Who's excited for an ARM Mac mini?

[deconstruct60]

I actually feel that AS will give the mini new life (and with it faster release cycles) because all Apple has to do is drop in their custom A14X chip in the existing mini, update to TB4, and call it a day.

A14X probably doesn't have the PCI-e v3 lanes to provision TBv4. Likely also doesn't have the display subsystem to handle 4 video streams. ( unless Apple backslides down to just two TBv4 ports).
[Thunderbolt v4 requires video on each port. If want to cut corners there is USB4. ]

Another SoC with substantively more capable I/O subsystems would be be part of the "update to" , but Mini has higher port needs than Apple's low end laptop line up. TBv4 is going to be a bit more than just a "ta da" we are done. Especially if Apple wants to match (or better) the port options on the current Mini.

If Apple is completely dumping Intel for Thunderbolt also then they have even bigger question marks. Won't be going to a vendor with better experience at it.

Case in point, the developer kit was a mini. They don't have to redesign it (considering their server market wants to same form factor)

Server folks not only want the same form factor. A substantive number of them probably want the same basic ports also. ( e.g., if want to add a second 10GbE port to SAN, then will need to use TBv3. ). The DTK actually backslides on ports. No TBv3. Two less USB-C. Two less USB-A. Probably not much more than two 3.0 USB-A ports worth of throughput. Probably no more than 1-2 video out.

In terms of testing apps it was/is sufficient. In terms of testing I/O it is a circa 2017 (or earlier ) status.

I could be wrong though, since they did re-release the 2018 model with upgraded storage and label it as "New".

 

[iAssimilated]

A14X probably doesn't have the PCI-e v3 lanes to provision TBv4. Likely also doesn't have the display subsystem to handle 4 video streams. ( unless Apple backslides down to just two TBv4 ports).
[Thunderbolt v4 requires video on each port. If want to cut corners there is USB4. ]

Another SoC with substantively more capable I/O subsystems would be be part of the "update to" , but Mini has higher port needs than Apple's low end laptop line up. TBv4 is going to be a bit more than just a "ta da" we are done. Especially if Apple wants to match (or better) the port options on the current Mini.

If Apple is completely dumping Intel for Thunderbolt also then they have even bigger question marks. Won't be going to a vendor with better experience at it.



Server folks not only want the same form factor. A substantive number of them probably want the same basic ports also. ( e.g., if want to add a second 10GbE port to SAN, then will need to use TBv3. ). The DTK actually backslides on ports. No TBv3. Two less USB-C. Two less USB-A. Probably not much more than two 3.0 USB-A ports worth of throughput. Probably no more than 1-2 video out.

In terms of testing apps it was/is sufficient. In terms of testing I/O it is a circa 2017 (or earlier ) status.

It was probably a bad idea to call out the A14X by name, since I know that is not going to be the desktop/laptop equivalent SoCs, but was just making an example of the work Apple would need to do to upgrade the mini (ie not much). I am fairly confident if Apple made the decision to go AS in their lineup they have a product that will meet the need.

If Apple can offer the exact same ports they did with the 2018 mini with AS, then I will be happy. TB3 has been enough throughput for me. I agree, the server folks would want the same basic ports. I think the reason the DK backslid on ports was due to the fact that Apple didn't want people getting a preview of their actual product and needed something to give developers that was compatible with their upcoming systems.

 

[Coolkiwi]

Yes ??‍♂️

 

[TrevorR90]

I just got the Mac Pro 2019 and I sort of feel cheated. But I'm excited to see what ARM allows Apple to do!

 

[dmccloud]

My guess is that the first machines to be upgraded to ARM will be the base 13" MBP (possibly replaced with a 14"?), the 21" iMac, and the Mac Mini. Apple will take the time before replacing their Pro lineup (Mac Pro, iMac Pro, MBP) because those are the users who will honestly be impacted the most by the transition. That also gives the 3rd party app developers additional time to get their apps ready for ARM. While Adobe and Microsoft clearly have an advantage with the transition, they have also been doing something similar on the Windows side with the Surface Pro X (which also runs ARM instead of x86). That lead time actually simplifies the transition, especially for Microsoft since both the Windows and Mac versions of Office use the same code base.

