apple’s *BEEN* making its own GPUs for 10+ years now and they’re part of the À-series silicon. They’re in every iDevice.
Also, they won’t be stock A-series chips (I.e., A14X), because the SoC has to also support Mac frameworks that are deprecated (e.g., OpenCL, OpenGL, etc.) for Rosetta 2 and for othe macOS apps in Big Sur. At wwdc they said it would appear as a Metal Family GPU , which is close BUT NOT IDENTICAL to an A-series Metal soC.
There won’t be external-to-the-SoC GPUs from Apple. There may bot even be RAM and Flash storage external to it.
That's been my hypothesis as how to best launch this architecture change, and that was years before Apple Silicon was ever announced publicly. I hope we get Si Macs sooner than the stated time frame, just like with the last transition, because I am not upgrading again until they do.
Conspicuous by its absence in the article and comments is any mention of a beefed up neural engine. Apple made an announcement last week building up their ability to perform ML tasks. Better Natural Language Processing would be an early goal.
Apple has already stated for the record that they are going to use their own GPU. Why is this written as speculation?
They have also stated that they are designing Mac specific SoCs. So no, it won’t be an A14X. Though it might use the same core design but the number of CPU and GPU cores are going to be Mac specific.
Because they haven't said that at least in the context of discreet GPU's. Some information was deciphered to suggest that that was the case but it can be seen that the information can be interpreted in more than one way. The discussion is a bit ridiculous anyways as every A series Apple SoC comes with a GPU, so all Macs using an A series chip will have an Apple GPU even if a discreet is included. So the question becomes does Apple use a discreet Apple GPU in addition to its integrated GPU in the initial Macs and frankly we can't say. There are still good reasons to stick with AMD at the high end, mainly because of performance, but even here Apple has options to mix AMD with Apple GPU's.
With respect to the A14X in a Mac Book revival; that is easy to understand and has no impact on the idea that Mac specific SoC's are coming. The very nature of Macs will require very different chips for the various classes of machines. Personally if Apple doesn't implement SMT they will need a 24 core laptop machine by the end of 2021 to remain competitive. On the Mac Pro they are screwed if they can't match what AMD is already offering which means offering at least 64 cores and 128 cores will be needed if they can't get SMT working. AMD has some of the most compelling workstation offerings on the market right now so a Mac Pro is going to really stretch Apples capabilities.
I would imagine that the MacBook would continue to use the SoC style GPU just like the iPad. It will be interesting to see what Apple does for higher end hardware like the MBP or iMac line. Will it still be integrated, or will they do a discrete version of the GPU?
I would suspect that most chips will ship with integrated GPU support. Having a base GPU across all platforms gives Apple considerable advantages. The only big question in my mind is how they address the Mac Pro.
By the way I can see Apple going the chiplet route like AMD and offering up the GPU as a chiplet that sits along side the processor chiplet. This could explain some of the conflicting info we are getting. A chiplet approach would make it very easy to scale the number of ARM processors in a Mac.
I would imagine that the MacBook would continue to use the SoC style GPU just like the iPad. It will be interesting to see what Apple does for higher end hardware like the MBP or iMac line. Will it still be integrated, or will they do a discrete version of the GPU?
I have a question, as a GPU layperson.
If they’re using their own tech for GPUs then what is the advantage of a discrete GPU?
The number one issue is that GPU's scale very well and if you make a big GPU chip you have far more computational power.
Apple has already stated for the record that they are going to use their own GPU. Why is this written as speculation?
They have also stated that they are designing Mac specific SoCs. So no, it won’t be an A14X. Though it might use the same core design but the number of CPU and GPU cores are going to be Mac specific.
I was going g to make a separate comment on that too. I keep reminding people that John, in his presentation, which had been given some time in June, stated very definitely that Apple would not be using the iOS series of chips in its Macs, and that a special line of SoCs for the Macs would be used. But I’m amazed at how many people keep saying that the A14x will be used. This is everywhere. It’s all wrong too.
I would imagine that the MacBook would continue to use the SoC style GPU just like the iPad. It will be interesting to see what Apple does for higher end hardware like the MBP or iMac line. Will it still be integrated, or will they do a discrete version of the GPU?
I have a question, as a GPU layperson.
If they’re using their own tech for GPUs then what is the advantage of a discrete GPU?
Most likely Apple won’t be able to compete on speed and features when it comes to high end GPUs such as the Radeon or Nvidia ones, and they’ll offer a replacement for what we consider built-in graphics GPUs (such as the Intel ones).
Perhaps the MacBook Pro Apple SoC version will contain an Apple integrated GPU plus a discrete ATI GPU, similar to laptops as of today but with Intel GPUs. As Apple advances GPU design they will more likely replace the mid to high-end with their own ones as well.
Apple already does IG graphics in their SoC. That’s nothing new. In fact, it was Apple that convinced Intel to do IG in the first place.
the entire reason why Apple is going to use their SoC IG graphics is because it leverages the other features of the SoC, such as the machine learning module, the neural engine module, the ISP and others, such as the shared memory. With a separate GPU, that can work with an Apple SoC, but how it would work with an x86 chip is harder to understand. Maybe that rumor is wrong, and Apple will be using their own chips there too.
