Exponential gains of M1

[dmccloud]

What exponential gains? Apple chips have never been delivering exponential increases in performance. They have been delivering rapid, but linear increases in power every year that has far outpaced anything from Intel.

The exponential gains which people are expecting is because until now, Apple has achieved this extremely impressive but linear increase in performance installed in power and thermally constrained devices, like iPhones, iPads and (now) small form factor laptops.

Once you remove those constraints and put them into larger form factor and mains power connected devices, they can start pumping in more power, increasing clock speeds, core counts and active cooling and you'll really start to see these things fly.

Just to piggyback on this with regards to the OP, the so-called "exponential" growth you refer to is in relation to the Intel processors that were replaced, not the A-series Apple units. If you look at that graph, you can draw a single straight line between each and every one of the Apple results, while the Intel has both 5th and 11th gen CPUs as outliers on their overall pattern. If you look at that graph, Intel only managed to increase performance by roughly 50% in six years, while Apple increased the performance of their A-series processors by over 300%, and in a linear fashion at that. There is one big factor working in Apple's favor that Intel does not have, and that is the TSMC partnership. While Apple is already using their 5nm process for both the A14 and M1, TSMC is already developing both 3nm and 2nm processes, which would serve to continue this linear path of improvement for the A-series. Meanwhile, Intel delayed their full shift to 10nm by two years because they can not get a 10nm desktop to work reliably enough to be labeled a Celeron, let alone anything in the Core series.

Knowing the way Apple operates the M1 is not the first Mac SoC Apple has developed, it is just the first SoC deemed ready for prime time. Furthermore, Apple probably has the next 2-3 years for both the A and M series already mapped out with the 2021 chips nearing production, especially if the rumors of an early 2021 release of more new Macs is true. The other reason Apple is likely to continue on this steady path of improvement on a year to year basis is that since they build both the OS and SoC, they can design both components to work well together, just like they have done with the iPhone and iPad since their initial launch.

 

[JohnnyGo]

Well like I mentioned, two of the M1 devices are power constrained because they operate on battery and all three are constrained because of package size Vs available area on the mainboard. The MacBook Air is completely thermally constrained because it is passively cooled and the other two have very small fans and cases.

Apple is claiming 10W for the M1 in the MacBook Air (I've read up to 15W from other sources).

Even with this power consumption the single core performance is up there with some of the best from AMD and Intel. Suppose they do this for the next iMac: Increase the power input, implement more efficient active cooling and then raise the core clock speed. Get rid of the efficiency cores and then double the amount of cores to 8 performance cores.

They have so much head room in terms of power consumption and thermals in the larger form factor Macs it's not funny.

If we don't see enormous gains in these machines I believe it will simply be Apple purposefully being incremental so they can "wow" us each year.

Anyway, time will tell.

Apple will deliver enormous gains over the next 36 months.

Be it with doubling/quadrupling core counts (8-16-32), be it from separating the GPU and going for 32-64 GPU cores

Be it by following TSMC down to 2nm over the next 4 years, be it by investing heavily on MacOS optimizations in its chip designs

Maybe after this initial flurry Apple will slowdown to a more pedestrian 20% gain every year. But hey, we’ve been watching Intel slow to a crawl, delivering 20% only every 3 years.

 

[SlCKB0Y]

Apple will deliver enormous gains over the next 36 months.

Be it with doubling/quadrupling core counts (8-16-32), be it from separating the GPU and going for 32-64 GPU cores

Be it by following TSMC down to 2nm over the next 4 years, be it by investing heavily on MacOS optimizations in its chip designs

I absolutely agree with this. They just have so much room to scale up, I can't emphasise this enough.

 

[thingstoponder]

Some people seem to have the idea that the M1 released by Apple is only the start, and by next year we will see exponential improvements like 2x faster and 20% less wattage usage. I feel like we have seen the exponential performance improvements, this is it. Any thing in the future will be spec bumps. We will not see improvements on the M1 chip like we have seen on the A-series on the iphones every couple of years. Of course, they have not released their PRO chips yet which should be much more powerful but once those are released its spec bumps from there.

Is there any indication that this is not the case and this is only the first generation which are usually weak(like on original iPad) and will see crazy boosts in the future?

Does anyone think it will be 2x faster? They’re crazy. Will it be 20% faster per core every year like the iPhone has been the last few years? I think so.

If not then only now would they be plateauing on year over year improvements. But I’ve seen no reason to believe they are.

