thadec

nubus said:

Both iMac and Studio should have received M2 by now. Those are premium products. With MBP 13 and mini - the two low-end products on M2 (and the mini using M2 Pro) iMac and Studio are in need upgrades this quarter. What could be the reasons for Apple to let their premium desktop products stay on old tech?

To understand why Apple didn't "update" their premium desktops to the M2, you need to leave the Apple bubble a bit. All of the Apple fans - which includes pretty much everyone in the tech and mainstream media except sites committed to Windows - are raving over how improved the M2 is over the M1. Benchmarks: 
M1 benchmarks. Single core: 1651 Multicore: 5181
M2 benchmarks. Single core: 1951 Multicore: 9003
M2 Pro benchmarks. Single core: 1952 Multicore 15013.
Fantastic right? Actually, no. 

To clarify:
Mx = Intel Core i3 and AMD Ryzen 3 (Apple has an advantage here)
Mx Pro = Intel Core i5 and AMD Ryzen 5 (a wash)
Mx Max = Intel Core i7 and AMD Ryzen 7 (Apple loses their advantage)
Mx Ultra = Intel Core i9 and AMD Ryzen 9 (Intel and AMD are clearly ahead)
Mx Extreme = Intel Xeon W and AMD Threadripper (the Extreme doesn't exist yet so ...)

I will grant you: the $599 Mac Mini is currently the best deal in computing. But despite TechRadar's "Apple now has no serious rivals in the computing space" claim ... the gap is a river, not an ocean or even a lake. An AMD Ryzen 5 6600H (6nm) benchmarks 1472 single core, 8054 multicore, offers RDNA2 graphics and you can get it in a mini-PC with 16 GB RAM for $589. And the AMD Ryzen 7040 series that AMD used to call out Apple over? It's 4nm Ryzen 5 7640HS with RDNA 3 graphics will be available in systems soon that will be in direct competition with the Mac Mini in a few months. (I won't list Intel here because Intel's Iris Xe graphics won't become competitive with Apple and AMD until Meteor Lake releases 4Q. More on that later.)

The AMD Ryzen 5 7600X loses to the M2 Pro in multicore (15013 to 11000) but it beats it in multicore: 2200 to 1952. The Intel Core i5-13600K beats the M2 Pro in both: single core as high as 2270 and multicore as high as 17300 on Geekbench. And that is a 10nm chip. Any guesses as to how fast the 7nm Core i5-14600K that launches 4Q2023 will be? And yes, Core i5/Ryzen 5 systems still cost less than the $1300 for an M2 Pro Mac Mini even when they have midlevel discrete GPUs like the Nvidia GeForce RTX 3060.

This is why it is better that the last batch of 5nm chips go to Apple's "low end products": the AMD and Intel competition in 2023 is cheaper and better. You may not want the premium products to wait on the 3nm M3 so that they won't be outclassed, but you can bet that Apple does. And before anyone brings up the "Apple gets much better power per watt" yes I am aware that tons of tech media types choose to take seriously "light thin devices with 12 hour battery life that barely need their fans are more important than single core and multicore performance because we say so" when it comes to laptops. With desktops? Nobody cares. If Mac desktops and workstations don't do software engineering optimization, rendering, 3D modeling, CAD etc. faster than Core i7 and Threadripper machines, then that is what the competition is going to buy.

retrogusto said:

I’ll be interested to see if anyone does benchmarks comparing the top two processors available in the new Mac Mini. I’m wondering if the fastest one will have any issues with throttling due to overheating in that small enclosure. 

What small enclosure? I was pointing this out in a "how was Apple able to drop the price on the Mac Mini" article on PCMag yesterday ... the Mac Mini enclosure is HUGE. It is 86 cubic inches, essentially unchanged from 2010. You have Intel Core i7 and AMD Ryzen 7 mini-PCs these days that have 30 cubic inch enclosures - practically 1/3 the size - that include heatsinks and fans. And you have Pentium and Celeron mini-PCs that are 18 cubic inches. An example: the Intel Core i7-1255U is a 10nm chip with 10 cores, and it is in the Acer Chromebox CXI5 that is so small that it will be able to fit inside the mounting bracket of a monitor (where the monitor stand attaches to the back of the monitor screen) to create an All-in-One: https://www.xda-developers.com/acer-chromebox-cxi5/

Yes, the M2 Pro's 1952 single core Geekbench score is faster than the i7-1255U's 1739 ... but not by a whole lot. (By the way .... the i7-1255U is for thin and light notebooks, which is why it ised used in mini-PCs. Their performance Core i7 chip, the i7-12800HX, used for workbooks and lower end workstation laptops, is a wash in single core and actually beats the M2 Pro in multicore. And this does not get into the Core i7 chips that actually are for desktops instead of laptops like the i7-12700K.) 

