Review: M1 Max MacBook Pro After Three Months

[VArase]

I'm not sure about that first assertion. Just like an x86 vm running within parallels on an Intel machine doesn't run as well as on the bare metal I think it would be the same for emulating on arm. You are going to have the WoW emulation stack in either situation on ARM, true, but with a boot camp style solution you remove the paralllels emulation layer which should make things faster.
You're dead on about how it is probably easier for boot camp on Intel to be developed though. Apple will have to build pretty much every peripheral driver which is a lot of work. I suppose there is an advantage in that it is all in house and getting bugs fixed is easier but it is still significant effort.

Remember: you're not emulating ARM - you're running ARM instructions natively with system service calls being intercepted and translated by the hypervisor. The only time you emulate client instructions in a VM is when you're running a non-native instruction set.

It's why SoftPC ran so much slower on PowerPC machines than VMWare or Parallels ran on x86.

As for writing drivers for boot camp - it just wouldn't be worth it. Apple may allow a ARM64 boot camp, but most of the graphical stuff just wouldn't run. ARM Windows probably uses the hardware in the Qualcomm SoCs anyway - betcha it's the hypervisor that's intercepting graphical calls at the service routing level and is transliterating them to Metal.

Microsoft wants to make it's own chips, but doesn't have the silicon chops to do so so leave it up to their silicon proxy Qualcomm. Qualcomm wants to do Microsoft's bidding but doesn't have the processor design chops so leaves it up to their newly acquired silicon proxy Nuvia. Qualcomm hopes that the Nuvia folks left Apple with workable IP, but the Nuvia folks know if they use Apple IP they're be sued into a hole so deep their descendants will need archeologists to find their remains.

So ... no one knows what ARM Win will look like, if Nuvia will create ARM processors compatible with Apple designs, and if ARM Win will run on Apple's ARM instruction set in the future. Writing entire complex and expensive subsystems for a future that iffy would be a waste of resources on Apple's part, considering how malleable the Microsoft ARM Win/Qualcomm/Nuvia future may be - definitely not a fixed set-in-stone target like x86.

When you come right down to it, Apple split from the Wintel homogeny when they moved their platform to their own silicon, and unless and until Apple goes back to using 3rd party GPUs - not bloody likely - boot camp is a dead issue except for systems specifically designed to run on Apple hardware.

 

[VArase]

Welcome! you haven't tried other browsers have you?

I have tried other browsers and find Safari to be much faster.

There are compatibility issues as a lot of web developers seem to think that Chrome is the web standard, but I've found Chrome to be problematic - especially with stuff like when the Chrome maintenance daemon goes looping while getting other tasks blamed for the high CPU utilization (like WinderServer or AV daemons).

 

[mzeb]

Remember: you're not emulating ARM - you're running ARM instructions natively with system service calls being intercepted and translated by the hypervisor. The only time you emulate client instructions in a VM is when you're running a non-native instruction set.

It's why SoftPC ran so much slower on PowerPC machines than VMWare or Parallels ran on x86.

As for writing drivers for boot camp - it just wouldn't be worth it. Apple may allow a ARM64 boot camp, but most of the graphical stuff just wouldn't run. ARM Windows probably uses the hardware in the Qualcomm SoCs anyway - betcha it's the hypervisor that's intercepting graphical calls at the service routing level and is transliterating them to Metal.

Microsoft wants to make it's own chips, but doesn't have the silicon chops to do so so leave it up to their silicon proxy Qualcomm. Qualcomm wants to do Microsoft's bidding but doesn't have the processor design chops so leaves it up to their newly acquired silicon proxy Nuvia. Qualcomm hopes that the Nuvia folks left Apple with workable IP, but the Nuvia folks know if they use Apple IP they're be sued into a hole so deep their descendants will need archeologists to find their remains.

So ... no one knows what ARM Win will look like, if Nuvia will create ARM processors compatible with Apple designs, and if ARM Win will run on Apple's ARM instruction set in the future. Writing entire complex and expensive subsystems for a future that iffy would be a waste of resources on Apple's part, considering how malleable the Microsoft ARM Win/Qualcomm/Nuvia future may be - definitely not a fixed set-in-stone target like x86.

When you come right down to it, Apple split from the Wintel homogeny when they moved their platform to their own silicon, and unless and until Apple goes back to using 3rd party GPUs - not bloody likely - boot camp is a dead issue except for systems specifically designed to run on Apple hardware.

