M1 Macs Reviewed

[Sorenson]

1 hour ago, A51UK said:

Having look up geekbench, it look like it be really bad benchmark for years. We should wait for more benchmarks, It best to not go by one benchmark anyway. I hope we find out today or tomorrow, also be careful with youtube channel you look at for review, it good idea not to go on pro-apple or pro-pc but more in the between like Linus or Game Nexus. people who love computer but not pro anything. 

Dave2d did more real world testing and it holds up shockingly well. It’s looking pretty clear that it’s a huge step up over the intel chips in native apps and a decent increase in Rosetta 2 apps.

[n0stalghia]

4 minutes ago, Aidoneus said:

I’m not sure the emulation of x86 to AArch64 will be having all that much impact on that number. RoTR already had a native Mac x86 version, so presumably all the graphics APIs are Metal anyway, so no translation is necessary for M1. It’s only any CPU calls that would need translating, and I don’t think games like RoTR are traditionally very CPU heavy. 

Oh, right, that makes sense. CPU calls are emulated, GPU runs natively.

 

In that case, not insane performance, but rather a very good integrated GPU. I mean, it's trading blows with dedicated mobile GPUs, Intel afaik isn't even remotely close to that yet. 

[Aidoneus]

4 minutes ago, n0stalghia said:

Oh, right, that makes sense. CPU calls are emulated, GPU runs natively.

 

In that case, not insane performance, but rather a very good integrated GPU. I mean, it's trading blows with dedicated mobile GPUs, Intel afaik isn't even remotely close to that yet. 

Oh yeah, I’m not knocking the performance of the chip at all- it’s really impressive.

 

Just not sure that ‘and this is UNDER EMULATION!’ Is a great modifier when looking at a game with an existing native x86 version.

[Commodus]

The more reviews I see, the more I suspect x86 (and particularly Intel) has been holding Apple back, if not the entire industry.

 

I don't think Apple will always have a commanding lead or make the most sense (particularly not when dedicated GPUs are involved). But if the M1's performance translates to higher-end chips, Intel and AMD have good reason to be worried. So does Microsoft, for that matter — it's whiffing on ARM at the moment, and its old obsession with legacy support might come to haunt it. The last thing Microsoft wants is to be stuck on x86 while ARM blows past.

[Bombastinator]

What I am seeing here:

 

the m1 machines seem quite capable for thin and light laptops.  There seem to be some laptops that are faster but they are not thin-and-lights, and whether this can be scaled beyond thin-and-light remains unknown.  
 

Notable is the devices currently have effectively near zero expansion capacity, which is low even for Apple.  If Apple can produce scaling and expansion capability this could be a massive thing, but such has not been shown yet.  one thing I’m quite curious about is how the devices do for gaming across a range of PC games. Gaming on thin-and-lights has previously been painful when possible at all.  If this has changed a large section of the non thin-and-light laptop market has opened for Apple

Edited November 17, 2020 by Bombastinator

[Spindel]

4 minutes ago, Commodus said:

The more reviews I see, the more I suspect x86 (and particularly Intel) has been holding Apple back, if not the entire industry.

 

I don't think Apple will always have a commanding lead or make the most sense (particularly not when dedicated GPUs are involved). But if the M1's performance translates to higher-end chips, Intel and AMD have good reason to be worried. So does Microsoft, for that matter — it's whiffing on ARM at the moment, and its old obsession with legacy support might come to haunt it. The last thing Microsoft wants is to be stuck on x86 while ARM blows past.

In theory MS could always pay for some software exclusivity of certain apps or some ”killer apps” for some industries. A bit like it is on the gaming console buissiness. 

[LAwLz]

@leadeater Told you so.

The M1 is fantastic.

Slightly better than Zen3 for single core performance.

Way better than anything Intel has to offer.

Only gets beaten in heavily threaded applications when compared to 6+ core chips.

 

Also, the SPEC2006 and SPEC2017 results are more or less the same. Can we please stop hating on SPEC2006 because "hurr durr it's old so therefore it's bad" now?

