First Geekbench Score Surfaces for MacBook Pro M1 Max With 2x Faster Multi-Core Performance Compared to M1

[VArase]

I think that's due to the chip supply situation, that probably they would have hoped to release them sooner. But now I fear that they'll delay M2 not to steal the light from the M1Xs, because M2 macs will be faster for most people than those Pro. So either apple plays marketing-wise, and they delay M2Xs to 2023, risking to be late vs intel, or they'll do the best technically meaning M2Xs next year and regain their absolute lead. But I fear they'll choose marketing approach, especially in those chip constrained times.

Rumor had it M1 Pro and Max were ready to go around WWDC, but they had problems getting the mini-LED panels (which is the reason for the delay).

I think they'll probably release the new redesigned MacBook Air early next year with the M2 SoCs.

The real question is what will they put in the iMac replacement.

 

[nathansz]

Apple chips have always been a single core leader, that's the main reason why stuff like UI, apps and webpages are snappier on apple devices. Apple always had better core design than its competitors.

which chips and core design when?
ppc was a long time ago

 

[bobcomer]

Apple chips have always been a single core leader, that's the main reason why stuff like UI, apps and webpages are snappier on apple devices.

Single thread doesn't really help that, and Macs UI isn't snappier, it's actually quite slow to me (maybe because I'm used to more cores on Windows machines). As for browsers, they all suck speed-wise on every platform. (it's how they work, it's mostly interpreted stuff)

It's just weird so see them make no progress on this, sure they could cram more cores into the package, but they use exactly the same core? A year of no core design improvments? That sounds like what intel had been doing.

Like I said, it's the same core. I would think that they wouldn't make any big changes, as it simplifies the design and keeps cost down. Their next core is the A15's core.

You could clock it different, but Apple seems to have chosen not to for whatever reason. (probably a yield thing, but who knows) Since Apple's on their won little island CPU-wise, there's not much pressure to change single core performance, but multicore, yep, Pro's needed more than the base M1 provided, so that's what they did. Increasing single core performance really wouldn't help them near as much as giving them more cores. (and the new iGPU)

A year of no core design improvments? That sounds like what intel had been doing.

Wow. It's not nearly so easy as you think, it takes time, real time.

We will see. But given the ambition and the hype train on apple silicon, I expected better from apple than this.

I expected better than the notch too, but we have what we have. (or not, we don't have to buy it!)

A lot of the hype train on Apple silicon is here in these type forums and the media, and the benchmarking community, outside in the real world, not so much. They don't even have a double digit market share in the PC space.

 

[VArase]

I hope this is fake results because it is pretty underwhelming. Why is nobody mentioning this?
There is pretty much 0% improvement on single core performance, really? After 1 year apple has made no progress on that?
The multi-core improvement is also not 70% better than M1 as apple said in the event.
I think this is fake. I am expecting 2000+ single and 13000 on multi.

I think these were originally suppose to be released much earlier but were delayed due to a lack of mini-LED panels - one of my primary reasons for getting this machine.

It's not like the Apple Silicon Team has been sitting on their hands - the A15 has about a 10-15% increase in speed, though a lot of that is probably due to use of TSMC's N5P node.

They fiddled with caches, but a lot of the improvements went into efficiency improvements (especially in the efficiency cores) which is one of the reasons the iPhone 13 has such great battery life. You can only go so wide before going wider is not going to yield much benefit - expect better numbers for A16/M3 when there will be another process shrink down to 3nm or 4nm, depending on whether 3nm is ready when it comes time for production.

There appears to have been a lot of work on the non-CPU aspects of the SoC with faster video encode/decode blocks and support for ProRes as well as improvements in the speed of the NPU and GPUs.

These speed and energy efficiency improvements should bear fruit in the M2 which is expected to increase the GPU and NPU core counts, and in the M2 SoCs which will benefit from A15's advancements in the high performance Avalanch and and high efficiency Blizzard cores.

