Report: Super-Lightweight 12-inch MacBook Powered By Apple Silicon to Launch This Year

[Nicole1980]

Heres what I dont get: with the intel 8 and 10 core imacs people have criticized intel for just 'throwing more cores at the problem'. And yet, here we supposedly have the first apple processor coming in at 12 cores. Why isnt apple criticized for 'throwing more cores at the problem'?

 

[AppliedMicro]

So my guess is that the new ARM 12" MacBook (if it happens) is actually slightly heavier than the current one.

This would be a wasted opportunity to market as the "most lightweight ever" MacBook.
They could even tout it the most light and longest battery-life MacBook.

My guess is that the reported A14X is a great CPU but not nearly as power efficient when clocked at higher frequencies for a laptop.

Does it need to be clocked higher?

The good thing for Apple is that they'll be able to taylor this chip to their needs. Things like video en-/decoding and encryption will be hardware-accelerated - and the low-performance cores might provide substantial benefits for the myriad of low-power processes running on a full desktop OS.

It might not be as efficient on raw computational power - but the latter might not be needed very much for everyday tasks on a ultraportable notebook.

 

[farewelwilliams]

It will be interesting to see how they give the Macbook a decent processor, but not allow it to be as good as the MacBook Pro. Will they start giving processor indicators, such as A14x 1.7 and then an A14x 2.8 for a MBP etc?

20 hours of battery life... Definitely interested to see what the MBP comes in at, and how it holds up to the stuff I need to run on it.

20 hours might be too much. i'd think they would cut back on the battery a bit to target 14-16 hours since not many need a 20 hour battery. most would be asleep before their macbook ran out of juice.

 

[EugW]

Does it need to be clocked higher?

The good thing for Apple is that they'll be able to taylor this chip to their needs. Things like video en-/decoding and encryption will be hardware-accelerated - and the low-performance cores might provide substantial benefits for the myriad of low-power processes running on a full desktop OS.

It might not be as efficient on raw computational power - but the latter might not be needed very much for everyday tasks on a ultraportable notebook.

If A14X they could clock it significantly lower and it would still be a good performer. People forget just how fast A12Z is. A12Z is faster than the vast majority of Intel laptops in use today, and moreover, it’s faster than most new Intel laptops currently for sale.

No, A12Z won’t come anywhere close to competing against the latest i7 or i9, but 99% of the population doesn’t need those, and 99% of the population wouldn’t want chips with that amount of power utilization anyway.

In fact, purely from a CPU performance point of view, they could get away with using just A14. The main problems I see with using A14 in this context is that the GPU is weaker (but they could have accounted for this in the new design), and marketing. Some people could complain about getting “just” a smartphone SoC in their MacBook, even though that smartphone SoC is the fastest in the world.

 

[theorist9]

This rumor is false, because of simple physics.

For quoted “maximum battery runtime” scenarios, the main battery drain is the screen.

So even if the CPU (essentially the only thing being changed) drained 0 W, getting 20 hours out of the same laptop (same weight, therefore same battery) that previously got 10 hours just won’t happen. Don’t get your hopes up.

I was thinking along the same lines, but wanted to investigate this with actual numbers. Initally I found general numbers for power usage for LED-backlit LCD screens and Intel chip TDPs, but wanted something more concrete. Eventually I found this table, at https://reductionrevolution.com.au/...top-power-consumption-wattage-macbook-pro-air:

Let's consider the 2018 MBA. They said the CPU consumes 8W with the screen at max brightness and everything at idle. If we assume that everything except the screen consumes 1 W, that means the screen consumes ~7 W at max brightness. If we further assume that typical light-load power consumption from the current Intel CPU is, say, 5W, and non-screen/non-CPU power is 1 W then switching to a CPU that can do the same tasks with half the power would increase battery duration by about 30%:

(7 W + 5 W+ 1 W)/(7 W + 2.5 W + 1 W) x 100 – 100 ~ 24%

OTOH, if you are doing heavier work, then the CPU's efficiency has a greater impact on battery life. E.g., suppose your work averages 15 W from the current Intel CPU, and non-screen/non-CPU power power is 2W. Then doubling its efficiency would increase your battery life by:

(7 W + 15 W + 2 W)/(7 W + 7.5 W + 2 W) x 100 – 100 ~ 45%

If we re-did these calculations at a lower screen brightness (say, 5 W instead of 7 W), we'd get 29% and 52%, respectively.

