I think this is notable for 2 things particularly:
1) The use of a SOC design which shows that the time for the ISA is (finally and thankfully) over. SOC means better integration, power management, and not insignificantly, better security
2) We are finally seeing an OS able to schedule across disparate CPU cores which the LITTLE.big architecture has enabled on smartphones for years. This is a big advance in allowing an OS to more efficiently use machine resources.
Could some explain why they were not able to update the higher end MacBook Pros with the M1?
Because:
the M1 is currently Apple’slow end chip. That it is being compared with, and can compete with, higher end alternative chips is just marketing gravy.
That clearly sells the chip short. The M1 is quite simply the highest performing laptop CPU in the world. By some way (looks like approx 50% over a 10-gen i9).
That said, the following limitations means that this won't be powering the higher-end laptops without another SOC package 1) Limitations in Thunderbolt channels 2) SOC memory capacity limited to 16GB 3) Monitor support limited to 1 addition.
Could some explain why they were not able to update the higher end MacBook Pros with the M1?
In theory, Apple could increase the clock speed of the M1 for higher computation performance with the same number of cores. In practice, power usage (and waste heat) grows in a nonlinear relationship to clock speed, so it’d be more like an Intel processor for power usage.
I have a late 2019 16” MacBook Pro, 64 GB RAM, 4 Thunderbolt 3 ports, 2.3 GHZ I9. I/O hardware takes more space due to power issues than pure CPU logic, and then there’s a matter of whether they have license to use that in their own chips. I’d not buy a 16” with only 2 ports. I’d also not buy with lower-end GPU performance you expect to get with a smaller machine, or with as few cores as the smaller machine.
More CPU and GPU cores require larger chips. The 5 nm process node is brand new, and even mature process nodes tend to have a certain percentage of completely defective chips (note option to buy MacBook Air with 7 GPU cores, likely that’s from those defects). There is a nonlinear percentage of defective chips on any wafer based on their size, which decreases wafer and chip yield. 16 billion transistors is HUGE for complexity and area on any node, especially brand new node. New process nodes are more expensive initially while working down defect rate.
In theory, Apple could have gone with larger chips to support greater GPU/CPU/ML cores and binned them for less powerful machines, but in practice this would be expensive for all the chips as a result. Less powerful computers are expected to be notably cheaper. Their approach makes good business and physics sense precisely because they blew away the Intel chips for performance with their low-end initial release.
"In theory, Apple could increase the clock speed of the M1 for higher computation performance with the same number of cores. In practice, power usage (and waste heat) grows in a nonlinear relationship to clock speed, so it’d be more like an Intel processor for power usage."
Correct, but also important to remember that this is actually Apple's 8th generation of 64bit microarchitecture (since the Cyclone A7 core). The microarchitecture of the M1 is very impressive, from what little information can be gleaned by specialist sites (note Apple don't disclose anything about it). As an example, leveraging the fixed length instruction set of ARM, they run an 8-wide decode which is twice what anyone else is doing on Arm, AMD and Intel.
In "old money" that's like running a 4-barrel Holley into your small-block over and above the old 2-barrel Rochester.
And, the other interesting thing about this chip over the A14 (which also uses 5nm) is the almost 50% increase in transistors. That is like an upgrade from a v6 to a v8 in old money. ;-)
The Apple M1 is imho indicative of a new generation of personal computers. Just like we talk about Fourth Generation jet fighters and Fifth Generation jet fighters, Apple’s M1 is the plus-one over whatever the current top of the line generation of personnel computers are considered. I’m not talking workstations, just personal computers. The workstation evolution to the next generation will no doubt be coming soon, and I expect Apple to be leading the way.
What specific things will the Apple Silicon Mac’s not be able to do that the preceding Mac’s could? Obviously we won’t be able to BootCamp and run 32-bit applications with an older system, but what else will not be possible? Will we still be able to boot off an external device? I assume with the integration of system memory that memory upgrades on desktop Macs will be a thing of the past too. So much for getting around Apple’s overpriced memory premiums, or will this be still possible somehow? What might this mean for PCI based expansion cards? Will these still work when in a thunderbolt enclosure or directly installed in a Mac Pro?
