Intel Unveils 10th-Gen Processors Suitable for Next 16-Inch MacBook Pro With Wi-Fi 6 and Turbo Boost Speeds Above 5GHz

[deconstruct60]

The T2 Does all Encode/Decode HEIC/HEVC, and much more. It also houses the bootloader.

The HEVC/HEIC are free side effects of the T2 handling the webcam. And driving the touchbar. The touchbar is more than a bit of a "side show" from the fundamental security duties. ( and in no way is an overlap to a PCH chipset. )

It really isn't the bootloader. it is the firmware loader. It's missing to stop unauthorized mutations of the underlying firmware. The x86 CPU only ever sees a copy, the master is always kept secured away from reach from x86 code. That is a SMC function.

There is some "ride along" stuff comes from it being a hand me down iPhone chip with some narrow mods stuffed into a Mac. The whole "Much , much more" really should stop if security is really the core missing. All that side show stuff is just more complexity presenting an opportunity to inject more holes if direct paths to the outside of the system.

10nm Intel fab is not for consumers. It's a low yield fab process with lots of issues. You won't see 10nm for the next 12-18 months--only server space.

What are you smoking? The new MBA 2020: 10nm CPU.

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

Dell XPS 13 ... 10nm , Microsoft Surface Laptop 3 (most models ) ... 10nm. HP 2-in-1 laptop in same class... 10nm. etc. There are more than a few volume laptops than could find on the showfloor in a BestBest (if the store isn't on pandemic lockdown). If being sold at BestBuy ... it is in the consumer space.

 

[falainber]

@4jasontv and @falainber

You guys are both making a great case for Apple to be running their own chip designs where they can design the internals fully around their desired enclosures and form factors.

Sure. Those designs most likely will be even worse in terms of performance but they will be a truly Apple creation.

 

[high heaven]

Again, failure to maintain max clock speed in a MBP falls on Apple not intel so long as it can perform in another condition.

Not even gaming laptops can achieve 5ghz due to overheating.
[automerge]1585857066[/automerge]

That's MacBook limitation not the CPU's. It's a result of Apple emphasis on "slick" designs.

Still, not even gaming laptops can achieve that speed to sustain. What about power consumption and heat?

TDP 45W? I doubt that.

 

[turbineseaplane]

Sure. Those designs most likely will be even worse in terms of performance but they will be a truly Apple creation.

Just like how their phone and tablet chips are always way behind the competiton
Excellent point.

 

[roland.g]

Call me in 2-3 years or more when Wifi 6 is everywhere. I got a new mesh system in the fall and after looking into Wi-Fi 6 routers which were limited and new, I decided to save the $$. We’re a little ways off from having them in enough devices.

 

[PickUrPoison]

The T2 is primarily a security chip not a PCH . Dumping the "kitchen sink" of PCH functionality onto the security chip is just plain dubious. There is argument has already got too much of a hodgepodge of stuff hanging off of it already. Even more would just open up more vectors . Especially if those vectors are outside the main system and entirely random external devices. No way doing physical information isolation at that point.

Technically , directly hard wired security sensors ( TouchID scanners , FaceID cameras, Webcam cameras (to prevent low level hijacking) , microphones ( Siri voice identification ) can be attached to the T2 as a security issue. ( the speakers coming along with the microphones for the ride. Although Siri on T-series could evenutally get smart enough to talk to. ) . But all of those are highly secured connections because primary hardwired ( and mostly difficult to tamper with).

Similar since Apple is assigning the T2 SMC/PMIC like issues to secure the initial boot the Fans and SMC connections make sense.

Apple bought their major PMIC solution. More of that will probably get weaved into the T-series over time.

Apple bought a celluar modem solution too.


The T2 is more so indicative of the "hand me down" where Apple simply takes an iPhone SoC and tweaks it for a another roll. Whether that is round peg in square or not. Just not going to run off and designing something completely different for the relatively (to iPad or iPhone ) small scale , far more fragmented Mac market. ( the T2 is a 'one size fits all for the Mac market' not a differentiated , custom processor for each Mac product. )

Apple is much more likely going to be focused on a A-series + Modem variant from the current A-series than in making several low volume SoC for a differentiated, diverse Mac line up.

