Thursday, November 12th 2020

Apple's M1-Based MacBook Air Benchmarked

When Apple announced that they are going to switch their Mac lineup from Intel-based x86 processors to the custom "Apple Silicon," everyone was wondering how the new processors will look and perform. To everyone's luck, Apple has just a few days ago announced its first Apple Silicon custom processor for MacBook. The M1, as the company calls it, is their first processor designed for higher-power and performance tasks The M1 features eight CPU cores (four high-performance and four-high efficiency) paired with eight cores dedicated to the graphics. On the die, there is also a 16-core neural engine made to accelerate machine learning tasks found in the new applications.

Today, we are getting the first GeekBench 5 CPU benchmarks that showcase just how far Apple has come with its custom design. What we have is the M1 processor found in MacBook Air. This Mac model features a passive cooling system, cooling a CPU with a base frequency of 3.2 GHz. The system scored 1719 points in the single-core result, and 6967 points in the multi-core result. The single-threaded results measure itself with some of the highest-end offerings from Intel and AMD, while the multi-threaded results are very good given the mix and match of small and big cores.
Source: GeekBench 5
Add your own comment

115 Comments on Apple's M1-Based MacBook Air Benchmarked

#77
hurakura
This shitty benchmark means absolutely nothing.
Posted on Reply
#78
ypsylon
Geekbench is worth less than bag of sand. However slowly real life performance is trickling down. In Affinity Photo it looks impressive.

[MEDIA=twitter]1326866126635143169[/MEDIA]
Posted on Reply
#79
Endeavour
Initialised
www.cpu-monkey.com/en/cpu-apple_m1-1804

Shows the M1 as the fastest Single threaded CPU in Geek Bench.
Actually the new Ryzen 5xxx processors are reporting single core scores over 2000 in MacOS
browser.geekbench.com/v5/cpu/singlecore

Which means the results are heavily influenced by the OS. Compared to those, the 1700 result from the M1 also in MacOS is way less impressive.

But in the end it is still a great processor with really good performance* and amazing low power consumption. It’s indeed a formidable opponent for both Intel and AMD, at least on this segment. They still have to show how well it scales, but I’m confident they will also be competitive there in a couple of years.

*my prediction for cinebench r23 is 1300-1400 single core and 6000-6500 multi core. Graphics will probably be on par with the new Intel Xe.
Posted on Reply
#80
Just Some Noise
Apple is so extremely big. And they know, if they f.... up this launch, they lose marketshare. I bet you the M1+ or however the bigger ones are called, will be real performer. I also ordered a MacBook Pro 13 with an M1 chip, so i can try Logic Pro X with it. Oh, and a heads up for the audio folk, Image Line also announced that FL Studio is in the works for the M1 ARM Mac here. This one is gonna be interesting. So happy to see the computer as a whole transform and change in its fundamentals again after a long long time!
Posted on Reply
#81
londiste
Just Some Noise
This one is gonna be interesting. So happy to see the computer as a whole transform and change in its fundamentals again after a long long time!
Academically and technically M1 is awesomely interesting.
Unfortunately in the real world it is very likely that this simply means increasingly walled walled garden of Apple.

I do hope machines have enough access that custom OS can be easily booted and SoC along with each part of it are well documented and supported - for example in Linux or FreeBSD. This hope probably means I am a fool...
Posted on Reply
#82
Mescalamba
Vya Domus
That has to be ironic, SPECint has absolutely nothing to do with the real world, it is purely a synthetic benchmark :


00.perlbenchCPerl Programming LanguageDerived from Perl V5.8.7. The workload includes SpamAssassin, MHonArc (an email indexer), and specdiff (SPEC's tool that checks benchmark outputs).
401.bzip2CCompressionJulian Seward's bzip2 version 1.0.3, modified to do most work in memory, rather than doing I/O.
403.gccCC CompilerBased on gcc Version 3.2, generates code for Opteron.
429.mcfCCombinatorial OptimizationVehicle scheduling. Uses a network simplex algorithm (which is also used in commercial products) to schedule public transport.
445.gobmkCArtificial Intelligence: go playingPlays the game of Go, a simply described but deeply complex game.
456.hmmerCSearch Gene SequenceProtein sequence analysis using profile hidden Markov models (profile HMMs)
458.sjengCArtificial Intelligence: chess playingA highly-ranked chess program that also plays several chess variants.
462.libquantumCPhysics: Quantum ComputingSimulates a quantum computer, running Shor's polynomial-time factorization algorithm.
464.h264refCVideo CompressionA reference implementation of H.264/AVC, encodes a videostream using 2 parameter sets. The H.264/AVC standard is expected to replace MPEG2
471.omnetppC++Discrete Event SimulationUses the OMNet++ discrete event simulator to model a large Ethernet campus network.
473.astarC++Path-finding AlgorithmsPathfinding library for 2D maps, including the well known A* algorithm.
483.xalancbmkC++XML Processing