It’s an exciting new step but I will like to know how these things will be restricted too. Faster is good, but is it still extendable like our current and past Macs or are these more commoditized devices that will run Mac software but not the same types of Hardware?
1. External boot Yes
2.Expandable memory Yes. Enevtually
3. eGPUs, Yes
This is all contingent on their ability to expand the number of PCI lanes,,, which really shouldn't be a problem.
I still fear for the Mac long term though. There is no reason these low end M1 chips couldn’t end up in an iPad Pro. Then the iPad Pro would get thunderbolt.
What's wrong with that? Thunderbolt 3 is also known as USB4. Why wouldn't you expect the iPad Pro to go from USB3 to USB4?
M1 MACs put their mark right out of the gate. Apple's direction not just vertically integrate hardware and software tightly, but more external components getting on SOC resulting in overall better performance,low power consumption,lower cost, smaller motherboard allowing larger battery to fit in. The list can go on. Bottom line, every iteration of M1 and MacOS will result in greater benefits to customers in price/performance/value and Apple with increased market share, higher hardware, software and service revenue..
What specific things will the Apple Silicon Mac’s not be able to do that the preceding Mac’s could? Obviously we won’t be able to BootCamp and run 32-bit applications with an older system, but what else will not be possible? Will we still be able to boot off an external device? I assume with the integration of system memory that memory upgrades on desktop Macs will be a thing of the past too. So much for getting around Apple’s overpriced memory premiums, or will this be still possible somehow? What might this mean for PCI based expansion cards? Will these still work when in a thunderbolt enclosure or directly installed in a Mac Pro?
It’s an exciting new step but I will like to know how these things will be restricted too. Faster is good, but is it still extendable like our current and past Macs or are these more commoditized devices that will run Mac software but not the same types of Hardware?
First off no matter what anybody else says, this is Apples low end chip. It is basically the cell phone solution beefed up a bit. It should be obvious that it is low end by the machines Apple debuted today and the limitations of those platforms. Now low end does not mean low performance in this case. In any event to address some of your concerns:
The lack of support for 32 bit apps is a good thing here as it means apple only had to implement one instruction set keeping the ALU's lean. 32 bit apps are history now and any developer listening to Apple would have known that this was coming at least 3 years ago.
In one video I did hear something to the effect that external booting would be possible.
there is no reasonable possibility of a RAM upgrade. So buy the big machine. Frankly I wouldn't be surprised to find out that the POP approach is giving Apple a speed advantage here.
This SoC is not designed for systems supporting PCI expansion. As for future machines that is a different issue and frankly we can't project what Apple will be using for an external bus. Considering the nature of high performance GPU's support for some sort of plug in card will be required. However Apple implements this though is a secret that has yet to slip out. They could team up with AMD and use Infinity Fabric for an interface to a second socket for all we know.
The machines released today do not support eGPU's at all.
I'm not sure what you mean by the last question. The machines released today are Apple lowest end machines. They will not support anything more that the previous generation and in some cases less.
Remember these are the lowest end machines Apple currently sells. Apple has repeatedly said that there are more machines coming using other chips as they develop a family of chips to cover all the performance classes of their machines. Nobody has any idea at all what these other chips will be like!!! As for future chips we can speculate but I've found that there are far to many people that don't understand the word speculate.
I was quite surprised by the Thunderbolt/USB4 and RAM limitations. I see in the slide that they did show “Up to 16 GB” which I missed during the presentation. But the single external display and only 2 TB/USB4 ports came as a real surprise.
The performance of these SoCs is lovely and industry changing but I would love an explanation why they only wanted to release low end SoCs for the first round. It doesn’t have to be detailed but it shouldn’t look like they can’t do much better right now which is what it does look like. Apple’s dedication to secrecy doesn’t seem particularly helpful right now. Either you assume that they will come out with a higher end SoC for the next round which means it isn’t very secret or you assume that they can’t which is worse.
I was pretty dissatisfied with the level of technical detail in the keynote. They could have done better.
First let me say that the Keynote is perhaps the most disgusting thing I've seen from Apple corporate. I've never seen them shovel so much useless garbage in one presentation (this after years of keynotes and WWDC vidoes). So yeah a 5 year old could have done better.