Apple very well skim off the lower part of the Mac laptop market. And perhaps come up with a Chromebox like lowest end offering the desktop since moved the Mac Mini out its "entry level" price point. But more likely that would be through largely "hand me down" chips. Probably the iPad Pro's.

iPadOS has trackpad (cursor ) and keyboard support. All Apple has to do at this point is put the A14X in the recently retired Macbook case and have a system . Like a number of Windows/Chromebook vendors they could sell the same basic hardware in two OS variants if they wanted. ( two birds with one stone of fixed R&D ). No huge resource splitting of the ARM design teams of the major volume drivers that pay for chip development. Apple doesn't want to be everything for everybody. Same parts used in multiple products is one their standard practices.


Apple can drop Intel and still be in x86-64 land for the upper part of the Mac line up with non "hand me down" parts.

I assume Apple will make more performant CPU/GPU than the A-series SoCs, for use in Macs. They’ll be higher wattage chips for the most part, so heat dissipation will be a key requirement.

The Tx is the logical place to add some additional I/O. They could easily add more bandwidth with another four PCI lanes to the CPU, and it would be trivial to add USB3 and Ethernet, for instance, to a T3 PCH/security chip.

 

[smulji]

The T2 Does all Encode/Decode HEIC/HEVC, and much more. It also houses the bootloader.
[automerge]1585854715[/automerge]


10nm Intel fab is not for consumers. It's a low yield fab process with lots of issues. You won't see 10nm for the next 12-18 months--only server space.

And low-power ultrabooks. The new Retina MBA and and Dell XPS 13" both use Intel's 10nm Ice Lake processors.

 

[deconstruct60]

@deconstruct60

Apple won’t be “loosing” anything and honestly, they don’t care about most of those things you mentioned.

So Apple is going to announce at WWDC 2020 that they are dumping Bootcamp because it was really never not useful for anybody. Right?

if had it and then don't have it anymore then it would be loosing something. Your are just handwaving because the functionality doesn't matter to you. It matters to a substantive set of users. That is why Apple put it there in the first place. Windows isn't going away and is a still a substantially bigger market than macOS.

If they drop x86 support they will loose a substantial number of developers who need to do mutliplatform x86 work. Apple could toss that subset of developers under the bus and survive. But that will be a hit to macOS platform growth curve. Not just in terms of revenue lost, but going to loose applications to the ecosystem also.

They are going to do what’s best for them, and what’s best for Apple is being in control of as much of the stack and supply chain for their products as possible.

What is best for Apple is more highly driven by what is profitable for Apple. Control purely and soley for control sake doesn't make Apple more money. Apple subcontracts out goobs of work to vendors. They don't have a desire to do everything themselves. Where they can be a profitable value add then often yes.

The Mac line up has 5-6 products with about 6 different processor variations in them. They aren't completely different but still 6 variants to design and validate. Some of those are very small run rates. ( probably in the less than 100k/yr range for each of the iMac Pro and Mac Pro ... if not both of those combined. ). Low run rates are as profitable unless crank up the price.

In contrast all new iPhone 12 models will get the same SoC. The iPads will likely use hand me down iPhone SoCs. The A14x (or A12Z) has pretty good chance of landing in a AppleTV. That control but it is also profitable.

They’ve given Intel so much rope that they hung themselves with it.

But myopic rope here is that Intel is the only possible option. They aren't.

AMD started to catch up to Intel when they stopped trying to do everything that Intel was doing. Intel trying to sell everything to everybody has been one source of their problems over the last 4-5 years. It has made them more money short term, but strategically they are also a bit weaker because it now also. They have their fingers in so much stuff they are easily distracted. One of the things they are fumbling is Apple as a satisfied customer.

AMD isn't on a laptop's first development path. Their mobile versions are coming out at the end of each iteration. So a mix of server/desktop completeness and then to mobile last. That doesn't make them the best option for entry level Mac laptops, but the order in which they are working is exactly where Apple is weakest with its offerings.

Apple tends to incrementally add to their strongest stuff. They are 'winning' in phone space because they do about 1 SoC a year and most of their phone competitors do 2-8 designs. Apple could dilute their talent base across the same wide breath of implementations. I'm sure their competitors in those spaces would be more than happy to see fixed max simulator time not be put into optimizations of the one product and diluted into multiple ones like theirs are. That will help even the playing field.

 

[smulji]

If they drop x86 support they will loose a substantial number of developers who need to do mutliplatform x86 work. Apple could toss that subset of developers under the bus and survive. But that will be a hit to macOS platform growth curve. Not just in terms of revenue lost, but going to loose applications to the ecosystem also.

