Apples new M1 microprocessor

[tszaboo]

https://www.techpowerup.com/275102/alleged-apple-m1x-processor-specifications-surface

Apple's silicon design team has recently launched its "fastest" CPU core ever, found inside the company's M1 processor designed for laptops and mini-PCs. Featuring an eight-core processor, where four cores are represented by low power small configurations, and four big, high-performance design cores, the M1 processor proved to be extremely fast. However, the Apple Silicon processor doesn't seem to cover anything higher than the 13-inch MacBook Pro. And that is about to change. When it comes to higher-end models like the 16-inch MacBook Pro, which provides more cooling area, it is logical that the processor for those designs is a higher performance design.

Enter the world of the Apple M1X processor. Designed for high-end laptops and the most demanding workloads, the new processor aims to create a new performance level. Featuring a 12-core CPU with eight big and four small cores, the M1X processor is going to deliver much better performance than M1. The graphics and memory configuration are currently unknown, so we have to wait and see how it will look like. The M1X is set to arrive sometime in Q1 of 2021, according to the source of the leak, so be patient and remember to take this information with a grain of salt.

Well, if that's true, it should be quite fast.

[bd139]

Saw this. The leaker is not well respected and generally throws shit at the wall and some of it sticks because it's logical not because it's correct information. It is feasible though.

Regarding the earlier network leak issues, I've seen a full review on Big Sur now and the pf firewall and system-wide VPNs are still effective. The issue is the filter API they added (NEFilterDataProvider) is bypassed by Apple applications as there is a genuine mistake in the policy for these which is apparently going to be fixed in the next Big Sur release. Apparently under MDM control you can specify per-application VPNs but the policy is incorrectly set by default. The issue may or may not be intentional and is feasible that it missed testing or they thought it was sensible (to some degree - think MITM apple services) but they got buggered hard for it.

[DiTBho]

According to some random benchmarks on the web, it seems the new M1 is getting some gain (4X? really?) with Final Cut, which definitely sounds a good point to buy a new MAC-M1 for January 2021; there is and there will never be anything like BootCamp, but Parallels is getting ready with a new release for the new year able to "run" Windows 10 in a (x86 emulated) "Fusion Box" (that's amazing!), Microsoft is thinking about releasing a Windows10/M1 version (will it happen? won't it happen? who knows ...(1)), and yesterday Linus Torvalds said he wants an Apple-M1 laptop running Linux (will it happen with kernel v6? I wonder).


edit:
(1) Microsoft has *already* released a Universal build, containing x86_64 and ARM64 binaries, of its Mac Office 2019 beta, but there's not yet an M1-native general release of Office, and the code editor Visual Studio Code has an experimental ARM64 build, with a Universal build planned for the end of this month.

Will Microsoft just provide "applications" for Apple/M1, or will they also release an OS for M1? That's precisely what I wonder.

[bd139]

Linux port is likely going to be blocked on peripherals, particularly GPU driver I reckon. I bet we'll see a headless box with ethernet up fairly quickly and then it'll stall.

The final cut performance is because it has dedicated encoder / decoder hardware. That's a good thing!

[DiTBho]

Linux port is likely going to be blocked on peripherals, particularly GPU driver I reckon. I bet we'll see a headless box with ethernet up fairly quickly and then it'll stall.

Probably. However, what makes me thinking, is that Apple has already made it clear they won’t be blocking you from installing other OSs.

So ... Sweet dreams are made of this, who am I to disagree? 

[dr.diesel]

Probably. However, what makes me thinking, is that Apple has already made it clear they won’t be blocking you from installing other OSs.

Wow really?  I expected Apple to lock it down hard, very interesting...  Hope this holds true with the up and coming higher performance chips.

