Rumored pro chip

[cmaier]

Fair enough, but we're in early days with a new architecture and, as you noted, the first models Apple has been replacing are the "consumer" models where the needs are not as great and the limitations of that first generation are not as cramping.

"M1X" will have more Thunderbolt controllers so it will be able to support more TB ports and, by extension, more displays.

Yet M1 already has the same number of thunderbolt controllers as intel macs have, but M1 machines have half the number of ports. So *shrug.*

 

[Stephen.R]

Yet M1 already has the same number of thunderbolt controllers as intel macs have, but M1 machines have half the number of ports. So *shrug.*

Thats also a factor, but only if you're using all-data.

If you use (on an Intel machine) one port for DP/HDMI out, and one for data, there's no "loss" by having 2 ports on 4 lanes of PCIe.

 

[CWallace]

Yet M1 already has the same number of thunderbolt controllers as intel macs have, but M1 machines have half the number of ports. So *shrug.*

This is because Apple's current TB controllers drive one port whereas Intel TB controllers drive two ports.

Not sure why Apple went this way: could be performance, could be licensing, could be expediency. Could be a mix.

 

[m.x]

The more I think of it, the more the added ‘Pro’ seems better. M1 Pro/M2 Pro. They’ve built the branding and positioning.

Maybe they call it P1, X1 or Z1 or something like that? Then the different classes of chips would be clearly distinguishable for the „average“ person. If they call it M1X and introduce the M2 MacBook Air in the fall, many people will think „well 2>1 so M2>M1X“, that wouldn’t be the case with M2 vs. X1. Also, if they call these chips M1 Pro, how do they call the - presumably much more capable - SOC in the Mac Pro? I think this would be too confusing for the average guy („the iPad Pro is a Pro device, why is there a M1 in it and not a M1 Pro?“)…

 

[nothingtoseehere]

Not sure why Apple went this way: could be performance, could be licensing, could be expediency. Could be a mix.

As much as I like Apple products in general: The handling of ports is one of Apple‘s biggest weaknesses IMO.

• it took years over years for them to accept that you sometimes need to stick something into an iPad (Pro),
• that silly dongle hell on Macbooks since 2016 was a bad idea (hopefully to be reverted this year),
• why on earth are there Macbooks with only two ports (or just one with the MacbookNoSuffix)?
• why are here so many threads describing monitor problems on M1 Macs? Even with the Mac mini that obviously needs a monitor (at least in most use cases)! I have (smaller) problems myself, not that bad but so unneccesary,
• And not only M1 Macs; the handling of external monitors with the Intel MBP 16“ seems to be a PITA for many users.

Therefore my resumé is that Apple has a somewhat cultural problem with ports. Apple does not care enough about them.

 

[cmaier]

This is because Apple's current TB controllers drive one port whereas Intel TB controllers drive two ports.

Not sure why Apple went this way: could be performance, could be licensing, could be expediency. Could be a mix.

Based on some tests I’ve seen, looks to me like the bus that drives the controllers has a little less bandwidth than it needs to keep the two ports fully saturated. The Intel Macs can keep two ports fully saturated but not 4. I guess my point is that to add 2 more ports Apple likely needs to increase the bus bandwidth, and they don’t need to add more TB controllers. (Though they may need to multiplex the outputs of the 2 they have).

 

[neilw]

The more I think of it, the more the added ‘Pro’ seems better. M1 Pro/M2 Pro. They’ve built the branding and positioning.

I was thinking along those lines as well, but maybe they'll save those names for the chips that go in the Mac Pro (and maybe iMac Pro, if there is to be such a thing with Apple Silicon). Unless they give those chips a different letter (P1?)

 

[Rickroller]

I was thinking along those lines as well, but maybe they'll save those names for the chips that go in the Mac Pro (and maybe iMac Pro, if there is to be such a thing with Apple Silicon). Unless they give those chips a different letter (P1?)

I guess they could use the same wording as they do for their Mac Pro GPUs. If they’re getting rid of AMD altogether, they can repurpose the Duo name.

M2 Pro

M2 Pro Duo

M2 Pro Quad

Multipliers are at least less confusing than increments like what we have now with 8,12,16,24 and 28 cores. Isn’t the Apple way to separate products into ‘Good, Better, Best’ ...?

 

[Rickroller]

iMac vs iMac Pro would be separated by the inclusion of XDR alone Just on pricing. I’m assuming they’re able to scale the 12.9 iPad Pro screen to the 32 inches of the XDR Display. It seems simpler and therefore hopefully cheaper to build than the current XDR monitor.

$2499 for base iMac + $2499 for a 32 inch XDR Display

$4999 iMac Pro.

