Apple Silicon M1 MacBook Pro Earns 7508 Multi-Core Score in Cinebench Benchmark

[echodriver]

And while it may be rated at 125W I think you’ll find it’ll pull more like 200-300W at 100% util., if you let it. And as long as you can cool it.

At this point, comparing to Intel's crap makes no sense anyways - the 65W AMD Ryzen 5 5600X blows about the 10900K (which can indeed draw 200+W) in a lot of uses.

Just as in the mobile space, the real competitor to the M1 - if we're talking perf/watt - is against the Ryzen 4600U/4800U where they are quite close.

In the end, those saying "x86 is dead" would be right if Intel were the only game in town, but Intel already lost the crown to AMD and that's where the comparisons should be made

Comparing to an Intel desktop CPU doesn't make any sense. Desktop CPUs (and the Intel K-Ones especially) are barely tuned for efficiency, and Intel currently can't compete there anyway.

Comparing to an AMD Ryzen CPU with 15W TDP is fair, where the M1 leads by ~10-20% in multicore performance. This makes the M1 a great CPU, but certainly not the miracle some people have hoped for, especially considering the more advanced process node on the M1.

Where does it lead by 10-20% in multicore performance? Cinebench R23 with the 4800U at 15W (see Anandtech) says they are trailing in multi-core, but GB5 flips it.

Anyways, if/when Zen 3 APU's come out, it will be interesting to see. They nailed a +19% IPC uplift on the desktop side, and they're still on 7NM.

Honestly, it goes to show you how under the radar AMD is, when everyone talks about the M1 beating Intel - it's AMD that's the real leader on the x86 side of things today.

 

[smulji]

No its not....

Intel And AMD x86 Mobility CPUs Destroy Apple’s M1 In Cinebench R23 Benchmark Results

This was Anandtech's conclusion on the first look of Apple's M1

"The performance of the new M1 in this “maximum performance” design with a small fan is outstandingly good. The M1 undisputedly outperforms the core performance of everything Intel has to offer, and battles it with AMD’s new Zen3, winning some, losing some. And in the mobile space in particular, there doesn’t seem to be an equivalent in either ST or MT performance – at least within the same power budgets."

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

 

[echodriver]

The M1 is a 10 watt chip and the 4800U is a 15 watt chip according to the internet,

There is still overhead of the PSU, fan, possibly other things. On the other hands, the MacBook Pro is pretty much silent under sustained load with not a whole lot of air moved around, which would likely mean it's pretty closed to the 15W envelope. However I'd be interested to know how the noise is under both CPU and GPU load.

Looks like ~15W in Cinebench:

https://twitter.com/x/status/1328777333512278020

So perf/watt, the M1 is significantly better than the other offerings in single core, but trails the AMD Zen 2 APU's in multi-core. With the caveat that Zen 2 APU's are on 7 NM process, and the Zen 3 on 7NM+ blows Zen 2 away

Either way, it is clear that Intel is way behind in 3rd place at the moment.

This was Anandtech's conclusion on the first look of Apple's M1

"The performance of the new M1 in this “maximum performance” design with a small fan is outstandingly good. The M1 undisputedly outperforms the core performance of everything Intel has to offer, and battles it with AMD’s new Zen3, winning some, losing some. And in the mobile space in particular, there doesn’t seem to be an equivalent in either ST or MT performance – at least within the same power budgets."

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

Admittedly, this is nitpicking, but I'm not sure how Anandtech came to that conclusion when they bench'd the M1 against the Zen 2 4800U locked at 15W TDP and saw it trade blows with one another, winning some (particularly in ST) and losing others (with more MT).

 

[Piggie]

Found a hidden camera view of Intel's Management right now

 

[PickUrPoison]

If i just want to send email, will this info help me?

It depends; how fast do you want to send it?

 

[Botts85]

Has anyone seen a Cinebench R23 result with on an Apple Silicon Mac running the X86 and ARM version of Cinebench?

I'd love to see in Cinebench what the Rosetta 2 impact is.

 

[Zen_Arcade]

I'd say a ~7500 multicore score for the low-end Mac is pretty remarkable. The higher end models could be amazing overall.

 

[IG88]

you havent read the previous 15 pages where this is compared?

Yes, except it's difficult to sludge through all of the useless fawning on Apple comments.

I should have been more specific, I apologize.

Can someone make a useful and worthwhile comment on these graphs?