Apple hasn't been real open about how virtualization works on the M-series procs. I looked but docs are thin. Speaking to x86 virtualization what you say is true that things are faster but still not native speeds. Technologies like vt-x and vt-d did a lot to speed things up by allowing hypervisors to pass instructions and I/O calls directly to hardware but they aren't perfect. They have limitations that bare metal does not. For instance, hyper-threading only works from within a VM and the process queues can't be intertwined so can't optimize with each other. The same is true of the queueing of microcode instruction. Directed I/O works best when a whole disk is handed off to it and there is no virtual disk being written to on top of it but rarely does anyone use VMs this way at home or in the enterprise. In theory it is near native but rarely does it end up this way. A simple test: run the black magic disk benchmark on a VM and on the bare metal. It's way better than it used to be but still not as good as running on hardware.

I think you're right about the cost effectiveness of driver development from Apple, though even for Intel machines Apple ended up writing a lot of drivers for cameras, track pads, mice and other peripherals so they could write for arm.

As for Qualcomm not having the chip design chops I have to disagree. The snapdragon series of processors has rivaled or surpassed the A-series chips in raw numbers. What they lack is Apple's integration between hardware and software because they don't build Android. If they were to move to building laptop class procs in a more serious way there is no doubt in my mind that they could compete with the M-series Apple silicon in numbers at least.

You are right that MS doesn't have the chip design expertise but I know you're not right about Windows on ARM having any sort of deep integration with Qualcomm chips. That would all be up to the compiler when binaries are built. Starting in Vista we had a strong move away from allowing deep ties with OEM hardware to keep Windows platform and architecture agnostic. This was because at the time of development x86, AMD64, Itanium, Alpha and PPC were all potential platforms even today. Each including their own chipsets. This agnosticism is even more true today. Yes, there are some custom drivers but those are all built by Qualcomm with assistance from MS. MS does not want to do anything to tailor an OS to platform and lock themselves in.

That said, there is a scenario I think you're missing where Boot Camp on Macs becomes a thing. The contract between MS and Qualcomm for Windows on ARM is just about up. What if MS bought chips from Apple instead for their systems? Apple has good laptop and soon desktop class ARM procs. It'd just make sense if MS can work that deal out with Apple. I'll admit that's a big "if" as Apple has not been playing nice with other lately but MS very much has and it makes a lot of sense for them. Maybe the chips are marketed the same but putting Apple Silicon in a Microsoft branded machine would make apple a lot of cash potentially. This is a scenario more likely than we think I suspect.

 

[gank41]

That said, there is a scenario I think you're missing where Boot Camp on Macs becomes a thing. The contract between MS and Qualcomm for Windows on ARM is just about up. What if MS bought chips from Apple instead for their systems? Apple has good laptop and soon desktop class ARM procs. It'd just make sense if MS can work that deal out with Apple. I'll admit that's a big "if" as Apple has not been playing nice with other lately but MS very much has and it makes a lot of sense for them. Maybe the chips are marketed the same but putting Apple Silicon in a Microsoft branded machine would make apple a lot of cash potentially. This is a scenario more likely than we think I suspect.

What a great perspective I haven't seen yet: MS using Apple for their SoC's.. What a lucrative business that could be for them if it was true. For both parties.

 

[BWhaler]

Really, any problems are mostly due to Monterey ...

You understand that was the OP's entire point.

 

[mzeb]

What a great perspective I haven't seen yet: MS using Apple for their SoC's.. What a lucrative business that could be for them if it was true. For both parties.

Yep, folks often forget that these companies are out for money first and foremost. If Apple stands to make bank they'd be happy to sell chips to whoever.

 

[VArase]

Yep, folks often forget that these companies are out for money first and foremost. If Apple stands to make bank they'd be happy to sell chips to whoever.

They actually would not - Apple knows a competitive advantage when the see it.

If Apple started selling Apple Silicon SoCs to Microsoft, Windows Surface machines would run all over Macs.

They learned their lesson when the licensed the Mac OS to OEMs.

 

[VArase]

Apple hasn't been real open about how virtualization works on the M-series procs. I looked but docs are thin. Speaking to x86 virtualization what you say is true that things are faster but still not native speeds. Technologies like vt-x and vt-d did a lot to speed things up by allowing hypervisors to pass instructions and I/O calls directly to hardware but they aren't perfect. They have limitations that bare metal does not. For instance, hyper-threading only works from within a VM and the process queues can't be intertwined so can't optimize with each other. The same is true of the queueing of microcode instruction. Directed I/O works best when a whole disk is handed off to it and there is no virtual disk being written to on top of it but rarely does anyone use VMs this way at home or in the enterprise. In theory it is near native but rarely does it end up this way. A simple test: run the black magic disk benchmark on a VM and on the bare metal. It's way better than it used to be but still not as good as running on hardware.