[bd139]

Yeah it's true. They aren't blocking third party OS installations. Confirmed already. In fact they said "it's entire up to microsoft if they want to run windows on it" (while Craig is going bwuhahahaha while taking off his human skin suit)

It's a matter of motivation and time now. I wonder if anything useful will appear in: https://opensource.apple.com/

[bd139]

Well I bought an ass end M1 mac mini. Arriving Thursday from Amazon. Have someone lined up to buy it off me for £650 if I'm not happy with it already

[Mr. Scram]

Well I bought an ass end M1 mac mini. Arriving Thursday from Amazon. Have someone lined up to buy it off me for £650 if I'm not happy with it already

Make up your mind! 

[tooki]

Probably. However, what makes me thinking, is that Apple has already made it clear they won’t be blocking you from installing other OSs.

Wow really?  I expected Apple to lock it down hard, very interesting...  Hope this holds true with the up and coming higher performance chips.

It’s not surprising to people who actually know Apple, rather than “know” the stereotype of Apple from sensationalist media or from people who haven’t actually given Apple a fair chance. :/

[bd139]

Well I bought an ass end M1 mac mini. Arriving Thursday from Amazon. Have someone lined up to buy it off me for £650 if I'm not happy with it already

Make up your mind! 

I have nothing else to do at the moment so need something to piss around with 

[bd139]

Probably. However, what makes me thinking, is that Apple has already made it clear they won’t be blocking you from installing other OSs.

Wow really?  I expected Apple to lock it down hard, very interesting...  Hope this holds true with the up and coming higher performance chips.

It’s not surprising to people who actually know Apple, rather than “know” the stereotype of Apple from sensationalist media or from people who haven’t actually given Apple a fair chance. :/

To be fair this does blur the lines between their iOS SoC's and desktops so it was a perfectly valid concern.

[tooki]

Probably. However, what makes me thinking, is that Apple has already made it clear they won’t be blocking you from installing other OSs.

Wow really?  I expected Apple to lock it down hard, very interesting...  Hope this holds true with the up and coming higher performance chips.

It’s not surprising to people who actually know Apple, rather than “know” the stereotype of Apple from sensationalist media or from people who haven’t actually given Apple a fair chance. :/

To be fair this does blur the lines between their iOS SoC's and desktops so it was a perfectly valid concern.

But being locked down (or not) has nothing to do with the choice of CPU/SoC. If Apple had wanted Macs to be locked down like iPhones and iPads, they could have done so with Intel CPUs. Conversely, iPhones have been locked down right from day 1, and they weren’t using custom Apple SoCs back then.

[olkipukki]

I promised some results of my M1 Mac testing so here you go.

Early days yet, but very different (open-source programmer oriented) tests than you'll find elsewhere. Includes building serious open-source software (binutils, gcc, gdb, newlib). Includes arm64 Ubuntu in a VM.

http://hoult.org/arm64_mini.html

A bit of exotic with default and tuned ^{(SME/SEV/SMT=off, BOOST=On, Determinism Slider=Performance)} settings

cat /proc/cpuinfo
model name      : AMD EPYC 7551P 32-Core Processor


1. Perl Summing Numbers Benchmark

default / tuned

real    0m0.083s / 0m0.075s
user    0m0.078s / 0m0.074s
sys     0m0.004s / 0m0.001s

2. Primes Benchmark

gcc -O -o primes primes.c

./primes         

default result:

Starting run
3713160 primes found in 4553 ms
244 bytes of code in countPrimes()


tuned result:

Starting run
3713160 primes found in 4536 ms
244 bytes of code in countPrimes()

3. Building the RISC-V toolchain

default / tuned

real    14m44.759s / 11m55.438s
user    103m4.967s / 73m23.795s
sys     15m40.658s / 9m1.816s


OS: SUSE SLED 15 SP2

[olkipukki]

Well I bought an ass end M1 mac mini. Arriving Thursday from Amazon. Have someone lined up to buy it off me for £650 if I'm not happy with it already

A bit early to open an champagnie, make sure they didn't send i3 

[bd139]

I actually used an i3 mini as my main desktop for a bit. Not as slow as i thought it was going to be but it ran stupid hot.