Nearly all of those are completely irrelevant to real world workloads performed by your average user. Even for me, someone who knows what all of those tests entail it still means nothing.




Huh ? Wouldn't it be strange if I were to hate on something that I will buy ? :kookoo:
Well, I find pretty much all of it relevant. Guess different target audience.
Posted on Reply
#83
Caring1
Nobody else think there should be a comma in the title?
How do you air benchmark a laptop? :wtf:
Posted on Reply
#84
windwhirl
Caring1
How do you air benchmark a laptop?
Install it in a wind tunnel, and ta-da! There you have your air benchmark. :D :laugh:
Posted on Reply
#85
Vayra86
foedecide
Oh wait, you guess, so you are basing this on your own personal opinion and no real knowledge???



Sorry for asking you to provide actually evidence to your personal claims, I have learned not to challenge people who claim to know better, thank you.
Benchmarks have always been misunderstood and misrepresented as overall performance progress. In the case of Geekbench that is made worse by the change logs between versions. Its no good when CPUs get placed on different tiers between different versions. And this happens every single time while geekbench itself is not even indicative of ANY performance level.

Simple questions Geekbench can never answer or shine light on: " how capable is thr device at X or Y" in terms of real use cases.

That alone makes it pure synthetic... but then why does every version effectively invalidate all scores before it?

Some benchmarks just exist as industry marketing machines and geekbench is a sorry attempt at such a thing. It suits the mobile progress crowd, stuck in more or less walled gardens and stuck in an online reality by design. The moment ARM cpus start becoming real CPUs for multitasking on varied stuff, they dangle along the bottom end of laptop CPUs in most real scenarios. They lose their efficiency edge or run into power limits. Benches dont show that.
Posted on Reply
#86
Initialised
Just Some Noise
So happy to see the computer as a whole transform and change in its fundamentals again after a long long time!
Those Archimedes RISC enthusiasts weren't, it just wasnt time for a RISC based architecture for desktop class workloads untill now. My guess is that Apples projections on where to get pwoer efficiency from was to ditch CISC based x86 in favor of RISC based Arm.

For me the main limitation that remains is the lack of eGPU and the 16GB memory limit as for a lot of tasks faster, lower latency, higher bandwidth RAM is no substitue for more RAM.
Posted on Reply
#87
r9
Using SoC, pack CPU, GPU, Chipset, modems and even the memory and it supposed to compete with the best AMD and Intel can offer .... haha what a joke.
Their new series will be bunch of glorified tablets.
It will do great for Apple for one they will able to distance themselves from apple to apple comparisons to the PCs as they are lower spec'd but higher priced, and they will get away from that by building a whole line of products that will be even lower spec'd. Pure Genius!
None of this matters as the Apple users don't care about any of that they just want the glitter.
Posted on Reply
#88
mechtech
Will is run steam/games?

Will it run autocad?
Posted on Reply
#89
Gungar
Rahnak
Geekbench is worthless for comparing between different architectures.
Yeah well look at all those people still comparing them... I get SO angry when i see that :(
Posted on Reply
#90
Searing
Vayra86
You realize this disqualifies it to begin with? If a new version is so radically different in how it scores, it immediately invalidates everything that came before it..... until Geekbench 6.
... that is a silly argument if I've ever seen one, technology changes, benchmarks must change also
Posted on Reply
#91
dyonoctis
Well, looks like the m1 isn't all mighty in every case, but those numbers aren't bad for such low clocks and power :


So while it looks great for photo editing and hardware video encoding, it's not the "king" for brute force 3D rendering. I wonder just how much more performance a desktop m1 could tap into to get on the same level as zen 3 ?
Posted on Reply
#92
windwhirl
dyonoctis
Well, looks like the m1 isn't all mighty in every case, but those numbers aren't bad for such low clocks and power :