I think you're right about the cost effectiveness of driver development from Apple, though even for Intel machines Apple ended up writing a lot of drivers for cameras, track pads, mice and other peripherals so they could write for arm.

As for Qualcomm not having the chip design chops I have to disagree. The snapdragon series of processors has rivaled or surpassed the A-series chips in raw numbers. What they lack is Apple's integration between hardware and software because they don't build Android. If they were to move to building laptop class procs in a more serious way there is no doubt in my mind that they could compete with the M-series Apple silicon in numbers at least.

You are right that MS doesn't have the chip design expertise but I know you're not right about Windows on ARM having any sort of deep integration with Qualcomm chips. That would all be up to the compiler when binaries are built. Starting in Vista we had a strong move away from allowing deep ties with OEM hardware to keep Windows platform and architecture agnostic. This was because at the time of development x86, AMD64, Itanium, Alpha and PPC were all potential platforms even today. Each including their own chipsets. This agnosticism is even more true today. Yes, there are some custom drivers but those are all built by Qualcomm with assistance from MS. MS does not want to do anything to tailor an OS to platform and lock themselves in.

That said, there is a scenario I think you're missing where Boot Camp on Macs becomes a thing. The contract between MS and Qualcomm for Windows on ARM is just about up. What if MS bought chips from Apple instead for their systems? Apple has good laptop and soon desktop class ARM procs. It'd just make sense if MS can work that deal out with Apple. I'll admit that's a big "if" as Apple has not been playing nice with other lately but MS very much has and it makes a lot of sense for them. Maybe the chips are marketed the same but putting Apple Silicon in a Microsoft branded machine would make apple a lot of cash potentially. This is a scenario more likely than we think I suspect.

I didn't say Qualcomm doesn't have chip design know how - I said they don't have processor design know how - something entirely different.

It's relatively simple to design a SoC - not so much so designing wide computational processors with deep pipelines. The Firestorm cores in M1 are eight wide with a 690 instruction execution queue, a massive reorder buffer, redundant arithmetic units, and an out-of-order execution unit capable of executing eight instructions simultaneously.

That's seriously great CPU design.

Qualcomm uses standard ARM designed cores and adjust caches and other components on their SoCs. Even their strongest chips like the 888 use standard ARM designed cores with a Cotex-X1 at the top of the pyramid.

Really, in the ARM space there are only two companies designing processors today: ARM and Apple. Maybe add Nuvia if they get their act together.

Apple has no desire to become a commercial processor developer - owning the whole stack is how Apple was able to deprecate ARMv7 (32 bit mode with condition code checking instructions), write it out of the iOS, and remove the logic from their processors in something like two or three years. If they were selling processors to others their customers would be caterwauling about the lack of ARMv7 for their legacy OSes and applications.

Oh, and the only place where Qualcomm has exceeded Apple SoCs in recent years has been in a multicore metric - and that only because their target market wants a zillion cores to bump up their paper specs - but it's really the single core speed which determines user responsiveness, both on phones and on PCs.

Most processes can't be easily multithreaded.

 

[Asbow]

For those considering a purchase, Dan says that he wishes that he had gone with a 2TB SSD over a 1TB because of his video editing needs, and he's not sure that he's happy with the weight and size of the 16-inch MacBook Pro when the same functionality is available with the smaller and more portable 14-inch MacBook Pro.

My major bugbear with laptop reviews is this ‘the 16-inch is a tad heavy compared to the 14’. State the obvious why don’t you.

 

[Blowback]

I love my MacBook Pro. I love the new design however at times I do miss the Touch Bar. In my opinion, this is definitely the product of the year. Hands down ?

View attachment 1946631

I too have always appreciated the Touch Bar...will miss it when I next purchase a MBP.

 

[widEyed]

It's nice that it has an HDMI port but I don't have a use for it…. — Dan

Sorry but how can I take a video editor review of this machine seriously when the video editor only uses the 16" for editing videos in the office/home-office? A $300 ext 27" Display will make you approx 2x more productive, Dan. 2x 27" 3x more productive. Throw in a tablet while you're at it.

Pay a little more like $400 for a USB-C connector on a Dell so you can plug USB devices into the monitor and have them all return to the MBP on one cable. or pay $700 for a factory colour calibrated DELL.

 

[widEyed]

That said, there is a scenario I think you're missing where Boot Camp on Macs becomes a thing. The contract between MS and Qualcomm for Windows on ARM is just about up. What if MS bought chips from Apple instead for their systems? Apple has good laptop and soon desktop class ARM procs. It'd just make sense if MS can work that deal out with Apple. I'll admit that's a big "if" as Apple has not been playing nice with other lately but MS very much has and it makes a lot of sense for them. Maybe the chips are marketed the same but putting Apple Silicon in a Microsoft branded machine would make apple a lot of cash potentially. This is a scenario more likely than we think I suspect.