 

[Joe The Dragon]

We have a divergent view of this. I don't think Apple will move the iMac Pro to the W2200, as that will be discontinued, as an opinion. They will not prolong a broken vendor relationship by continuing to use newer Intel designs, especially if the 30" AS Mac can come close to the performance of the current iMac Pro. When the W2100 CPUs run out, the iMac Pro is history. Apple will not sink any more R&D money inot any product that is Intel based going forward. That can be see in the latest 27" iMac announcement, where the iMac not only wasn't upgraded in any meaningful way (including no nanotextured screen), it was barely mentioned.

While I agree that the MacPro's SoC will appear last, I don't see any Intel upgrades being made. The next MacPro will be an AS Mac. I also am fairly certain that the MacPRo's SoC exists, and has for a while. It is probably currently undergoing validation tests and the software guys are starting to work with it. None of this happens in two years, it takes longer, in many cases, as I suspect the MacPro SoC to be one case, it will take 4-5 years from start of design to final product, 3 years at least for the SoC, and another year for software and logic board/system design and test (including life testing). If my thoughts are anywhere near true, we are at maximum, less than half a year away from the finalized MacPro SoC, perhaps less. And this assumes the worst case of Apple waiting for the full two years (as stated in the WWDC announcement) befor the Mac Pro makes the AS SoC transition. I personally think that the transition will not take 2 years, maybe 16-18 months at most, which working backwards, means that the MacPro is somewhere around 75-80% complete, today. So transitioning to a newer generation Intel CPU doesn't make sense. My best guess is that there will be a AS MacPro being sold before the end of 2021, and at that time, there will not be any Intel CPUs in any Macs being sold by Apple.

apple can bump cpu's on the mac pro easy with like no R&D

 

[Kostask]

Do you honestly expect Apple to just drop in a new CPU into the Mac Pro (the machine that professionals are depending on for income on some of the biggest and most complex applications on the Mac platform) and call it a day? This is something we very much disagree on. Apple would get a number of the new CPUs in a number of MacPros of differing configurations, and run them day and night, perhaps even throught the Pro gauntlet that they used to develop the MacPro originally, and after months of testing, would be willing to release it. You must think that Apple runs like a gamer in their Mom's basement.

Or, they could just jump to the AS Mac Pro, and be done with Intel once and for all.

 

[deconstruct60]

apple can bump cpu's on the mac pro easy with like no R&D

There is nothing socket compatible coming for the Mac Pro. The Mac Pro has "end of the line" CPU and socket and chipset. Apple can 'coast' on that infrastructure for another 1-2.5 years, but there are no easy upgrades there.

Intel revealed some of the aspects of Ice Lake SP ( which later will probably be used in a w-3300 series )

Hot Chips 2020 Live Blog: Next Gen Intel Xeon, Ice Lake-SP (9:30am PT)

www.anandtech.com

Even if Intel did chop off some of the memory channels to stuff it into a 3647 socket the DMI connection is different to a different chipset. The power management is different. PCI-e v4 is different (which is going to cause some board re-flows to deal with).


The iMac Pro with W-2200 and some "Pro Vega II" GPUs would be a low cost R&D bump to 'kick the can' into probably 2022 for a replacement. Plus the nano texture screen that is around anyway due to the iMac having it. And a bump of the Thunderbolt v3 controller to support DPv1.4 (might need some board trace upgrades for DPv1.4 )

 

[Boil]

There is nothing socket compatible coming for the Mac Pro. The Mac Pro has "end of the line" CPU and socket and chipset. Apple can 'coast' on that infrastructure for another 1-2.5 years, but there are no easy upgrades there.