 

[thingstoponder]

This is what I am thinking, they probably have some powerful CPUs to be released but after that by next year I don't think we will see the generational jump we saw with this year's MBA.

Nobody thinks that.

People seem to think this is only the beginning and we will have exponential jumps by next year and the year after that.

When people say “this is just the beginning” they mean we’ve only seen a 4+4 core model. The M1 is already punching above its weight and the higher end ones to replace higher end machine will be even better.

 

[thingstoponder]

Its not year-over-year but if you look at the graph from A11 to A14 there is near double the performance and from A9 to A14 its 3x+ the performance. On the Intel side, there is not even double. The biggest jump is a mere 50% in the past 7 years. I am not counting the jump since A4 either.

it’s entirely possible the M4 will be twice as fast per core. It would be quite the feat, but as long as TSMC delivers 3nm and 2nm like they’re saying and Apple keeps iterating then it’s possible.

 

[SlCKB0Y]

Does anyone think it will be 2x faster? They’re crazy. Will it be 20% faster per core every year like the iPhone has been the last few years? I think so.

They were passively cooled devices which had to meet certain battery life requirements with a tiny mainboard. You're still thinking about these chips as if they a mobile only SoC with all the constraints that go along with that. Constraints being physical size, power consumption and the resulting heat. They were getting that 20% increase year on year despite of these constraints. Constraints which are now gone for their desktop Macs.

They can scale now scale these cores both vertically (clock speed) and horizontally (core count).

I'd like to hear your explanation as to why, right now, Apple couldn't create a "desktop" grade CPU/package for an iMac - say increase power consumption, bump the clock speed by 10-20% and put anywhere from 8 to 32 performance cores in it? Forget about 2x faster, this configuration would absolutely destroy that.

Intel can put these kind of core counts on their CPUs and they do not have anywhere near the thermal performance of the M1.

Seriously - provide a single technical reason why this would not be possible right now? Just ONE.

 

[armoured]

That was going to be MY post.

I can't stand it when precise terminology enters the masses and is misused.

There is nothing exponential here.

My complaint is that exponential does not mean just super rapid growth. It's a functional form, not a synonym of 'big'.

One percent a year is exponential, too.

 

[Fomalhaut]

Does anyone think it will be 2x faster? They’re crazy. Will it be 20% faster per core every year like the iPhone has been the last few years? I think so.

If not then only now would they be plateauing on year over year improvements. But I’ve seen no reason to believe they are.

This sounds more reasonable. Maybe what the other poster meant was that Apple will add more cores in each iteration, so an 8+4 or 12+4 core SoC could easily be twice as fast (for multi-core workloads).

It will be really interesting to see how far Apple can scale vertically. Larger dies have a greater chance of failures, and there comes a point where the GPU would need to be on a separate die, or for the CPU cores to be placed on separate chiplets similar to AMDs Epyc & Rome CPUs

AFAIK the ARM Neoverse N1 that is used by Amazon (for their Graviton 2), Marvell and Ampere, supports up to 128 cores on a single die, so this probably isn't an issue for Apple.

I am really curious about Apple's path for matching external GPU performance for the iMac and Mac Pro. Will they support external GPU cards via PCIe, or have a custom interconnection between GPU and CPU and RAM on a single package? Will we see the end of the huge 300W dGPU cards in PC desktops and the Mac Pro?

Here an image to the AMD-designed SoC in the Microsoft XBox X:

Notice how the GPU is huge compared to the CPU cores (8 core/16 thread). If Apple wants to compete on GPU as well as CPU cores, *plus* all the custom features on Apple Silicon, either the die would need to be gigantic, or the GPU needs to be separated.

 

[SlCKB0Y]

If Apple wants to compete on GPU as well as CPU cores, *plus* all the custom features on Apple Silicon, either the die would need to be gigantic, or the GPU needs to be separated.

I think we will see this separation when the new Mac Pro comes out. Also, assuming they use a dedicated card, they don't need to put this card on a standard PCIE interface, they could come up with something proprietary to get better performance.

This would also allow them to maintain the upgradeability and vertical integration of the Mac Pro whilst still forcing users to purchase what will almost certainly be expensive Apple GPUs.

 

[FlyingTexan]

Some people seem to have the idea that the M1 released by Apple is only the start, and by next year we will see exponential improvements like 2x faster and 20% less wattage usage. I feel like we have seen the exponential performance improvements, this is it. Any thing in the future will be spec bumps. We will not see improvements on the M1 chip like we have seen on the A-series on the iphones every couple of years. Of course, they have not released their PRO chips yet which should be much more powerful but once those are released its spec bumps from there.