Very hard to see a corporation as large and used to getting it's own way as MS putting their future in the hands of Apple. That is a classic corporate situation where "supply constraints" and "administrative delays" can be used by the supplier company to shaft the customer corporation. Seen it happen many times.

Even if they could buy a licence and manufacture them through independent chip/SoC manufacturers, they'd still be dependent on Apple for future release versions of the ARM/SoC/subsystem software.

 

[mzeb]

Very hard to see a corporation as large and used to getting it's own way as MS putting their future in the hands of Apple. That is a classic corporate situation where "supply constraints" and "administrative delays" can be used by the supplier company to shaft the customer corporation. Seen it happen many times.

Even if they could buy a licence and manufacture them through independent chip/SoC manufacturers, they'd still be dependent on Apple for future release versions of the ARM/SoC/subsystem software.

Delays are built into most contracts these days as a fine on the supplier. It can still happen but it hurts. This hit intel hard when they couldn’t meet their promises to Dell and HP for their 6th skylake chips. MS is savvy enough to work that contract if they wanted to. I’m sure this kind of clause is built into their current contract with Qualcomm. It wouldn’t be all that different.

That said, I disagree on your assessment of MS. They certainly used to operate as you say but they have become a different beast. Their collaboration in the open source world including Linux and WebKit even with flagship products like SQLServer no longer being windows only would have been unheard of a decade ago. Heck, Edge is based of the same tech as chrome.

 

[VArase]

Sorry but how can I take a video editor review of this machine seriously when the video editor only uses the 16" for editing videos in the office/home-office? A $300 ext 27" Display will make you approx 2x more productive, Dan. 2x 27" 3x more productive. Throw in a tablet while you're at it.

Pay a little more like $400 for a USB-C connector on a Dell so you can plug USB devices into the monitor and have them all return to the MBP on one cable. or pay $700 for a factory colour calibrated DELL.

Is there something that would preclude using a 27" monitor attached to the MBP at home or in the office while he uses the builtin display at a center table at Starbucks?

 

[lewisflude]

I've been consistently getting the clamshell sleep kernel panic for a while now on Monterey. I even did a total firmware reinstall/clean wipe, and am continuing to get a kernel panic/crash whenever the machine goes to sleep while in clamshell mode.

Is this a known issue?

 

[VArase]

You understand that was the OP's entire point.

Sorry, I took his tone as a deprecation of macOS in general.

Perhaps I was hasty.

 

[gank41]

I've been consistently getting the clamshell sleep kernel panic for a while now on Monterey. I even did a total firmware reinstall/clean wipe, and am continuing to get a kernel panic/crash whenever the machine goes to sleep while in clamshell mode.

Is this a known issue?

It’s a known issue in my house

 

[kingtj1971]

I've been consistently getting the clamshell sleep kernel panic for a while now on Monterey. I even did a total firmware reinstall/clean wipe, and am continuing to get a kernel panic/crash whenever the machine goes to sleep while in clamshell mode.

Is this a known issue?

I believe it is. My 16" does this regularly too. I mean, it's not so bad I'm likely to catch it doing it if I'm right in front of it and it goes to sleep and I immediately wake it again? But it happens frequently enough so if I just leave it on and let it sleep overnight each night? I do come back fairly often to find it had a kernel panic and restarted itself.

I'm really hoping the new OS X update addresses this.

 

[PaunchyPirate]

I’ve got my 14” MBP all set up and working in clamshell mode with an external Dell 4k monitor using a USB-C to USB-C cable and Apple Magic Keyboard with Touch ID and a Magic Mouse. The MBP is powered by the USB-C cable from the external monitor. Once up and running it all works great.

At the end of the day, I have 2 options:

1. If I put the MBP to sleep and then shut off the external monitor, the next day I can then turn on the monitor And I can hear the MBP detect the power from the now running monitor. I can the touch the space bar on my keyboard and this wakes up the laptop and I can use the Touch ID to authenticate in. Perfect.

2. If I shutdown the MBP… when I turn the monitor on, I can hear the chime on the MBP indicate it’s getting power. But the MBP doesn’t turn on. I have to open the lid to power it up. I could have sworn that people had theirs working to boot up as soon as the power from the monitor started. Should it? What am I doing wrong.