Apple, like any other silicon designer, will inevitably run into Moore's wall - and when that happens you can keep batting your head against the wall or can examine the target workflows you're trying to optimize and find other areas where process improvements will yield more productive results. Efficiency improvements will yield better battery life, but will also allow you to include more cores in a cluster without boosting your heat signature.

A15 was able to boosts clocks a tad without the consequent exponential rise in energy consumption experienced by its rivals.

Apple is perhaps fortunate that its biggest products are the iPhone and iPad, which allows them to concentrate on the the efficiency which keeps them from seeing boosted clocks as the cure for all ills - which allows their hardware designs to concentrate on more than boosting clocks (which increases energy use exponentially) to achieve speed only to then turn around and expend more energy to move the heat off the chip and out of the chassis. This is the trap that the Wintel homogeny have been in for a good while now - which is why there is legislation in several states mandating computer efficiency in these changing times of diminishing hydroelectric power and fragile power grids.

Here is the Anandtech article describing the changes in the A15.

 

[bowen1506]

I think these were originally suppose to be released much earlier but were delayed due to a lack of mini-LED panels - one of my primary reasons for getting this machine.

It's not like the Apple Silicon Team has been sitting on their hands - the A15 has about a 10-15% increase in speed, though a lot of that is probably due to use of TSMC's N5P node.

They fiddled with caches, but a lot of the improvements went into efficiency improvements (especially in the efficiency cores) which is one of the reasons the iPhone 13 has such great battery life. You can only go so wide before going wider is not going to yield much benefit - expect better numbers for A16/M3 when there will be another process shrink down to 3nm or 4nm, depending on whether 3nm is ready when it comes time for production.

There appears to have been a lot of work on the non-CPU aspects of the SoC with faster video encode/decode blocks and support for ProRes as well as improvements in the speed of the NPU and GPUs.

These speed and energy efficiency improvements should bear fruit in the M2 which is expected to increase the GPU and NPU core counts, and in the M2 SoCs which will benefit from A15's advancements in the high performance Avalanch and and high efficiency Blizzard cores.

Apple, like any other silicon designer, will inevitably run into Moore's wall - and when that happens you can keep batting your head against the wall or can examine the target workflows you're trying to optimize and find other areas where process improvements will yield more productive results. Efficiency improvements will yield better battery life, but will also allow you to include more cores in a cluster without boosting your heat signature.

A15 was able to boosts clocks a tad without the consequent exponential rise in energy consumption experienced by its rivals.

Apple is perhaps fortunate that its biggest products are the iPhone and iPad, which allows them to concentrate on the the efficiency which keeps them from seeing boosted clocks as the cure for all ills - which allows their hardware designs to concentrate on more than boosting clocks (which increases energy use exponentially) to achieve speed only to then turn around and expend more energy to move the heat off the chip and out of the chassis. This is the trap that the Wintel homogeny have been in for a good while now - which is why there is legislation in several states mandating computer efficiency in these changing times of diminishing hydroelectric power and fragile power grids.

Here is the Anandtech article describing the changes in the A15.

I must say I was quite hyped about apple silicon and the prospect that it might overtake the somewhat stagnant x86 and PC market. Maybe I also got affected by the media hype and apple's events full of ambition and confidence. I was hoping for another at least semi big improvement in the CPU department. And I do think if apple wants to fill their ambition and truly make their apple silicon the best chip in the world as they say and a household name, they really gotta build up steam and deliver that improvement year over year. Now it seems that the reality sinks in, same CPU and GPU cores, just more of them packed together. I don't care about the encoders they added and how many 8k videos stream it can render at the same time. It seems like this macbook is still only good for the niche market of "creative pros". I've already got a very powerful PC and esxi server at home. For the prices of a reasonably spec'ed MBP M1 pro/max I could buy like 3 base model razerbook 13.

 

[paulchiu]

I have my doubts about the 28 core thing, but M1 Max should have two ProRes encode/decode blocks and can encode whereas from what I understand the afterburner could not.