Is there any indication that this is not the case and this is only the first generation which are usually weak(like on original iPad) and will see crazy boosts in the future?

There’s a 32core chip out there in the wilds. Guessing for the Mac Pro

 

[thingstoponder]

You're right, Apple has seen a 300% increase in performance over the same period that Intel has seen 50%.

BUT, Apple's performance increases ARE linear. Moore's Law is exponential. Apple has not been keeping up with Moore's law until now because Apple's performance gains have been seriously constrained by the power and thermal limitations (and therefore core count and CPU frequency) and die/package size due to the fact that the processors have been purposefully optimised for battery life, passive cooling on size constrained mainboards. They don't have these constraints anymore.

Once their desktop devices (Mac mini excluded) are introduced it gets even better - they have the luxury of much more efficient active cooling than a laptop/SFF can provide, permanent mains power and a big mainboard to take a larger package with lots of CPU, GPU cores and other co-processors.

Moore’s law has nothing to do with performance. It has to do with transistor density which Apple has no control over as they don’t manufacture integrated circuits.

They were passively cooled devices which had to meet certain battery life requirements with a tiny mainboard. You're still thinking about these chips as if they a mobile only SoC with all the constraints that go along with that. Constraints being physical size, power consumption and the resulting heat. They were getting that 20% increase year on year despite of these constraints. Constraints which are now gone for their desktop Macs.

They can scale now scale these cores both vertically (clock speed) and horizontally (core count).

I'd like to hear your explanation as to why, right now, Apple couldn't create a "desktop" grade CPU/package for an iMac - say increase power consumption, bump the clock speed by 10-20% and put anywhere from 8 to 32 performance cores in it? Forget about 2x faster, this configuration would absolutely destroy that.

Intel can put these kind of core counts on their CPUs and they do not have anywhere near the thermal performance of the M1.

Seriously - provide a single technical reason why this would not be possible right now? Just ONE.

When I talk about a CPU getting faster I’m talking about single core in a fixed powerenvelope. It doesn’t impress me if you just add cores, it makes it faster in some tasks but not basic performance than a user would perceive. When I said 20% I meant the increases we’ve seen year over year in the iPhone since the a11. They’ve had the same 2+4 core arrangement and every year they get about 20% faster in single core. That’s great, but not 2x.

 

[thingstoponder]

You're right, Apple has seen a 300% increase in performance over the same period that Intel has seen 50%.

BUT, Apple's performance increases ARE linear. Moore's Law is exponential. Apple has not been keeping up with Moore's law until now because Apple's performance gains have been seriously constrained by the power and thermal limitations (and therefore core count and CPU frequency) and die/package size due to the fact that the processors have been purposefully optimised for battery life, passive cooling on size constrained mainboards. They don't have these constraints anymore.

Once their desktop devices (Mac mini excluded) are introduced it gets even better - they have the luxury of much more efficient active cooling than a laptop/SFF can provide, permanent mains power and a big mainboard to take a larger package with lots of CPU, GPU cores and other co-processors.

Moore’s law has nothing to do with performance. It has to do with transistor density which Apple has no control over as they don’t manufacture integrated circuits.

They were passively cooled devices which had to meet certain battery life requirements with a tiny mainboard. You're still thinking about these chips as if they a mobile only SoC with all the constraints that go along with that. Constraints being physical size, power consumption and the resulting heat. They were getting that 20% increase year on year despite of these constraints. Constraints which are now gone for their desktop Macs.

They can scale now scale these cores both vertically (clock speed) and horizontally (core count).

When I talk about a CPU getting faster I’m talking about single core in a fixed powerenvelope. It doesn’t impress me if you just add cores, it makes it faster in some tasks but not basic performance than a user would perceive. When I said 20% I meant the increases we’ve seen year over year in the iPhone since the a11. They’ve had the same 2+4 core arrangement and every year they get about 20% faster in single core. That’s great, but not 2x.

I'd like to hear your explanation as to why, right now, Apple couldn't create a "desktop" grade CPU/package for an iMac - say increase power consumption, bump the clock speed by 10-20% and put anywhere from 8 to 32 performance cores in it? Forget about 2x faster, this configuration would absolutely destroy that.

Intel can put these kind of core counts on their CPUs and they do not have anywhere near the thermal performance of the M1.

Seriously - provide a single technical reason why this would not be possible right now? Just ONE.