 

[dvdweyer]

OWC Thunderbolt Dock from 2018, two monitors (LG 27” 5K via Thunderbolt, Dell 32” 4K mostly via HDMI) connected and no crashes. Also runs Windows 11 under Parallels Desktop with no issues. I absolutely love the lack of fan noise and the display is brilliant.

 

[kingtj1971]

OWC Thunderbolt Dock from 2018, two monitors (LG 27” 5K via Thunderbolt, Dell 32” 4K mostly via HDMI) connected and no crashes. Also runs Windows 11 under Parallels Desktop with no issues. I absolutely love the lack of fan noise and the display is brilliant.

Yeah... I'm not using an actual Thunderbolt dock. I have one of those multi-port dongle type adapters on mine that provides a gigabit Ethernet port, HDMI port, several USB ports and a pass-thru USB-C connector. It was $60 or so off Amazon. Looks nice, in that it matches the space grey color of my notebook .... but definitely could be part of what triggers the crashes after it sleeps with the lid closed.

 

[lewisflude]

As a small update, the crash when sleep bug is still occurring.

panic(cpu 0 caller 0xfffffe002444e6bc): Sleep transition timed out after 35 seconds while calling power state change callbacks. Suspected bundle: com.apple.driver.AppleBCMWLANCore. Thread 0x37300.

 

[mustbmac]

It's now been a few months since the M1 Pro and M1 Max MacBook Pro models launched in October, and MacRumors video editor Dan Barbera has been using one of the new machines since they debuted. Over on the MacRumors YouTube channel, Dan has shared a three month review of his MacBook Pro to see how it has held up over time and how it's changed his workflow.

Subscribe to the MacRumors YouTube channel for more videos.​

Dan is using the 16-inch MacBook Pro that features an M1 Max chip with 10 compute cores and a 32-core GPU, aka the top of the line chip. The machine features 32GB RAM and a 1TB SSD, and it is the high-end pre-configured version that Apple offers.

In usage, the MacBook Pro is close to perfect, but Dan does go over a handful of complaints in his video. The MacBook Pro is subject to crashes and restarts when used in clamshell mode, which could be attributed to third-party Thunderbolt accessories.

The SSD can also malfunction randomly when transferring footage from an external drive. At times, transfer speeds slow to a crawl, an issue fixed with a restart. These are the two major issues that Dan has experienced, and he has otherwise had no problems. Performance overall is excellent, the display is beautiful, and the extra ports are useful.

For those considering a purchase, Dan says that he wishes that he had gone with a 2TB SSD over a 1TB because of his video editing needs, and he's not sure that he's happy with the weight and size of the 16-inch MacBook Pro when the same functionality is available with the smaller and more portable 14-inch MacBook Pro.

Make sure to watch Dan's video for his full review, and let us know in the comments how you're liking your 2021 M1 Pro/M1 Max MacBook Pro if you have one.

Article Link: Review: M1 Max MacBook Pro After Three Months

As a stand alone, the new Macbook with M1 Chip is great. However, when connected to a NAS Synology Server the new Macbooks with M1 chip are a DESASTER. At least that is our experience with several MacBooks Pro with M1 chip. For example: opening PDF's on the server doesn't work: the Macbook Pro with M1 crashes. Looking up files with the Finder on the NAS Synology server? The Macbook Pro crashes. Our older IMacs and Macbooks do not have these problems. The problem is clearly with the new Macbook with M1 chip. Theses issues are mentioned in several fora. But for some reason, Apple continues ignoring them.

 

[MrMJS]

As a small update, the crash when sleep bug is still occurring.

Same with me... Daily crashes and wasted time. Tired of Apple using their customer base as beta testers. Fix your junk software/Hardware, please... I wish I hadn't bought an M1 Macbook "Pro"

panic(cpu 3 caller 0xfffffe001c482bf8): Sleep transition timed out after 35 seconds while calling power state change callbacks. Suspected bundle: com.apple.driver.AppleT8103PCIeC. Thread 0x37dcd.

 

[lilkwarrior]

As a stand alone, the new Macbook with M1 Chip is great. However, when connected to a NAS Synology Server the new Macbooks with M1 chip are a DESASTER. At least that is our experience with several MacBooks Pro with M1 chip. For example: opening PDF's on the server doesn't work: the Macbook Pro with M1 crashes. Looking up files with the Finder on the NAS Synology server? The Macbook Pro crashes. Our older IMacs and Macbooks do not have these problems. The problem is clearly with the new Macbook with M1 chip. Theses issues are mentioned in several fora. But for some reason, Apple continues ignoring them.

I would say that it's very possible it's a Software Engineering bug that's not necessarily on Apple instead of Synology's software engineers.