The iMac Pro with W-2200 and some "Pro Vega II" GPUs would be a low cost R&D bump to 'kick the can' into probably 2022 for a replacement. Plus the nano texture screen that is around anyway due to the iMac having it. And a bump of the Thunderbolt v3 controller to support DPv1.4 (might need some board trace upgrades for DPv1.4 )

I would see Apple going for whatever Pro version of Big Navi AMD has for them in an iMac Pro final Intel version refresh...

W2200-series Xeon CPUs
Big Navi / RDNA2 GPUs
Nano texture display option
HOPEFULLY, a RAM access door...?!?

 

[deconstruct60]

I would see Apple going for whatever Pro version of Big Navi AMD has for them in an iMac Pro final Intel version refresh...

W2200-series Xeon CPUs
Big Navi / RDNA2 GPUs
Nano texture display option
HOPEFULLY, a RAM access door...?!?

Not in the context of minimizing R&D. RAM door means a trip to Industrial Design. That will cost lots of money while they go off and "discover" and meditate on what needs to be done. Big Navi is more work than dropping in GPUs for which there drivers are already done with. While there may be some Big Navi HBM models it is another huge timelline slip potential.

Also up in the air whether "big navi" is going to be a good thermal fit to the iMac Pro. ( AMD took better perf/watt and threw it at more performance; not necessarily better thermals. )

 

[Boil]

Not in the context of minimizing R&D. RAM door means a trip to Industrial Design. That will cost lots of money while they go off and "discover" and meditate on what needs to be done. Big Navi is more work than dropping in GPUs for which there drivers are already done with. While there may be some Big Navi HBM models it is another huge timelline slip potential.

Also up in the air whether "big navi" is going to be a good thermal fit to the iMac Pro. ( AMD took better perf/watt and threw it at more performance; not necessarily better thermals. )

I did say HOPEFULLY, and it would be a compelling change for some to make that "final Intel" Mac purchase, where they may have been on the fence & thinking to just wait it out with whatever gear they currently have & see what Apple silicon Macs brings...?

But if Apple decides to upgrade the CPUs in the iMac Pro (not just a spec shuffle as they have done most recently), I would hope they upgrade the GPU as well, and that would be Big Navi (meaning whatever custom version Apple specs to AMD / PCB manufacturers)...

And if Apple does a decent overhaul on the iMac Pro for the last Intel model, maybe they go big & have it be a 30" display (with Nano texture option)...

But then the argument could be made that if they just let the iMac Pro fade away as it currently sits (spec shuffled CPU edition), it would be a lower performance factor to compare a possible future iMac Pro with Apple silicon inside...?

 

[Codeseven]

Bring it on!

 

[projectle]

My dream would be the return of Xgrid, and clustering computers together via Thunderbolt 4 to “direct attach” to a control node. When you need more processing power, you can add an extra SoC in much the same way we can an external GPU today. After all, the 40Gbps PCIe interface we have today with Thunderbolt 3 is already in the vicinity of RAM performance speed.

 

[deconstruct60]

My dream would be the return of Xgrid, and clustering computers together via Thunderbolt 4 to “direct attach” to a control node. When you need more processing power, you can add an extra SoC in much the same way we can an external GPU today.

Ethernet over Thunderbolt Only really does point to point . depending upon drivers not that well .
basically shaving costs on switches more than anything else . A 10GbE switched backhaul network would cost more but would be far more robust .

The notion of Mini as a lego block, ‘Single image’ computer wont work . A network to dispatch Mostly independent batch Jobs too can work .

After all, the 40Gbps PCIe interface we have today with Thunderbolt 3 is already in the vicinity of RAM performance speed.

First, Thunderbolt is 40 Gbps , PCE-e v3 ( x4 ) is not. The bandwith of thunderbolt is that which encodes and transports. PCie v3 x4 is 32Gbps.

