Apple launches new Mac mini

[Jamie McKane]

Apple launches new Mac mini

Apple has unveiled a new iteration of its Mac Mini desktop computer, which it said delivers "staggering performance" in an ultra-compact design.

Alongside a new MacBook Pro and MacBook Air, the 2020 Mac Mini is one of the first Apple systems to feature the company's own M1 System-on-Chip (SoC), which includes an 8-core CPU and 8-core GPU.

 

[system32]

Looks wonderful.

Has Apple purposefully blocked Windows and Linux installation with a locked bootloader?

 

[RudderVator]

Looks wonderful.

Has Apple purposefully blocked Windows and Linux installation with a locked bootloader?

Don't think it would even be possible to run windows at all with the new architecture?

 

[system32]

Don't think it would even be possible to run windows at all with the new architecture?

Windows 10 (and Windows 8) is available on ARM.
Windows 10 ARM runs on on RPI4 and Surface Pro X (SQ2 ARM processor)

From a technical perspective, it should be relatively simple to make Windows 10 work.

 

[RudderVator]

Windows 10 (and Windows 8) is available on ARM.
Windows 10 ARM runs on on RPI4 and Surface Pro X (SQ2 ARM processor)

From a technical perspective, it should be relatively simple to make Windows 10 work.

Guess you learn something new each day

 

[Itsa Trap]

I'd pass on this generation, and wait to see feedback from Apple users, especially of Adobe products.

 

[backstreetboy]

Windows 10 (and Windows 8) is available on ARM.
Windows 10 ARM runs on on RPI4 and Surface Pro X (SQ2 ARM processor)

From a technical perspective, it should be relatively simple to make Windows 10 work.

Windows 10 on ARM doesn't support 64 bit apps yet whereas Apple went 64 bit only yonks ago...

 

[Johnatan56]

Performance numbers so far seem to be based on an old Geekbench benchmark which had the issue of anything with a larger cache not giving an accurate depiction of performance, since a short burst for anything with a very large cache would score similarly.

Supposedly though the emulation is pretty good, rumors state similar performance to Intel's 10W parts when using x64 emulation. Should probably be credible since Intel's chips are more designed to be at the 20-25W mark, plus the huge process node lead (14nm Intel vs 5nm TSMC).

Original source from: https://semiwiki.com/semiconductor-...foundry/8157-tsmc-and-samsung-5nm-comparison/

TSMC has a 4x density advantage.

Note TSMC 7nm is ~97, 7nm FFP is 114. AMD for example is still on the 97 one for the new Zen architecture AFAIK, so Apple should be using a 50% higher density process. Note it's not a linear scale for density vs performance.
TSMC 5nm should be +15% performance at the same power usage, or 30% less power at the same performance.

That Apple is about the same performance is quite a big deal, but a lot of that is from the process node probably. Once AMD released their next gen on 5nm, Zen will widen the gap significantly.

But these machines are great for casual users, browsing web, watching videos.

Don't think it would even be possible to run windows at all with the new architecture?

Also interesting that .Net Core/5 now are all cross platform, including ARM and I think MacOS. So if Apple allows .Net to run natively on it, they will have a pretty easy app adoption process, but then they'd have to start giving up their control a little bit.

 

[G.A.S]

Windows 10 (and Windows 8) is available on ARM.
Windows 10 ARM runs on on RPI4 and Surface Pro X (SQ2 ARM processor)

From a technical perspective, it should be relatively simple to make Windows 10 work.

As far as I am aware, Windows 10 ARM is OEM only. And that is assuming you can even access the bootloader.

 

[system32]

Performance numbers so far seem to be based on an old Geekbench benchmark which had the issue of anything with a larger cache not giving an accurate depiction of performance, since a short burst for anything with a very large cache would score similarly.