[brucehoult]

https://www.techpowerup.com/275102/alleged-apple-m1x-processor-specifications-surface

Apple's silicon design team has recently launched its "fastest" CPU core ever, found inside the company's M1 processor designed for laptops and mini-PCs. Featuring an eight-core processor, where four cores are represented by low power small configurations, and four big, high-performance design cores, the M1 processor proved to be extremely fast. However, the Apple Silicon processor doesn't seem to cover anything higher than the 13-inch MacBook Pro. And that is about to change. When it comes to higher-end models like the 16-inch MacBook Pro, which provides more cooling area, it is logical that the processor for those designs is a higher performance design.

Enter the world of the Apple M1X processor. Designed for high-end laptops and the most demanding workloads, the new processor aims to create a new performance level. Featuring a 12-core CPU with eight big and four small cores, the M1X processor is going to deliver much better performance than M1. The graphics and memory configuration are currently unknown, so we have to wait and see how it will look like. The M1X is set to arrive sometime in Q1 of 2021, according to the source of the leak, so be patient and remember to take this information with a grain of salt.

Well, if that's true, it should be quite fast.

I guarantee this guy knows nothing.

Everyone in the world with a brain knows the M1 isn't the last or the best chip Apple will make. They've *said* they're transitioning the entire Mac line to their own chips over the next two years. They wouldn't say that if they didn't have chips suitable for them in the works. QED

[brucehoult]

I promised some results of my M1 Mac testing so here you go.

Early days yet, but very different (open-source programmer oriented) tests than you'll find elsewhere. Includes building serious open-source software (binutils, gcc, gdb, newlib). Includes arm64 Ubuntu in a VM.

http://hoult.org/arm64_mini.html

A bit of exotic with default and tuned ^{(SME/SEV/SMT=off, BOOST=On, Determinism Slider=Performance)} settings

cat /proc/cpuinfo
model name      : AMD EPYC 7551P 32-Core Processor

Cool!

Interesting the gains they made from one year extra development from 7551P to 2990wx.

It's really close on the perl test, but significantly slower on the others, even after the tuning.

What annoys me is the 12 and 16 core 5900X / 5950X now match or blow away my 1st gen Threadripper at much lower cost. My consolation is I got a couple of years use out of it first. And it's dependable. Even if it's 20% slower than the latest thing on some test, you can just keep throwing more and more stuff at it.

Thanks for your effort!

[olkipukki]

It's really close on the perl test, but significantly slower on the others, even after the tuning.

How come? epyc-tuned and your ripper is really close to each other on the build test, that's surprised me... or am I'm looking 'wrong' numbers?

Don't forget this EPYC max 2.55GHz only for all cores (and a single up to 3GHz)

Interesting the gains they made from one year extra development from 7551P to 2990wx.

What annoys me is the 12 and 16 core 5900X / 5950X now match or blow away my 1st gen Threadripper at much lower cost. My consolation is I got a couple of years use out of it first. And it's dependable. Even if it's 20% slower than the latest thing on some test, you can just keep throwing more and more stuff at it.

My desktop Ryzen 3600X easily beat this EPYC on single-threads/user interactive apps... and I don't need to look on passmark single thread performance benchmarks, although a ratio between is almost true there.
If I paid RRP for 7551P, I pulled all my hair out by now 

TBH, glad to see that Apple M1 performs quite well

 

[brucehoult]

It's really close on the perl test, but significantly slower on the others, even after the tuning.

How come? epyc-tuned and your ripper is really close to each other on the build test, that's surprised me... or am I'm looking 'wrong' numbers?

Don't forget this EPYC max 2.55GHz only for all cores (and a single up to 3GHz)

It's still 10% slower 10m51s vs 11m55s.

The 2990wx is 3.0 GHz base. lscpu gives readings all over the place for it. When it's running the ./configure parts it's around 2.1-2.2 (IO limited I guess), When there are a lot of cc1's running it seems to be about 3.2.

I think the EPYC may have a better RAM system?

TBH, glad to see that Apple M1 performs quite well

Between AMD's latest and Apple, we're in for some interesting times in desktop/server land in the next few years.

