I feel like i'm alone to dislike M1/what is it good for (Merged Mega Thread)

[Fomalhaut]

Yes, but what is there to configure on an M1 mini... from 8 to 16GB RAM is not enough... and that is what the OP (as well as myself) are complaining about. I personally do not need a Mac Studio M1 Max/Ultra for just higher RAM, is a complete overkill for my needs. Just need more RAM... Oh well.

It will be interesting to whether the base M2 SoC allows 32GB RAM. That would be a significant improvement, but I suspect it will not. I also don't think we'll see an M1/M2 Pro in the Mac Mini - Apple wants to push people to the entry-level Mac Studio I think.

As a thought experiment I specc'd a hypothetical M1/2 Mac Mini with 32GB RAM & 512GB SSD, the same as the entry level Mac Studio. It would cost $200 to upgrade the base $699 model to 512GB, and another $200 to get to 16GB RAM. Following the MBP pricing, it would cost another $400 to go to 32GB RAM, so a total of $1499. The upgrade from MBP14 7/14 core binned to M1 Max (24-core) is $500 which would get you the same price as the entry-level Studio. But this is from a binned M1 Pro, not an M1, so there doesn't seem to a lot of wiggle room in the pricing structure in which you could fit an M1 Pro Mini.

If the M2 base gets close to the M1 Pro in performance through faster CPU and GPU cores and an increase in the number of GPU cores, and has an option for 32GB, then we could probably live with the jump from M2 to Mac Studio with M1 Max....but then there will eventually be an M2 Max....so who knows!

 

[Fomalhaut]

But ther comes a point when your little 1 bed bungalow turns into a monstrocity of you adding on , adding on etc.
You get more worth - just selling up and living in a better neighbourhood.

Let's try another example.
My first car was a Yugo 45 ( No idea if it sold in the US - Google it ).

Now - I could replace the engine , gearbox etc etc. ad infinitem or buy something else fit for my purpose.

We do it with pretty much everything else we buy.

I have no idea how much work it is for a tech company to ensure that if they release some kit - it must be able to work optimum with every other bit of stuff someone decides to plug into it.

Keep the damned thing sealed and you get less hassle as a manufacturer.

When is the last time you wanted to physically get hands on in upgrading the internals of your coffee maker, tumble dryer, bedside lamp etc?

Just because - that is where we came from - it doesn't need to be the future that where we want to be.

There is a market for those who want to play and upgrade hands on.
There is also a market for those who understand that if "you want a faster car - buy a faster car"

There is room for both.

I am now in the camp of "If I want a better and faster car..... I'll just buy it"
( and I used to spend a lot of my weekends getting oily in just running / beefing up my cars. )

I'm turning into my folks.

Ah yes, the Yugo...I always imagined you had to perform maintenance on them like this:

East European cars may have had the last laugh though. I bought a Skoda (essentially VW/Audi components with a cheaper exterior), and it has been an excellent car

 

[Unregistered 4U]

If the M2 base gets close to the M1 Pro in performance through faster CPU and GPU cores and an increase in the number of GPU cores, and has an option for 32GB, then we could probably live with the jump from M2 to Mac Studio with M1 Max....but then there will eventually be an M2 Max....so who knows!

The M2 base will likely surpass the M1 Pro in single threaded operations, but not likely much else. Mainly because the M2 base will have fewer CPU and GPU cores, just like the M1 base has fewer CPU and GPU cores than the M1 Pro, Max and Ultra.

 

[JonathanX64]

The concept of «upgradeability» is stupid, it has never truly worked, and people only want it back because Apple charges too much for the parts.

I'm dead serious on this one. Desktop PCs are not upgradeable either.

Let's say a person has an Intel Coffee Lake system. It won't run NVMe drives at PCIe 4.0 speeds because chipset is limited to 3.0 anyway, it won't run faster RAM because memory controller is limited to DDR4-2667 standard, and it won't allow to install a more efficient CPU because newer CPUs are incompatible with LGA1151v2. The only possible upgrade options in this case are to replace one 8th gen CPU with another 8th gen CPU, or to replace one PCIe 3.0 SSD with another PCIe 3.0 SSD. But this computer still has old and inefficient platform from 2017.

People only ever did upgrades to save money. Instead of paying Apple $100 for 2 more gigs of RAM, people bought third-party DDR3 sticks for $50 and installed it themselves, and so on.

Any request for Mac upgradeability is nothing more than a simple cry for a discount. As soon as Apple lowers prices for non-stock Mac configurations, this subject will be long gone, forgotten, and will never pop up again in discussions.

