Mac Mini DTK A12Z faster than MBP 16" i9 in native benchmarks.

[Homy]

8-core A12Z 17% faster than 8-core 2.3 - 4.8GHz i9 in native benchmarks. Rosetta means a 52% performance hit. It's like a 15" MacBook Pro from 2012.

Mac mini with A12Z chip benchmarked

But the scores don't show how well native apps made for the new Apple Silicon processor will run

www.macworld.co.uk

UPDATE!
They've deleted all the evidence, probably after Apple's request. The original German article about the test on Macwelt, Macworld's translation and Macwelt's video have all been removed. As you can see the link to Macworld's article directs you to and old article but in the thumbnail you can see the original article was different. I read somewhere that developers are not allowed to share benchmarks of the DTK so maybe that's the reason. The results are from a CPU intensive test Macwelt developed by themselves. Fortunately Google still has cached images of the results and I found another report that is still online. I share them below.

Apple Silicon Speed Compared To MacBook Pro

Apple Silicon Mac versus Intel Core i9 Macbook Pro

screenrant.com

ROSETTA 2

INTEL MBP 16"

MAC MINI DTK A12Z NATIVE TEST

 

[NotTooLate]

Lets hope for "one last thing" in the upcoming September event , i think everyone are ready to see how the new Macs will run.

 

[Homy]

Lets hope for "one last thing" in the upcoming September event , i think everyone are ready to see how the new Macs will run.

Or in the iPhone 12 October event.

 

[antipodean]

The iPhone 12 screenshot leaks showing a 240FPS 4K video setting suggest that any Apple silicon based on an A14 (A14X?) should have some very impressive hardware encoding abilities. How about encoding a 30fps 4K video in HEVC at 8x faster than real time?

I think the FCPX demo that Apple did with the mac mini DTK at WWDC is just a teaser. The FCPX performance of an Apple Silicon MBP should be quite something.

 

[Aenean144]

Yes, Apple Silicon will be very very good. A14 derived chips are going to outclass A12 derived chips by a lot!

However, Macworld UK is writing some spurious conclusions regarding Rosetta performance and A12Z performance. Every processor will have their strengths and weaknesses depending on what type of algorithm or math is being used or how much they stress the memory subsystem, and which part of the memory subsystem; and, how good the compiler is. Geekbench or SPEC has a suite of 15 to 30 tests, stressing various parts of the chip, and their results represent an aggregate performance comparison.

Macwelt ran only 3 tests: a random number generator, an FFT, and a ray tracer. The A12Z could do those really well, while doing other things not as well. They really need to have more tests before they can make generalized, aggregate performance statements.

 

[ChromeCloud]

I'm looking forward to the reveal of the first Apple Silicon MacBook Pro.

I'd love to see a demo during the keynote where the new MacBook Pro completely destroys the fastest Intel PC laptop available on the market. While drawing half the power.

 

[Azrael9]

8-core A12Z 17% faster than 8-core 2.3 - 4.8GHz i9 in native benchmarks. Rosetta means a 52% performance hit. It's like a 15" MacBook Pro from 2012.

Mac mini with A12Z chip benchmarked

But the scores don't show how well native apps made for the new Apple Silicon processor will run

www.macworld.co.uk

So there is a strong possibility that AS chips (With a 50%+ speed boost over Intel chips...) could run emulation the same or faster...and native AS apps will burn rubber.

Azrael.
[automerge]1598654674[/automerge]

Yes, Apple Silicon will be very very good. A14 derived chips are going to outclass A12 derived chips by a lot!

The A14. We await it's performance leap over the A12 with great interest.

Azrael.

 

[Homy]

So there is a strong possibility that AS chips (With a 50%+ speed boost over Intel chips...) could run emulation the same or faster...and native AS apps will burn rubber.

Azrael.
[automerge]1598654674[/automerge]


The A14. We await it's performance leap over the A12 with great interest.

Azrael.

Yes, A14 is said to be 30-50% faster than A13 and if they make an A14Z it could run as fast as MBP i9 in Rosetta and burn rubber in native apps.

 

[Azrael9]

Yes, A14 is said to be 50% faster than A13 and if they make an A14Z it could run as fast as MBP i9 in Rosetta and burn rubber in native apps.