You misunderstood me. I never said they couldn’t and I’m not a skeptic at all. I was saying that anyone who thinks they will increase the performance by 2x every year is crazy. The OP is claiming something no one ever said. Anyone who is realistic knows they will improve at a more gradual rate of about 20% every year just like the phones. Again, I’m talking about single core speed.[/quote]

 

[leman]

I am really curious about Apple's path for matching external GPU performance for the iMac and Mac Pro. Will they support external GPU cards via PCIe, or have a custom interconnection between GPU and CPU and RAM on a single package? Will we see the end of the huge 300W dGPU cards in PC desktops and the Mac Pro?

One of Apple Silicon main features is unified memory, which is not possible with a classical PCI-e dGPU design. We might see a multi-chip package where CPU/GPU is connected to the same system-level cache and memory controller (like AMD does now with Zen3). I can imagine some interesting packaging technology where the IO die is stacked together with RAM for ultra-low latency.

Notice how the GPU is huge compared to the CPU cores (8 core/16 thread). If Apple wants to compete on GPU as well as CPU cores, *plus* all the custom features on Apple Silicon, either the die would need to be gigantic, or the GPU needs to be separated.

I would suppose that a gaming console needs to be balanced differently — the GPU is certainly the main focus. The die shot you link is vey interesting though. The GPU appears to occupy at least 50% of the die area, that's 180mm2 or 50% more than the entire M1. In comparison, Apple GPU is quite small — not just in absolute terms, but in relative terms. The 8-core M1 GPU with 1024 ALUs occupies less than 1/4 of the chip (that would be around 30mm2). The Xbox chip has 3584 ALUs occupying 180mm2. An Apple GPU with equivalent performance would need 32 cores — and it would fit into 120mm2 — still 50% smaller than the Xbox GPU. If we keep the same basic proportions for the rest of the chip (for more CPU cores, larger cache etc.), an Apple SoC with such a GPU would be 400-480mm2 in size — probably still feasible to manufacture (modern GPU dies are larger).

I wonder whether the relatively small size of M1 GPU is just because of the smaller process, or because Apple GPUs use a more efficient internal design? Navi's "dual compute unit" seems to be very similar to an Apple "core": both contain four 32-wide ALUs...

 

[SlCKB0Y]

When I talk about a CPU getting faster I’m talking about single core in a fixed powerenvelope. It doesn’t impress me if you just add cores

But once we get these CPUs out of a passively cooled, battery powered machine, they can increase the clock speed significantly. Secondly, if they can add a bunch of cores and still have the thermal overhead to not decrease the individual core clock speeds (too much), that should impress you. Historically, the number of cores and their clock speed have been inversely proportional. It will be interesting to see what Apple can do with this.

I was saying that anyone who thinks they will increase the performance by 2x every year is crazy.

Ok, I did misunderstand you - completely. I was referring to initially, say over the next 24 months. I'll predict they double their single core performance within 2 years, in time for the Mac Pro before they level out to their previous rate.

 

[dmccloud]

I think that you have to place the M1 on its own chart/line. It is the modern day equivalent to the A9 in that chart comparing Apple's processors to the Intel lineup. The question is how will the performance increase from generation to generation compared to either Intel or Apple's A Series. Since none of the remaining models to be upgraded to the M series have the same thermal constraints as the MacBook Air, iPhone, or iPad, that trend could lead to bigger performance gains between generations. Apple could also trend conservatively and maintain the same incremental rate of increase as the A series, which would still put Apple Silicon on pace to rapidly surpass Intel across the board.

But once we get these CPUs out of a passively cooled, battery powered machine, they can increase the clock speed significantly. Secondly, if they can add a bunch of cores and still have the thermal overhead to not decrease the individual core clock speeds (too much), that should impress you. Historically, the number of cores and their clock speed have been inversely proportional. It will be interesting to see what Apple can do with this.



Ok, I did misunderstand you - completely. I was referring to initially, say over the next 24 months. I'll predict they double their single core performance within 2 years, in time for the Mac Pro before they level out to their previous rate.

I think that the comment about the historical correlation between clock speeds and core counts is not as relevant to Apple's chips, because it is a completely different approach to design. Intel and AMD use a homogeneous design, where each core is structurally and functionally identical. Apple's chips use the BIG.little setup, where some cores are designed for performance and the rest are focused on efficiency. As a result, each set of cores can run at different speeds. Assuming the M1X (or whatever Apple calls it) is either an 8/4 or 12/4 configuration (performance/efficiency), they could easily maintain current clock speeds without requiring a significant boost in either power demand or cooling requirements because these systems are much more power efficient than their Intel counterparts.