Changing from SPEC2006 to SPEC2017 changes things a little, but not to the point where one chip might look really good and then in the newer/older test it looks bad all of a sudden. It might changes a couple of percentages here and there but that's about it. Easily the kind of differences you can see in real world applications as well.

 

SPEC2006 int:

M1 - 69.40

Zen3 - 68.53

M1 performance compared to Zen3 - 101%

 

SPEC2017 int:

M1 - 6.66

Zen3 - 7.29

M1 performance compared to Zen3 - 91%

 

SPEC2006 fp:

M1 - 104.10

Zen3 - 94.08

M1 performance compared to Zen3 - 111%

 

SPEC2017 fp:

M1 - 10.37

Zen3 - 9.79

M1 performance compared to Zen3 - 106%

[captain_to_fire]

34 minutes ago, Commodus said:

The more reviews I see, the more I suspect x86 (and particularly Intel) has been holding Apple back, if not the entire industry.

 

I don't think Apple will always have a commanding lead or make the most sense (particularly not when dedicated GPUs are involved). But if the M1's performance translates to higher-end chips, Intel and AMD have good reason to be worried. So does Microsoft, for that matter — it's whiffing on ARM at the moment, and its old obsession with legacy support might come to haunt it. The last thing Microsoft wants is to be stuck on x86 while ARM blows past.

Well Intel blew off their chances...

• Paul Otellini, a past Intel president admitted that he regrets of not accepting Apple’s offer to build the chip for the first iPhone 
• An ex Intel engineer said that Apple is the number one who finds bugs in Skylake 
• Intel promised 10 nm transistors much earlier, it took the years to have one

[Kisai]

58 minutes ago, n0stalghia said:

Wait, the Mac mini has higher FPS in Tomb Raider even while emulating the game through Rosetta than a mobile 560X? On 20W TDP? what

It doesn't take much to beat iGPU parts. The 560X is also 3 years old on 14nm, like the 1050Ti.

 

Like I wouldn't call this awesome/impressive, just an interesting comparison when there's a 4th option.

 

Nobody really wants to play games at 1080p unless it's 1080p60+ and Intel/AMD iGPU and nVidia 1050Ti/ AMD 560X parts are just barely within that performance envelope. So in most cases the iGPU parts will fall short, and even the dGPU parts in laptop's will fall short. Nobody really builds "macmini" type systems other than Apple, so when the MacMini beats a laptop, that's likely due to higher thermal envelopes being possible. The Macbook Air on the otherhand couldn't even outperform the iPad of the same size with the Intel iGPU starting around the A10.

 

Which is to say, it's interesting to see the benchmarks, but keep your skeptic hat on, because until there is a lot of software optimized for the ARM cpu, these benchmarks are going to be deceiving. Anyone who used the PPC and Intel mac's can tell you that it's not just the translation layer that increases the load on the cpu, it's also the how other software interacts with translated binaries. Running your older x86 licenced software on a new ARM may work, but you may also need to buy new versions of software that worked perfectly fine on x86 for 10 years but hasn't been optimized for the ARM cpu, leading you to throw money at solutions that are half baked.

 

Like if everything you need and use (and Apple does provide enough software to do this, to be fair) is available from Apple itself, and the few vendors that actually had time to make the migration then jumping on to the ARM Mac's may be fine and you have nothing to worry about. However if you require a specific program, then you need to wait until that program has native ARM binaries, because a 20% performance hit is also a 20% productivity hit. It's the kind of thing that is ok to overlook when you're running Office or a Web Browser, but not something you can overlook for graphics and video work.

[williamcll]

Very interesting, I wonder how well it runs on other OS

[Curufinwe_wins]

47 minutes ago, LAwLz said:

@leadeater Told you so.

The M1 is fantastic.

Slightly better than Zen3 for single core performance.

Way better than anything Intel has to offer.

Only gets beaten in heavily threaded applications when compared to 6+ core chips.

