Home » Technology » The MacBook Air with the Apple M1 processor trampled the Intel version to the ground

The MacBook Air with the Apple M1 processor trampled the Intel version to the ground

<!—->

History repeats itself a bit. In 2006, computers switched from IBM PowerPC processors to Intel processors. According to Steve Jobs, the then “roadmap” of Intel processors better matched the construction and design of the computers that Apple wanted to bring to market.

It was a big deal, the applications had to be optimized for a new platform, rock fans were crying. I wasn’t surprised then, at school we had several older iMacs, on which, for example, the video ran as the same old PC could seem to make.

Fifteen years later, Apple is embarking on another change, switching from Intel processors to chips of its own production. The Apple Silicon M1 can be found in the new MacBook Air, MacBook Pro and Mac Mini, and more are being prepared for generational changes. Almost no one is crying this time. The new “em number one” is rolling the competition in performance and consumption, and the list of optimized, or at least via emulator executable applications is growing rapidly.

We borrowed two 13 “MacBook Airs from the Alza.cz store, one in a configuration with an Apple M1 chip, the other with an Intel Core i5 1030NG7, and decided to try the differences between the machines. Except for the processor, they are the same, both have 8 GB of RAM , 500 GB SSD, screen resolution
2560 x 1600 pixels and a battery with a capacity of 4,382 mAh. The one with the Apple chip was sold for CZK 36,281 at the time of the loan, and with Intel for CZK 33,490.

Video performance and encoding

Apple Silicon M1 is a processor with ARM architecture, so it is a processor technically related to those in smartphones. Applications and programs written for use on Apple computers with Intel’s x86 processors cannot be run on this new chip.

“Just like that” you can run applications that have been either directly created for use on a new processor, or redesigned and optimized. There is already a fairly long list of such programs and more are being added quickly. You can also run many apps written for iPads and iPhones.

Because it will take some time to redesign many native run applications to M1, there is the so-called Rosetta 2. It is a special application layer, you can think of it as a reduction that mediates communication between x86 application code and new hardware. As we have tried, some applications work very well in this mode, others not at all, some so halfway. This is a temporary solution before application developers release modified code variants (and for running applications that are no longer being developed).

We maintained this division when testing performance. All benchmarks in the following table have already been optimized to run on the Apple M1, so they take full advantage of its potential.

Tests with native M1 support
Apple Silicon M1 Intel Core i5
GeekBench 5.3 – one core 1 727 1 176
GeekBench 5.3 – all cores 7 660 3 990
GeekBench 5.3 – OpenCL 18 552 7 781
GeekBench 5.3 – Metal 20 863 9 028
CineBench R23 – one core 1 492 919
CineBench R23 – all cores 7 115 2 540
PassMark – one core 3 547 2 033
PassMark – all cores 15 287 7 256
Blackmagic Disk Speed ​​- write 2 289 MB / s 1 357 MB / s
Blackmagic Disk Speed ​​- reading 2 855 MB / s 1 262 MB / s

The difference is literally abysmal. We were surprised by the difference in the speed of working with the disk (there is the same SSD with the Apple logo).

The following table lists tests that have not yet been optimized. While it runs natively on Intel, it runs on the M1 using Rosetta.

Tests without native M1 support
Apple Silicon M1 Intel Core i5
Novabench – overall score 2 289 1 542
Novabench – CPU 1 170 939
Novabench – GPU 605 243
Novabench – RAM 281 234
Novabench – Disk 233 126
GeekBench 4.4 – one core 5 759 5 198
GeekBnech 4.4 – all kernels 23 311 14 939
GeekBnech 4.4 – OpenCL 53 287 31 232

Even in this mode, the power advantage of the M1 over the Core i5 is quite significant.

Apple computers are often used for multimedia creation, so we tried how to handle video encoding. In the Handbrake application, which is native to both Core i5 and M1, we created a queue of twelve 4K HEVC videos with a total length of 24 minutes 14 seconds (the source files had a total of 10.2 GB) and had them converted to Full HD resolution and MPEG4 format. (HQ 1080p30 preset).

Video encoding
Apple Silicon M1 Intel Core i5
Handbrake 63 minutes 148 minutes

The difference in transmission speed is abysmal. In addition, emotionally enhanced by the fact that the machine with the intel spun the fans to the maximum and was quite noisy, the passively cooled machine from M1 was, of course, completely noiseless at all times.