 

[jido]

Has anyone seen a Cinebench R23 result with on an Apple Silicon Mac running the X86 and ARM version of Cinebench?

I'd love to see in Cinebench what the Rosetta 2 impact is.

It is on Anandtech.

 

[matt_k]

Looks like ~15W in Cinebench:

https://twitter.com/x/status/1328777333512278020

So perf/watt, the M1 is significantly better than the other offerings in single core, but trails the AMD Zen 2 APU's in multi-core. With the caveat that Zen 2 APU's are on 7 NM process, and the Zen 3 on 7NM+ blows Zen 2 away

Either way, it is clear that Intel is way behind in 3rd place at the moment.


Admittedly, this is nitpicking, but I'm not sure how Anandtech came to that conclusion when they bench'd the M1 against the Zen 2 4800U locked at 15W TDP and saw it trade blows with one another, winning some (particularly in ST) and losing others (with more MT).

I honestly would want to see more benchmarks of the 4800U. The 15W performance is suspiciously close to 4800H 45W.

 

[Taz Mangus]

These stand alone benchmark tests are well and good but I would like to see a 4K/8K video render test of a laptop with the AMD Zen 3 4800U with 8-core iGPU vs the Apple M1 with 8-core iGPU MacBook Air or MacBook Pro.

 

[PickUrPoison]

No its not....

Intel And AMD x86 Mobility CPUs Destroy Apple’s M1 In Cinebench R23 Benchmark Results

A chart from that blog post:

I read that post and it’s nothing more than fanboy crap. It’s a complete joke. Blogger loves the word “thrashed” lol. Since when does thrashed mean “no perceptible difference” lol?

1498 vs 1538? And AMD’s highest score of 1284—though most AMD CPUs are in the 1100s—is AMD “destroying” the M1?

 

[PickUrPoison]

The true capability of CPU is meaningless if you don't have thermal capacity to sustain it. If you look at MacBook Air and Pro geekbench, you might expect that they're going to perform same, which definitely isn't the case. Maybe M1 is the 10k cinebench MC CPU if you increase voltage, overclock and cryo cool it. But that hardly means anything.

No, I wouldn’t expect that at all. I would suspect that they’re the same CPU though. That’s my point. And OP’s point.

But you’ve again made my point for me: attempting to characterize the performance of a CPU, if said CPU is in a thermally constrained enclosure, is foolish. You will get a false reading as the CPU is forced to throttle due to the enclosure hampering true performance.

It seems like you do understand that, since you keep posting examples of it. Are you just feigning ignorance?

 

[matt_k]

No, I wouldn’t expect that at all. I would suspect that they’re the same CPU though. That’s my point. And OP’s point.

But you’ve again made my point for me: attempting to characterize the performance of a CPU, if said CPU is in a thermally constrained enclosure, is foolish. You will get a false reading as the CPU is forced to throttle due to the enclosure hampering true performance.

It seems like you do understand that, since you keep posting examples of it. Are you just feigning ignorance?

This is such a waste of time. Talking about "performance of CPU" without taking into account the thermal and power constraints in which the CPU must perform is simply meaningless. CPUs don't exist in vacuum, there is always a context they need to be evaluated in. You can put 5950x in a 13" laptop and geekbench will likely give you amazing result, disregarding that as soon as you try to actually stress the CPU all performance will go to hell. It doesn't give you the whole picture.

Case in point: i7-1185G7 has 1585 single core score, 6000 multicore. You could say that it's pretty close to M1. Except to sustain this under load, it needs 2 - 3 times as much power as M1 does. That's a mind boggling difference in overall experience (heat, noise, battery life), and geekbench will not tell you anything about this.

 

[Robospungo]

This is impressive much like first generation Ryzen was. The start of being competitive but not quite there yet.

It took two more generations for AMD to really beat out Intel and then one more to embarrass Intel, it will be interesting to see how Apple silicon matures.

No its not....

Intel And AMD x86 Mobility CPUs Destroy Apple’s M1 In Cinebench R23 Benchmark Results

Sorry, I meant U series. I’m not taking about 45w H series chips or desktop chips.

The M1 beats everything Intel’s top of the line i7 15w chip offers in multicore, graphics, and battery, and is equal with Intel’s best in single core.

The only AMD chips that beat the M1 in multicore are the 4600U and the 4800U, and both get stomped by the M1 in single core, graphics, and battery life.