As for the higher end chips there are things to consider. First M1 is obviously derived from the A series which probably explains some of the limitations we are seeing. To support Apples higher performance machines they will likely need an entirely different architecture, so this implies another series of chips that may take longer to develop. Then you have the reality of TSMC 5nm process and how early Apple is on this process, there is a lot to be said for being conservative here making sue the ramp goes well. Don't forget Covid. speaking of which there is now a $500 bonus to encourage engineers to travel to Taiwan (or China no sure), and sit in quarantine so there is an impact due to Covid.
As for secrecy well if I'm right and the high end SoC is dramatically different, then yeah they need to keep that secret. A lot of people seem to think that they will just throw more cores into M1 to make an M2 for the high performance machines. That is totally possibly but I'm not convinced that they will. The problem is a good portion of their customer base works in industries where more cores are better no matter how good those cores are. They will need a way to address this and frankly they will 20+ cores in the high end laptop by the end of 2021. The design of M1, a beefed up portable chip, does not allow this.
Could some explain why they were not able to update the higher end MacBook Pros with the M1?
The M1 is a low end chip derived from the A series. The fact that it performs well doesn't fix the problem of limited capability to handle the work loads of more demanding users.
Could some explain why they were not able to update the higher end MacBook Pros with the M1?
Because:
the M1 is currently Apple’slow end chip. That it is being compared with, and can compete with, higher end alternative chips is just marketing gravy.
That clearly sells the chip short. The M1 is quite simply the highest performing laptop CPU in the world. By some way (looks like approx 50% over a 10-gen i9).
That said, the following limitations means that this won't be powering the higher-end laptops without another SOC package 1) Limitations in Thunderbolt channels 2) SOC memory capacity limited to 16GB 3) Monitor support limited to 1 addition.
But it is Apples low end Mac chip!!!! This is what gets me excited because it implies that the high end chips will be even more impressive. I may not be very positive on Apple for other reasons but this chips is a great achievement.
You reasons are interesting but the reasons this isn't in higher end laptops has many more justifications.
There simply is not enough high performance cores. I'm not talking adding a couple cores to this chip but rather the need to add at least 20 cores by the end of 2021. This requires a new architecture.
The GPU in the M1 is kinda sucky. This will require significant improvements or support for a bus to an external GPU.
We are not really sure what the RAM limitation is, it could simply be nothing available for higher capacities in a PoP solution. Beyond that I'd be very surprised to see future solutions using DDR4 of any type.
The current SoC doesn't support the I/O a modern Pro computer requires.
In the end I can't see Apple adding enough cores to M1 to make a reasonable processor solution for a pro laptop. Pros simply have come to expect that they will have hardware to run their highly threaded work loads. This gets even more important on the iMac and Mac Pro.
Comments
That clearly sells the chip short. The M1 is quite simply the highest performing laptop CPU in the world. By some way (looks like approx 50% over a 10-gen i9).
That said, the following limitations means that this won't be powering the higher-end laptops without another SOC package
1) Limitations in Thunderbolt channels
2) SOC memory capacity limited to 16GB
3) Monitor support limited to 1 addition.
I have a late 2019 16” MacBook Pro, 64 GB RAM, 4 Thunderbolt 3 ports, 2.3 GHZ I9. I/O hardware takes more space due to power issues than pure CPU logic, and then there’s a matter of whether they have license to use that in their own chips. I’d not buy a 16” with only 2 ports. I’d also not buy with lower-end GPU performance you expect to get with a smaller machine, or with as few cores as the smaller machine.
More CPU and GPU cores require larger chips. The 5 nm process node is brand new, and even mature process nodes tend to have a certain percentage of completely defective chips (note option to buy MacBook Air with 7 GPU cores, likely that’s from those defects). There is a nonlinear percentage of defective chips on any wafer based on their size, which decreases wafer and chip yield. 16 billion transistors is HUGE for complexity and area on any node, especially brand new node. New process nodes are more expensive initially while working down defect rate.
In theory, Apple could have gone with larger chips to support greater GPU/CPU/ML cores and binned them for less powerful machines, but in practice this would be expensive for all the chips as a result. Less powerful computers are expected to be notably cheaper. Their approach makes good business and physics sense precisely because they blew away the Intel chips for performance with their low-end initial release.