Rayz2016 said: I have a question, as a GPU layperson.
If they’re using their own tech for GPUs then what is the advantage of a discrete GPU?
My understanding is that it has to do with power draw, i.e., the really high end GPUs require too many watts to be feasible for an integrated design.
That’s certainly part of it, but they’re also some of the biggest chips being made, with the most transistors in the higher end models. Apple’s GPU can’t compete on its own. But we’re getting the information that it’s Apple’s integration that going to give the GPU the boost it gets. How would a separate GPUs work? It’s hard to tell, but possibly Apple can’t really do that.
I would imagine that the MacBook would continue to use the SoC style GPU just like the iPad. It will be interesting to see what Apple does for higher end hardware like the MBP or iMac line. Will it still be integrated, or will they do a discrete version of the GPU?
I would suspect that most chips will ship with integrated GPU support. Having a base GPU across all platforms gives Apple considerable advantages. The only big question in my mind is how they address the Mac Pro.
By the way I can see Apple going the chiplet route like AMD and offering up the GPU as a chiplet that sits along side the processor chiplet. This could explain some of the conflicting info we are getting. A chiplet approach would make it very easy to scale the number of ARM processors in a Mac.
I expect the Mac Pro, the iMac Pro, and maybe the top non-Pro iMac will use scale-out chips rather than scale-up. That is, I expect them to increase performance not by going faster in one chip, but by having multiple chips on the board.
I also expect the initial round of these scale-out chips will not have integrated GPU cores (or if they do, to use them only for compute rather than video output). If you have four chips on a board, which one runs video? Do they elect one? Do they all share video processing? How would that work? Much simpler to just cut the GPU out of the SoC, devote more SoC area to CPU cores, give the chips some interprocessor bus and PCIe, and use a PCIe video card.
Apple has already stated for the record that they are going to use their own GPU. Why is this written as speculation?
They have also stated that they are designing Mac specific SoCs. So no, it won’t be an A14X. Though it might use the same core design but the number of CPU and GPU cores are going to be Mac specific.
Because they haven't said that at least in the context of discreet GPU's. Some information was deciphered to suggest that that was the case but it can be seen that the information can be interpreted in more than one way. The discussion is a bit ridiculous anyways as every A series Apple SoC comes with a GPU, so all Macs using an A series chip will have an Apple GPU even if a discreet is included. So the question becomes does Apple use a discreet Apple GPU in addition to its integrated GPU in the initial Macs and frankly we can't say. There are still good reasons to stick with AMD at the high end, mainly because of performance, but even here Apple has options to mix AMD with Apple GPU's.
With respect to the A14X in a Mac Book revival; that is easy to understand and has no impact on the idea that Mac specific SoC's are coming. The very nature of Macs will require very different chips for the various classes of machines. Personally if Apple doesn't implement SMT they will need a 24 core laptop machine by the end of 2021 to remain competitive. On the Mac Pro they are screwed if they can't match what AMD is already offering which means offering at least 64 cores and 128 cores will be needed if they can't get SMT working. AMD has some of the most compelling workstation offerings on the market right now so a Mac Pro is going to really stretch Apples capabilities.
If AMD really makes the most compelling workstation CPU, they won’t release their Threadripper Pro just right now.
Reading all these predictions are nice, but I’d really prefer seeing a timeline diagram with AppleInsider’s estimated release dates for each family of Macs. Because this is too many words.
If a 450-word article is "too many words" for you, you might be one of the LIVs or LINVs I've been reading about. This is an article-based website, so guest what ... words! Nuance! Depth!
Anyway, there's no value in AI (or anyone) putting out "estimated release dates" because there is simply zero information on which to base guesses ... and that's what they would be. We've been told already that the first Apple Mac will be out by the end of the year, and that the entire transition will take about two years. That's it, in terms of solid information. Going beyond speculation about what the first one will be is a waste of time. Make your own guesstimate timeline, since it will be every bit as valid as any that any other source comes up with.
The bottom line based on the info we have is that Apple is likely to continue to put out another generation of most Macs with an Intel chip in it (the iMacs are already done, of course) across the next year or so, and those machines will get their normal support cycle in terms of software updates and service, so I think if one is dependent on the unique Intel capabilities (like the ability to run Windows) and/or need to keep running older software that is unlikely to be updated for Apple Silicon, now through next year is probably the time to buy the last Intel Macs as any new Intel models get introduced.
Since we already know that the transition will be complete in two years, we also know that there will be an Apple Mac of one's preferred form factor available sometime in the next two years. If you're eagerly looking forward to getting one of those, start saving for it now -- you don't have terribly long to wait.
I would imagine that the MacBook would continue to use the SoC style GPU just like the iPad. It will be interesting to see what Apple does for higher end hardware like the MBP or iMac line. Will it still be integrated, or will they do a discrete version of the GPU?
I have a question, as a GPU layperson.
If they’re using their own tech for GPUs then what is the advantage of a discrete GPU?
