Apple VPs talk new M1 Mac development, Intel relationship, and more
A pair of Apple vice presidents offered more details about the development and performance of the company's M1 Apple Silicon chip in a new interview.

Credit: The Tech Chap
Apple VP of Platform Architecture Tim Millet and Apple VP of Worldwide Marketing Bob Borchers sat down via video call with YouTube channel The Tech Chap to discuss some of the intricacies of the M1 and the new Apple Silicon lineup.
"M1 is a huge step forward in terms of performance for Macs," said Borchers. "And in very real ways, whether's it battery life, graphics performance, or just CPU. We are talking multiples increase over previous generations."
Millet, a part of the team who works on Apple Silicon, explained how the gains made with M1 stemmed from the Platform Architecture team's work on Apple A-series chips.
"What we knew going in when we started building chips for the phone was these were incredibly thin and light beautiful works of art that our industrial design team creates," Millet said. "We don't get to tell them to increase the size of the product, increase the battery to deliver better performance."
The job of Millet's team then was to "figure out how to deliver that performance within the constraints of the system," adding that performance that doesn't translate to the real world within those constraints "doesn't matter."
When asked about the potential confusion for consumers about the differences between Apple Silicon and Mac chips, Borchers said the first step with M1 was to bring the proprietary chip design to some of Apple's most popular models.
"It is a transition that will take time, so there are places where we wanted to continue to offer Intel processors and additional choice," said Borchers.
Some of that additional choice could be users who need four ports on their Mac portables, or users who want to upgrade internal memory up to 32GB.
When asked about whether the M1 would be a 10W or 15W chip, Millet said "I think you'll see across the different array of machines that we announced yesterday, you're gonna see M1 at its best in every one of those enclosures."
As an example, the M1 is able to fit in the enclosure of the MacBook Air. But with the addition of a fan in the 13-inch MacBook Pro and Mac mini models, users will see "a different level of performance," Millet added.
At another point in the interview, Millet also expanded on how the M1 chip's unified memory carries a number of benefits over traditional RAM.
"The CPU gets a much wider memory system to allow multi-threaded applications across our eight cores. They're going to be unconstrained by memory bandwidth," Millet added. "And the GPU has a high-capacity memory system getting its high bandwidth."
When asked about whether Apple is still "friends with Intel," Borcher said that the two companies still have a "great relationship" and that Apple still plans to ship "amazing Intel systems."
The full interview is about half an hour and is worth a watch for anyone interested in some of the more technical details about the M1.

Credit: The Tech Chap
Apple VP of Platform Architecture Tim Millet and Apple VP of Worldwide Marketing Bob Borchers sat down via video call with YouTube channel The Tech Chap to discuss some of the intricacies of the M1 and the new Apple Silicon lineup.
"M1 is a huge step forward in terms of performance for Macs," said Borchers. "And in very real ways, whether's it battery life, graphics performance, or just CPU. We are talking multiples increase over previous generations."
Millet, a part of the team who works on Apple Silicon, explained how the gains made with M1 stemmed from the Platform Architecture team's work on Apple A-series chips.
"What we knew going in when we started building chips for the phone was these were incredibly thin and light beautiful works of art that our industrial design team creates," Millet said. "We don't get to tell them to increase the size of the product, increase the battery to deliver better performance."
The job of Millet's team then was to "figure out how to deliver that performance within the constraints of the system," adding that performance that doesn't translate to the real world within those constraints "doesn't matter."
When asked about the potential confusion for consumers about the differences between Apple Silicon and Mac chips, Borchers said the first step with M1 was to bring the proprietary chip design to some of Apple's most popular models.
"It is a transition that will take time, so there are places where we wanted to continue to offer Intel processors and additional choice," said Borchers.
Some of that additional choice could be users who need four ports on their Mac portables, or users who want to upgrade internal memory up to 32GB.
When asked about whether the M1 would be a 10W or 15W chip, Millet said "I think you'll see across the different array of machines that we announced yesterday, you're gonna see M1 at its best in every one of those enclosures."
As an example, the M1 is able to fit in the enclosure of the MacBook Air. But with the addition of a fan in the 13-inch MacBook Pro and Mac mini models, users will see "a different level of performance," Millet added.
At another point in the interview, Millet also expanded on how the M1 chip's unified memory carries a number of benefits over traditional RAM.
"The CPU gets a much wider memory system to allow multi-threaded applications across our eight cores. They're going to be unconstrained by memory bandwidth," Millet added. "And the GPU has a high-capacity memory system getting its high bandwidth."
When asked about whether Apple is still "friends with Intel," Borcher said that the two companies still have a "great relationship" and that Apple still plans to ship "amazing Intel systems."