Actually, M1 Max should have 2 video encode/decode blocks for H.264 and H.265 also, so encodes/decodes/transcodes should be pretty fast.

Of course, nobody has anything (to my knowledge) in their hot little hands yet so we'll have to see: but if things work out like Apple says - and they've been pretty good about delivering unbelievable results with Apple Silicon - there will be a lot of looonng nights and TUMS in store for anyone involved in the Wintel homogeny.

$INTC stock today, ouch!

 

[bowen1506]

I think that's due to the chip supply situation, that probably they would have hoped to release them sooner. But now I fear that they'll delay M2 not to steal the light from the M1Xs, because M2 macs will be faster for most people than those Pro. So either apple plays marketing-wise, and they delay M2Xs to 2023, risking to be late vs intel, or they'll do the best technically meaning M2Xs next year and regain their absolute lead. But I fear they'll choose marketing approach, especially in those chip constrained times.

That's exactly what I thought too. Because the architecture improvement generation is really the time when people should spend the money, not a core reuse but added some pro-res encoder generation. I think they've meticulously played this game to always give people the good yet imperfect product so people always have a reason to buy now and also buy next year's product, and I am not liking it. Last year we got a new chip but same old design, this year we got a new design but same old core from last year? What if people spend big on this long waited 14/16 inch redesign MBP only to have it obsolete next year with M2 chip when they "regain the absolute lead" as you said? I've waited 2 years for a good apple silicon macbook to try out and I am not even a mac user lol. Not to mention if apple plays this game it gives time for intel to catch up, so apple silicon might not be that superior anyways.

 

[VArase]

I must say I was quite hyped about apple silicon and the prospect that it might overtake the somewhat stagnant x86 and PC market. Maybe I also got affected by the media hype and apple's events full of ambition and confidence. I was hoping for another at least semi big improvement in the CPU department. And I do think if apple wants to fill their ambition and truly make their apple silicon the best chip in the world as they say and a household name, they really gotta build up steam and deliver that improvement year over year. Now it seems that the reality sinks in, same CPU and GPU cores, just more of them packed together. I don't care about the encoders they added and how many 8k videos stream it can render at the same time. It seems like this macbook is still only good for the niche market of "creative pros". I've already got a very powerful PC and esxi server at home. For the prices of a reasonably spec'ed MBP M1 pro/max I could buy like 3 base model razerbook 13.

Took at look at a Razorback Pro 13 and it looks like more of a competitor to the MacBook Air than a competitor to the new MacBook Pros.

At $1699 at B&H, it sports a core-i7, 16 GB RAM, and integrated graphics and likely wouldn't do well in a show off against an M1 MacBook.

At 3*$1699 you'd be talking $5097 which is waaayyy more than I paid for a M1 Max MacBook Pro 16" with 32 GPU cores, 32 GB RAM, and a 2 TB SSD.

The new MacBooks aren't just a marketing term for a nice laptop - they're really designed for a pro or heavy prosumer customer. Single core-wise they're not really that much faster than an M1 MacBook pro for normal consumer computing since they both use the same high performance Firestorm cores. The M1 machines are just about ideal for Harry in accounting or Sally in legal. They probably have all the aspirational power those two would need without having to spend more to buy a prosumer model.

These new models do feature heavy lifting capabilities and are targeted at those who do heavy lifting on a regular basis - but they will be just about the best laptop for media creation and consumption. 1000 NIT maximum brightness with peak brightness of 1600 NITs for HDR content with 120 hz panels will make them capable even of use outdoors.

They have SSDs capable of 7.4 GB/sec and memory access of over 400 GB/sec, probably using 512-bit LPDDR5-6400 with 32 channels giving them faster storage access than pretty much anything out there - including the computers you have at home.