 

[kaioshade]

The concept of «upgradeability» is stupid, it has never truly worked, and people only want it back because Apple charges too much for the parts.

I'm dead serious on this one. Desktop PCs are not upgradeable either.

Let's say a person has an Intel Coffee Lake system. It won't run NVMe drives at PCIe 4.0 speeds because chipset is limited to 3.0 anyway, it won't run faster RAM because memory controller is limited to DDR4-2667 standard, and it won't allow to install a more efficient CPU because newer CPUs are incompatible with LGA1151v2. The only possible upgrade options in this case are to replace one 8th gen CPU with another 8th gen CPU, or to replace one PCIe 3.0 SSD with another PCIe 3.0 SSD. But this computer still has old and inefficient platform from 2017.

People only ever did upgrades to save money. Instead of paying Apple $100 for 2 more gigs of RAM, people bought third-party DDR3 sticks for $50 and installed it themselves, and so on.

Any request for Mac upgradeability is nothing more than a simple cry for a discount. As soon as Apple lowers prices for non-stock Mac configurations, this subject will be long gone, forgotten, and will never pop up again in discussions.

This is...certainly a take.

Your analogy is a bit flawed however. (And I am the proud owner of a 14" M1 Pro)

Your comparison infers you have to throw out the entire PC. Power Supplies, RAM, Hard drives, GPUs, are all reusable, since they are not soldered to the board. Get a new motherboard and CPU, and you are good to go. Certainly easier (provided you have the knowledge to do so) and cheaper than buying an entire new desktop.

So to say Desktop PCs cannot be upgraded is woefully ignorant.

 

[easy4lif]

It would be nice if things were upgradable, but the fact they can’t be shouldn’t be a surprise anymore. They have been that way for a long time.

The compatibility issue will resolve itself over time. Overall I’ve been quite impressed with how smoothly it’s gone.

U summed it up nicely, unfortunately, your also in the minority.

The vast amount of people I’ve recommended macs to just want a computer that takes them from A to B. Most came from windows after leaving a bad taste in their mouths. These same people wouldn’t know the difference between a intel or arm chip. It’s prob why apple is so successful, it just works.

 

[Fomalhaut]

Good thing that the SSD isn’t soldered? Just socketed.

With the commoditising of devices and computers, no wonder device as a service is likely coming.

You might argue that "device as a service" is already here with the plethora of cloud desktop services available. It's certainly quite popular with enterprises because of the control and cost savings it can bring to the table. Of course you still need a computer to access these services, so it's somewhat debatable how much it saves unless you have some spare lower spec computers, or an employer has a BYOD policy and gets the users to provider their own client machines.

I've not used any of the virtual gaming desktop services, but I hear they are quite good for what they cost.

My own experience with Windows desktop services is that they are "not bad", but still a bit clunky compared to having a good physical machine on your desk. Network bandwidth and latency are still an barrier for a seamless experience that is indistinguishable from a physical machine.

In the enterprise apps and server world, there is a strong move toward cloud services of course.

 

[Fomalhaut]

The Register of UNIX® Certified Products

And as for dead...You might want to tell HP that given that they still update HP-UX...

Interesting that Solaris in no longer on that list (if it ever was?)

In any case, it's a shorter list than I expected, especially when compared to the list of Linux distros

 

[Fomalhaut]

The M2 base will likely surpass the M1 Pro in single threaded operations, but not likely much else. Mainly because the M2 base will have fewer CPU and GPU cores, just like the M1 base has fewer CPU and GPU cores than the M1 Pro, Max and Ultra.

I don't think M2 base will surpass M1 Pro, but it may be getting close to the double-binned version of the Pro.

If CPU cores stay the same and have a 30% multi-core improvement, so 130% of the M1, then a 6 core M1 Pro is maybe 140% better than M1 (50% more perf cores, but only half the efficiency cores). For GPU if the M2 get 10 cores with 30% improvement, then it's 163% of the M1, and the M1 Pro (14-core GPU) is 175% on the M1.

I've no idea if 30% improvements on both CPU and GPU cores are likely, but it is conceivable.

A Mini with M1/M2 Pro would be a sweet spot, but I think Apple is more likely to just point interest buyers to the Mac Studio....

 

[Unregistered 4U]

I don't think M2 base will surpass M1 Pro, but it may be getting close to the double-binned version of the Pro.

If CPU cores stay the same and have a 30% multi-core improvement, so 130% of the M1, then a 6 core M1 Pro is maybe 140% better than M1 (50% more perf cores, but only half the efficiency cores). For GPU if the M2 get 10 cores with 30% improvement, then it's 163% of the M1, and the M1 Pro (14-core GPU) is 175% on the M1.