Supposedly though the emulation is pretty good, rumors state similar performance to Intel's 10W parts when using x64 emulation. Should probably be credible since Intel's chips are more designed to be at the 20-25W mark, plus the huge process node lead (14nm Intel vs 5nm TSMC).
View attachment 950774
Original source from: https://semiwiki.com/semiconductor-...foundry/8157-tsmc-and-samsung-5nm-comparison/
View attachment 950776
TSMC has a 4x density advantage.

Note TSMC 7nm is ~97, 7nm FFP is 114. AMD for example is still on the 97 one for the new Zen architecture AFAIK, so Apple should be using a 50% higher density process. Note it's not a linear scale for density vs performance.
TSMC 5nm should be +15% performance at the same power usage, or 30% less power at the same performance.

That Apple is about the same performance is quite a big deal, but a lot of that is from the process node probably. Once AMD released their next gen on 5nm, Zen will widen the gap significantly.

But these machines are great for casual users, browsing web, watching videos.


Also interesting that .Net Core/5 now are all cross platform, including ARM and I think MacOS. So if Apple allows .Net to run natively on it, they will have a pretty easy app adoption process, but then they'd have to start giving up their control a little bit.

Anandtech has some benchmarks, and indeed M1 seems to be faster:

Apple Announces The Apple Silicon M1: Ditching x86 - What to Expect, Based on A14

www.anandtech.com

Spoiler: Anandtech Performance Leadership?

Apple claims the M1 to be the fastest CPU in the world. Given our data on the A14, beating all of Intel’s designs, and just falling short of AMD’s newest Zen3 chips – a higher clocked Firestorm above 3GHz, the 50% larger L2 cache, and an unleashed TDP, we can certainly believe Apple and the M1 to be able to achieve that claim.

This moment has been brewing for years now, and the new Apple Silicon is both shocking, but also very much expected. In the coming weeks we’ll be trying to get our hands on the new hardware and verify Apple’s claims.

Intel has stagnated itself out of the market, and has lost a major customer today. AMD has shown lots of progress lately, however it’ll be incredibly hard to catch up to Apple’s power efficiency. If Apple’s performance trajectory continues at this pace, the x86 performance crown might never be regained.

 

[Johnatan56]

Anandtech has some benchmarks, and indeed M1 seems to be faster:

Apple Announces The Apple Silicon M1: Ditching x86 - What to Expect, Based on A14

www.anandtech.com

Spoiler: Anandtech

Performance Leadership?
Apple claims the M1 to be the fastest CPU in the world. Given our data on the A14, beating all of Intel’s designs, and just falling short of AMD’s newest Zen3 chips – a higher clocked Firestorm above 3GHz, the 50% larger L2 cache, and an unleashed TDP, we can certainly believe Apple and the M1 to be able to achieve that claim.


This moment has been brewing for years now, and the new Apple Silicon is both shocking, but also very much expected. In the coming weeks we’ll be trying to get our hands on the new hardware and verify Apple’s claims.


Intel has stagnated itself out of the market, and has lost a major customer today. AMD has shown lots of progress lately, however it’ll be incredibly hard to catch up to Apple’s power efficiency. If Apple’s performance trajectory continues at this pace, the x86 performance crown might never be regained.

Those tests are all cache skewed and most comments are based on perf/watt, and as said, SPEC 2006 is a bit outdated as cache sizes have increased quite a bit over the last few years, before the test was made: https://research.spec.org/icpe_proceedings/2015/pabs/p11.pdf (take a note of cache miss hit sensitivity on the misses per size increment)
Spec2006 and 2017 are also heavily skewed towards larger L2 caches (Apple's one is 12MB vs new Zen's 8MB for example).
An interesting read of 2006 vs 2017: https://uweb.engr.arizona.edu/~tosiron/papers/2018/SPEC2017_ISPASS18.pdf

The issue here is that the Spec benchmark is not really indicative of most real-world loads.
And I repeat again: the Apple chip is definitely not bad chip, just that the benchmarks paint it in a better light than it is, and next gen Zen if it moves to 5nm will probably be substantially better.