There's nothing stopping Apple from providing virtual machine support on ARM processors.

 

[PickUrPoison]

The maximum clock speed is already an example. Can you sustain the clock speed at 5ghz with 16-inch MBP? The sustainable clock speed is actually 3.3 ghz or near, not even close to 5ghz.

If you’re talking about the top of the line i9, that’s a 2.4GHz part. Do you think it’s supposed to run at 5GHz when 8 cores are at 100%? Do you have any idea how much power that would draw?

As it is, Apple allows that chip to draw upwards of 70W, far in excess of its 45W rating. It’ll sustain a 30% higher clock speed than its rated frequency with all-cores at 100%.

Sure, Apple could make an even higher performing machine, but it would be significantly louder and consume much more power, leading to drastically reduced run times.

Or maybe I missed your point?

 

[neuropsychguy]

yeah.. and AMD announced theirs which is more energy efficient and seems to beat intels top of the line mobile cpu in single and multi threaded benchmarks...

AMD’s new processors, as good as they are, almost certainly won’t beat these. AMD’s chips are likely more energy efficient, as you said, but these new Intel processors should keep Intel ahead.

The good news is we can’t go wrong with either of these new generations of processors. It’s great AMD is killing it on the desktop front and looking to kill it on the mobile front. Intel needs competition.

 

[high heaven]

If you’re talking about the top of the line i9, that’s a 2.4GHz part. Do you think it’s supposed to run at 5GHz when 8 cores are at 100%? Do you have any idea how much power that would draw?

As it is, Apple allows that chip to draw upwards of 70W, far in excess of its 45W rating. It’ll sustain a 30% higher clock speed than its rated frequency with all-cores at 100%.

Sure, Apple could make an even higher performing machine, but it would be significantly louder and consume much more power, leading to drastically reduced run times.

Or maybe I missed your point?

lol, I know how it works. Then again, did Intel even mention anything about the maximum clock speed with how many cores? No. What about the power consumption base on the clock speed and core uses? No info. What's the point of having a maximum clock speed which can not be used mostly?

The point is Intel's advertisement with 14nm sucks and they are outdated. How long do we have to use that for?

 

[avtella]

Seems we might get 8 core 10nm (7nm TSMC equivalent) Tigerlake late 2020 - early 2021 which would be a great addition. This would have integrated Thunderbolt like Icelake, something AMD doesn’t have currently. Also remember, we’ve had Intel 10nm for nearly a year now on mobile.

No, the architectural improvements of Ice Lake will carry over to the 14nm Rocket Lake for the 6-8 Core ~45 Watt CPUs. As far as I know/heard Tiger Lake will be 10nm yes, but still limited to low power 4 Core SKUs like Ice Lake. I don't think you will see anything less than a 14nm part for the 6-8+ Core consumer CPUs till 2022.

 

[smulji]

No, the architectural improvements of Ice Lake will carry over to the 14nm Rocket Lake for the 6-8 Core ~45 Watt CPUs. As far as I know/heard Tiger Lake will be 10nm yes, but still limited to low power 4 Core SKUs like Ice Lake. I don't think you will see anything less than a 14nm part for the 6-8+ Core consumer CPUs till 2022.

You're probably right.

"There have been recent rumors and talks regarding Intel back porting a 10nm++ product (Tiger Lake) to 14nm+++ (Rocket Lake). Substantial evidence has been found, but since the product is aiming a 2021 launch, there's no official word from Intel on the matter. But, given that this roadmap talks about back porting, we may indeed see Rocket Lake CPUs featuring a back port of the Willow Cove cores that are to utilize a 10nm++ node on the mobility platform. "

Intel Lays Down 2019-2029 Manufacturing Roadmap - 1.4nm in 10 Years, 2 Year Cadence with Back Porting on Advanced ++ Nodes

Intel's Process roadmap for 2021-2029 has been unveiled, showcasing 10nm, 7nm, 5nm, 3nm, 2nm, 1.4 nm and their respective optimized nodes.

wccftech.com

 

[PickUrPoison]

lol, I know how it works. Then again, did Intel even mention anything about the maximum clock speed with how many cores? No. What about the power consumption base on the clock speed and core uses? No info. What's the point of having a maximum clock speed which can not be used mostly?

The point is Intel's advertisement with 14nm sucks and they are outdated. How long do we have to use that for?