I've no idea if 30% improvements on both CPU and GPU cores are likely, but it is conceivable.

A Mini with M1/M2 Pro would be a sweet spot, but I think Apple is more likely to just point interest buyers to the Mac Studio....

For single threaded operations only, I do expect M2 base will surpass M1 Pro. The main reason why is because, unlike AMD and Intel, Apple has delivered all M1 solutions with similar (impressive) single threaded operation performance.

The M1 8 core only scores 132 less than the Ultra. If the M2 base is 20% faster than the M1 base at single threaded operations, that’d put it at 4516, which is better single core performance than the current M1 Ultra. Even if Apple’s only able to obtain a 5% improvement on the base M1 performance, that would still yield a single core score of 3952, which is still better than the M1 Ultra.

That 4516 score would actually put it above Intel’s i9-12900KF as well. Consider that the base single threaded performance of any M2 system could potentially start that high up on the chart. And, those would be Apple worst performing M2‘s! Interesting times, indeed!

 

[Fomalhaut]

For single threaded operations only, I do expect M2 base will surpass M1 Pro. The main reason why is because, unlike AMD and Intel, Apple has delivered all M1 solutions with similar (impressive) single threaded operation performance.
View attachment 1983924
The M1 8 core only scores 132 less than the Ultra. If the M2 base is 20% faster than the M1 base at single threaded operations, that’d put it at 4516, which is better single core performance than the current M1 Ultra. Even if Apple’s only able to obtain a 5% improvement on the base M1 performance, that would still yield a single core score of 3952, which is still better than the M1 Ultra.

That 4516 score would actually put it above Intel’s i9-12900KF as well. Consider that the base single threaded performance of any M2 system could potentially start that high up on the chart. And, those would be Apple worst performing M2‘s! Interesting times, indeed!

What benchmark is this? Geekbench 3 / 4? It's better to use the de facto standard of Geekbench 5 for comparison, because these numbers don't look like anything I've seen in the last few years.

But to your point; yes, it is a given that single-core results are likely to be in the high 1900s or better (GB5), which will probably be noticeable (M1/M1 Pro/Max is about 1760)

M2 multi-core and Metal benchmarks will no doubt be better, but I don't think anyone really knows by how much. It would be (pleasantly) surprising if it beat the M1 Pro in any metric, but even if it doesn't, it will close the gap between...until the M2 Pro is released of course

That said, I don't think it will convince a large proportion of M1 owners to upgrade so soon.

 

[root42]

1) You cannot get Low Power RAM modules in any other package. To move away from soldered RAM, they’d need to lose performance and increase power consumption.

Honestly, this is I think rather a choice than a technical limitation. I don't see why the RAM couldn't be soldered onto an S0-DIMM or similar module. But someone with more knowledge in RAM technology might correct me, if I'm wrong...

 

[Richu]

You can disagree, but the facts speak for themselves. x86 shipments are dropping and Arm shipments are going up

View attachment 1983678

The biggest driver for Arm will be the cloud - cloud providers need the lowest powered, high performing architecture available - and that’s Arm. There’s already a massive number of container images that have been ported to Arm - and more turn up every day.

Many Linux distro’s are already Arm based, heck Microsoft are getting on the act with Windows 11 and their ‘version’ of Rosetta for x86 and x64 apps.

And if that wasn’t enough - we’re still finding critical issues with Intel architecture - years after Spectre and Meltdown first crashed onto the scene - all of which makes betting on an Intel future very dodgy.

Sure x86 isn’t going to die tomorrow - it’s probably got another 10-15 years of life in it. But don’t kid yourself that its growing because it’s not.

There’s a ‘new’ kid on the block and its Arm.

And while Apple may be leading this public charge with their “Apple Silicon”, Arm themselves have some amazing chips coming out - such as the Neoverse chips, and Amazon’s take on the v2 - the Graviton3.

Those estimation charts are just some analysts excel formula… I’ve done those as well. It’s an opinion, not facts

 

[Fomalhaut]

Those estimation charts are just some analysts excel formula… I’ve done those as well. It’s an opinion, not facts

The graph seems to be just a (somewhat optimistic) estimate, but I do agree that we will see more ARM-based computers over the next 5 years. I expect a lot of those to be in the server space because of the significant cost savings for data centers in terms of $/Watt for a given performance level.

I think adoption on consumer devices (other than phones & tablets) will be quite slow though. Apart from Apple, no other vendors seem to be putting much of a focus on ARM-based laptops or desktops. Maybe Microsoft or Samsung will surprise us by unveiling a competitor to the M1...but I'm not holding my breath.