If the laptop you need is something like that Razor, you might consider looking at the MacBook (when it comes out probably in the first half of 2022) with an M2 processor which will probably feature 10-15% increased speed with greater efficiency, a few additional GPU cores, and a couple of more NPU cores and a redesigned chassis.

The M2 will be based on the A15 which lives in the iPhone 13 and has Avalanch high performance and Blizzard high efficiency cores. Here's an Anandtech article profiling these cores in the A15.

Come to think of it, if you were to buy the lowest level 14" MacBook Pro, you'd be getting a M1 Pro processor, with 6 Firestorm high performance cores, 2 high efficiency Icestorm cores, a 14 core GPU, 16 GB unified 200 GB/sec 256-bit LPDDR5-6400 16 Channel RAM, a 16 core Neural engine, the liquid retinal XDR display, a great force touch trackpad, 512 GB high speed SSD, a 67 watt charging brick - which I'd recommend upgrading for $20 to a fast charging 96 watt brick, three Thunderbolt 4 ports, an SDXC port, an HDMI port, a Magsafe 3 port, a high impedance audio jack for $1999.

If you take the +$20 96 watt charging brick and a +$200 SSD upgrade to 1 TB it'd run you $2219 (the SSD is high speed 7.4 GB/sec which you're not going to find in your Razer). You'd be getting a much better machine than that Razorback though it wouldn't run x86 games natively.

This thing would be great for media creation or consumption and would beat the bejesus out the Razorback performance-wise, have a better bigger brighter 14.2" 254 PPI screen, have full HDR DCI-P3 10 bit support for a billion colors with up to 1600 NITs peak brightness when in HDR mode 1000 NIT sustained with 120 hz Promotion support (which is more efficient and varies from 24 hz to 120 hz depending on activity) and a million to one contrast ratio, a 1080p webcam which will produce superior video through the M1 Pro's ISP, great audio with force cancelling woofers, and better battery life under load, and hardware support for H.264, H.265, and ProRes video both encode and decode.

A superior laptop in just about every way except the ability to play x86 games natively which the Razorback can kinda do (though not well). The memory access speed is still probably faster than your home machines since Wintel machines are usually limited to 6 channel if they have really fast memory access.

The multicore speeds would probably lose out to your home desktop/server machines depending on the core count, and may lose out single core depending on how high your home machines push up clocks but keep in mind that this is a laptop - not a desktop machine which can do this on battery outside in the bright sun.

 

[VArase]

$INTC stock today, ouch!

Oh, I don't know. Looks like they closed up 63¢ (but fell $4.86 in after hours trading).

Looks like they're being hit with component shortages - analysts don't give a whit about performance of chips.

Also looks like their margins are dropping and they've abandoned their attempt to acquire SiFive - which I thought was going to be their attempt at expansion into an architecture capable of really wide CPUs (RISC-V).

We'll see how they stand after Thursday's close when they report third quarter earnings.

 

[theorist9]

Here's a few more GB scores, posted today, with about the same single-core score, but a 9-11% higher multi-core score, than what MR posted. They all list a base frequency of 3.22 GHz rather than 24 MHz so it appears GB has resolved that issue. They also differ from the one posted by MR in that they're running 12.0.1 instead of 12.4.

Sources:https://browser.geekbench.com/v5/cpu/10536410
https://browser.geekbench.com/v5/cpu/10536594
https://browser.geekbench.com/v5/cpu/10537943

 

[0134168]

Here's a few more GB scores, posted today, with about the same single-core score, but a 9-11% higher multi-core score. They lists a base frequency of 3.22 GHz rather than 24 MHz so it appears GB has resolved that issue. They also differ from the one posted by MR in that they're running 12.0.1 instead of 12.4.

Sources:https://browser.geekbench.com/v5/cpu/10536410
https://browser.geekbench.com/v5/cpu/10536594
https://browser.geekbench.com/v5/cpu/10537943


View attachment 1874373

View attachment 1874364

View attachment 1874352

Any GB of the M1 Pro?