You say you know how it works, but you seem confused about why the CPU doesn’t sustain 5GHz. (It obviously has nothing to do with MBP thermals, as another poster suggested.) That CPU has a base freq. of 2.4GHz. 5GHz is the max Turbo freq.

If you expect the chip to “mostly” run at 5GHz, that misunderstanding is on you. Nothing Intel has said should have led you to think that.

 

[deconstruct60]

There's nothing stopping Apple from providing virtual machine support on ARM processors.

The ARM instruction set has VM assist instructions. The instructions aren't the issue what is being virtualized is. They aren't virtualizing x86 images. Next to nobody is virtualizing Window ARM ( and the secure boot sequencing of Windows ARM is not quite the same either. ) . Linux ARM virtualizing is growing but not anywhere a demand market as x86 instances. If working with a VM instance pulled from an x86 cloud farm then working/testing with what you have to deploy on; not what Apple wished you were deployed on.


If talking about layering an emulator on top of virtualization. Again Apple does have an emulator and last time did one was back in the 90's. They aren't proven experts in this area. Even if did make it work ARM isn't that better that the emulation overhead isn't being sucked away ( can claw much overhead back with JIT compiles )

It isn't a matter of "ooh, let's spend a large chunk of Apple's cash pile on a cool project". Apple could build and ship an xMac . Are they going to do it in 2020-2021 ? Very probably not.

Apple has pointed to the expanse "Pros" definition that they are making systems for "Pro"s all along. One aspect of that "Pro" category they are pointed to explicitly in the past is developers. They could have meant "Oh only XCode developer ... the rest can jump in lake". But they'd have to be drinking more than a little Cupertino kool-aid to think that those are the only driver of the developer pro growth usage in the ecosystem over the last 9-12 years. There are already some grumbling in several subcategories of developers at the CUDA situation ( and 4 years to fix the keyboard and touchbar esc key ) . Dumping x86 VM would only be throwing gas on the fire for them.

 

[Zdigital2015]

If you’re talking about the top of the line i9, that’s a 2.4GHz part. Do you think it’s supposed to run at 5GHz when 8 cores are at 100%? Do you have any idea how much power that would draw?

As it is, Apple allows that chip to draw upwards of 70W, far in excess of its 45W rating. It’ll sustain a 30% higher clock speed than its rated frequency with all-cores at 100%.

Sure, Apple could make an even higher performing machine, but it would be significantly louder and consume much more power, leading to drastically reduced run times.

Or maybe I missed your point?

Not this again...Turbo Boost and TVB is icing on the cake, not the cake itself. The MacBook Pro would need to be 3” thick with liquid cooling to keep the CPU at boost all the time, and even then I don’t think it could then. Not to mention the two hours of battery life the user would end up getting since the part is drawing 105w-115w while on Boost.

I cannot for the life of me understand why users here and other forums think Boost is the core frequency, it drives me absolutely nuts.

At this point Intel threw away meaningful TDP about 2-3 years ago...now it’s just another parlor trick to disguise all the diminishing returns of being slaves to x86 compatibility.

 

[deconstruct60]

No, the architectural improvements of Ice Lake will carry over to the 14nm Rocket Lake for the 6-8 Core ~45 Watt CPUs. As far as I know/heard Tiger Lake will be 10nm yes, but still limited to low power 4 Core SKUs like Ice Lake. I don't think you will see anything less than a 14nm part for the 6-8+ Core consumer CPUs till 2022.

Alder Lake will be more than 4 cores and is queued up to arrive in 2021. Intel is adding public notification for the new instruction when means it likely isn't 2022. Prep for that is 2020 work.

https://www.anandtech.com/show/1568...-for-alder-lake-also-bf16-for-sapphire-rapids

may not like the other 4 cores but more than four.
[automerge]1585861631[/automerge]

Not this again...Turbo Boost and TVB is icing on the cake, not the cake itself. The MacBook Pro would need to be 3” thick with liquid cooling to keep the CPU at boost all the time, and even then I don’t think it could then. Not to mention the two hours of battery life the user would end up getting since the part is drawing 105w-115w while on Boost.
...

Battery life at 105-115W ? The MBP 16" chassis doesn't even draw 105-115W from the wall when plugged in. It would need a fixed , dedicated power port. ( USB PD only goes to 100W )

 

[high heaven]

You say you know how it works, but you seem confused about why the CPU doesn’t sustain 5GHz. (It obviously has nothing to do with MBP thermals, as another poster suggested.) That CPU has a base freq. of 2.4GHz. 5GHz is the max Turbo freq.