Most likely Apple won’t be able to compete on speed and features when it comes to high end GPUs such as the Radeon or Nvidia ones, and they’ll offer a replacement for what we consider built-in graphics GPUs (such as the Intel ones).
Perhaps the MacBook Pro Apple SoC version will contain an Apple integrated GPU plus a discrete ATI GPU, similar to laptops as of today but with Intel GPUs. As Apple advances GPU design they will more likely replace the mid to high-end with their own ones as well.
I don’t think they’ll have problems replacing the discrete GPU on laptops, especially when they’ll be using the 5nm process instead of current 7nm. External solutions won’t go away anytime soon as it always puts more performance most of the time. You also have to consider cooling and battery life, there’s no way to accelerate a CPU though.
neilm said: A new 12" Retina MacBook would certainly be an easy introductory Apple Silicon model from a performance standpoint, whereas the expectations for a MacBook Pro replacement would be much higher.
Think of it as a clamshell iPad with a keyboard and running MacOS instead of iPadOS. Current iPads already have much better performance than the previous 12" MacBook, so that part wouldn't be a challenge. Apple might even get away with not having Thunderbolt 3, although I'd hope for more than the old MacBook's miserable single USB-C port.
The MacBook "plain" is also a currently vacant slot in Apple's lineup, although it could be argued that a new MacBook might cannibalize iPad sales to some degree. To this day I've never seen the old 12" MacBook in the field, whereas Airs and Pros are everywhere.
I would hope that if they stick to the one-port form factor, they'd at least do a TB3 port as you mention. But of course they don't have to -- clearly two TB3/USB-C ports as seen in the base MBP model has proven more popular/useful. That said, my experience with MBs and chats with people who owned one was that the one port thing wasn't a big issue ... the battery lasted quite a long time, so the one port was generally free at nearly all times. This surprised me, but my observation of students and other MB users in cafes was that they didn't have anything attached to it at all most of the time. You saw more chargers out on them in the evening, after a day of school/work.
Nowadays, I have a 15-inch MBP with four of those ports, and when I bring it out with me, I'm typically not using the charger at all because I don't plan to be out more than the six or so hours I'm expecting to get from the battery, and the other ports are -- as usual -- not in use. I do bring a little USB-C hub in case someone hands me a thumb drive or I need an HDMI port for a presentation, but I have never found myself using more than half of the ports on this 4-port model simultaneously ... ever.
The article notes that the term "MacBook" could be a generic, meaning Apple may not be reviving the line we all know as "MacBook (non-Pro)" at all. I wouldn't mind seeing that model return, particularly if the price can be brought down significantly (and unspoken but inferred part of the whole transition). The MacBook of old was a popular choice in college towns and with travellers, but the price was a big barrier compared to the MBA, which was just a better machine on nearly every level but weight. Give me a (let's say) $799 MB and it would be a real tossup between it and an iPad Pro with a decent keyboard in terms of super-light on-the-go machine.
Comments
apple’s *BEEN* making its own GPUs for 10+ years now and they’re part of the À-series silicon. They’re in every iDevice.
With respect to the A14X in a Mac Book revival; that is easy to understand and has no impact on the idea that Mac specific SoC's are coming. The very nature of Macs will require very different chips for the various classes of machines. Personally if Apple doesn't implement SMT they will need a 24 core laptop machine by the end of 2021 to remain competitive. On the Mac Pro they are screwed if they can't match what AMD is already offering which means offering at least 64 cores and 128 cores will be needed if they can't get SMT working. AMD has some of the most compelling workstation offerings on the market right now so a Mac Pro is going to really stretch Apples capabilities.
By the way I can see Apple going the chiplet route like AMD and offering up the GPU as a chiplet that sits along side the processor chiplet. This could explain some of the conflicting info we are getting. A chiplet approach would make it very easy to scale the number of ARM processors in a Mac.
Apple already does IG graphics in their SoC. That’s nothing new. In fact, it was Apple that convinced Intel to do IG in the first place.
the entire reason why Apple is going to use their SoC IG graphics is because it leverages the other features of the SoC, such as the machine learning module, the neural engine module, the ISP and others, such as the shared memory. With a separate GPU, that can work with an Apple SoC, but how it would work with an x86 chip is harder to understand. Maybe that rumor is wrong, and Apple will be using their own chips there too.
That’s certainly part of it, but they’re also some of the biggest chips being made, with the most transistors in the higher end models. Apple’s GPU can’t compete on its own. But we’re getting the information that it’s Apple’s integration that going to give the GPU the boost it gets. How would a separate GPUs work? It’s hard to tell, but possibly Apple can’t really do that.
But, I might make an exception for the Apple SoC. It's only money.
I also expect the initial round of these scale-out chips will not have integrated GPU cores (or if they do, to use them only for compute rather than video output). If you have four chips on a board, which one runs video? Do they elect one? Do they all share video processing? How would that work? Much simpler to just cut the GPU out of the SoC, devote more SoC area to CPU cores, give the chips some interprocessor bus and PCIe, and use a PCIe video card.
Your AMD fanboys are obnoxious.