Where a 2-fold increase in efficiency would have a more striking effect is in thermal management (the power consumption for the screen doesn't enter into this, since its waste heat is dissipated into the lid). This would allow either a thinner and lighter laptop with the same performance, or a laptop of the same size but with significantly increased performance. Apple could go for either, or some combination of the two.

 

[Nicole1980]

If A14X they could clock it significantly lower and it would still be a good performer. People forget just how fast A12Z is. A12Z is faster than the vast majority of Intel laptops in use today, and moreover, it’s faster than most new Intel laptops currently for sale.

No, A12Z won’t come anywhere close to competing against the latest i7 or i9, but 99% of the population doesn’t need those, and 99% of the population wouldn’t want chips with that amount of power utilization anyway.

In fact, purely from a CPU performance point of view, they could get away with using just A14. The main problems I see with using A14 in this context is that the GPU is weaker (but they could have accounted for this in the new design), and marketing. Some people could complain about getting “just” a smartphone SoC in their MacBook, even though that smartphone SoC is the fastest in the world.

Until shown otherwise, I'll remain skeptical of sustained high performance of the apple chips.
Everyone only references geekbench. But geekbench testing is a sprint ... It gives no indication how the chip will do with a sustained high workload.
I come at this as a professional video editor.
I dont want a chip that can go balls to the wall for 2 minutes, but wilts during a 45 minute render.
So I guess we'll see how apple chips shake out.

Then theres the question of how much software will be optimized to the apple chips, and how well.
So all this incessant cheerleading on these forums about apple chips seems a tad premature to me.

 

[EugW]

Until shown otherwise, I'll remain skeptical of sustained high performance of the apple chips.
Everyone only references geekbench. But geekbench testing is a sprint ... It gives no indication how the chip will do with a sustained high workload.
I come at this as a professional video editor.
I dont want a chip that can go balls to the wall for 2 minutes, but wilts during a 45 minute render.
So I guess we'll see how apple chips shake out.

This is strictly dependent upon power utilization and cooling. If an A14 non-X were used with a proper heatsink in a MacBook, even fanless it would probably do very well at least for many minutes. However, if you really want a machine that will survive at 100% with a 45 minute render, you're better off getting a machine with a proper fan. A12Z with a decent fan would not be a problem at all. I suspect the same will be true with A14X as well, if there is a fan. Fanless is a different story though.

However, I wouldn't be recommending a 12" MacBook as a video rendering machine in the first place anyway.

 

[Nicole1980]

This is strictly dependent upon power utilization and cooling. If an A14 non-X were used with a proper heatsink in a MacBook, even fanless it would probably do very well at least for many minutes. However, if you really want a machine that will survive at 100% with a 45 minute render, you're better off getting a machine with a proper fan. A12Z with a decent fan would not be a problem at all. I suspect the same will be true with A14X as well, if there is a fan. Fanless is a different story though.

However, I wouldn't be recommending a 12" MacBook as a video rendering machine in the first place anyway.

I agree it follows logic to assume the a14 might do well under extended stress with proper cooling.
But at the moment, thats still just speculation.
As is stands, I know my 2019 8-core, vega 48 iMac can handle heavy sustained loads for as long as I ask it to.
Soon enough we'll know for sure whether apple chips can do the same.
All I'm saying is - before getting into a spazzy lovefest with apple silicon, lets see some real world proof.

I just feel like all this cheerleading is way out of line based on the current evidence. Evidence which is mostly speculative.

 

[chucker23n1]

Heres what I dont get: with the intel 8 and 10 core imacs people have criticized intel for just 'throwing more cores at the problem'. And yet, here we supposedly have the first apple processor coming in at 12 cores. Why isnt apple criticized for 'throwing more cores at the problem'?

So, the Comet Lake CPU in the 2020 iMacs is the process node from 2014 (Broadwell, 14nm) with the microarchitecture from 2015 (Skylake), with some minor improvements.

It’s a bit like Apple taking the A9 and making an A9Z variant which adds more cores, in 2020.