Dude, I know how it works. Do you really think I dont know that?

If you expect the chip to “mostly” run at 5GHz, that misunderstanding is on you. Nothing Intel has said should have led you to think that.

Mostly, people dont know that. They believe it is better. would you buy 2.4~4ghz or 2.4~5.3ghz even if they have the same sustainable clock speed?

 

[ls1dreams]

Comet lake is a joke. IMO Apple will wait until Tiger Lake for a Q3-Q4 16" refresh.

It is possible the 13" models get this gen though in the next couple months.
[automerge]1585862853[/automerge]

I have a hunch the Ryzen 9 4900HS is a much suitable cpu for the mac currently, although I stopped caring about CPU anymore on these machines.
For laptops they are powerful enough these days, and I am more interested in power consumption, gpu and making these darn machines less hot.

I also don't care as much about IPC/cpu on laptops, but I care a LOT more about other feature sets. i.e. Wi-fi 6, AV1 hardware decoding, and other standards. Obviously keyboard/screen matter a lot too.

 

[Falhófnir]

Comet lake is a joke. IMO Apple will wait until Tiger Lake for a Q3-Q4 16" refresh.

It is possible the 13" models get this gen though in the next couple months.
[automerge]1585862853[/automerge]


I also don't care as much about IPC/cpu on laptops, but I care a LOT more about other feature sets. i.e. Wi-fi 6, AV1 hardware decoding, and other standards. Obviously keyboard/screen matter a lot too.

Tiger Lake H isn't a thing as far as I'm aware?

 

[smulji]

Tiger Lake H isn't a thing as far as I'm aware?

It is. But I highly doubted it's coming out this year. Most likely some time next year

Intel Rocket Lake-S Desktop CPUs Feature 8 Cores, GT1 Xe Graphics - Rocket Lake-U 6 Core at 15W, Tiger Lake-U 4 Cores at 15W & Tiger Lake-H 8 Cores at 45W Base TDPs

The latest details surrounding Intel's 11th Generation Tiger Lake & 12th Generation Rocket Lake CPUs have leaked out.

wccftech.com

 

[Zdigital2015]

Battery life at 105-115W ? The MBP 16" chassis doesn't even draw 105-115W from the wall when plugged in. It would need a fixed , dedicated power port. ( USB PD only goes to 100W )

I meant if the Core i9-9980HK was allowed to operate without any constraints, as some seem to think it should with this idiocy that the 9980HK should run at Turbo frequencies all the time or else Apple is throttling it. Everybody is throttling the 9980HK, if they are using it at all. The power draw is enormous and ridiculous for a mobile chip.

 

[Falhófnir]

It is. But I highly doubted it's coming out this year. Most likely some time next year

Intel Rocket Lake-S Desktop CPUs Feature 8 Cores, GT1 Xe Graphics - Rocket Lake-U 6 Core at 15W, Tiger Lake-U 4 Cores at 15W & Tiger Lake-H 8 Cores at 45W Base TDPs

The latest details surrounding Intel's 11th Generation Tiger Lake & 12th Generation Rocket Lake CPUs have leaked out.

wccftech.com

Huh. It wasn't on any roadmaps I've seen, it was meant to be Rocket Lake after Comet Lake - maybe Intel came to the conclusion they couldn't rehash the 14nm design any further so they just need to get something else out. It will be interesting to see if it can outperform the 14nm equivalent like Ice Lake struggles to, though what I've heard about Tiger Lake U makes it sound like finally a solid, if not particularly groundbreaking 10nm release.

 

[Zdigital2015]

lol, I know how it works. Then again, did Intel even mention anything about the maximum clock speed with how many cores? No. What about the power consumption base on the clock speed and core uses? No info. What's the point of having a maximum clock speed which can not be used mostly?

The point is Intel's advertisement with 14nm sucks and they are outdated. How long do we have to use that for?

You have to ask Intel. They seems to think that information is a state secret, which it sort of is until it works i their marketing favor. Anandtech has provided a grid of how the core speeds break down (Cores/Turbo Boost), but I don’t think it’s easy to do for each CPU every time. Basically, Intel obfuscates until something works in their favor and then the PR flacks tout their “unique advantages” to those who will actually listen. That number goes down every time there is a new launch.