 

[PickUrPoison]

This is such a waste of time. Talking about "performance of CPU" without taking into account the thermal and power constraints in which the CPU must perform is simply meaningless. CPUs don't exist in vacuum, there is always a context they need to be evaluated in. You can put 5950x in a 13" laptop and geekbench will likely give you amazing result, disregarding that as soon as you try to actually stress the CPU all performance will go to hell. It doesn't give you the whole picture.

Case in point: i7-1185G7 has 1585 single core score, 6000 multicore. You could say that it's pretty close to M1. Except to sustain this under load, it needs 2 - 3 times as much power as M1 does. That's a mind boggling difference in overall experience (heat, noise, battery life), and geekbench will not tell you anything about this.

It’s really not all so complicated. But apparently you really don’t understand that a shorter benchmark, that doesn’t cause the CPU to throttle, is better than running longer ones that induce throttling, when you’re comparing CPUs or trying to characterize a chipset’s performance 🤷‍♂️

Those longer runs only serve to demonstrate how insufficient any particular enclosure may be for the CPU’s thermal requirements.

It seems like all this word salad and recasting of the argument you’ve been flailing about with is nothing more than an attempt to distract from the fact that you’re wrong, and you don’t like being corrected.

Deal it: the only way to find out what performance the M1 is capable of is to drop it in a mini enclosure that has plenty of cooling—not to stuff it in a fanless Air and see how bad it throttles. We’re trying to determine the performance of the M1, not the (in)effectiveness of any particular cooling system 🙄 But you know that.

Once you find out the max raw performance, only then will you be in a position to start evaluating systems in various form factors that contain that chipset/SoC.

And since the max performance has already been properly determined, you’re now in a perfect position to know which systems let the CPU reach its potential vs. which systems provide insufficient cooling and let it down. That’s where the longer benchmarks are useful.

 

[throttlemeister]

Except that Geekbench tells you exactly nothing, zip, nada, zero about actual cpu or system performance. The only thing static benchmarks like this are good for is to generate random numbers for virtual pissing contests and fights on the internet. The only thing that makes sense for true performance comparisons are dynamic benchmarks simulating real world application usage. And that is exactly the one thing we have not seen yet.

And to be honest, I would expect Apple to be quite impressive there. Apple has always been pretty darn good at real world performance.

 

[throttlemeister]

These stand alone benchmark tests are well and good but I would like to see a 4K/8K video render test of a laptop with the AMD Zen 3 4800U with 8-core iGPU vs the Apple M1 with 8-core iGPU MacBook Air or MacBook Pro.

Not that easy. What are you going to use? Premiere on AMD and FCP on Apple? Then you are comparing editing suites, not systems. Premiere on both? You'd be comparing native x86 vs emulated ARM. Premiere x86 on AMD and beta ARM premier on Apple? Mature product vs feature incomplete and unoptimized.

The most reliable means would be to test M1 Mac vs x86 Mac on MacOS and using the same software, but then you wouldn't be able to compare to AMD systems.

 

[UCDHIUS]

I honestly would want to see more benchmarks of the 4800U. The 15W performance is suspiciously close to 4800H 45W.

imo a direct x86 AMD comparison would be the 4700U it’s 8c/8t.

I’ve posted some R23 results previously (The M1 smokes the 4700U in single core) and the 4700U (multi core) is only 23 points stronger.

it’s my personal daily driver laptop that I bought a few months ago.


1150 singlecore

7531 multicore

Actual CB screen shot is in the post.

it’s a impressive start for apple, being that this is the lowest end weakest chip they will produce and it will only get better from here.

 

[matt_k]

imo a direct x86 AMD comparison would be the 4700U it’s 8c/8t.

I’ve posted some R23 results previously (The M1 smokes the 4700U in single core) and the 4700U (multi core) is only 23 points stronger.

it’s my personal daily driver laptop that I bought a few months ago.


1150 singlecore

7531 multicore

Actual CB screen shot is in the post.

it’s a impressive start for apple, being that this is the lowest end weakest chip they will produce and it will only get better from here.

I think the really impressive part is that during cinebench MacBook Pro stays quiet, while all the other laptops (i7, H4800U) have to move lot of air around to keep up the performance. That means there's quite a bit of thermal headroom even in the single fan MacBook Pro body, which could potentially be used by additional cores (M1X).

 

[Alan Wynn]

Not that easy. What are you going to use? Premiere on AMD and FCP on Apple? Then you are comparing editing suites, not systems. Premiere on both? You'd be comparing native x86 vs emulated ARM. Premiere x86 on AMD and beta ARM premier on Apple? Mature product vs feature incomplete and unoptimized.

