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

[djjeff]

They don't, they just ditched the technical debt that is the x86 instruction set.

You mean like what the PPC processors didn't have when they migrated to the technically laden debt of the x86 instruction set?

 

[jclo]

Juli, again, there is no “2020 16-inch MacBook Pro”. This clarification is important given recent speculation that there may be such an official release. For now, there is only a 2019 16-inch MacBook Pro.

Apologies, fixed that.

 

[djjeff]

And spec as you mentioned too, a test that many in the industry have respect for.

Yep, spec too. You will never see me recommending a system based on either one. Instead you'll find me recommending systems based on a users needs, not some lame benchmark.

 

[ValueArb]

Cinebench R23 (Single-Core)​

Cinebench R23 is the successor of Cinebench R20 and is also based on the Cinema 4 Suite. Cinema 4 is a worldwide used software to create 3D forms. The single-core test only uses one CPU core, the amount of cores or hyperthreading ability doesn't count.

	AMD Ryzen 5 5600X 6x 3.70 GHz (4.60 GHz) HT	1572
	Intel Core i7-1185G7 4x 3.00 GHz (4.80 GHz) HT	1538
	Intel Core i7-1165G7 4x 2.80 GHz (4.70 GHz) HT	1504
	Intel Core i9-10900K 10x 3.70 GHz (5.30 GHz) HT	1418
	Intel Core i9-10900KF 10x 3.70 GHz (5.30 GHz) HT	1418
	AMD Ryzen 9 3950X 16x 3.50 GHz (4.70 GHz) HT	1371
	Intel Core i9-10850K 10x 3.60 GHz (5.20 GHz) HT	1367

Wow, all those 28-65 watt parts selling for $300+ barely outpace the M1 that costs Apple $65?

 

[Kabeyun]

I can't think of how this makes Intel look good in any light.
The M1 is also fanless, no?

Computer design affects whether or not a chip has a fan. It’s not intrinsic, but depends on things like how its power is managed and the thermal characteristics of its enclosure. In the case of the M1, the MacBook Air does not have a fan, while the MacBook Pro does.

 

[januarydrive7]

You mean like what the PPC processors didn't have when they migrated to the technically laden debt of the x86 instruction set?

To be fair, PPC processors were beasts when they migrated to intel -- it wasn't an issue of performance, but performance per watt. PPC was more interested in server demands than consumer, so Apple got the short end of that stick.

 

[matt_k]

So take a cloud Windows pc, problem solved. And seems Parallels is working for Windows support, or just keep your Intel one for the next years.

Again, it's not just Windows. Every now and then I need to fix and obscure bug that only happens on older macOS versions. I can run those in parallels without any issues. That's no longer possible with M1. For many people ARM migration is going to be a huge hassle. Just because it's not you it doesn't mean these people don't exist. So going through all that knowing that if Apple chose AMD instead it would get comparable performance with none of the compatibility issues would suck. I'm hopeful that's not the case.

And no offense but only Zen3 and that R7 4800U is nice, previous versions where not that impressive. Apple made this call years ago when AMD sucked really hard.

R7 4800U is still zen 2. We have yet to see what Zen on laptops will look like.

 

[Someyoungguy]

Well, I just ran Cinebench on my mid-2011 base iMac (633/1949) and I think I would absolutely love the M1 in a computer with a 20 hour battery life and no fan that fits in a mail slot!

 

[vadimyuryev]

See you tomorrow on the Tube

 

[Kung gu]

Just to be clear, for anyone who has used Cinebench in the past— this is the new version, so the scores have different values from before.

My 10900k in my PC just benched 14,217 multicore. So... M1 is not the PC crusher yet, in Cinebench at least.

The TDP of the 10900k is 125W TDP, the TDP of M1 is 10-18W. The M1 scored 7500 in multicore, whiles sonsuming A LOT LESS power. It scored 50% less still, so i don't what to say to u.

If the M1 had 12 CPU cores and a 45W TDP it would smoke ur 10900k,and we might get that in the 16" 2021 next year

 

[hot-gril]

You mean like what the PPC processors didn't have when they migrated to the technically laden debt of the x86 instruction set?

Yes. They were still at the mercy of chipmakers that were starting to fall behind in the early 2000s. For a while Apple was indeed putting the fastest CPUs in their machines.

 

[hugodrax]

A $600 CPU requiring 125 WATTS should outpace the M1 by quite a bit more than that, shouldn't it?

