M1 Macs Reviewed

[leadeater]

2 minutes ago, Bombastinator said:

You’re saying the 2006 and 2017 versions of the benchmark were not similar? Or that the 2017 test was not run?  Someone on a previous page said it was run, and that they were similar. 

The benchmark results are not similar, if you actually care about performance analysis. The delta between the two are 10% INT and 5% FP, sure close if you don't care.

[Curufinwe_wins]

11 minutes ago, Bombastinator said:

You’re saying the 2006 and 2017 versions of the benchmark were not similar? Or that the 2017 test was not run?  Someone on a previous page said it was run, and that they were similar. 

LAwLz views it as similar results. I don't think I would and leadeater doesn't either. It's 10% difference relative in int workloads, and 5% difference relative in fp loads between the two benchmark sets (how they rank the processors). Given that Apple itself only pushed a 5% ipc improvement this generation, and Intel has commonly come in beneath that gen on gen, it's pretty signficant. Now it wouldn't be a problem in a large testing suite, but it's a bad benchmark to use by itself and make sweeping generalizations about the processors with. Again, Anandtech doesn't do this with literally any other benchmark suite. I get lawls statement about ram limitations, but that should be disclosed and stated every time it comes up as to why they are using an obsoleted version of the benchmark.

[Curufinwe_wins]

7 minutes ago, leadeater said:

 

 

So I've just gotten out of bed so I have not read any of the review or really the thread but reading this post I'd like to just double check, so the M1 doesn't run at 3GHz rather 3.2GHz for many SC test? Also the power draw of the M1 (in this Mac Mini) is greater than 10W? Both these two bits of information indicate to me that the M1 is a lot less of the halo some have been trying to make out it is. Don't get me wrong it's still very good, but not the performance per watt metric people were using.

 

Did I understand what you said wrong?

Correct, that's what Anandtech found, and I have no reason to doubt them.

 

https://www.anandtech.com/show/16252/mac-mini-apple-m1-tested

 

In the finally more detailed testing analysis.

 

Quote

During average single-threaded workloads on the 3.2GHz Firestorm cores, such as GCC code compilation, we’re seeing device power go up to 10.5W with active power at around 6.3W. The active power figure is very much in line with what we would expect from a higher-clocked Firestorm core, and is extremely promising for Apple and the M1.

In workloads which are more DRAM heavy and thus incur a larger power penalty on the LPDDR4X-class 128-bit 16GB of DRAM on the Mac mini, we’re seeing active power go up to 10.5W. Already with these figures the new M1 is might impressive and showcases less than a third of the power of a high-end Intel mobile CPU.

In multi-threaded scenarios, power highly depends on the workload. In memory-heavy workloads where the CPU utilisation isn’t as high, we’re seeing 18W active power, going up to around 22W in average workloads, and peaking around 27W in compute heavy workloads. These figures are generally what you’d like to compare to “TDPs” of other platforms, although again to get an apples-to-apples comparison you’d need to further subtract some of the overhead as measured on the Mac mini here – my best guess would be a 20 to 24W range.

 

 

[Bombastinator]

2 minutes ago, leadeater said:

 

 

So I've just gotten out of bed so I have not read any of the review or really the thread but reading this post I'd like to just double check, so the M1 doesn't run at 3GHz rather 3.2GHz for many SC test? Also the power draw of the M1 (in this Mac Mini) is greater than 10W? Both these two bits of information indicate to me that the M1 is a lot less of the halo some have been trying to make out it is. Don't get me wrong it's still very good, but not the performance per watt metric people were using.

 

Did I understand what you said wrong?

As for me I don’t remember even talking about wattage.  The only thing i know about the m1 wattage is that it’s really low compared to desktop stuff.  How low I couldn’t tell you.  There seems to be a bunch of different opinions about that one.  It may have been I was making a rebuttal to another comment.

[Bombastinator]

4 minutes ago, Curufinwe_wins said:

LAwLz views it as similar results. I don't think I would and leadeater doesn't either. It's 10% difference relative in int workloads, and 5% difference relative in fp loads between the two benchmark sets (how they rank the processors). Given that Apple itself only pushed a 5% ipc improvement this generation, and Intel has commonly come in beneath that gen on gen, it's pretty signficant. Now it wouldn't be a problem by itself, but it's a bad benchmark to use by itself and make sweeping generalizations about the processors with. Again, Anandtech doesn't do this with literally any other benchmark suite. I get lawls statement about ram limitations, but that should be disclosed and stated every time it comes up as to why they are using an obsoleted version of the benchmark.

So it was run but you are saying that 5-10% is not sufficiently similar and that the benchmark is worthless anyway because it’s just a benchmark. 
 

