Apple M1 ARM Performance With A 2020 Mac Mini

[PerformanceExpert]

It outperform the 4500U Sometime

And the difference is only 1.2w (13.8 for the M1, 15 for the tested Zen 3 chip). If it's what you call "fraction"
Now imagin that that same AMD chip where made in the same 5nm process... It would have a lower power consumption than the M1. (should come soon)

On native code it beats the 4500U easily - it even outperforms the fastest Zen 3! And 4500U uses up to 45 Watts when running the Phoronix testsuite. So your point was?

5nm Zen 4 won't be anytime soon, likely mid 2022, but it won't help anyway: 7nm Zen 2 doesn't have anywhere near the same power consumption as 7nm A13.

[baryluk]

For anybody interested: It looks like due to stringer security defaults on Apple Silicon, iOS / macOS, qemu doesn't currently work on Apple Silicon. There is a patch to fix that: https://lists.nongnu.org/archive/htm.../msg02132.html , this allows qemu TCG to emit and execute executable code, as required by QEMU TCG JIT. Once that is applied running x86_64 / i386 (and other guests) should be possible on Apple arm64 under macOS. It looks like the review is progressing smoothly, and it should be merged into qemu master branch soon (week, maybe two). It will not make for qemu 5.2 final release tho unfortunately (5.2 is already in freeze).

I wonder how it compares to Rosetta in general. Is is 2 or 20 times slower

[deppman]

Mixed bag when comes to performance and terrible software. No, thanks.

I agree. The performance is good but overstated because somehow Apple have convinced all reviewers to compare only against systems where it looks more impressive than it actually is. Part of their trickery was to use only 9th gen intel CPUs and throttle them by 20% in the Macs where they switched, thereby ensuring comparison with the prior models would be favorable. When compared against top-line CPUs that can be had at the same cost like the i7-10875h, the M1 mostly loses outside of the single core performance. And discrete GPUs of course destroy its graphics.

The power savings are impressive, but how does x86 stack up against ARM at 5nm? Given how well 7nm Ryzen looks, my guess is the node is more important than the micro arch. I expect ARM to remain more power efficient, but battery life isn't everything!

And what is the cost of this decent performance? No expandability - 16GB total RAM vs 72GB for real pro laptops. No eGPU, CUDA, cuDNN, or OptiX. MacOS. Even greater incompatibility with x86 deployment environments. Apple lock-in taken to new heights and costs. The walled garden becomes a padded cell.

As much as I despise the unholy hardware oligarchy of AMD, Nvidia, and Intel, at least there is the ability to innovate and build within the x86 ecosystem.

[sverris]

Nice, yet by far not good enough for me. Performance is one thing, but there are so many others things that are important, too. So many other things, where Apple miserably fails.

[blackshard]

I agree. The performance is good but overstated because somehow Apple have convinced all reviewers to compare only against systems where it looks more impressive than it actually is. Part of their trickery was to use only 9th gen intel CPUs and throttle them by 20% in the Macs where they switched, thereby ensuring comparison with the prior models would be favorable. When compared against top-line CPUs that can be had at the same cost like the i7-10875h, the M1 mostly loses outside of the single core performance. And discrete GPUs of course destroy its graphics.

Anandtech is comparing against 10th gen intel cpus (which IMHO has the same IPC of 9th gen, but with more aggressive turbo due to power levels) and Zen3, do you have any valid proof of what you say?

Also anandtech review is saying exactly the opposite of what you say: single core performance is great on M1 (https://www.anandtech.com/show/16252...le-m1-tested/4), multicore is less great. All of this without any turbo boost and other crappy trickery mistakes Intel put in the field to mask the limits of x86 and AMD needed to follow.

The power savings are impressive, but how does x86 stack up against ARM at 5nm? Given how well 7nm Ryzen looks, my guess is the node is more important than the micro arch. I expect ARM to remain more power efficient, but battery life isn't everything

Power efficiency is not battery life matter, it's overall architecture a productive process advantage. When you get a 25W part that has the same performance of a 65W part, what happens if you decide that you want to rearrange your part to target 45W ? I tell you: you get the advantage in performance.
But I don't think Apple is going to raise performance of their SoC so much, probably they will switch more number crunching software to GPUs instead of making a huge and inefficient SoC like Intel is doing. That's an advantage of having their own hardware and software ecosystem.

[BillBroadley]

I agree. The performance is good but overstated because somehow Apple have convinced all reviewers to compare only against systems where it looks more impressive than it actually is. Part of their trickery was to use only 9th gen intel CPUs and throttle them by 20% in the Macs where they switched, thereby ensuring comparison with the prior models would be favorable. When compared against top-line CPUs that can be had at the same cost like the i7-10875h, the M1 mostly loses outside of the single core performance. And discrete GPUs of course destroy its graphics.