Now, Intel is working on fixing this: Ice Lake and Tiger Lake are 10nm, and each add major microarchitecture changes. Rocket Lake, some time this winter, will then take some of Tiger Lake and retrofit it in another 14nm CPU. So that’s when, hopefully, Intel will gain some major performance on their desktop (and high-end laptop) CPUs again. But not today.

To be clear, I also think praising how many cores the first ARM Macs are rumored to have is a little misguided and premature. Perhaps Apple will do something similar to Turbo Boost, but if not, I would expect an 8+4-Core arrangement on a low-end MacBook to have a lot of idling cores. The code that takes heavy advantage of parallelism simply isn’t there. In specialized applications like compiling and video editing, sure, but those aren’t typical uses of a 12-inch MacBook.

My guess is either these rumors are wrong, or there’s another shoe to drop, like Turbo Boost.

 

[Nicole1980]

So, the Comet Lake CPU in the 2020 iMacs is the process node from 2014 (Broadwell, 14nm) with the microarchitecture from 2015 (Skylake), with some minor improvements.

It’s a bit like Apple taking the A9 and making an A9Z variant which adds more cores, in 2020.

Now, Intel is working on fixing this: Ice Lake and Tiger Lake are 10nm, and each add major microarchitecture changes. Rocket Lake, some time this winter, will then take some of Tiger Lake and retrofit it in another 14nm CPU. So that’s when, hopefully, Intel will gain some major performance on their desktop (and high-end laptop) CPUs again. But not today.

To be clear, I also think praising how many cores the first ARM Macs are rumored to have is a little misguided and premature. Perhaps Apple will do something similar to Turbo Boost, but if not, I would expect an 8+4-Core arrangement on a low-end MacBook to have a lot of idling cores. The code that takes heavy advantage of parallelism simply isn’t there. In specialized applications like compiling and video editing, sure, but those aren’t typical uses of a 12-inch MacBook.

My guess is either these rumors are wrong, or there’s another shoe to drop, like Turbo Boost.

Thanks for the reply, but that still doesn’t seem to answer the crux of my question, which is : Regardless of the die shrink size, certain commentators imply that Intel is kind of ‘cheating’ by adding more cores.
So Again, why isn’t the fact that Apple is using 12 cores, even if it’s on a smaller die process, considered ‘cheating’ just the same?

 

[EugW]

Thanks for the reply, but that still doesn’t seem to answer the crux of my question, which is : Regardless of the die shrink size, certain commentators imply that Intel is kind of ‘cheating’ by adding more cores.
So Again, why isn’t the fact that Apple is using 12 cores, even if it’s on a smaller die process, considered ‘cheating’ just the same?

You are complaining about something that isn’t really a thing these days. Just about nobody with a clue thinks multiple cores is bad.

 

[Nicole1980]

You are complaining about something that isn’t really a thing these days. Just about nobody with a clue thinks multiple cores is bad.

I’m not complaining. I’m simply asking a question.
Many commentators such as Marco Armani and Renee Richie have both said derisively that intel is trying to make up for slow processor improvements by ‘throwing more cores at the problem’. They don’t however, level the same gripe against apple which is using 12 cores in their a14 chip.
So I’m simply asking why they seem to criticize intel for doing it while making no such dig at apple.

 

[EugW]

I’m not complaining. I’m simply asking a question.
Many commentators such as Marco Armani and Renee Richie have both said derisively that intel is trying to make up for slow processor improvements by ‘throwing more cores at the problem’. They don’t however, level the same gripe against apple which is using 12 cores in their a14 chip.
So I’m simply asking why they seem to criticize intel for doing it while making no such dig at apple.

It sounds like you have already discovered to whom you should not be listening.

 

[chucker23n1]

Thanks for the reply, but that still doesn’t seem to answer the crux of my question, which is : Regardless of the die shrink size, certain commentators imply that Intel is kind of ‘cheating’ by adding more cores.

Cheating, I don’t know. Only making progress in very specific scenarios, yes. And also in part by temporarily turning heat way above the regular design point.

So Again, why isn’t the fact that Apple is using 12 cores, even if it’s on a smaller die process, considered ‘cheating’ just the same?

If Apple uses the same process and same microstructure and stuffs 12 cores in there’s and says “look how much faster this is (but only for brief bursts and in specialized apps)”, then yup, I think that’s cheating-ish.