 

[Homy]

TFLOPS is not everything but the rumored 128-core GPU would be crazy fast. It would be faster than any GPU on the market, including GF 3090!!

M1 8 GPU cores 2.6 TFLOPS
M? 16 GPU cores 5.2 TFLOPS
M? 32 GPU cores 10.4 TFLOPS
M? 64 GPU cores 20.8 TFLOPS
M? 128 GPU cores 41.6 TFLOPS

Radeon Pro 5700 6.2 TFLOPS
Radeon Pro 5700 XT 7.7 TFLOPS
Radeon Pro Vega II 14.06 TFLOPS
Radeon Pro Vega II Duo 2x14.06 TFLOPS
GF RTX 3060 14.2 TFLOPS
GF RTX 3060 Ti 16.2 TFLOPS
Radeon RX 6800 16.2 TFLOPS
GF RTX 3070 20.3 TFLOPS
Radeon RX 6800 XT 20.7 TFLOPS
Radeon RX 6900 XT 23 TFLOPS
GF RTX 3080 29.8 TFLOPS
GF RTX 3090 35.6 TFLOPS

 

[Pressure]

This sounds more reasonable. Maybe what the other poster meant was that Apple will add more cores in each iteration, so an 8+4 or 12+4 core SoC could easily be twice as fast (for multi-core workloads).

It will be really interesting to see how far Apple can scale vertically. Larger dies have a greater chance of failures, and there comes a point where the GPU would need to be on a separate die, or for the CPU cores to be placed on separate chiplets similar to AMDs Epyc & Rome CPUs

AFAIK the ARM Neoverse N1 that is used by Amazon (for their Graviton 2), Marvell and Ampere, supports up to 128 cores on a single die, so this probably isn't an issue for Apple.

I am really curious about Apple's path for matching external GPU performance for the iMac and Mac Pro. Will they support external GPU cards via PCIe, or have a custom interconnection between GPU and CPU and RAM on a single package? Will we see the end of the huge 300W dGPU cards in PC desktops and the Mac Pro?

Here an image to the AMD-designed SoC in the Microsoft XBox X:
View attachment 1689034

Notice how the GPU is huge compared to the CPU cores (8 core/16 thread). If Apple wants to compete on GPU as well as CPU cores, *plus* all the custom features on Apple Silicon, either the die would need to be gigantic, or the GPU needs to be separated.

360mm2 is not gigantic by any means. It just shows Apple have ample of room to expand the 120mm2 M1 chip.

 

[SlCKB0Y]

Apple's chips use the BIG.little setup, where some cores are designed for performance and the rest are focused on efficiency.

But would they still stick to this model in machines with mains power? Could they not just use only performance cores in the iMac and MacPro? or at least not have them in a 1:1 ratio?

 

[leman]

But would they still stick to this model in machines with mains power? Could they not just use only performance cores in the iMac and MacPro? or at least not have them in a 1:1 ratio?

Efficiency cores are useful in a desktop machine as well. They occupy very little die space and they barely register on the thermal budget of the chip. Without efficiency cores you will need to run low-priority tasks on the main cores, taking computational resources away from where they are needed more.

Personally, I doubt that Apple will include more than 4 efficiency cores, as that's more than plenty for doing background work. My guess is that we will see 8+4 configuration next, and then potentially up to 32+4 for the Mac Pro.

 

[dmccloud]

But would they still stick to this model in machines with mains power? Could they not just use only performance cores in the iMac and MacPro? or at least not have them in a 1:1 ratio?

It wouldn't be a 1:1 ratio, that's where the 8/4 and 12/4 configurations come from. Because of the way the ARM ISA works, BIG.little actually works better than the homogeneous cores used in x86 CPUs. There is a focus on performance for sure, but power efficiency is also a hallmark of ARM architecture. Anybody can get more performance by bumping up the wattage pushed to the CPU , look at Intel's product line over the last 4 years for proof of that.

 

[MacBH928]

This sounds more reasonable. Maybe what the other poster meant was that Apple will add more cores in each iteration, so an 8+4 or 12+4 core SoC could easily be twice as fast (for multi-core workloads).