The M2 Pro has 2 more cores - both performance - as well as way more GPU cores than the i7-1255U, but it is also on a TSMC 5nm process optimized for efficiency - 5NP - so it should definitely be able to fit into that Chromebox case. So even though the Mac Mini "could" be as small as an Apple TV 4K, designing a new exterior for it would have driven up the price. But this means that throttling and overheating definitely won't be a problem. 

But down the line, Apple does need to make the Mac Mini smaller. If the 10nm Intel Core i7 CPU mini-PCs are that small, you can only imagine how small they will be  in 1Q2025 when the 5nm (Intel calls it 20A) chips will be in mini PCs. When that happens, tiny mini-PCs that offer the same CPU and graphics performance as these massive Mac "Minis" will be an embarrassment. (Yes, the graphics performance thing will happen. Starting with 13th gen, Intel is going to have their discrete GPUs, which are already being manufactured by TSMC, manufactured as "tiles" that will become integrated GPUs.) So a redesign will need to happen, even if Apple can't meet the $599 price point.

keithw said:

Not to be ungrateful, but where are the Mac Studio updates with the M2 Pro, Max and Ultra options?

Dude, what is up with you? I am not even an Apple guy - as you can probably tell from my comment history, my job is to be a contrarian when applicable - but why would you even expect an annual upgrade cycle from a company that makes the chips, OS and computers themselves? 

You get annual updates with Dell and HP workstations because they only have to make the PC.
They get the chips from Intel and AMD which are refreshed annually.
They get the OS from Microsoft (and for workstations the various entities that provide the Linux distros).
They only have to take those parts and use them to build "new" computers that rarely meaningfully change from year to year ... they just need to swap out last year's Intel/AMD CPU with this year's and keep everything else the same. I still remember the freakout that Intel caused when they stated that 12th gen and higher wasn't going to be backwards compatible with most older motherboards (even if you could get it to work it wouldn't be supported or covered by warranty). 
That the PC manufacturers get everything else supplied to them that they only have to use to make - or more accurately have Foxconn and other white box types make for them - the PCs practically eliminates the R&D and manufacturing costs that Apple has to cover by themselves. That means that it makes no financial sense for Apple to refresh their Macs every year like they do their smartphones. 

Another thing that makes refreshing smartphones each year sensible? That Apple sells hundreds of millions of them each year. Note that Apple TV, which sells in far lower volumes, gets refreshed much less frequently. Same deal with Macs. Apple sells 30 million Macs in a great year, 25 million in a good year, 20 million in a normal one. That means that the market for any annually refreshed device but the MacBook Air and MacBook Pro - their biggest sellers, and even that is only the entry level MBA and the 13" and 14" MBP - won't be big enough to sustain it profitably. This is not like PCs, which outsells Macs by a factor of 10, meaning that there will always be a decent number of people looking for a new HP or Acer in any given year. 

The Mac Studio is a $2000 device that only makes sense for video and photo editors and musicians. For literally everyone else it is a terrible product because you can either get similar performance for way less than $2000 or a lot more performance for that $2000. Yes, that AMD Threadripper or Intel Core i9 workstation will be a lot bigger, noisier and draw a lot more power but it will do everything that doesn't need the prores codecs - again everything but photo, video and music stuff - faster and cheaper. To expect Apple to upgrade it any more often than every 2 years makes no sense. At least if it gets upgraded in 2023 then it will be a major upgrade from a 5nm M1 Ultra to a 3nm M3 Ultra. Merely going from a 5nm M1 to a 4nm M2 is lame. It only made sense to do it for the Mac Mini, MacBook Air and 13" MacBook Pro because they were still on the original 2020 M1, not the improved M1 that came out in 2021. 

No, AMD's competition is not Intel because they can't beat Intel at raw power, which is what the x86 people care about. So, AMD is attempting to go after Apple fans by proclaiming great power per watt. 

AMD is not going to convince the x86 crowd that - all of a sudden after all these decades - the ability to make thin and light devices that have 12 hour battery life is more important than single core performance. AMD has played this game for ages too, and everyone would know that if they shift gears after all this time it will be an admission that they can't win it; that they have no answer for things like this: https://wccftech.com/intel-core-i9-13900ks-worlds-first-6-ghz-cpu-now-available-for-699-us/

So they are going to the "on second thought power-per-watt is what REALLY matters crowd" and saying "hey, can beat Apple on power per watt too!" No, their chips won't beat the M1 Max or M1 Ultra, but who cares. A laptop - these are laptop chips so they won't compete with the Mac Studio - with the M1 Max starts at $2900! AMD Ryzen 9 systems start at $1500. 

They are letting people who have been paying $2000 for entry level 14" and 16" MacBook Pros that they can get the same power per watt for hundreds less. That argument isn't going to work against Intel, because no one who buys Intel cares about the latest Ryzen 9 having great power per watt. They care about whether the Ryzen 9 can beat the i9-12900HX in horsepower. As AMD has never matched Intel's single core performance, and as thanks to big.LITTLE the Intel Core i9 has 16 cores to the Ryzen 9's 8 (and 24 threads to the Ryzen 9's 16) it can't match the multicore/SMT performance either, it can't. 