 

[Larsvonhier]

I would seriously be interested in what emulator you run and what drawbacks you’ve found. I’m assuming that if you have found some then they aren’t serious, and if I move to a new M platform I may want to try it.

The only drawback atm is that the legal status is somewhat "murky", as Microsoft does offer developer builds of Win10 and Win11 for ARM that run fine on the M1 processors. But it is unclear if they ever un-bundle the OS for buyers that do not obtain it with a Windows tablet or ARM laptop. The silence of Parallels is deafening now (they initially tried to negotiate with M$ about a bundle of PD with Windows ARM). edit: see post below!
UTM is also fine running Win(ARM), but system integration into macOS is less "magic" than Parallels´.

No other drawbacks found, I use the combo (Parallels, Win10/11 and Monterey on M1 Air) even with heavy apps like Altium Designer for ECAD. GPU performance is also better than any other machine´s here (incl. 12core MP5,1 with metal capable cards) - I also just for the fun of it ran marble marcher (mandelbulb based mini-game) in a good framerate and resolution on it.

 

[Larsvonhier]

Until about 6 months ago that involved signing up to a "Windows Insiders Program", agreeing to a lot of T&Cs and downloading a "preview" version of Windows 11 that was only licensed for test and evaluation purposes (whether or not you managed to activate it with an existing key). You probably weren't likely to get sued, but good luck if you use it for work and get hit by a software audit.

Then, with no fanfare beyond an addendum to a post on a Parallels support forum, it became possible to download the release version of Windows 11 for ARM, buy a license from Microsoft with no mention of joining any evaluation-only "insiders" program and activate it on a M1 Virtual machine. However, Microsoft don't list M1 as a supported processor (have fun debating whether that refers to virtualisation or not) so, presumably, if it breaks that's just tough.

That make it fine for personal use (not like you were counting on MS for support anyhow), but really non-ideal for business/professional use.

...and, yeah, the x86 emulation is great if it "works for you" but it's not really a replacement for people running more demanding x86 windows apps, probably under Bootcamp (which ain't gonna happen on Apple Silicon).

Personally, I could live without x86 - all I'm saying is that if you actually need to run x86 OSs for anything non-trivial then Apple Silicon may not be great for you.

Thanks for the update on Parallels Desktop and Win licence. Did not hear about that, so a good hint.
Heavy lifting apps do run, even faster than on other Macs on Win (bootcamp). Altium Designer latest revision is such a case, i.e.
Found it way more usable than on a 12core Mac Pro...
As for the "good luck" section of your advice/opinion: We rely on Macs (even unsupported ones through Open Core and DosDude patches) for our business since almost 10 years now. And I can say, luck never let us down. In contrary, other branches of our company that do traditional PC computing have way more hassle. (But are not such techies, I admit).

 

[Lihp8270]

Honestly, this is I think rather a choice than a technical limitation. I don't see why the RAM couldn't be soldered onto an S0-DIMM or similar module. But someone with more knowledge in RAM technology might correct me, if I'm wrong...

You need significantly more power to compensate for the socketed connection. You can’t run the sockets at the voltages that LPDDR uses.

LPDDR4 can be run at voltages as low as 0.6v

 

[jdb8167]

Bear in mind that some of the earliest consumer home computers were also just appliances, with little upgrade capability, e.g. Sinclair Spectrum, Commodore VIC20/64, BBC Micro etc. - although you could plug peripherals into them (including "RAM packs" that must have had horrible performance). In most cases, if you wanted more internal RAM you had to buy a new computer...same as today! It wasn't until the advent of the Apple II and IBM PC clones that the idea of expansion really took off (I'm sure there were some other options out there too).

You’ve got your computer timelines a bit messed up. The Apple II was introduced in 1977 and is one of the oldest commercial home computers and predates everything else in your list. The Sinclair Spectrum was 1982, the VIC20 was 1980 and the BBC micro was 1981.

All early personal/home computers had slots for upgrades because almost nothing was standard on those machines. Floppies, printer boards, and display cards were very commonly found in slots. Sorry, you got it backwards. Upgrades were first and appliances came later.

 

[russell_314]

This is...certainly a take.

Your analogy is a bit flawed however. (And I am the proud owner of a 14" M1 Pro)

Your comparison infers you have to throw out the entire PC. Power Supplies, RAM, Hard drives, GPUs, are all reusable, since they are not soldered to the board. Get a new motherboard and CPU, and you are good to go. Certainly easier (provided you have the knowledge to do so) and cheaper than buying an entire new desktop.

So to say Desktop PCs cannot be upgraded is woefully ignorant.

I think it's a more complicated answer than just they cannot be upgraded.