It will be really interesting to see how far Apple can scale vertically. Larger dies have a greater chance of failures, and there comes a point where the GPU would need to be on a separate die, or for the CPU cores to be placed on separate chiplets similar to AMDs Epyc & Rome CPUs

AFAIK the ARM Neoverse N1 that is used by Amazon (for their Graviton 2), Marvell and Ampere, supports up to 128 cores on a single die, so this probably isn't an issue for Apple.

I am really curious about Apple's path for matching external GPU performance for the iMac and Mac Pro. Will they support external GPU cards via PCIe, or have a custom interconnection between GPU and CPU and RAM on a single package? Will we see the end of the huge 300W dGPU cards in PC desktops and the Mac Pro?

Here an image to the AMD-designed SoC in the Microsoft XBox X:
View attachment 1689034

Notice how the GPU is huge compared to the CPU cores (8 core/16 thread). If Apple wants to compete on GPU as well as CPU cores, *plus* all the custom features on Apple Silicon, either the die would need to be gigantic, or the GPU needs to be separated.

Could we see an Apple Silicon GPU based on the M1 technology? Or that works different and we need x86?

When I talk about a CPU getting faster I’m talking about single core in a fixed powerenvelope. It doesn’t impress me if you just add cores, it makes it faster in some tasks but not basic performance than a user would perceive. When I said 20% I meant the increases we’ve seen year over year in the iPhone since the a11. They’ve had the same 2+4 core arrangement and every year they get about 20% faster in single core. That’s great, but not 2x.

How do they increase 20% yearly? I thought to increase speed you need to add more transistors, cores, or make them closer by nanometers? What I am asking is what do they do yearly that makes it 20% faster that they couldn't do last year?

 

[brig2221]

Some people seem to have the idea that the M1 released by Apple is only the start, and by next year we will see exponential improvements like 2x faster and 20% less wattage usage. I feel like we have seen the exponential performance improvements, this is it. Any thing in the future will be spec bumps. We will not see improvements on the M1 chip like we have seen on the A-series on the iphones every couple of years. Of course, they have not released their PRO chips yet which should be much more powerful but once those are released its spec bumps from there.

Is there any indication that this is not the case and this is only the first generation which are usually weak(like on original iPad) and will see crazy boosts in the future?

I completely agree with you and is one of the main reasons why I decided to jump in Gen 1. I think two things are going to happen going forward. The spec bumps/updates will be every 18-24 months, and they will offer gains in both power and efficiency, but certainly nothing monumental (IMHO).

Point being, if you want a new MBA or MBP now, get it now, as it more than likely won't have been worth it to wait an additional 18-24 months for M2, which I think will only be incrementally better.

 

[acidfast7_redux]

I completely agree with you and is one of the main reasons why I decided to jump in Gen 1. I think two things are going to happen going forward. The spec bumps/updates will be every 18-24 months, and they will offer gains in both power and efficiency, but certainly nothing monumental (IMHO).

Point being, if you want a new MBA or MBP now, get it now, as it more than likely won't have been worth it to wait an additional 18-24 months for M2, which I think will only be incrementally better.

All of these improvements are linear. Prove me wrong.

 

[brig2221]

All of these improvements are linear. Prove me wrong.

I'm not attempting to prove anyone wrong, simply laying out my opinion. I guess my point is, the MBA or MBP is my sweet spot computing device. I don't do any heavy video or photo editing, so an iMac and/or the 4 Thunderbolt variants of the 13" and 16" MBP's are overkill for me and extremely expensive to boot, so I'm narrowing my thoughts here specific to the M1 chip, not any of the larger more powerful chips that are coming next year.

That being the case, I don't think we see an M2 chip until mid 2022 at the earliest, and I think it will offer performance gains in the 15-25% range. That's no small potatoes, but I view it a lot like I do the A13 to the A14 mobile chip. I have an iPhone 11 with the A13 chip, and the thing still FLIES and will do so for the next couple of years. Design overhaul aside, I think it would have been silly for me to delay purchasing an iPhone 11 for the future prospect of an iPhone 12 based on guesstimated gains in chip performance of the A14.

I think the same ultimately will be said about the M1 respective to the M2 in that it will still fly on be more than capable of running MACOS and most applications blazingly fast like it does today, which I thin will be more than enough for people with my use case, which is a vast majority of the market in my opinion (people not on these forums).

To wrap it all up, my point was for those in the market for a MBA or MBP now or soon, I think they would be doing themselves a disservice to wait an additional 18 months for an M2 that won't "change the game again" in my humble opinion.

I'm excited at the prospects of Apple Silicon's future and love the conversation!