 

[Sydde]

They can. But 4 such cores can manage more than 2 such cores. And there are hundreds of low priority tasks running at any given time. Again, it’s pretty clear to me that the new efficiency cores will have a different microarchitecture than the icestorms, and will be able to handle higher workload at equivalent power dissipation.

What I wonder about is, if the Firestorm use twice the power of an Icestorm to run 4 times as fast, can you design the process daemon so that it can identify the jobs that a Firestorm can just chew right through and exit, as opposed to the long-running ones? If such a strategy is employed, the cost of using a high-performance core is mitigated by it getting a job done quickly. If it takes a third the time of an efficiency core using twice the power, you end up a little bit ahead. And Apple has some OS programmers who could all but surely make such a strategy pay off, thus reducing the need for as many low-power cores.

 

[cmaier]

What I wonder about is, if the Firestorm use twice the power of an Icestorm to run 4 times as fast, can you design the process daemon so that it can identify the jobs that a Firestorm can just chew right through and exit, as opposed to the long-running ones? If such a strategy is employed, the cost of using a high-performance core is mitigated by it getting a job done quickly. If it takes a third the time of an efficiency core using twice the power, you end up a little bit ahead. And Apple has some OS programmers who could all but surely make such a strategy pay off, thus reducing the need for as many low-power cores.

MacOS (and iOS) use a ”QoS” technique - in our code, when we create threads, we assign threads to specific queues based on how critical they are. The process/thread scheduler then uses that information to decide what to do.

 

[Krevnik]

What I wonder about is, if the Firestorm use twice the power of an Icestorm to run 4 times as fast, can you design the process daemon so that it can identify the jobs that a Firestorm can just chew right through and exit, as opposed to the long-running ones? If such a strategy is employed, the cost of using a high-performance core is mitigated by it getting a job done quickly. If it takes a third the time of an efficiency core using twice the power, you end up a little bit ahead. And Apple has some OS programmers who could all but surely make such a strategy pay off, thus reducing the need for as many low-power cores.

The power difference between Firestorm and Icestorm is more along the lines of 10x more power, 4x as fast. So using Firestorm to “race to sleep” won’t result in efficiency high enough to beat Icestorm.

You aren’t completely off here, though. “Race to sleep” isn’t a new idea, and has been a mantra for over a decade when optimizing mobile devices for power consumption. The next level of this concept is task coalescing to cut down on how many times the CPU is woken up, and avoid underutilizing cores that are woken up for background tasks to avoid waste. Both Windows Phone and iOS have employed these techniques. Android I’m not certain of, but I’d be surprised if they didn’t have some version of it.

That said, these “Little” cores provide more options and flexibility than can be had with these two previous techniques alone.

 

[dgdosen]

While the estimated die sizes for these new chips is small enough to fit within TSMC's packaging, does anyone think Apple and TSMC might still be working together to become a 'model' of using TSMCs 3D stacking (silicon on silicon)?

Might there be easy footprint/perf/power gains (and challenges) from going more 3D?

Oh man - AMD is talking about 3D stacking with TSMCs "Fabric". I think AMD is 2nd fiddle to Apple form TSMC's strategic point of view. Can you imagine Apple doing something similar and managing thermals in 12-layered chips?

 

[Kpjoslee]

Oh man - AMD is talking about 3D stacking with TSMCs "Fabric". I think AMD is 2nd fiddle to Apple form TSMC's strategic point of view. Can you imagine Apple doing something similar and managing thermals in 12-layered chips?

It is possible Apple's future SoC will implement some form of 3D stacking, but I don't see Apple being in any hurry to implement that kind of design when the memory is already integrated very close to the SoC.

 

[leman]

Oh man - AMD is talking about 3D stacking with TSMCs "Fabric". I think AMD is 2nd fiddle to Apple form TSMC's strategic point of view. Can you imagine Apple doing something similar and managing thermals in 12-layered chips?

Apple has tons of patents with related technologies, so it wouldn’t surprise me if we see something similar in the upcoming chips. In terms of memory microarchitecture and cache technology, M1 is already fairly impressive.

 

[leman]

It is possible Apple's future SoC will implement some form of 3D stacking, but I don't see Apple being in any hurry to implement that kind of design when the memory is already integrated very close to the SoC.

Apple will need more cache and the die space is precious. In fact, cache is more important to Apple than to anyone else given their focus on heterogeneous computing and the fact that their systems have lower aggregate memory bandwidth than traditional systems with similar performance.

Contrary to the popular belief, the only significant advantage of Apple integrating the RAM on package is reduced power consumption (and probably material cost). There is no latency or bandwidth advantages to what Apple is doing.