 

[theorist9]

Any GB of the M1 Pro?

Not as many, but here's a couple. I wouldn't read too much into the lower multi-core vs. the M1 Max at this point, since these are just two data points (plus it's only Geekbench ).

The Max machines are 18,2, while the Pro's are 18,1. A quick Google search didn't tell me whether these are the 14" or 16" models.

 

[0134168]

Not as many, but here's a couple. I wouldn't read too much into the lower multi-core vs. the M1 Max at this point, since these are just two data points (plus it's only Geekbench ).

The Max machines are 18,2, while the Pro's are 18,1. A quick Google search didn't tell me whether these are the 14" or 16" models.




View attachment 1874379

View attachment 1874380

Thank you so much.

 

[creediddy]

I mean, what is the point to compare a desktop to a laptop anyways.

Obvious troll is obvious.

I am no troll. I own 4 iPad's, 2-Macbook Airs, we use 4 iPhones in our house, 2 Apple TV's 4k, AirPods...I am an Apple fanboy. I don't see value of justifying $3500-$6000 for a laptop. Again I prefer Windows desktop over MacOS. I have built over 12 PC's, including dual graphics card setup through SLI and set them up in RAID configuraitons. You get way more value with PC builds than Apple. This has been a fact for years. Mac users use their laptop/PC much differently than Windows users. Sorry no trolling going on...

 

[creediddy]

No more expensive than high end Windows laptops - and probably a vastly superior experience.

Probably beats out the vast majority of desktop computers too. That's a lot to say for a laptop.

1000 NIT normal and 1600 NIT brightness playing HDR content with a billion colors at 120 hz. M1 Max has twin video encode/decode blocks for H.264, H.265, and ProRes so this is gonna be a beast for encoding/decoding/transcoding. The best speakers in the business (these are improved on the 2019 16" which already had the best speakers in the business). This is gonna be a great machine for creation and consumption.

Like any troll you feel like you've got to spec it up to the max so you can then complain about how expensive it is.

Do you have a laptop with 8 TB of expensive 7.4 GB/sec SSD? Why not?

Does your laptop have 64 GB of expensive 400GB/sec memory?

Mid level - huh. Wanna know where the median Windows machine is? Probably still down in el-cheapo ultrabook territory. The machines you're talking about occupy maybe the top 1% or 2% of the Windows market (excluding servers).

You're somehow trapped yourself in the top tier build-your-own bubble, and you have no idea what the laptop universe is like. Most of the PCs in use are commodity levels machines from HP or Dell, and in enterprises Wintel computers are being rapidly replaced by linux-based thin clients.

Oh, and here are some games Andrew Tsai has been playing on his 8 GB 8 GPU core M1 MacBook Air (with no fan).

My 16" MacBook Pro will have a full cooling system, 8/2 CPU cores instead of his 4/4, 32 GPU cores instead of his 8, and 32 GB RAM instead of his 8 GB, 1000 NIT display with 1600 in HDR mode and a 120 fps display instead of his 400 NIT non-HDR 60 hz display. You think that will change the experience?

You don't have to justify a $3500 purchase with me. I understand your argument as I provided the range of $3500-$6000 to configure the MBP. I didn't say the MBP cost $6000. Again I am an Apple fanboy. I don't have to agree that the M1 Max is the wholly grail and is the best in the world like you. You can get the same performance in a Windows laptop for half the price. Game for game most highend Windows laptops can run 4k games effectively and efficiently on a GTX3070. Whats impressive is that Apple has managed to place the CPU and GPU on a single subtrate rather than a dedicated CPU and GPU. No need to talk with me about display brightness, etc...bottom line, a $1700-$2000 windows laptop with GTX3070/3080 will perform just as well as the MBP at $3500. Sorry!