So while it looks great for photo editing and hardware video encoding, it's not the "king" for brute force 3D rendering. I wonder just how much more performance a desktop m1 could tap into to get on the same level as zen 3 ?
The single thread is kinda expected, I guess. MT, though, is Cinebench using all 8 cores or only the high-performance ones?
Posted on Reply
#93
dyonoctis
dyonoctis
Well, looks like the m1 isn't all mighty in every case, but those numbers aren't bad for such low clocks and power :


So while it looks great for photo editing and hardware video encoding, it's not the "king" for brute force 3D rendering. I wonder just how much more performance a desktop m1 could tap into to get on the same level as zen 3 ?
Wait nevermind, those results are the same as the A12Z, hexus.net made a mistake :(, well tomorow is the day where people are supposed to receive their first machines, so we'll finally know.
dyonoctis
It's for the A12z, but unless the m1is at least x 2 times faster in multithread, the multicore score of the m1 in geekbench won't be representative for every workloads. The m1 is supposed to be as faster than the 9880h in mt...


Posted on Reply
#94
Vayra86
Searing
... that is a silly argument if I've ever seen one, technology changes, benchmarks must change also
Change? They're supposed to present an increased load so they can best show the performance gaps between CPUs in a stack. Not change in terms of favoring specific architectures.

Good benchmarks 'change' neutrally, and if they don't, they are a new sort of benchmark.

3DMark versions present different loads and are up front about it. They test against an API and they present new tests to stress different parts of it. Unigine gives us a new benchmark for every different type of stress test. You see version changes, that add options. But the bench itself remains the same.

Geekbench is nothing like that, especially not in how it is handled. Compare it to how we handle Cinebench results. Versions are noted with each one because there are some differences. Geekbench scores are one big clusterfuck in that sense. Geekbench is just one big score that is supposed to tell us... ehm... what exactly? And it won't even have value in relative score comparisons because new versions completely change them across the board. So what's the actual use of a Geekbench score in practice?
Posted on Reply
#95
Searing
Vayra86
Change? They're supposed to present an increased load so they can best show the performance gaps between CPUs in a stack. Not change in terms of favoring specific architectures.

Good benchmarks 'change' neutrally, and if they don't, they are a new sort of benchmark.

3DMark versions present different loads and are up front about it. They test against an API and they present new tests to stress different parts of it. Unigine gives us a new benchmark for every different type of stress test. You see version changes, that add options. But the bench itself remains the same.

Geekbench is nothing like that, especially not in how it is handled. Compare it to how we handle Cinebench results. Versions are noted with each one because there are some differences. Geekbench scores are one big clusterfuck in that sense. Geekbench is just one big score that is supposed to tell us... ehm... what exactly? And it won't even have value in relative score comparisons because new versions completely change them across the board. So what's the actual use of a Geekbench score in practice?
It isn't about changing to favor Apple, it was about removing the memory dependency, that made it a better benchmark (better than Spec that way). You can view all the sub scores in the test.
Posted on Reply
#96
Endeavour


Results are out... and it's really good, a bit better than expected!
Posted on Reply
#97
windwhirl
Endeavour


Results are out... and it's really good, a bit better than expected!
Now that's interesting! Considering how high the ST score is, I'd bet that's the high-performance core speaking. It also would indicate that the low-performance cores are really low performance. The MT score is somewhere between the 3300X / i5 8400 and the 3600.

What's the source of this, by the way?
Posted on Reply
#98
Endeavour
windwhirl
Now that's interesting! Considering how high the ST score is, I'd bet that's the high-performance core speaking. It also would indicate that the low-performance cores are really low performance. The MT score is somewhere between the 3300X / i5 8400 and the 3600.

What's the source of this, by the way?
Source is this twitter thread. [MEDIA=twitter]1328453127209504771[/MEDIA]
Posted on Reply
#99
Sandbo
dyonoctis
Well, looks like the m1 isn't all mighty in every case, but those numbers aren't bad for such low clocks and power :


So while it looks great for photo editing and hardware video encoding, it's not the "king" for brute force 3D rendering. I wonder just how much more performance a desktop m1 could tap into to get on the same level as zen 3 ?
Endeavour


Results are out... and it's really good, a bit better than expected!
Why do these two look so different?
Are they referring to different processors?
Posted on Reply
#100
windwhirl
Sandbo
Why do these two look so different?
Are they referring to different processors?
The first one was wrong, it was from an A12, apparently.
Posted on Reply
Add your own comment