Now...that's what I'm talking about.

*Der-rool...der-rool.

*A14X iPad 12.9 on shopping list.*

The Mother of all iPad updates is upon us.

Azrael.

 

[MalcolmH]

Yes, Apple Silicon will be very very good. A14 derived chips are going to outclass A12 derived chips by a lot!

However, Macworld UK is writing some spurious conclusions regarding Rosetta performance and A12Z performance. Every processor will have their strengths and weaknesses depending on what type of algorithm or math is being used or how much they stress the memory subsystem, and which part of the memory subsystem; and, how good the compiler is. Geekbench or SPEC has a suite of 15 to 30 tests, stressing various parts of the chip, and their results represent an aggregate performance comparison.

Macwelt ran only 3 tests: a random number generator, an FFT, and a ray tracer. The A12Z could do those really well, while doing other things not as well. They really need to have more tests before they can make generalized, aggregate performance statements.

Yes, and the fft test used the Accerate framework which is going to be tuned to use the Neural Engine (?). I’m sure the A14 Mac chip will be great .. but I’m not so sure that this test is extensive enough to really draw the conclusions.

 

[Yebubbleman]

8-core A12Z 17% faster than 8-core 2.3 - 4.8GHz i9 in native benchmarks. Rosetta means a 52% performance hit. It's like a 15" MacBook Pro from 2012.

Mac mini with A12Z chip benchmarked

But the scores don't show how well native apps made for the new Apple Silicon processor will run

www.macworld.co.uk

That YouTube clip is pulled. I also don't see where you can conclude that the A12Z beats the 9th Gen 8-core i9, seeing as the A12X couldn't do it and the only difference between the two is an additional GPU core. Can it toast a 16" MacBook Pro on a hexa-core i7? I'd say it's likely. But Apple can't pump out a 16" MacBook Pro on Apple Silicon until it is able to say it is categorically faster than the Intel counterpart it is replacing.

 

[Homy]

That YouTube clip is pulled. I also don't see where you can conclude that the A12Z beats the 9th Gen 8-core i9, seeing as the A12X couldn't do it and the only difference between the two is an additional GPU core. Can it toast a 16" MacBook Pro on a hexa-core i7? I'd say it's likely. But Apple can't pump out a 16" MacBook Pro on Apple Silicon until it is able to say it is categorically faster than the Intel counterpart it is replacing.

Well, it wasn't something I made up but I can understand your scepticism since they've deleted all the evidence, probably after Apple's request. The original German article about the test on Macwelt, Macworld's translation and Macwelt's video have all been removed. As you can see the link to Macworld's article directs you to and old article but in the thumbnail you can see the original article was different. I read somewhere that developers are not allowed to share benchmarks of the DTK so maybe that's the reason. The results are from a CPU intensive test Macwelt developed by themselves. Fortunately Google still has cached images of the result and I found another report that is still online. I share them in my first post.

 

[Donga120]

That YouTube clip is pulled. I also don't see where you can conclude that the A12Z beats the 9th Gen 8-core i9, seeing as the A12X couldn't do it and the only difference between the two is an additional GPU core. Can it toast a 16" MacBook Pro on a hexa-core i7? I'd say it's likely. But Apple can't pump out a 16" MacBook Pro on Apple Silicon until it is able to say it is categorically faster than the Intel counterpart it is replacing.

The A12Z from the DTK is just an off the shelf CPU for Apple and developers to test and doesn’t represent anything heading to Macs, although one would assume they will be a lot more powerful than the DTK

Intel have been out the CPU game for years now, one of Apple’s reason to go for its own Silicone. Like for like Apple silicone will destroy Intels equivalent. We already know how good their mobile chips are, combined into a proper desktop with a heatsink Intel stand no chance. Bearing in my Apple are going to be on 5nm by the and of the year and intel can’t even manage 10nm yet until next year at the earliest.

 

[DeanL]

The original article is available on the Web Archive: https://web.archive.org/web/2020082...eats-macbook-pro-in-speed-comparison-3794690/

Here is the full article in German:https://web.archive.org/web/2020082...16-im-Geschwindigkeitsvergleich-10873144.html

 

[Homy]

The original article is available on the Web Archive: https://web.archive.org/web/2020082...eats-macbook-pro-in-speed-comparison-3794690/

Here is the full article in German:https://web.archive.org/web/2020082...16-im-Geschwindigkeitsvergleich-10873144.html

Great! Thanks!