 

[JahBoolean]

You don't have to justify a $3500 purchase with me. I understand your argument as I provided the range of $3500-$6000 to configure the MBP. I didn't say the MBP cost $6000. Again I am an Apple fanboy. I don't have to agree that the M1 Max is the wholly grail and is the best in the world like you. You can get the same performance in a Windows laptop for half the price. Game for game most highend Windows laptops can run 4k games effectively and efficiently on a GTX3070. Whats impressive is that Apple has managed to place the CPU and GPU on a single subtrate rather than a dedicated CPU and GPU. No need to talk with me about display brightness, etc...bottom line, a $1700-$2000 windows laptop with GTX3070/3080 will perform just as well as the MBP at $3500. Sorry!

Account for Form factor, display tech .. and yes you pay the apple tax. The upper bound is when you overpay for memory (noone is arguing that 2k for an SSD is theft).No one with any sense or a most niche use case getting more than 2 TB memory so the upper bound realistically is 3900-4600.

Why do you feel the need to apologize, macs are premium products, nothing new under the sun.

Could you link those 3080 laptops ?

 

[theorist9]

No need to talk with me about display brightness, etc...bottom line, a $1700-$2000 windows laptop with GTX3070/3080 will perform just as well as the MBP at $3500. Sorry!

Except:
1) For me the performance is as much about the software as the hardware. And I'm more productive, and have a far more pleasant user experience, working in MacOS than Windows (I currently use both). In that sense, a Windows PC will never be able to perform at the same level as the Mac, regardless of hardware. That's a lot of what you're paying for when you buy a Mac instead of a PC. Looking only at hardware misses this.

2) We'll know more when these machines are released into the wild, but: While a 3070/3080 laptop should approximately equal the GPU performance of the M1 Max, I don't think there is any mobile Intel or AMD processor that equals its CPU performance. And the discrepancy in performance should become even larger when you are running on battery.

3) Performance is also about usability, including noise and battery life. I expect any 3070/3080 laptop will be much noisier, and have lower battery life, than the M1 Max.

4) Mac support is far better than the support offered by any PC company.

So I would qualify your statement by saying if you only care about GPU performance, then PC laptops provide a much better value (and have a higher performance ceiling, since you can get a 3090, which should outperform the M1 Max). Further, if your main criterion for a laptop is GPU performance, then there's a decent chance you're a high-end gamer, in which case you should probably be getting a PC anyways (and probably a desktop PC instead of a laptop, unless you travel to game with friends, for competitons, etc.).

Also, is it really the case that you can get a 3080 laptop with the M1 Max's 400 GB/s memory bandwidth and 7.6 GB/s SSD for the $1700-$2000 you quote? [The memory bandwith and SSD speed might be available PC laptops, but I suspect they'd have to be pretty high-end devices.]

 

[VArase]

You don't have to justify a $3500 purchase with me. I understand your argument as I provided the range of $3500-$6000 to configure the MBP. I didn't say the MBP cost $6000. Again I am an Apple fanboy. I don't have to agree that the M1 Max is the wholly grail and is the best in the world like you. You can get the same performance in a Windows laptop for half the price. Game for game most highend Windows laptops can run 4k games effectively and efficiently on a GTX3070. Whats impressive is that Apple has managed to place the CPU and GPU on a single subtrate rather than a dedicated CPU and GPU. No need to talk with me about display brightness, etc...bottom line, a $1700-$2000 windows laptop with GTX3070/3080 will perform just as well as the MBP at $3500. Sorry!

Actually, one of the biggest draws for me outside of performance is that bright mini-LED display.

That great memory and SSD access speed in addition to bypassing the Wintel copying overhead will lead to a nice, fast graphics pipeline.

I know there are OLED panels out there which purport to do HDR, but they're generally limited to about 600 NITs vs. 1600 peak brightness and 1000 sustained, and I don't know if any of them are 120 hz adaptive refresh. Great for media creation and consumption.

That in combination with a 1080p webcam makes for a great teleconferencing experience.