[AntiProtonBoy]

Between AMD's latest and Apple, we're in for some interesting times in desktop/server land in the next few years.

I reckon AMD is going to pull a few tricks out of the hat. Meanwhile Intel is shitting its bed.

[bd139]

I don’t think Intel are crapping themselves yet. They’re still the only major supplier which hasn’t got supply problems and can fulfil orders for the large OEM builders. Can’t sell Ryzen if you can’t make them quick enough and M1 cores are not offered to other OEMs.

[olkipukki]

It's still 10% slower 10m51s vs 11m55s.

The 2990wx is 3.0 GHz base. lscpu gives readings all over the place for it. When it's running the ./configure parts it's around 2.1-2.2 (IO limited I guess), When there are a lot of cc1's running it seems to be about 3.2.

10% diff is nothing... EPYC 7551P has 2GHz base 

I guess if you turn off HT and Virtualisation settings in BIOS, your ripper might perform better than now

I think the EPYC may have a better RAM system?

Yes, indeed, 2x memory bandwidth compare to the ripper and the system fully filled with 2666 DIMMs, but imho - irrelevant/nonsignificant for the current test cases, you can see above results before and after tunning.

Not sure if you can apply on non-epyc cpus  http://developer.amd.com/wp-content/resources/56420.pdf

Between AMD's latest and Apple, we're in for some interesting times in desktop/server land in the next few years.

If keep same analogy, M1 like Ryzen 1.
Hopefully Apple will figure out how to increase a screen to 14" in the 13" macbookpro case in the next iteration  or introduce something like Mac Pro Lite with M2+ 

[wizard69]

Thanks for the post & performance benchmarks.   I just got a MBA and have been installing from Mac Ports.    So far it has been a mixed bag (kinda expected) with some stuff not compiling.

I'm actually very pleased with this machine, it literally performs better than any laptop that I've owned personally or at work.    After just reading a good portion of this thread I suspect that many are woefully underestimating just what Apple has achieved with M1.   

I promised some results of my M1 Mac testing so here you go.

Early days yet, but very different (open-source programmer oriented) tests than you'll find elsewhere. Includes building serious open-source software (binutils, gcc, gdb, newlib). Includes arm64 Ubuntu in a VM.

http://hoult.org/arm64_mini.html

Ongoing. Not everything works with the combination of MacOS AND arm64 yet. Will add more tests. Suggestions welcome if not too difficult to do.

[wizard69]

Hi Guys;   Just started reading this thread about an hour ago.    I picked up a Mac Book Air a couple of days ago and hope to have a lot of free time with it on the holiday!   So far though this MBA has really impressed me and for the most part is living up to my expectations.

For those not familiar the MBA series is Apples LOWEST end laptop and thus has never been a performance champ.  With the M1 chip in this machine that is an entirely different story as the machine is snappy from the users stand point.   I don't think Apple is being dishonest when they compare these machines to higher end solutions.   This is a passively cooled machine that can compete with machines that could never run passively cooled.

A little background, between work and home I've used or owned a number of laptops over the years.   The last one I personally owned died over a year an a half ago and frankly I got pissed off with the high cost of laptops (to get decent performance) and the short life spans, this resulted in a new Linux desktop build.   Ever since that HP died I've been looking for an ARM based laptop done right.   Why ARM - long battery life, a laptop with an advertised 4 hours of life on battery that dies after 2 is useless to me.

So whom decides to lead the industry to ARM, good old Apple.   This is not a crap low performance machine even if it is Apples introductory model.   It also meets my #1 criteria which is good life on battery plus it can run Python no problem so I have no issue there.   I already have a desktop so travel is an important element in this machines usage, considering that it can run off USB-C charging it anywhere should not be an issue.   For me it is a much better solution for travel than a tablet, that mainly due to having a real file browser and access to the file structure.

I hope that this doesn't sound too much like an advertisement but I'm trying to get across why I went out and purchased this machine.    I haven't tried to do much benchmarking yet as that has never been my thing.   

If any one has questions please ask.   I'm by no means an expert when it comes to a Mac so take that into mind.