Have you been reading this thread? Sure the M1 doesn't win everything, but it's drawing comparisons between it and crazy more expensive chips that use many times more power. Personally my interest is in floating point. Take a look at the Anandtech chart reposted at the bottom of page 13 on this thread. The M1 got 38.71. Keep in mind the M1 has a GPU, neural processor, and image processing acceleration and consumes around 10 watts, possibly as high as 20 when actively cooled. The ryzen 3600 has no GPU, consumes way more power and gets 33 at FP. The intel I7-10700k has no GPU, is a high end chip and gets 34.98. The ryzen 7 3700x, one of the best desktop chips for the money, out from 6 months ago finally pulls ahead, barely at 40.20 by 3.8%.

The first significant winner is the Ryzen (zen3) 5600x that has a clear win (almost 15%), not that a normal user is going to notice 15%. So in a $699 mac mini, a $1000 mb air, or a $1250 mbp you can get nearly the 5600x performance, awesome battery life, an integrated GPU, machine learning acceleration, acceleration for image editing, etc. That's not just evolutionary, it's revolutionary. The i9-10900k and Ryzen 7 5800X are faster yet, increases of 25-35% over the M1. But they are using MUCH larger number of cores, crazy expensive CPUs, consuming around 10x the power and still only win by 25-35%. Keep in mind the M1 only has 4 fast cores, and 4 very slow cores. Future M1 iterations might well double the fast cores and double the memory channels ... all without changing the pin interface to the mac mini, mbp, or mba. Sure it would consume double the power (still much less than any of the high end chips it would be compared to).

When the high end Intel and AMD chips are only winning by 25-35% despite 5-10x the power, and 2 to 2.5x the fast cores they should be worried. The M1 is the low end, on under $1000 laptops and desktops. Keep in mind that the mac book mini is physically smaller than the power supply that you would find in an average desktop with a ryzen 5800x or i9-10900k.

As for the GPUs, yes the mac is not a top of the line GPU killer, but the M1 managing to beat a gtx 1050 and trading blows with a gtx 1650 (win some and lose some) is quite impressive for a cheap m1 chip with an integrated GPU. A fair chunk of the market is happy with that level of performance and given more GPU cores and more than 10-20 watts of power the next apple implementation might well double that performance. I'm sure Apple's going to learn the strengths and weaknesses of the M1 and improve on them with the next implementation.

I will however admit that there's likely not much frequency room left for the M1. Numerous implementation details, length of pipelines, cache design, etc means that it's not going to be easy to scale the M1 up to 5GHz like the competition, but the M1 does just find per core at 3.2 GHz, often beating the 5 GHz competition on a per core basis..... that's what should terrify the competition more than anything else. The M1 has 4 fast cores, and 4 very slow cores, yet is within 30% of so of the 8-10 core competition.

[blacknova]

There is another interesting point in all of this. If I remember correctly NVIDIA have purchased ARM just this September...
We might see some new hardware from the soon.

[PerformanceExpert]

I will however admit that there's likely not much frequency room left for the M1. Numerous implementation details, length of pipelines, cache design, etc means that it's not going to be easy to scale the M1 up to 5GHz like the competition, but the M1 does just find per core at 3.2 GHz, often beating the 5 GHz competition on a per core basis..... that's what should terrify the competition more than anything else. The M1 has 4 fast cores, and 4 very slow cores, yet is within 30% of so of the 8-10 core competition.

It already beats the fastest Intel and AMD CPUs at just 3.2GHz so there is really no need to scale up to 5GHz - like most Arm CPUs it has been designed to be fast at 3GHz. Higher frequencies use significantly more power, so it's just not worth it. And yes, if the next generation has 8 big cores...

[PerformanceExpert]

There is another interesting point in all of this. If I remember correctly NVIDIA have purchased ARM just this September...
We might see some new hardware from the soon.

That hasn't finished and might take 2 years to complete. They recently announced Ampere Altra servers with Ampere accelerators: 160x 3GHz Arm cores plus 2x A100 for HPC.

[lutel]

I've never been a fan of Apple software, and for 25 years i was enjoying building PCs. I'm shocked what Apple had just delivered. I'd really like them to not be evil, and to not only open the platform to Linux, but actively support Linux, Windows and other 3rd party developers to quickly onboard software to new architecture. Maybe it is time to let go fun of tinkering with hardware, a bit sad for me, but at the end it is performance/efficiency which matters. How comfortable the workflow is. I'm wondering about gaming community, does Apple have any strategy for gaming market?

I think we are all in shock, probably even Apple is to some extent. They developed Wunderwaffe, hopefully they have a wisdom on how to use it to make the computing market better, not only to monetize on what they did. In 10 years from now the world could be running on Apple Silicon, and it takes great responsibility from Apple not to abuse their monopoly. Anyway big thanks to engineers and brave souls at Apple who had given us something amazing after decade of stagnation on x86 market.

And you know what amazes me most? That Apple had developed it for years, Intel and AMD engineers must knew about it. And they did nothing. Did they really think they can make money on 1-5% improvement per year? They were lazy, conservative, thinking they can sit in comfort zone with mocked competition. They deserve this fate.

I have some wishes for incoming years
- Apple will move to passive cooling in every ARM platform
- Game devs/engines will get support from Apple to jump on the Apple wagon
- Apple will open and support the hardware to Linux dev
- Apple will give solid Machine Learning support for training models