You can upgrade a desktop PC within the short period that hardware is supported. After that it just doesn't make sense. Most Windows laptops aren't upgradable except the SSD and RAM and many don't even support that.

CPU... Most of the time not worth it. With Intel CPU's and how they love to change sockets it's likely you will a new motherboard for a new CPU.

GPU... Maybe worth it if you must stay current but if you're running a CPU more than a few generations old then it doesn't really make sense especially with the inflated prices. On a laptop this was never a realistic thing. Some brands attempted to use it as a selling point but no real upgrade path ever existed.

RAM can be upgraded providing it's the same type. If you configured it right from the start it shouldn't be needed but it's one area that can make sense. Unfortunately I don't think it's possible to get the performance Apple is getting with removable RAM. At least not with the current DDR5 but who knows what will come out in the future.

Hard drive. I'm guessing you mean SSD. I will say that's the one component that makes sense to upgrade! I wish it was easier on Apple computers but I think it would mean a speed hit.


Basically if you have a 5+ year old desktop PC it's just not worth spending money other than maybe an SSD to upgrade. Everything about it is soon to become almost unusable for it's original purpose. Sure you can browse the web on it but you can do the same on a Chromebook. This is why the whole it's upgradable or spending a fortune when buying new to make it "future proof" is such a bad idea...

My opinion is people should buy something that suits their wants and needs. The market will supply what people will buy. If upgradable is a factor in what people look for when shopping for a PC then it's going to be something manufacturers pay attention to. I don't think it is outside of a few tech nerds. I guarantee if I ask my friends if their computers are upgradable the only ones that will even have a clue are the gaming nerds I know. My non nerdy friends would say IDK when it breaks I'll get a new one. To them it's a tool to get a job done rather than a hobby to tinker with.

My prediction is computers are going to get much smaller and have less components. Maybe they will keep a little bit of the current Lego system to build a PC just to satisfy that market but I don't think it's going to be the norm. This trend has been going on since the PC came out and I don't see it changing directions.

 

[bobcomer]

It will be interesting to whether the base M2 SoC allows 32GB RAM. That would be a significant improvement, but I suspect it will not. I also don't think we'll see an M1/M2 Pro in the Mac Mini - Apple wants to push people to the entry-level Mac Studio I think.

As a thought experiment I specc'd a hypothetical M1/2 Mac Mini with 32GB RAM & 512GB SSD, the same as the entry level Mac Studio. It would cost $200 to upgrade the base $699 model to 512GB, and another $200 to get to 16GB RAM. Following the MBP pricing, it would cost another $400 to go to 32GB RAM, so a total of $1499. The upgrade from MBP14 7/14 core binned to M1 Max (24-core) is $500 which would get you the same price as the entry-level Studio. But this is from a binned M1 Pro, not an M1, so there doesn't seem to a lot of wiggle room in the pricing structure in which you could fit an M1 Pro Mini.

If the M2 base gets close to the M1 Pro in performance through faster CPU and GPU cores and an increase in the number of GPU cores, and has an option for 32GB, then we could probably live with the jump from M2 to Mac Studio with M1 Max....but then there will eventually be an M2 Max....so who knows!

You can't forget about cooling something like the more advanced M1's -- the mini doesn't have near the cooling capacity of the studio.

 

[yabeweb]

To me Apple silicon makes Macs Mac again, while I loved the speed bump from G5 to Intel, it always felt wrong.

Loving my M1 Pro 14 inches,sure I do not need windows, but I think it is quite compatible and seamless, even more so than G series to Intel.

 

[PsykX]

1/ Zero upgrade possiblity.

You should pay 2X or 3x times the normal price with our configuration."

Of course it's an SoC. You can't have the advantages of an SoC without its disadvantages. It cannot be upgraded. Shame you cannot change the SSD though.

But the price of these Apple Silicon machines is actually agressive. Every Apple Silicon machine can be compared with similarly spec'ed PCs and the price will usually be very similar, but with a Mac you get... better audio, display, trackpad, webcam (sometimes), materials, design, build quality, and so on.

 

[Tyler O'Bannon]

I believe the shortcomings, especially compatability, will be addressed quickly. Next Gen will probably solve a lot, by 3rd Gen, these issues will be all but history.

 

[ct2k7]

I remember upgrading my MacBook Pro with an SSD years after it AppleCare ended. Those were the good times.

 

[4743913]

The M1 is an awesome computer, but its only HALF of a computer. No Boot Camp (whatever I dont care whose fault it is, apple, microsoft) really cripples a Mac. I hope this gets corrected in the future. Parallels and Crossover are poor solutions.