 

[aednichols]

This is awesome. The fact that it was pulled makes me think it's real.

Going from my 2016 12" to the ARM 13" Pro will be the upgrade of the century. Hopefully this fall!

 

[deconstruct60]

..... We already know how good their mobile chips are, combined into a proper desktop with a heatsink Intel stand no chance. Bearing in my Apple are going to be on 5nm by the and of the year and intel can’t even manage 10nm yet until next year at the earliest.

There is nothing definitive that Apple is shipping a desktop SoC in 2020. The first Apple Silicon Mac is likely going to be a laptop. (whether it is a Macbook , MacBook Air, smaller MacBook Pro is debatable. That the first will be a desktop is dubious. )

Intel is already shipping 10nm mobile processors. the 4 port MacBook Pro 13" is a 10nm chip. So "next year" for that doesn't make sense. In fact, Intel is about to drop a "2nd generation" 10nm mobile chip ( 'Tiger Lake' Gen 11 ) processor next month left alone by the end of the year. Intel is also on track to ship server 10nm ('Ice lake' SP) chips this year also. [ Probably just to the major cloud vendors (not mere mortals) ]. The drama that Intel "cannot make 10nm work" is relatively worn out notion at this point. Mainstream desktop yeah won't ship in numbers until 2021 , but more a priority weighting than a tech gap.

Apple's own slide at WWDC 2020 demonstrated that they are more interested in perf/watt than just performance. Desktop wise they are likely more interested in addressing the thermal issues of the Mac Mini and iMac 21-27 than in winning some ultimate drag racing tech press ( tech porn ) benchmark war. For the first year or two , Apple's desktop offerings are probalby going to be better Mac processors than out to fill performance ranges in enclosures that Apple doesn't sell in very large numbers ( if at all. There are more than several desktop system variations that Apple does not sell. ).

Apple just refreshed the iMac 27" which means they could 'coast' on that for at least 10-12 months easy. ( Apple's track record on the pace of desktop updates is not speedy ). Moving to Apple Silicon isn't likely to drastically speed up the rate of updates. Apple's priority is still going to be iPhone and iPad Pro first , laptops next , and desktops somewhere behind that in their "spare time".

 

[deconstruct60]

Yes, and the fft test used the Accerate framework which is going to be tuned to use the Neural Engine (?). I’m sure the A14 Mac chip will be great .. but I’m not so sure that this test is extensive enough to really draw the conclusions.

Similarly the Rosetta 2 translator totally punts on any AVX code. So any Accelerate framework to cover FFT on x86_64 would have to punt completely out of any AVX optimizations. That makes that Rosetta translation doing more than relatively bad here also not particularly indicative of the translation norm. [ i.e. what seeing here is likely also as completely avoiding using vector instructions versus using them. ]

Between perhaps tapping into the Neural Engine ( or the ARM extensions to interface with Neural engine computations) and throwing the Rosetta translation "under the bus" this benchmark is quite likely more so a cherry picked example of the differences. Any benchmark which pushes lots of work into Apple's hand tuned acceleration libraries says more about Apple's code than the code that the general developer population produces.

 

[JohnnyGo]

Agree that the selected tests put a bad spin on Rosetta and glorify the A12Z performance.

But also know that Apple Silicon will be more than an A14 or A14X/Z soc. It will be its own class, maybe even two (laptop and desktop).

Aside from more cores, Apple’s new class of chips will be allowed to use 20-30-40 watts (in laptops) and 60-80-100 watts on desktops. The additional power will mean additional performance.

We already knew Rosetta will great a performance hit. But given the additional horsepower of the new AS SOCs it will likely mean same performance for most Intel apps as the latest intel Macs.

 

[Homy]

Apple just refreshed the iMac 27" which means they could 'coast' on that for at least 10-12 months easy.

Actually Apple, after announcing the transiton to Intel in June 2005, introduced the first Intel iMacs in Jan 2006, only three months after they released new iMac G5s in Oct 2005. They won't wait a year to release AS iMac. They discontinued the last iMac G5s after three months. These newly released Intel iMacs can meet the same fate this time too.