 

Also, the SPEC2006 and SPEC2017 results are more or less the same. Can we please stop hating on SPEC2006 because "hurr durr it's old so therefore it's bad" now?

Changing from SPEC2006 to SPEC2017 changes things a little, but not to the point where one chip might look really good and then in the newer/older test it looks bad all of a sudden. It might changes a couple of percentages here and there but that's about it. Easily the kind of differences you can see in real world applications as well.

 

SPEC2006 int:

M1 - 69.40

Zen3 - 68.53

M1 performance compared to Zen3 - 101%

 

SPEC2017 int:

M1 - 6.66

Zen3 - 7.29

M1 performance compared to Zen3 - 91%

 

SPEC2006 fp:

M1 - 104.10

Zen3 - 94.08

M1 performance compared to Zen3 - 111%

 

SPEC2017 fp:

M1 - 10.37

Zen3 - 9.79

M1 performance compared to Zen3 - 106%

Actually I'd say this is a pretty good reason to hate on it from the "don't just use this and then claim that Apple is the best" perspective. Again, no one was doubting that it's still rather strong. Still competitive. (I mean you did see my comment on page 1) 

 

It went from leading in everything to 10% lag in int and half the lead (5%) in fp.

 

Just that Andrei's crowning was and is premature from just spec2006 and GB5. We never said that it was worthless. 

 

And again, part of my personal problem is that Anandtech intentionally sunsets basically every single other obsoleted benchmark rather aggressively (they dont show GB3 or GB4 or R11 or R15 or their old SSD tests to go on and on and on). So holding on to this one seems very very strongly out of place for their ethos.

[randomhkkid]

25 minutes ago, Kisai said:

It doesn't take much to beat iGPU parts. The 560X is also 3 years old on 14nm, like the 1050Ti.

 

Like I wouldn't call this awesome/impressive, just an interesting comparison when there's a 4th option.

 

Nobody really wants to play games at 1080p unless it's 1080p60+ and Intel/AMD iGPU and nVidia 1050Ti/ AMD 560X parts are just barely within that performance envelope.

I think it's impressive we're getting a part that sits at 20 -25W doubling the GPU performance of a 4700U. I'd call that pretty awesome for the ultrabook space even if not of much consequence in the desktop.

[Curufinwe_wins]

2 minutes ago, randomhkkid said:

I think it's impressive we're getting a part that sits at 20 -25W doubling the GPU performance of a 4700U. I'd call that pretty awesome for the ultrabook space even if not of much consequence in the desktop.

I also agree that pushing the minimal acceptable spec forward is a big deal. A huge deal in fact.

 

Bigger for most people than even the high end desktop space. When you try to explain to someone you don't need a 1000 dollar graphics card to run rocket league or sc2 or league of legends etc... that's the biggest thing holding PC gaming back, the perception that it is p2w and only for the rich.

[LAwLz]

9 minutes ago, Curufinwe_wins said:

It went from leading in everything to 10% lag in int and half the lead (5%) in fp.

It went from a 1% lead to 9% behind, and 11% lead to 6% lead. You're making it sound like a way bigger difference than it is.

It is barely outside margin of error if you ask me, and not enough to actually change anyone's opinion. A couple of percentage points back or forth isn't going to make someone go from "wow it's amazing" to "it's alright" or "it's bad" to "it's good".

 

 

13 minutes ago, Curufinwe_wins said:

And again, part of my personal problem is that Anandtech intentionally sunsets basically every single other obsoleted benchmark rather aggressively (they dont show GB3 or GB4 or R11 or R15 or their old SSD tests to go on and on and on). So holding on to this one seems very very strongly out of place for their ethos.

I explained why they keep using SPEC2006 in the other thread.

SPEC2017 takes way longer to run and they run into memory issues on iOS because it is fairly limited in what programs are allowed to do. But since the scores are very similar anyway it doesn't make sense to spend hours upon hours to run a test where a different tests that takes a couple of hours gives similar results.