Second, nither one of those anywhere near memory speeds.
core i3 1000G4 is 53 GBPs ( I.e. 424 Gbps )

Intel® Core™ i3-1000G4 Processor (4M Cache, up to 3.20 GHz) Product Specifications

Intel® Core™ i3-1000G4 Processor (4M Cache, up to 3.20 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

even a lower power Modern CPU package

Lakefield. i3 L3G4 is 34 GBps ( i.e. 272 Gbps )

Intel® Core™ i3-L13G4 Processor (4M Cache, up to 2.8GHz) Product Specifications

Intel® Core™ i3-L13G4 Processor (4M Cache, up to 2.8GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

so off by order of magnitude . Latency wise even worse .

 

[Jorbanead]

I suspect the MacPro SoC to be one case, it will take 4-5 years from start of design to final product, 3 years at least for the SoC, and another year for software and logic board/system design and test (including life testing). If my thoughts are anywhere near true, we are at maximum, less than half a year away from the finalized MacPro SoC, perhaps less.

Yes, this would track with why the new Mac Pro took so long to make. I think the very beginnings of the Mac Pro SoC started when they redesigned the Mac Pro. I remember when everyone was moaning that Apple took 6 years to build a tower, and even with the impressive engineering, it shouldn’t have taken that long. But I think this new Mac Pro took so long to build because they wanted to build an enclosure that would last the next 10 years - and of course at that point they knew the Apple Silicon transition was going to happen, so they had to build it with the transition in mind which is why it took longer. I’m sure they already have Mac Pro SoC designs and even an Apple silicon Mac Pro revision 1 - and they work with Apple MPX modules too. We will probably see revision 2 or 3 when the Apple silicon Mac Pro comes out.

 

[yellowhelicopter]

I'm excited but concerned that still there aren't any leaks that Mac Mini AS will be amongst first AS products this fall. I just hoped it would be released this year.

 

[Populus]

Before I jump onto the next Mac mini, I need to know if Apple Silicon MacBooks will support touchscreen.

if ARM MacBooks do indeed support touchscreen then I will sell my iPad Pro and my Mac mini and get one of those machines. If MacBooks don’t support touchscreen then probably I’ll get an ARM based mini and keep my iPad Pro as a portable device, hoping that eventually iPad OS will be closer in functionality to macOS.

 

[MiniApple]

As far as I understood ARM uses less power, has longer battery life, generates less heat, less noise and is cheaper than Intel and Co.

Can people who know these tech/science things maybe elaborate in layman's terms and more confirm or correct my summary?

If so, would that theoretically allow for an dongle sized Mac Mini stick or at least for a much smaller Mac Mini for those that don't need the Pro / Server Mac Mini power?

I'd be very excited for either a Mac Mini stick or Apple TV sized Mini, if it would allow basic things (web browsing, storing documents, watching photos/movies, listen to podcasts/music)

 

[jazz1]

If the ARM mini has solid thermals, a strong gpu (No need for an egpu), and fixed Bluetooth and WiFi, I am in.

I would like a Mini, but these four things prevent me from jumping on.

Is Bluetooth and WiFi iffy on the current Mini? Just wondering?

 

[MisterMe]

Before I jump onto the next Mac mini, I need to know if Apple Silicon MacBooks will support touchscreen.

if ARM MacBooks do indeed support touchscreen then I will sell my iPad Pro and my Mac mini and get one of those machines. If MacBooks don’t support touchscreen then probably I’ll get an ARM based mini and keep my iPad Pro as a portable device, hoping that eventually iPad OS will be closer in functionality to macOS.

We don't know what we don't know. However, this is a question that has been asked for decades. For decades, the answer has been "No."

This is not an answer that Steve Jobs reached around and pulled out. The thing that people new to the Mac and the Apple Universe do not get is that Apple tests many more ideas than ever reach market. They actually build working prototypes and distribute them to Apple engineers who use them in their in their day-to-day lives. For devices that make it to market, this process helps identify unforeseen issues and to correct them. For devices that never make it to market, this process identifies the flaws in the device and determines that there is no reasonable way to reduce the flaws to a level that will turn it into a successful product.