The full interview is about half an hour and is worth a watch for anyone interested in some of the more technical details about the M1.
Comments
https://sneak.berlin/20201112/your-computer-isnt-yours/
So at this point it would be absolutely nuts for most people to buy an Intel based Mac. By the way one of the reasons I see M1 or Apple Silicon based Macs, as mandatory isn't so much the ARM cores but the fact that Neural Engine is in there. Neural Engine is already making an impact on recently released third party software. The reality is that M1 is pretty well balanced as far as the many subsections of the SoC goes. The only thing that would have been nice is more performance out of the GPU but even that is pretty much industry leading.
Very interesting, thanks.
Current entry level 21.5" iMac is limited to 16GB RAM, so this SoC could eventually make its way into that model, but I don't think we'll see an iMac with Apple Silicon until there's more than a single configuration.
I suspect that an M1X will come out in early Spring with 6/4 core CPU, 12 core GPU, 16/32 RAM. This SoC will be an upgrade option for the mini and 13" MacBook Pro (at which time, the Intel models will be dropped). It will also be used in the entry level 16" MacBook Pro. Both the M1 and M1X will be used in 21.5" iMac models.
I don’t see Apple dropping the new Mac Pro for at least 2-3 years from launch, but they might if a crazy Apple Silicon model offers bonkers performance at a higher margin for Apple and if there is developer support.
That would leave the Mac Pro as an expensive example of how we used to do PCs. Which, I think, was Apple’s plan all along.
I also think that support for the Intel machines will be a lot longer than what the PPC transition did. Most Macs these days get OS upgrades out 5-7 years and 8 years for Mac Pros desktops.
Obviously, we don't know know where or how Apple is taking this naming scheme, but I'd have to believe - based off the A-series - that each number, M1, M2, M3, etc., is a generation marker. So just as with the A13 to the A14, the jump in specs isn't going to be as great. What we'll see is updated IP blocks, CPU, GPU, ISP, I/O controllers, etc. The performance differences will come within SoC variants of each generation. M1, M1A, M1B, M1C, etc. (Although probably not that many in the first generation.)
So something along the lines of...
M1; 4/4 CPU, 8 GPU, 8/16 RAM, 8x PCIe
M1A; 6/4 CPU, 12 GPU, 16/32 RAM, 16x PCIe
M1B; 8/4 CPU, 16 GPU, 32/64 RAM, 16x PCIe
M1C; 10/4 CPU, 20 GPU, 64/128 RAM, 32x PCIe
And there might be [binning] variations within each of those just as we have 7/8 GPU in the M1. There also would not be a need to differentiate mobile from desktop because these SoCs are designed to be extremely efficient when they need to be; perform based on thermals (and power supply).
I also think that the SoCs for the iMac Pro and Mac Pro, could be an entirely different series; P1, P2, P3, etc. That are designed to support discreet RAM and GPUs. And of course, have extremely high performance; 32/48/64 CPU cores.
https://www.amd.com/system/files/documents/amd-cdna-whitepaper.pdf
https://www.amd.com/en/products/server-accelerators/instinct-mi100
Anyone complaining about Nvidia should just stop. The MI100 Radeon Instinct is Passively Cooled to boot. I would love to see a Dual Radeon Instinct on a single Card option by Apple allowing 4 of these connected [Infinity Fabric 3.0 on these allow up to 8 connected] but on the Mac you could have a Dual in on slot and the regulard GPU in the other slot, each with Apple's TB3 quad on-board outputs. The W6900X would support a series of the 6K Displays and the CDNA 1.0 dual card with 64 GB and a pure compute monster offers an extremely long term value proposition for the Mac Pro market.
The white paper explains it all.
2+4+4 : phone and lower end tablets
4+4+8 : pro tablets and lower end PCs
8+4+16 : midrange desktops and laptops
16+8+32 : high end laptops and desktops, Pro desktops
32+8+64 : Pro desktops
If they want to have 6+4+12 SKU, they can bin the 8+4+16 chip. If they want to have a 12+8+24 SKU, they can bin the 16+8+32 chip. It would be a very poor use of resources to have a SoCs that are only different in 2 CPU cores or 4 GPU scores.
Lastly, who knows really. If the GPU cores end up capable of doing 64 bit FP math, this could mean they won't be growing CPU core counts. Then, the NPU occupies a rather large part of the SoC chip. They could have added another 4 perf CPU or 4 GPU cores in place of it, yet they included the NPU. If applications can really take advantage of the NPU, they could devote the transistors to NPU instead of the GPU or CPU.