Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
On eGPUs: Now the question will be, with that Thunderbolt hardware support, which OEMs will provide drivers to interoperate between an ARM64 ISA and their hardware? Will AMD start creating ARM64 ISA Mac drivers? nVidia is out of the question I expect.

Thunderbolt is used for a whole lot more than just eGPUs, and while the hardware will be there, the driver support is what I question. Could very well be that Thunderbolt display and storage options will work seamlessly, but not much else. But, here's to hope!
So far a MacPro with dedicated GPU exists, there would be a possibility of having some drivers for some GPUs.
 
Good. Now all the nervous folks that freaked out over the DTK not having it can relax (even though it should have been obvious that Apple would include a feature that they themselves helped design).



You have to look at it in order to use it (which is really annoying for touch-typists), it's only available on some of their laptops, and it's useless if you use an external keyboard. It's a stupid feature that just raises the cost of their products. It should definitely be dumped. It actually makes me less productive when I'm using the laptop's built-in keyboard.

Also:
  • Goes to sleep, requiring you to touch it to wake it, then touch it again wherever the virtual key reveals itself
  • Unless you have a 16" MBP or later, you don't have a dedicated Esc key
I went back to my 2014 MBP, and it's a joy to use that row of physical keys that never fall asleep. The Emoji Bar is, at best, Stockholm Syndrome. The main thing is no one asked for this, and there's neither an option nor a discount for "can we not."
 
  • Like
Reactions: RandomDSdevel
On eGPUs: Now the question will be, with that Thunderbolt hardware support, which OEMs will provide drivers to interoperate between an ARM64 ISA and their hardware? Will AMD start creating ARM64 ISA Mac drivers? nVidia is out of the question I expect.

Thunderbolt is used for a whole lot more than just eGPUs, and while the hardware will be there, the driver support is what I question. Could very well be that Thunderbolt display and storage options will work seamlessly, but not much else. But, here's to hope!
Or better yet will Apple allow them to create kext for ARM hardware. Might be closed like the iPadOS and iOS.
 
On eGPUs: Now the question will be, with that Thunderbolt hardware support, which OEMs will provide drivers to interoperate between an ARM64 ISA and their hardware? Will AMD start creating ARM64 ISA Mac drivers? nVidia is out of the question I expect.

Thunderbolt is used for a whole lot more than just eGPUs, and while the hardware will be there, the driver support is what I question. Could very well be that Thunderbolt display and storage options will work seamlessly, but not much else. But, here's to hope!
Isn’t that issue completely taken care of if a developer is using DriverKit like they should be?
 
Personally it’s about having to pay the premium for a TouchBar to get the specs I need on the machine while having an inferior workflow.

I’m glad they added back the physical escape key but my model doesn’t have that. As a software engineer the Touch Bar provides zero acceleration to my workflow and is just a detriment.

Ive customized it quite a bit, but more so to reduce its negative impact. It doesn’t add anything but I’ve managed to make it less bothersome.
 
  • Like
Reactions: RandomDSdevel
I never look at the keyboard.
So for me it is a complete waste. And when I do want to change things like the volume, I have to look away from the screen.
I much prefer keys and find using a external keyboard a better experience than using the keyboard with Touch Bar.
At least now I have a physical escape key...

Have you tried BetterTouchTool? Some of the popular presets (or you can make your own) use a two-finger swipe to change volume and a three-finger swipe to change brightness. That's truly a no-look experience, even more than having dedicated keys for those purposes.

Also, it's possible to keep your favorite functions in one place no matter what you're doing, which also allows you to develop a muscle memory as you might have with fixed keys. The difference is that you can more readily customize those functions to fit your own needs.
 
So let's look at it like this. Apple's ARM is RISC so unless it's clocked higher than Intel, will be slower. It will be more power efficient. You won't be able to run full Windows; it will be Microsoft's ARM version. This will be missing DirectX and optimised Graphics drivers. On the Mac Side, you will need custom graphics to support whatever bus standard they decide to use. These drivers will unlikely be optimised and so graphics will be slower.

In general, if you want a power friendly laptop but don't need performance or graphics, the ARM based Macs will be interesting. I'm not really sure they are aiming for, to be honest. It's probably the same people who buy the iPad Pro's today, but need a bit of extra power.
On that last part, possible to merge the Mac with Apple’s mobile products.
 
You mean Thunderbolt 4 ... which also supports USB 4!

Thunderbolt 4 is Thunderbolt 3 plus USB 3.2 Gen 2x2 and other USB4 (no space between USB and 4, and this stupid thing is important as we know USB-IF's naming ability will always screw us up when people not typing the name exact as they decide) features.
So basically Thunderbolt 4 is ultimate version of USB4 with almost all optional features enabled.

It have same maximum speed of Thunderbolt 3.
 
Everything about Big Sur screams touch screen support. Fonts are bigger, there are settings to add more space in the menu bar, sliders every year, tons of padding between controls.

Its a dead give away that macOS devices are gonna come with touch screen.

Wow, how did I not see this? You're absolutely right. I watched the video thinking "more space around menu items? who's complained they can fit too much on the screen?" Now it makes sense...
 