You may rest assured that touchscreen Macs were tested back when they were desktops with CRTs, laptops with LCD displays, and desktops with LCD displays. They failed each time. Notice that iOS apps installed on ASi Macs will have their UIs mapped to standard Mac. There is no indication that you will interact with iPad apps on an ASi Mac the way that you interact with these apps on your iPad.

 

[kazmac]

Is Bluetooth and WiFi iffy on the current Mini? Just wondering?

I have no idea, as there were none on display I could test before lockdown.

Perhaps someone else will respond?

 

[Populus]

We don't know what we don't know. However, this is a question that has been asked for decades. For decades, the answer has been "No."

This is not an answer that Steve Jobs reached around and pulled out. The thing that people new to the Mac and the Apple Universe do not get is that Apple tests many more ideas than ever reach market. They actually build working prototypes and distribute them to Apple engineers who use them in their in their day-to-day lives. For devices that make it to market, this process helps identify unforeseen issues and to correct them. For devices that never make it to market, this process identifies the flaws in the device and determines that there is no reasonable way to reduce the flaws to a level that will turn it into a successful product.

You may rest assured that touchscreen Macs were tested back when they were desktops with CRTs, laptops with LCD displays, and desktops with LCD displays. They failed each time. Notice that iOS apps installed on ASi Macs will have their UIs mapped to standard Mac. There is no indication that you will interact with iPad apps on an ASi Mac the way that you interact with these apps on your iPad.

Yeah, that’s why they would never release a pencil... until they did. I use my iPad Pro as a laptop, and I got so used to it, that when I went to format my 2010 MacBook Pro yesterday (yeah, I’m not new to the Mac and the “Apple Universe”) I accidentally ended up touching the screen. Then I remembered that MacBooks don’t have touchscreen.

I’m sure they had their reasons to not release an Apple Silicon mac 10 years ago, as they remained with Intel. And they didn’t have a reason to release a touchscreen mac, as it would be unconfortable to use (granted, it’s not the most ergonomic), but having nowadays something like an iPad Pro with a keyboard, trackpad, and a touchscreen, I see there’s a possibility that Apple might release the new AS MacBooks with touchscreen.

I’m not saying there’s going to be one, in fact, if you read my comment, you’ll see I‘m only contemplating both scenarios. I don’t know yet which one I will follow. Regardless of touchscreen MacBooks, either I get an Apple Silicon Mac mini + my iPad Pro, or I sell everything to get an Apple Silicon MacBook (Pro, Air).

 

[macguru9999]

I just got the Mac Pro 2019 and I sort of feel cheated. But I'm excited to see what ARM allows Apple to do!

Honestly, its like the mac pro (xeon) team and the apple silicon team were they even talking to each other ??? I cannot see where the 2019 mac pro fits in the road map going forward . Are they going to have another total redesign ? I may move from a mac pro 2012 to a mac mini 2021 , who knows ....

 

[thenewperson]

Honestly, its like the mac pro (xeon) team and the apple silicon team were they even talking to each other ??? I cannot see where the 2019 mac pro fits in the road map going forward . Are they going to have another total redesign ? I may move from a mac pro 2012 to a mac mini 2021 , who knows ....

I can't imagine they weren't aware of the ASi transition coming up when designing it. What would make the MP not fit the road map?

 

[Boil]

Honestly, its like the mac pro (xeon) team and the apple silicon team were they even talking to each other ??? I cannot see where the 2019 mac pro fits in the road map going forward . Are they going to have another total redesign ? I may move from a mac pro 2012 to a mac mini 2021 , who knows ....

There will always be those who need the physical PCIe slots, for Pro Tools cards being the most common...?

Apple will also most likely be making their own GPUs, and those will need PCIe slots...

And there is the Apple Afterburner card...

So plenty of use for PCIe slots in an Apple silicon Mac world...?!?