It seems reasonable to say that treating a benchmark as more than a benchmark is silly.  I’m not sure that was done though.   There’s a certain amount of “that’s all there is to look at atm” that has to be dealt with.  As for the similar or not similar thing it may be rougher than is preferred, not sure that makes it totally worthless though.  Just less useful than would be preferred. 

[ne0tic]

1 hour ago, Curufinwe_wins said:

I mean yes, but also A14 only managed a 5% ipc improvement over A13. And A13 was less of an improvement over A12 than A12 was over A11.

I'm not sure where you got this information from but I'm pretty sure there was a 15% IPC improvement with the A14 compared to the A13. A12 over A11 wasn't that big at all, A13 over A12 was a really big improvement tho (about 20%) so I think you may have mixed those up. 

[Curufinwe_wins]

Just now, Bombastinator said:

So it was run but you are saying that 5-10% is not sufficiently similar and that the benchmark is worthless anyway because it’s just a benchmark. 
 

It seems reasonable to say that treating a benchmark as more than a benchmark is silly.  I’m not sure that was done though.   There’s a certain amount of “that’s all there is to look at atm” that has to be dealt with.  As for the similar or not similar thing it may be rougher than is preferred, not sure that makes it totally worthless though.  Just less useful than would be preferred. 

I didn't say it was worthless. I specifically said it wasn't worthless. I said making general sweeping generalizations in comparative analysis (esp statements that M1/A14 beats X and Y overall) was premature based on just those two tests. And that as a general rule I expect Anandtech to be much more consistent with removing legacy benchmarks as they do for everything else, and if not, explicitly discussing why they didn't (they made a long discussion about this in their last ARM server review, but didnt mention it at all in their A14 discussion and I do find that negligent). Or their A13 review.

[Curufinwe_wins]

5 minutes ago, ne0tic said:

I'm not sure where you got this information from but I'm pretty sure there was a 15% IPC improvement with the A14 compared to the A13. A12 over A11 wasn't that big at all, A13 over A12 was a really big improvement tho (about 20%) so I think you may have mixed those up. 

"For the rest of the workloads, the A14 generally looks like a relatively linear progression from the A13 in terms of progression, accounting for the clock frequency increase from 2.66GHz to 3GHz. The overall IPC gains for the suite look to be around 5% which is a bit less than Apple’s prior generations, though with a larger than usual clock speed increase."

 

From Anandtech. Clock speed changes along the way have changed things. IPC progression as percentage gains have slowed down significantly.

[GDRRiley]

2 hours ago, Bombastinator said:

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.

my mistake. its zen2 with RNDA2/3

right but they hit 2080ti GPU levels which are way faster than this on the GPU side

APUs are Q1 which means final testing now

[Curufinwe_wins]

End of the day, I'm not saying these are bad. I'm not saying they are anything short of an incredible effort and demonstration of prowess. Still some lingering questions, but that's true with any new launch.

 

My P1 comment is the best one for my overall thoughts. 

 

 

[ne0tic]

2 minutes ago, Curufinwe_wins said:

"For the rest of the workloads, the A14 generally looks like a relatively linear progression from the A13 in terms of progression, accounting for the clock frequency increase from 2.66GHz to 3GHz. The overall IPC gains for the suite look to be around 5% which is a bit less than Apple’s prior generations, though with a larger than usual clock speed increase."

 

From Anandtech. Clock speed changes along the way have changed things. IPC progression as percentage gains have slowed down significantly.

Okay, makes sense. Although that is based on if clock frequency scales this well, but it seems to be doing that! 

[LAwLz]

2 hours ago, Curufinwe_wins said:

I mean yes, but also A14 only managed a 5% ipc improvement over A13. And A13 was less of an improvement over A12 than A12 was over A11. I don't want to pretend like I'm expecting Apple to flounder, but they have been slowing down their CPU growth recently. GPUs and other dedicated silicon? Now that is flying ahead. But it's worth noting that it seems like Apple is also hitting some diminishing return situations. 

 

This may not actually be the worst relative position Apple has compared to incumbent x86, but that all depends on how much iterative performance from here out each side can deliver. And we've been waiting for quite a while still for Intel to pick back up. Obviously AMD has been punching yoy gains far and above anyone else atm.

 

 

 

 

 

Also also! This literally validates Linus's position IMO. M1 is only 10-15% faster than A14 despite being almost 4x the power (comparing peak to peak with iphone vs mac mini). I do hope that Apple can broaden their performance curves a bit more moving forward.

 

It is an iPad. Now turns out, iPads are pretty darn powerful though.

The M1 is like twice as powerful as the A14. Not sure where you're getting 10-15% from. 

 

Sure it might only be 10-15% faster in single core workloads but most of the high power budget is there because they doubled the amount of cores. 

You can't just say "it's only 15% higher performance at 20 watts" when you're using single core figures for performance but multi-core figures for power consumption. 