An Apple silicon cpu with a TDP of 125w will do 70000-75000 on that benchmark.

 

[hot-gril]

The TDP of the 10900k is 125W TDP, the TDP of M1 is 10-18W. The M1 scored 7500 in multicore, whiles sonsuming A LOT LESS power. It scored 50% less still, so i don't what to say to u.

If the M1 had 12 CPU cores and a 45W TDP it would smoke ur 10900k,and we might get that in the 16" 2021 next year

He knows. Point is, M1 doesn't win in multi-core, so it's not competitive in a big desktop tower *yet* if you care about multi-core performance a lot (which some day nobody will).

 

[now i see it]

Can't wait to build Cinema 4D models and render them out on my MacBook Air

/s

 

[EugW]

Yep, spec too. You will never see me recommending a system based on either one. Instead you'll find me recommending systems based on a users needs, not some lame benchmark.

Of course you just said you said you think Cinebench is a much better test than Geekbench. Well, for 90% of the computer buying population, Cinebench is even less relevant than Geekbench.

Sorry, you just contradicted yourself. You can’t can’t have it both ways.

 

[hot-gril]

Yep, spec too. You will never see me recommending a system based on either one. Instead you'll find me recommending systems based on a users needs, not some lame benchmark.

Xcode builds?

 

[smulji]

But in the meantime we should be honest that the M1 is competitive with Intel Tiger Lake and quite a bit behind AMD Ryzen 4000 mobile chips in multicore performance

The performance improvement figures that Apple touted during the November keynote, they were comparing it to the previous models that the new M1 Macs were replacing. So;

M1 MBA performance is compared to the previous dual-core / quad-core MBA
M1 MBA performance is compared to previous base-model 13' MBA
M1 Mac Mini performance is compared to previous base model quad-core i3 Mac Mini

If you go to each of the product pages on Apple's website and scroll to the bottom, it goes into more detail in the fine print. It isn't a secret.

 

[FilmIndustryGuy]

"but..but..what about..."

come up with something so that we can debate for 600 comments

 

[hugodrax]

Cant wait to see what the VUP score will be on this new processor.


(Vax Unit of Performance)

 

[djjeff]

Of course you just said you said you think Cinebench is a much better test than Geekbench. Well, for 90% of the computer buying population, Cinebench is even less relevant than Geekbench.

Sorry, you just contradicted yourself. You can’t can’t have it both ways.

I did? Where?

 

[djjeff]

Xcode builds?

I am not sure what it is you're attempting to say with this comment.

 

[joeguide4]

*only in Mac computers made by Apple... which are less than 10% of the industry.

Apple isn't selling these chips to anyone else in the industry.

Dell, HP, Lenovo and others will have to be satisfied with chips from Intel and AMD...

Actually, next year Apple will sell more Apple Silicon chips worldwide (A and M chips in computers, iPhone and iPad) than all the x86 chips (intel and AMD).

 

[avkills]

Nobody should be comparing the multi-core performance of their desktop class chip against the Apple M1. The M1 is a mobile chip. Comparing the single core performance is good because it can give us a good indication on what to expect from latter M series chips with more cores and larger memory pools.

Not a bad showing for Generation 1.

 

[matt_k]

That 60% difference in wattage is likely more important than you're giving credit for (also, scales on R20 to R23 are not identical, so it's more than a bit disingenuous to postulate here), not to mention the wattage on the M1 includes the entire SoC.

The difference in wattege is only 40%, not 60. And how is it disingenuous? Those extra 10W give 20% increase in R20, it's likely they would do the same in R23. There really isn't that much difference between those two benchmarks.

Additionally, for a pure look CPU, not including any throttling effects, Geekbench is likely the better benchmark, where M1 does handily outperform the 4800U.

Actually, no. The throttling effects are important. Geekbench is designed to prevent throttling, but that's only relevant if your real-life usage consists of short bursty loads. Case in point, my i7-1065g7 can easily go over 35W for short periods of time, but that's way beyond sustainable thermal envelope.

Finally, these are Apple's entry level chips -- a fairer comparison against AMD's higher performance laptop chips isn't available, yet. Those will be available likely in the next 6 months.

Indeed.

 

[vinegarshots]

Time to upgrade. Ryzen 5950X does 28,641 multi-core on Cinebench 23.

Yeah, that thing is a beast, but it wasnt available when I was building and I couldn't wait for it to be released