Look, stuff like this "By showing it can be in the same ballpark as Apple, AMD has laid down the challenge to Intel, threatening to take PC customers away with promises of performance and battery life. After all, if Apple can worry Intel, and AMD is vaguely in the same area, it stands to reason that AMD should be able to worry Intel too. If AMD can eat Intel's lunch by posturing against Apple" is a flat out refusal to deal with reality. Despite all of the grandiose claims since Apple Silicon made its debut in 2020, Macs have made no real increase in market share. According to all 3 major research firms - Gartner, IDC and Canalys - Apple had anywhere from 9% to 11% of the market last year. To put it another way, ChromeOS outsold macOS for the third year in a row. 

Get it through your heads. Apple Silicon is no threat to Intel. The only threat to Intel was Intel, meaning their badly mishandling their fab situation, meaning that they were stuck on 14nm for 6 years and then 3 years on 10nm. When Intel issues 7nm chips - possibly 3Q 2023 but no later than 1Q 2024 - that will be over. AMD knows this. They know that their advantages that were due to being 4 generations ahead of Intel on process node are going to be essentially over by 4Q 2024. Intel already adopted big.LITTLE on 12th gen. They are going to add tiles on 14th gen. And 15th gen they are going to be on a 5nm process. Their window for making the same inroads against Intel in laptops that they made against Intel for gaming desktops is over. They missed it. So they are pivoting to try to pick up some MacBook Pro consumers, because they know that at this point they aren't going to get the Dell and HP laptop customers. Several at CES, several manufacturers revealed that they are actually going Intel-only with their laptops in 2023, and they aren't going to offer any AMD variants at all. 

Intel is #1. AMD, Apple and I guess starting in 2024 Qualcomm (though to be honest ... they have no chance) are going to fight each other for #2. Good luck!

tht said:

Hardware wise, an M1 Pro in a Mac mini for $800 would be just fine as "gaming PC". If people want more, there's the Studio, but Apple will have to drop the price. Apple just aren't interesting in games for Macs.

Yeah, the cheapest device with the regular M1 costs $700 and the cheapest device with the M1 Pro costs $2000. And both those have only 8 GB RAM. Even the $400 Steam Deck has 16 GB of RAM. So A. no way Apple sells a device with an M1 Pro for $800 instead of twice that and B. it still wouldn't be "fine." It isn't just that you can frequently find a device with a Ryzen 5 or Intel Core i5 and an Nvidia RTX 3050 or AMD Radeon 6500 for about the $800 that you claim this M1 Pro Mac Mini would cost. It is that you can use that $800 x86 box as a base and significantly increase/improve the RAM, storage, GPU or even CPU over time as you can afford it in a way that you never can with an Apple Silicon device. Some gamers bought the Core i5 and Core i7 Mac Minis because you could do Windows on bootcamp with them and upgrade them. Not anymore.

Also, you are ignoring the real thing holding back gaming on macOS: small market share. 2022 was the best year for Mac market share since the combination of Windows 8 and people who fell for Apple's "an iPad can replace your PC" marketing campaign had x86 down in the dumps ... and it was still under 11%! (One analyst had it under 10%.) And most of that "less than 11%" are the entry level models running CPUs with 4 performance cores that - in gaming - would perform about as well as systems with Intel Core i5 in CPU performance and worse than those with NVIDIA GeForce 1650 in graphics (basically entry level gaming laptops). Developers aren't going to create ports of their games for such a tiny market share.

The only way that Apple is going to become a factor in gaming with their silicon is to create a console. But ... fat chance. Both Microsoft and Sony lose money on XBox and PlayStation consoles. Nintendo doesn't, but only because the Nintendo Switch is actually a smaller, cheaper version of the Nvidia Shield Android tablet from 2014. And the Shield was never a premium tablet ... the first version started at $299 and the K1 version cost $199. And the parts for the Shield and Shield K1 were state of the art in 2014 and 2015. When the Switch came out in 2017 they were already outdated. So now the current Switch is basically an 8 year old midrange Android tablet like the Nexus 9 (which also used the Nvidia Tegra K1).

At the very minimum, Apple would need to use an M1 in a console that costs only $500. Even that is debatable ... the PS5 and the XBox Series X cost $500 but give you 8K gameplay (in theory) and their AMD CPUs would thrash the M1, especially in graphics. The M1 is better than to the XBox Series S, which costs $300 but outputs only 1080p. So charge $400 for it and call it a day. Except Apple isn't going to lose $100-$150 on every M1 console sold under their current business model (where you lose money on the hardware in order to make it back on AAA exclusives made by studios that you own for $60-$70 apiece, not including DLC).