I would expect Apple’s own GPUs for the mobile ARM Macs such as laptops. For desktops, there is little advantage for saving power so they would probably use current GPU chips.
Two things to keep in mind, though:

1) For desktops, reduced power consumption means reduced cooling and thus quieter operation. One of the complaints about the iMac has been excessive fan noise under heavy load. You also get longer component life from reduced temperature.

2) Apple would like to keep everything under their own umbrella. So if they can replace AMD dGPUs with their own GPU, I think they will (plus some say there's a benefit from not switching between iGPU and dGPU, which is what Macs with dGPUs do now). Whether they've developed their AS GPUs to the point where they can out-compete the AMD dGPUs is, however, another question.
 
  • Like
Reactions: RandomDSdevel
And? Precisely that same comment can be applied to plenty of things Apple makes. You present it like it’s a unique reason not to build ARM-based Macs.
Point is, if they screw it up they'll lose people except for the fan boys.
 
Intel's chips are RISC. They have a hardware translation layer to convert x86 and x64 opcodes into an internal execution cache that operates on a completely different RISC-based microcode.

You can argue that an essential property of a RISC implementation is, that it does NOT break down instructions into microcode or doing microcode lookups - which makes your statement wrong. I mean thats the very idea of RISC - that the ISA reflects what the HW can essentially issue/execute with a single cycle delay.
 
Last edited:
  • Like
Reactions: RandomDSdevel
Isn’t that issue completely taken care of if a developer is using DriverKit like they should be?

As far as I know, there are no DriverKit implementations for proper GPUs — they use kernel extensions today.

I do expect Apple to have “dGPUs”, but it remains to be seen whether that will be AMD or something developed in-house (like a “graphics accelerator”-type thing similar to what they’ve did for the Mac Pro Afterburner add on card).

Nothing really stopping AMD or nVidia from working with Apple, but my guess is that they’ll end up having to write new custom drivers that Apple will have to bless. How much is that market worth to AMD? nVidia has already shown their hand unless they have a change of heart.
 
  • Like
Reactions: RandomDSdevel
You can argue that an essential property of a RISC implementation is, that it does NOT break down instructions into microcode or doing microcode lookups - which makes your statement wrong. I mean thats the very idea of RISC - that the ISA reflects what the HW can essentially issue/execute with a single cycle delay.
That doesn't work on any modern RISC processor. Once your CPU runs at 2.5 GHz, RISC or not, or as a desktop processor at 3.5 GHz, reading data from L1 cache is a 3 or 4 cycle operation already. On the other hand, the A12 can decode 7 instructions per cycle, and execute up to 10. There are plenty of instructions with long latencies.
[automerge]1594341158[/automerge]
If a new CPU to compete with Intel is wanted, develop a whole new architecture.
Apple doesn't want a CPU to _compete with Intel_, they want a CPU to run Macs faster, using less power, at the best possible price. And Apple's mobile chips, suitably modified, will do just that, because Apple has the experience from building 200 million+ chips per year already. A new architecture just for MacOS would throw away all of Apple's know how, and would only be built in numbers of 10-20 million per year.

And what you apparently don't realise is that ARM 64 bit _is_ a whole new architecture. In Apple's chips, all 32 hardware is completely gone, and Apple did have their fingers quite deep in the ARM 64 bit design.
 
Last edited:
So let's look at it like this. Apple's ARM is RISC so unless it's clocked higher than Intel, will be slower.
Wrong. Apple's ARM implementation beats Intel in instruction decode (7 instructions vs 4, and 7 microops vs 4). Apple's ARM implementation beats Intel in number of processing units (6 integer units, 2 load, 2 store, 3 vector vs. a lot less). ARM 64 bit beats Intel in architectural registers (30 available for general use vs. 16). Apple's ARM has faster floating point (division with a throughput of one per cycle and 8 cycle latency, three fused multiply-add throughput with 3 cycle latency). And Apple's ARM has massive caches (8MB L2 cache, plus 16MB cache between CPU and memory).
 
That doesn't work on any modern RISC processor. Once your CPU runs at 2.5 GHz, RISC or not, or as a desktop processor at 3.5 GHz, reading data from L1 cache is a 3 or 4 cycle operation already. On the other hand, the A12 can decode 7 instructions per cycle, and execute up to 10. There are plenty of instructions with long latencies.
[automerge]1594341158[/automerge]

Apple doesn't want a CPU to _compete with Intel_, they want a CPU to run Macs faster, using less power, at the best possible price. And Apple's mobile chips, suitably modified, will do just that, because Apple has the experience from building 200 million+ chips per year already. A new architecture just for MacOS would throw away all of Apple's know how, and would only be built in numbers of 10-20 million per year.

And what you apparently don't realise is that ARM 64 bit _is_ a whole new architecture. In Apple's chips, all 32 hardware is completely gone, and Apple did have their fingers quite deep in the ARM 64 bit design.
ARM in itself is still old, you can not deny that.
 
Will it interface to PCIe or some other bus? And, and how many total PCIe lanes available and how fast? Have been following ARM devices and they usually have limited or no PCIe.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.