 

[chucker23n1]

I’m not complaining. I’m simply asking a question.
Many commentators such as Marco Armani and Renee Richie have both said derisively that intel is trying to make up for slow processor improvements by ‘throwing more cores at the problem’. They don’t however, level the same gripe against apple which is using 12 cores in their a14 chip.

You’re making a as lot of assumptions based on a chip that doesn’t exist.

Odds are a 12-core A14 (if that’s a thing) will not only have more cores, but compared to last year’s A13 will be ~15% faster per core at the same clock.

Comet Lake is only faster due to higher clocks and more cores.

 

[open0]

we need more usb-c port for the next Macbook 12 inch based on A14x

 

[cmaier]

Heres what I dont get: with the intel 8 and 10 core imacs people have criticized intel for just 'throwing more cores at the problem'. And yet, here we supposedly have the first apple processor coming in at 12 cores. Why isnt apple criticized for 'throwing more cores at the problem'?

because their per-core performance trounces Intel. And because those 12 apple cores are heterogenous.

 

[Nicole1980]

because their per-core performance trounces Intel. And because those 12 apple cores are heterogenous.

Per core performance doesnt 'trounce' intel chips in the slightest. What fantasy numbers are you looking at?

 

[cmaier]

Per core performance doesnt 'trounce' intel chips in the slightest. What fantasy numbers are you looking at?

An A12 get around 1115 on single-core benchmark vs. 1024 for core i7 used in 16" MBP.
An A13 gets 1325 single-core geek bench.

 

[EugW]

Per core performance doesnt 'trounce' intel chips in the slightest. What fantasy numbers are you looking at?

To be honest, I think you’re being a little too defensive for your Intel Mac, consciously or not. It’s well known that Apple A series chips have excellent IPC, top tier in fact. This is not news.

The main unknowns right now include how high Apple can clock those chips, or how many cores they can include on a single package with decent power characteristics.

 

[reilenf]

Thats crazy, 12 to 13 have amolst not diff

 

[ascender]

I’m not complaining. I’m simply asking a question.
Many commentators such as Marco Armani and Renee Richie have both said derisively that intel is trying to make up for slow processor improvements by ‘throwing more cores at the problem’. They don’t however, level the same gripe against apple which is using 12 cores in their a14 chip.
So I’m simply asking why they seem to criticize intel for doing it while making no such dig at apple.

I‘m guessing its because of Intel’s well publicised struggles with getting to 10nm and beyond... while other chip makers are showing them up. So Intel keep increasing core count in the meantime...

More cores are great, but they’re not always equal so to speak.

 

[iPadCary]

I certainly wouldn't call <2.2 pounds "super lightweight" ....

 

[theorist9]

because their per-core performance trounces Intel...

An A12 get around 1115 on single-core benchmark vs. 1024 for core i7 used in 16" MBP.
An A13 gets 1325 single-core geek bench.

You and I have had this discussion before, but I'll reiterate, with respect, that these are (as you know) synthetic benchmarks, and thus may not correlate well with real-world performance. I recall a quote from Linus Torvalds specifically warning against using Geekbench for cross-platform comparisons.

You might wish to invoke comparisons done with SPEC06, but this is also a synthetic benchmark. CERN, for instance, found that SPEC06 showed good correlation to its actual workloads in 2009, but that was no longer the case in 2019. Further, SPEC CPU 2017 also didn't correlate well with its current workloads. [https://indico.cern.ch/event/773049...931/3200035/20191104-CHEP2019-BMK-AV-v005.pdf]

So we won't actually know if AS "trounces" Intel in per-core performance until we see a wide range of real-world benchmarks from independent sources. As an industry insider you may have access to that info., but the rest of us can only make judgements based on the data to which we have access; and no reliable (i.e., independent, real-world) comparative info. has yet been publicly released.

 

[deevey]

An A12 get around 1115 on single-core benchmark vs. 1024 for core i7 used in 16" MBP.
An A13 gets 1325 single-core geek bench.

A score that pisses all over all the latest Mac Mini's, with the Fastest i7 Mini only beating it in multicore by a marginal amount.

Hell even the old A9 iPad benches almost twice as fast as the newest Celerons in el-cheopo notebooks.

The A12 might not be fast enough for a Mac Pro, but I'd warrant that even that is more than adequate for a Notebook, MacMini or iMac.

Cant wait to see what an A14 system is capable of.