Correct, but also important to remember that this is actually Apple's 8th generation of 64bit microarchitecture (since the Cyclone A7 core). The microarchitecture of the M1 is very impressive, from what little information can be gleaned by specialist sites (note Apple don't disclose anything about it). As an example, leveraging the fixed length instruction set of ARM, they run an 8-wide decode which is twice what anyone else is doing on Arm, AMD and Intel.
In "old money" that's like running a 4-barrel Holley into your small-block over and above the old 2-barrel Rochester.
And, the other interesting thing about this chip over the A14 (which also uses 5nm) is the almost 50% increase in transistors. That is like an upgrade from a v6 to a v8 in old money.
;-)
1. External boot
Yes
2.Expandable memory
Yes. Enevtually
3. eGPUs, Yes
This is all contingent on their ability to expand the number of PCI lanes,,, which really shouldn't be a problem.
Expandability.
This is a Gen 1 chip... They probably shoehorned PCI Gen 4 lanes into it... This design is almost 3 years in the making
M2 will have proper support for PCI
- The lack of support for 32 bit apps is a good thing here as it means apple only had to implement one instruction set keeping the ALU's lean. 32 bit apps are history now and any developer listening to Apple would have known that this was coming at least 3 years ago.
- In one video I did hear something to the effect that external booting would be possible.
- there is no reasonable possibility of a RAM upgrade. So buy the big machine. Frankly I wouldn't be surprised to find out that the POP approach is giving Apple a speed advantage here.
- This SoC is not designed for systems supporting PCI expansion. As for future machines that is a different issue and frankly we can't project what Apple will be using for an external bus. Considering the nature of high performance GPU's support for some sort of plug in card will be required. However Apple implements this though is a secret that has yet to slip out. They could team up with AMD and use Infinity Fabric for an interface to a second socket for all we know.
- The machines released today do not support eGPU's at all.
- I'm not sure what you mean by the last question. The machines released today are Apple lowest end machines. They will not support anything more that the previous generation and in some cases less.
Remember these are the lowest end machines Apple currently sells. Apple has repeatedly said that there are more machines coming using other chips as they develop a family of chips to cover all the performance classes of their machines. Nobody has any idea at all what these other chips will be like!!!As for future chips we can speculate but I've found that there are far to many people that don't understand the word speculate.
As for the higher end chips there are things to consider. First M1 is obviously derived from the A series which probably explains some of the limitations we are seeing. To support Apples higher performance machines they will likely need an entirely different architecture, so this implies another series of chips that may take longer to develop. Then you have the reality of TSMC 5nm process and how early Apple is on this process, there is a lot to be said for being conservative here making sue the ramp goes well. Don't forget Covid. speaking of which there is now a $500 bonus to encourage engineers to travel to Taiwan (or China no sure), and sit in quarantine so there is an impact due to Covid.
As for secrecy well if I'm right and the high end SoC is dramatically different, then yeah they need to keep that secret. A lot of people seem to think that they will just throw more cores into M1 to make an M2 for the high performance machines. That is totally possibly but I'm not convinced that they will. The problem is a good portion of their customer base works in industries where more cores are better no matter how good those cores are. They will need a way to address this and frankly they will 20+ cores in the high end laptop by the end of 2021. The design of M1, a beefed up portable chip, does not allow this.
You reasons are interesting but the reasons this isn't in higher end laptops has many more justifications.
- There simply is not enough high performance cores. I'm not talking adding a couple cores to this chip but rather the need to add at least 20 cores by the end of 2021. This requires a new architecture.
- The GPU in the M1 is kinda sucky. This will require significant improvements or support for a bus to an external GPU.
- We are not really sure what the RAM limitation is, it could simply be nothing available for higher capacities in a PoP solution. Beyond that I'd be very surprised to see future solutions using DDR4 of any type.
- The current SoC doesn't support the I/O a modern Pro computer requires.
In the end I can't see Apple adding enough cores to M1 to make a reasonable processor solution for a pro laptop. Pros simply have come to expect that they will have hardware to run their highly threaded work loads. This gets even more important on the iMac and Mac Pro.