[DrMacintosh]

Looks like M1 Macs are incapable of emulating OpenGL and I'm still waiting to see an M1 Mac running DX9-12 games under a Windows 10 VM. 

[Orangeator]

Looks to be very impressive, I seriously cannot wait to see the high end of Apple Silicon. I want to see 16 full performance cores of this running at 85-120watts. I want to see this pushed to its limits lol. 

[LAwLz]

Just now, DrMacintosh said:

Looks like M1 Macs are incapable of emulating OpenGL and I'm still waiting to see an M1 Mac running DX9-12 games under a Windows 10 VM. 

Can the M1 Macs even run Windows VMs?

They don't support virtualization of x86, and I wouldn't be surprised if Windows on ARM didn't have the necessary drivers.

[DrMacintosh]

Just now, LAwLz said:

They don't support virtualization of x86

Oh, then these Macs really are dead to me. They are literally roided ChromeBooks

[Citadelen]

8 minutes ago, DrMacintosh said:

Oh, then these Macs really are dead to me. They are literally roided ChromeBooks

Chrome OS can now run Linux, and using Parallels, Windows apps

[DrMacintosh]

1 minute ago, Citadelen said:

Chrome OS can now run Linux, and using parallels, Windows apps

So ChromeBooks are more advanced than Apple Silicon Macs  

[Bombastinator]

9 minutes ago, DrMacintosh said:

Oh, then these Macs really are dead to me. They are literally roided ChromeBooks

Calling them roided chrome books is pretty inaccurate I think.  They do do Rosetta of some version which apparently does many of the things virtualization does. Not all of them though so it’s possible they won’t work for you at all.  It seems to be situational. 

[GDRRiley]

2 hours ago, Curufinwe_wins said:

Happy to see much more details in this piece from Anandtech. Looking pretty dope for a first attempt. Best efficiency and performance in the <35W market.

right now. but look what zen3+RDNA2/3 can do in the PS5/xobx series x/s.

Zen3 APUs are going to beat it if they get RDNA2 GPUS.

[Bombastinator]

1 minute ago, GDRRiley said:

right now. but look what zen3+RDNA2/3 can do in the PS5/xobx series x/s.

Zen3 APUs are going to beat it if they get RDNA2 GPUS.

Ps5/XboxX isn’t zen3. It seems to be more like zen1.75.  Sort of between zen+ and zen2 with rdna10+ GPUs. They also use more wattage.  That is not to say that zen3 and RDNA2 21/23 or something beyond that won’t crush it.  The future is the future though.  Apple would be able to do an “m2” or “Mx” or whatever a future chip might be called that could in theory meet it.  The complication is that while zen3 and RDNA2 of its various types has been shown to be functional, this has not yet happened with M series stuff.

[Commodus]

23 minutes ago, DrMacintosh said:

So ChromeBooks are more advanced than Apple Silicon Macs  

Parallels will offer VMs on Apple Silicon Macs! It is unfortunate that there's no OpenGL emulation, but I think Apple sees that as outdated tech regardless of your architecture. If it was going to adopt any open graphics tech now, it'd be Vulkan.

[DavidKalinowski]

2 hours ago, Commodus said:

The last thing Microsoft wants is to be stuck on x86 while ARM blows past.

I mean, M$ has tried to make windows on ARM more than once, and it fails almost every time. They are finally trying to get x64 emulation working with win10ARM and surface pro X but I have not kept up on that. (I also think M4 will simply create a linux distro that has native windows app support some time in the next 5 years, and roll it into the yearly "365" subscription ecosystem. They wont be charging for windows it self, but for the security updates, support, and emulation tech.) I think the only way M$ will need to worry about arm is if Intel and AMD start pushing it, which will force the PC industry to shift.  These macs, while impressive are still going to be niche at best. Apple would be smart to make the next version have at least expandable memory modules in the desktop versions. Even if its not needed; the marketing buzz alone would be worth it.