We wondered how it was inside machines with temperature. The Temp Monitor application was somewhat brief on the M1 machine, CPU Proximity reached 47 ° C, battery maximum 45 ° C. For the Core i5, the CPU Proximity was held at 76 ° C, the battery at 45 ° C. And while we didn’t find more detailed temperatures for the M1, for Intel the temperature of the chip itself reached 100 ° C, which is really a lot.

In the keyboard area, both machines were similarly hot after an hour of full load, but while the palmrests and touchpad area were cold on the blowing Intel version, the passive M1 spread the heat here and it wouldn’t be pleasant to work with. On the other hand, the MacBook Air is not a long full load machine, since it is the MacBook Pro and desktops.

Windows virtualization

Mac OS users usually can’t come up with a name for the Windows operating system (and vice versa), but it’s true that for some, the ability to run it on an Apple computer is sometimes quite important. This is not a problem on a MacBook with Intel, just install them as a second system and use BootCamp to switch between them. This is not the case with the M1 processor model.

This leaves another option to virtualize Windows within the running process
Mac OS – they are then launched in a window like a classic application. A popular choice is the Parallels Desktop virtualization system. And while everything is ready for systems with Intel, for the system on ARM it is still in trial form.

Windows 10 in Parallels Desktop on an Intel Processor System

Windows 10 in Parallels Desktop. Left on M1 system, right on Core i5

Parallels Desktop can be downloaded in the beta version as part of the trial version, and since Windows does not normally run on ARM processors, you need to install so-called Windows for ARM, which you can find pre-installed on Microsoft Surface Pro X (and several other machines). under Windows Insider download, which we did.

But the question is, how much does it make sense at all, because in Windows for ARM you do not run applications for x86 processors, or 64-bit ones at all, 32bit via the built-in emulator. We wrote more about it here: The unique notebook with the Microsoft SQ1 processor opens a new era of Windows.

Tests in Windows 10 via Parallels Desktop Virtualization
Apple Silicon M1 Intel Core i5
GeekBench 5.3 – one core 1540 958
GeekBench 5.3 – all cores 2863 1536
Hardlimit Benchmark – One thread 42,91 58,54
Hardlimit Benchmark – All trains 76,65 88,82

It is clear that the available performance is modest in both cases, especially in applications that are otherwise capable of using multiple compute cores. Graphics chip acceleration does not work here either, which is a problem for applications using OpenCL, for example. Many programs from the Microsoft Store, which are ready to run on ARM hardware, we still failed to run on the M1. For others, we encountered limited functionality.

Battery life

In the case of ARM processors, high expectations relate to energy consumption or endurance per battery charge. It is identical in both versions of the notebook, it has a modest capacity of 4,382 mAh.

With very easy use when streaming video and browsing the Internet, with a pleasant brightness setting and wi-fi connection, the notebook with Intel lasted a solid 9 hours and 50 minutes, the notebook with M1 an incredible 17 hours and 40 minutes.

On the other hand, at full CPU usage, Intel consumed 32 percent of the battery, M1 29 percent of the battery capacity. It is necessary to take into account that the M1 managed significantly more work per hour.

One cable for everything.  USB-C connection transfers 4K image and sound to the monitor and ...

One cable for everything.  USB-C connection transfers 4K image and sound to the monitor and ...

One cable for everything. The USB-C connection transfers 4K video and audio to the monitor while powering the notebook.

We tested charging with both the supplied 30W USB-C adapter and the 4K AOC U32U1 monitor. It offers both video and audio transmission over
USB-C and 65W power delivery via the same cable. All you have to do is connect the laptop to the monitor with one cable, which takes care of everything.

And while with the included adapter, laptops charged from empty to full for 3 hours and 20 minutes, with a cable from a monitor, they did it 50 minutes faster.

Conclusion

If you don’t necessarily need to switch between macOS and Windows, and if your most used applications already have an M1-optimized version, then the Apple Silicon M1 models make sense now. They excel in performance and energy intensity and thus in the production of heat, and therefore low (no) noise.

You can find lists of compatible applications on the Internet, for example, here isapplesiliconready.com , using Silicon application you can do a small audit to find out which applications you are using will already run on M1 without any problems.

All that remains is to keep your fingers crossed that Apple has allocated enough production capacity for chips, and laptops installed in this way were not as rare as graphics cards or game consoles.

<!—->
– .

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.