 

[eric89074]

There is nothing definitive that Apple is shipping a desktop SoC in 2020. The first Apple Silicon Mac is likely going to be a laptop. (whether it is a Macbook , MacBook Air, smaller MacBook Pro is debatable. That the first will be a desktop is dubious. )

Intel is already shipping 10nm mobile processors. the 4 port MacBook Pro 13" is a 10nm chip. So "next year" for that doesn't make sense. In fact, Intel is about to drop a "2nd generation" 10nm mobile chip ( 'Tiger Lake' Gen 11 ) processor next month left alone by the end of the year. Intel is also on track to ship server 10nm ('Ice lake' SP) chips this year also. [ Probably just to the major cloud vendors (not mere mortals) ]. The drama that Intel "cannot make 10nm work" is relatively worn out notion at this point. Mainstream desktop yeah won't ship in numbers until 2021 , but more a priority weighting than a tech gap.

Apple's own slide at WWDC 2020 demonstrated that they are more interested in perf/watt than just performance. Desktop wise they are likely more interested in addressing the thermal issues of the Mac Mini and iMac 21-27 than in winning some ultimate drag racing tech press ( tech porn ) benchmark war. For the first year or two , Apple's desktop offerings are probalby going to be better Mac processors than out to fill performance ranges in enclosures that Apple doesn't sell in very large numbers ( if at all. There are more than several desktop system variations that Apple does not sell. ).

Apple just refreshed the iMac 27" which means they could 'coast' on that for at least 10-12 months easy. ( Apple's track record on the pace of desktop updates is not speedy ). Moving to Apple Silicon isn't likely to drastically speed up the rate of updates. Apple's priority is still going to be iPhone and iPad Pro first , laptops next , and desktops somewhere behind that in their "spare time".

I think the first silicon Macs will be the air, 13/14” pro, and the Mac mini. I don’t think the 16” pro and iMac will be on silicon until the A15 variant is ready. No Mac Pro until the A16 is ready.

I’d be all over a mac mini with A14x/z, 16GB of memory, and SSD for $799.

 

[thunng8]

I think the first silicon Macs will be the air, 13/14” pro, and the Mac mini. I don’t think the 16” pro and iMac will be on silicon until the A15 variant is ready. No Mac Pro until the A16 is ready.

I’d be all over a mac mini with A14x/z, 16GB of memory, and SSD for $799.

Not really sure why we should wait for A15 when A14 (with the appropriate core counts) is likley to be faster than anything currently shipping for 16" macbook pro or the imac

 

[DeepIn2U]

Lets hope for "one last thing" in the upcoming September event , i think everyone are ready to see how the new Macs will run.

we don’t just want the Mac software to just run on Apple Silicon, we want them to “sing”!

 

[iPadified]

Has any reliable leaker refuted Kou´s predictions of an October release? I think not. The 21.5 inch iMac was not updated suggesting a release soonish. Strategically, Apple needs to release something that silence the performance crowd. It is not the tech press or macrumours uses that needs placating but the developers and uses of softwares requiring high compute. That is, is AS a viable platform for all types of computing. Buying decision for leaving MacOS is made every day and if an iPad with a fixed keyboard running MacOS is the only thing we see, there will be disappointments and insecurity.

An iMac 24 inch makes good sense and can likely be a higher clocked "laptop chip" with a performance of a the i9 in the MBP16. The rumour that the new iMac will borrow the iPad Pro design suggest a thinner and equally thick case all over. It means that the thermal restraints of a 24 inch iMac will be similar as a laptop has. There were actually many rumours of a iPad Pro type of case for iMac in the late spring/summer. Such rumours are too early for a release in 12 months.

Disclaimer: I want that iMac 24 inch before Christmas, hence my opinion is likely biased

 

[Tafkaeken]

From a marketing point of view, an air with better battery life and as good performance as the current 13 inch MPB and a 14 inch MBP with the better performance AND the battery life of the current MBA. Fits excellent with that diagram they showed at WWDC.

They don’t really need to go extreme to clear that bar and think they are going to save the extra gap they could make for a while so that they can clear the bar for years to come.