 

[cmaier]

Apple will need more cache and the die space is precious. In fact, cache is more important to Apple than to anyone else given their focus on heterogeneous computing and the fact that their systems have lower aggregate memory bandwidth than traditional systems with similar performance.

Contrary to the popular belief, the only significant advantage of Apple integrating the RAM on package is reduced power consumption (and probably material cost). There is no latency or bandwidth advantages to what Apple is doing.

Well, there’s always latency advantages. 6 picoseconds per mm, after all.

 

[Kpjoslee]

Apple will need more cache and the die space is precious. In fact, cache is more important to Apple than to anyone else given their focus on heterogeneous computing and the fact that their systems have lower aggregate memory bandwidth than traditional systems with similar performance.

Oh, I am aware, I was thinking more in the lines of integrating HBM2 on SoC package may be a better approach than stacking huge L3 cache on top.

Contrary to the popular belief, the only significant advantage of Apple integrating the RAM on package is reduced power consumption (and probably material cost). There is no latency or bandwidth advantages to what Apple is doing.

I think the physical distance between the RAM and the memory controller actually impacts the memory access latency substantially. M1's integrated DRAM itself has higher latency compared to typical DDR4 present in latest x86 systems so it is hard to make a comparison.

 

[CWallace]

So looks like the new SoC might indeed be the "M1X".

M1X MacBook Pro referenced in metadata for Apple's WWDC YouTube video - 9to5Mac

There were many rumors about a new MacBook Pro coming at WWDC, but it (sadly) didn’t happen. However, metadata attached...

9to5mac.com

Some commenters are claiming it was put in purely for SEO reasons because "naturally people are going to search for 'M1X'" and others noted that iPad OS had a reference to an "A14X" and yet the iPad Pro launched with M1, so it may yet not come true, but personally I think it will be called M1X.

 

[cmaier]

So looks like the new SoC might indeed be the "M1X".

M1X MacBook Pro referenced in metadata for Apple's WWDC YouTube video - 9to5Mac

There were many rumors about a new MacBook Pro coming at WWDC, but it (sadly) didn’t happen. However, metadata attached...

9to5mac.com

Some commenters are claiming it was put in purely for SEO reasons because "naturally people are going to search for 'M1X'" and others noted that iPad OS had a reference to an "A14X" and yet the iPad Pro launched with M1, so it may yet not come true, but personally I think it will be called M1X.

I think they’re trolling.

 

[CWallace]

I think they’re trolling.

Why would Apple troll people?

 

[Falhófnir]

So looks like the new SoC might indeed be the "M1X".

M1X MacBook Pro referenced in metadata for Apple's WWDC YouTube video - 9to5Mac

There were many rumors about a new MacBook Pro coming at WWDC, but it (sadly) didn’t happen. However, metadata attached...

9to5mac.com

Some commenters are claiming it was put in purely for SEO reasons because "naturally people are going to search for 'M1X'" and others noted that iPad OS had a reference to an "A14X" and yet the iPad Pro launched with M1, so it may yet not come true, but personally I think it will be called M1X.

M1X (16 core GPU) and M1Z (32 core GPU) has a nice ring to it, but I think it limits them in terms of adding more options down the line if future chip generations offer more options?

From what they've done so far, something like 'M1 with 8 performance cores' seems as likely as anything (in which case the existing chip would be retconned as 'M1 with 4 performance cores'). Currently they advertise the total core count, but if the more powerful chip offers only two high efficiency cores, 8 vs 10 would end up being a little misleading, perhaps?

 

[leman]

So looks like the new SoC might indeed be the "M1X".

Na, that's just an SEO optimization.

 

[Jorbanead]

M1X (16 core GPU) and M1Z (32 core GPU) has a nice ring to it, but I think it limits them in terms of adding more options down the line

I always saw them using 3 classes of chips. The M1X would be both the 16 and 32 core variants. You’d just specify which variant you want in checkout but Apple would always display “up to 32 GPU cores” in all their marketing. Just how they do with the 7 vs 8 on M1. The M1Z or M1Pro (whatever) would be an even higher-end chip likely for the Mac Pro and also have a few variants of its own.

Though by then it may be on the next gen architecture so M2X or M3 who knows.

Na, that's just an SEO optimization.

Totally could be for sure. That being said, you’d also think they would tag M2 as well. AFAIK Apple hasn’t really done SEO for rumored products before though. I could be wrong.

Edit: some say it’s been removed. Hmm ?

 

[leman]

Totally could be for sure. That being said, you’d also think they would tag M2 as well. AFAIK Apple hasn’t really done SEO for rumored products before though. I could be wrong.

At this point any guess is valid. I mean, if you have X equiplausable choices, you might as well as go with your personal preference