Hell, the 3080 probably performs better than the MacBook Pro - graphics-wise, not CPU performance-wise - unless you actually use it as a laptop and pull the plug out of the wall. Let's face it, that machine is not a laptop - it's a portable desktop.

 

[mjtomlin71]

I hope this is fake results because it is pretty underwhelming. Why is nobody mentioning this?
There is pretty much 0% improvement on single core performance, really? After 1 year apple has made no progress on that?
The multi-core improvement is also not 70% better than M1 as apple said in the event.
I think this is fake. I am expecting 2000+ single and 13000 on multi.

Single core should remain the same as it's the same cores as the M1, Firestorm (p-cores) and Icestorm (e-cores). Multi-core is up by 70% because Apple doubled the p-cores. (Newer Geekbench scores are showing this; ~12,800 is highest so far).

Apple did make progress over the year with their newest cores, the Blizzard (p-cores) and Avalanche (e-cores). They are both much more efficient (i.e. cooler; hence their names) and also boosted performance at the same time, especially in the e-cores. They are used in the A15, not in these M1's. They will make their appearance with the M2 generation, starting in redesigned models of last year's M1 systems. Those systems will most likely be released throughout next year.

I do believe at some point Apple will release the rumored M1 "Ultra" or "Extreme" (probably during WWDC) in a "pro" iMac and then a new Mac Pro at the end of the year (Fall). This SoC is supposed to be huge... 20/40 core CPUs and 64/128 core GPUs.

 

[mjtomlin71]

I must say I was quite hyped about apple silicon and the prospect that it might overtake the somewhat stagnant x86 and PC market. Maybe I also got affected by the media hype and apple's events full of ambition and confidence. I was hoping for another at least semi big improvement in the CPU department. And I do think if apple wants to fill their ambition and truly make their apple silicon the best chip in the world as they say and a household name, they really gotta build up steam and deliver that improvement year over year. Now it seems that the reality sinks in, same CPU and GPU cores, just more of them packed together. I don't care about the encoders they added and how many 8k videos stream it can render at the same time. It seems like this macbook is still only good for the niche market of "creative pros". I've already got a very powerful PC and esxi server at home. For the prices of a reasonably spec'ed MBP M1 pro/max I could buy like 3 base model razerbook 13.

First of all, this is the first generation of Apple's Mac SoC's. They are not going to take over the industry right off the bat... it took several years to build up their A-series and surpass the industry. Give it at least until their transition is over!

And yes, I think you've got it! Apple's "pros" are creatives! That's their core pro market and always has been. Their systems are going to be video/audio centric. People need to get over it.

And again, Apple is not sitting still... they just released the A15... With... Guess what? A new generation of processing cores! New CPU, GPU, ANE, ISP, etc cores.

The CPU's in the Pro/Max are 70% more performant than M1, that puts them in the range of higher end CPU's from AMD and Intel. Near the top 20 scoring CPU's on Geekbench... With a CPU that tops out at 30W!!! The comparable CPU is an Intel Core i9-7940X with a TDP 165W with "boost" levels jump well passed 200W.

 

[0134168]

First of all, this is the first generation of Apple's Mac SoC's. They are not going to take over the industry right off the bat... it took several years to build up their A-series and surpass the industry. Give it at least until their transition is over!

And yes, I think you've got it! Apple's "pros" are creatives! That's their core pro market and always has been. Their systems are going to be video/audio centric. People need to get over it.

And again, Apple is not sitting still... they just released the A15... With... Guess what? A new generation of processing cores! New CPU, GPU, ANE, ISP, etc cores.

The CPU's are 70% more performant, that's put them in the range of higher end CPU's from AMD and Intel. Near the top 20 scoring CPU's on Geekbench... With a CPU that tops out at 30W!!! The comparable CPU is an Intel Core i9-7940X with a TDP 165W with "boost" levels jump well passed 200W.

Wise words.

 

[VArase]

First of all, this is the first generation of Apple's Mac SoC's. They are not going to take over the industry right off the bat... it took several years to build up their A-series and surpass the industry. Give it at least until their transition is over!