One could use FFMPEG or Blackmagic’s DaVinci Resolve.

The most reliable means would be to test M1 Mac vs x86 Mac on MacOS and using the same software, but then you wouldn't be able to compare to AMD systems.

I guess it would depend on what you were trying to test. If one is trying to test how well the system would work for a particular workload, testing with the best available software on the platform would do that. It might change as software improves and is optimized for the platform, but it would represent the best I could do on an actual task.

 

[fourthtunz]

And without the hassle of having hardly any native software too.

and what isn't native runs quite fast through rosetta luckily.

 

[Alan Wynn]

And without the hassle of having hardly any native software too.

I am pretty sure I asked you this before (in another thread), but do not remember (nor can I find) a response. If you already answered, just post the reference.

1. What set of benchmarks will you consider as the basis for comparison between the released Apple Silicon Mac systems and competitive Intel/AMD machines?
2. When doing our comparisons between Apple Silicon-based hardware and AMD/Intel based hardware, how will you pick the AMD/Intel chip to compare? What objective metric would you use to define equivalent systems for comparison? Machines at the same price point? Machines with the same max TDP? Something else? The point of this question is that since Apple will not be selling its SoCs to others, one cannot do it purely on price of the chip, one needs some other objective metric to decide what two items should be compared.
3. What objective criteria would Apple Silicon have to meet to be a successful product vs. Intel/AMD’s chips? (10% faster? 25%? 10% better battery life? 25%? Something else?) Once Apple starts to deliver high-end GPUs, what are your answers on those same metrics for those?
4. When did you purchase your most recent Mac from Apple or a third party reseller that was currently shipping at the time you purchased it?
5. What would be required for you to purchase an Apple Silicon-based system?

 

[throttlemeister]

I am pretty sure I asked you this before (in another thread), but do not remember (nor can I find) a response. If you already answered, just post the reference.

1. What set of benchmarks will you consider as the basis for comparison between the released Apple Silicon Mac systems and competitive Intel/AMD machines?
2. When doing our comparisons between Apple Silicon-based hardware and AMD/Intel based hardware, how will you pick the AMD/Intel chip to compare? What objective metric would you use to define equivalent systems for comparison? Machines at the same price point? Machines with the same max TDP? Something else? The point of this question is that since Apple will not be selling its SoCs to others, one cannot do it purely on price of the chip, one needs some other objective metric to decide what two items should be compared.
3. What objective criteria would Apple Silicon have to meet to be a successful product vs. Intel/AMD’s chips? (10% faster? 25%? 10% better battery life? 25%? Something else?) Once Apple starts to deliver high-end GPUs, what are your answers on those same metrics for those?
4. When did you purchase your most recent Mac from Apple or a third party reseller that was currently shipping at the time you purchased it?
5. What would be required for you to purchase an Apple Silicon-based system?

Not the one you are asking to, but these are very good and interesting open questions that could create discussion beyond the typical "Apple is the best thing since sliced bread" and "Apple sucks and lost their way".

1. For me that would be real-world application workloads of a wide variety of applications
2. There could be multiple. System price can be a very valid criteria, as what you have in your pocket dictates what you can buy. But power usage is just as valid, albeit a more technical qualification. In an ideal world, one would compare as many qualifiers as possible with as many systems so you can select which criteria is most important for you and go from there. This is why you need websites other than MacRumors, that are dedicated to testing hardware, that have a documented standard way of testing and have a database of historical tests done to the same standards so you actually can compare to whatever criteria for you personally is most important. Mind you, in my opinion there is no bad choice, as long as it is your choice based on what you find important. And your choice does not need to be validated against someone else's. If you need validation, you didn't make a choice, you got what you felt was the right thing for the crowd you're hanging with; sheep behavior. But as long as you made a conscious choice, that choice is right for you and anyone that disagrees should be told to stick it where the sun don't shine.
3. None you mentioned. If Apple customers are happy with the product and they are happy buying it at the price it is offered for by Apple, it's a successful products. Everything else is irrelevant.
4. irrelevant
5. For my personal use case, Apple has moved way beyond the performance per $$ I am willing to spend years ago and I don't see that improving anytime soon. So basically, not happening. And that's got nothing to do with their architecture choices. I still like their products, I just feel they are way too expensive for my decision qualifiers.

 

[Spock1234]

My new M1 Air's scores in GeekBench 5