't make much business sense for them to build a faster dedicated desktop chip, since most of their business is mobile. But they are exploring ways to build more performant chips. One patent (published some time ago) mentions stacking two dies — potentially using different node sizes — to implement different components. E.g. compute could go on a 3nm die and most of the uncore, including memory controllers and caches, could be implemented on a cheaper 5nm die. Could be a nice solution that would allow them to optimally use the higher density of the new nodes while managing its higher cost.
Yeah - I was thinking they could build multi-chip just for desktops and keep their monolithic approach for mobile but I don't think the desktop market is perceived to be big enough to warrant such an approach.
I suppose that they have settled for their UltraFusion symmetric MCM's as a compromise. That said, a 4x chip (Extreme) package — if accompanied with a clock increase, should go a long way towards addressing Apple's performance concerns on high end desktop. Not that an average user will have anything out of it of course...
for what it’s worth, I’ve seen users here that have developed games state that MacOS isn’t really even a secondary priority because of the low RoI.
Which makes sense since Apples apparent strategy for games is to make porting easier (a la GPTK).
In many office jobs, people connect cheap laptops to multiple cheap monitors. They don't need fancy computers, and even the MBA is high-end from their perspective. They just need to have several documents visible at once.
Tight integration means making compromises, and the same compromise never works for everyone. Multi-monitor support is a low-end requirement, but it's particularly expensive for Apple, due to their insistence on very high resolutions and minimizing power consumption.