And yes, I think you've got it! Apple's "pros" are creatives! That's their core pro market and always has been. Their systems are going to be video/audio centric. People need to get over it.

And again, Apple is not sitting still... they just released the A15... With... Guess what? A new generation of processing cores! New CPU, GPU, ANE, ISP, etc cores.

The CPU's are 70% more performant, that's put them in the range of higher end CPU's from AMD and Intel. Near the top 20 scoring CPU's on Geekbench... With a CPU that tops out at 30W!!! The comparable CPU is an Intel Core i9-7940X with a TDP 165W with "boost" levels jump well passed 200W.

The A15 cores are not 70% more performant ... I believe they're more like 10-20% more performant (excluding efficiency increases). You've got to get over the performance increases we've seen due to the transition from Intel to AArch64: we're not going to see those numbers every year. They're a product of the transition, and you have to temper expectations to more reasonable levels.

Here is an article from Anantech detailing the changes in A15:

The Apple A15 SoC Performance Review: Faster & More Efficient

www.anandtech.com

The higher performance cores are already eight wide: you can't really go much wider and expect to derive any additional benefit. Caches are now much bigger, and the M1 Pro/Max have increased memory bandwidth to 200/400 GB/sec to the CPU and IP blocks which means 256/512-bit LPDDR5-6400 and 16/32 channel access to memory which is a big deal.

The M2 family will be based on the Avalanch high performance and Blizzard high efficiency cores, and will probably be built on TSMC's N5P node.

Expect bigger numbers with A16/M3 when Apple will be using TSMC 3nm or 4nm nodes (depending on whether TSMC can iron out the wrinkles in 3nm in time), where process shrink will produce a bigger boost in speed and efficiency.

Remember that the Apple Silicon Team (capitalized out of respect) does not just concentrate on CPU/GPU speed like standard CPU and GPU vendors - they meet regularly with software and product design teams and optimize workflows for future products. This means that they put work into NPUs and other IP blocks to attempt to optimize workflows in everything from video transcode speeds, memory access, Thunderbolt controllers, caches and all the other components which lead to faster and more efficient products.

I'm actually a bit surprised that M1 Pro/Max stayed with a monolithic design - 57 billion transistors is a heck of a big chip, and yields can't be all that great. I was expecting to see at least the GPU cores to be broken out into chiplets which would sit side by side with the main SoC and memory on the package-on-packge, but there are proximity advantages to using a monolithic design and binning does allow them to use SoCs with smaller flaws. I expect at some point Apple will need to break things up a bit - a monolithic design for the Pro seems like just too big of a chip to expect anything like decent yields.

The remaining Macs to be transitioned are the iMac and Pro and possibly the Mini - though I think the Mini's fate is TBD and depends on an adequate chip supply. Certainly if chip supply is limited Apple will want to ration this precious resource and use them in the highest margin machines they make - and the Mini is probably the lowest margin Mac.

As for power: Apple is fortunate that their SoC development always talks place on their Bionic chips for the iPhone and iPad where efficiency is paramount - it keeps them from falling into the x86 trap where higher clocks (and exponentially higher energy use) are the cure for all ills, though I'm wondering how Pro Mode on the M1 Max will be implemented. Is it just a fan boost to allow higher thermals, or a clock boost as well? I guess well see in a Monterey update along with Universal Control.

 

[Xtir]

R.I.P. Intel

say what?

 

[BeatCrazy]

say what?

Which laptop does the i9-9940X go into?

 

[mjtomlin71]

The A15 cores are not 70% more performant ... I believe they're more like 10-20% more performant (excluding efficiency increases). You've got to get over the performance increases we've seen due to the transition from Intel to AArch64: we're not going to see those numbers every year. They're a product of the transition, and you have to temper expectations to more reasonable levels.

I was referring to the CPU in the M1 Pro/Max versus the first M1, not the A15.