[Spindel]

Since people can’t stop saying ”well desktop CPU is as fast or faster in MC” (and it’s been like this since the first GB scores surfaced).

 

I think we can confirm Apples success with the M1. 
 

 

After seeing some reviews I fond it interesting that the M1 MBP seems to beat (only slightly) a top spec 2019 16” MBP running Premiere under Rosetta.  

[Bombastinator]

20 minutes ago, GDRRiley said:

my mistake. its zen2 with RNDA2/3

right but they hit 2080ti GPU levels which are way faster than this on the GPU side

APUs are Q1 which means final testing now

I still remember reading under full zen2 performance for cpu and hotted up navi10 rather than either Navi 21 or 23.  I remember seeing the term “navi15” used once. Part of the issue is both the ps5 and XboxX hardware has been around for testing purposes literally for years.  XboxX mules sent to game devs actually predate zen2 release. I’ve seen the new console chip referred to as a “2700 variant” though apparently an improved one. I don’t know how improved.  Might reach into the zen2 performance envelope.

[Curufinwe_wins]

2 minutes ago, LAwLz said:

The M1 is like twice as powerful as the A14. Not sure where you're getting 10-15% from. 

ST comparisons.  

https://www.anandtech.com/show/16252/mac-mini-apple-m1-tested/4 among others.

 

It does have 2x the Firestorm cores though so yes.

 

Looking right back at it though. I did mess up. Obviously single core power isn't going to be anywhere near the full 6W MT peak for A14 and it was only as high as 15W for M1 at that level as well. Same basic idea though that it's clearly far past the efficiency curve to run at these clocks. (Which tells you something about the mammoth efficiency of mobile phones at low power).

[Spindel]

Also this was kind of impressive

 

 

link with more info

https://www.macrumors.com/2020/11/17/apple-silicon-m1-compiles-code-as-fast-as-mac-pro/

[wall03]

5 hours ago, JoseGuya said:

I'm still very conflicted. Do I switch my aging 15 inch late 2013 MBP to the M1 MBP, or to the 2020 16 inch Intel? Do I wait for the 16 inch with Apple Silicon? WHAT TO DO

wait for gen 2, gen1 always has undiscovered problems

[Curufinwe_wins]

3 minutes ago, Spindel said:

Also this was kind of impressive

 

 

link with more info

https://www.macrumors.com/2020/11/17/apple-silicon-m1-compiles-code-as-fast-as-mac-pro/

It is impressive, but we really need to remember that Apple has been underdesigning their cooling solutions so badly for so many years now that performance of the same chip in their laptops is way worse than in anyone elses. Almost as if to intentionally make this comparison look stronger.... 

 

But change the order a bit, close the gaps, it's still impressive stuff. With or without the "win".

 

(I do think it's extremely important to note that Apple has literally been doing a disservice with how bad most of their chassis designs have been, and literally everyone in the industry knows it, so giving a free pass over that is not fair).

[Bombastinator]

6 minutes ago, Curufinwe_wins said:

It is impressive, but we really need to remember that Apple has been underdesigning their cooling solutions so badly for so many years now that performance of the same chip in their laptops is way worse than in anyone elses. Almost as if to intentionally make this comparison look stronger.... 

 

But change the order a bit, close the gaps, it's still impressive stuff. With or without the "win".

 

(I do think it's extremely important to note that Apple has literally been doing a disservice with how bad most of their chassis designs have been, and literally everyone in the industry knows it, so giving a free pass over that is not fair).

I have noticed some sandbagging via under cooling from Apple on their intel chips too.  As such I am more interested in comparisons from hackintoshes or win10.  It is important to keep size considerations in mind though. 

[LAwLz]

8 minutes ago, Curufinwe_wins said:

It is impressive, but we really need to remember that Apple has been underdesigning their cooling solutions so badly for so many years now that performance of the same chip in their laptops is way worse than in anyone elses. Almost as if to intentionally make this comparison look stronger.... 

 

But change the order a bit, close the gaps, it's still impressive stuff. With or without the "win".

 

(I do think it's extremely important to note that Apple has literally been doing a disservice with how bad most of their chassis designs have been, and literally everyone in the industry knows it, so giving a free pass over that is not fair).

Are you sure the cooling in the Mac Pro is bad? 

The cooling in the MacBook Pro was bad but I haven't heard complaints about the Mac Pro having poor thermal solutions. 

I think the bigger issue with that benchmark is that it doesn't specify which Mac Pro they were testing against. 

The Mac Pro goes all the way from an 8 core CPU to a 24 core CPU. The test is probably the 8 core model at 3.5GHz.

 

[Bombastinator]

7 minutes ago, LAwLz said:

Are you sure the cooling in the Mac Pro is bad? 

The cooling in the MacBook Pro was bad but I haven't heard complaints about the Mac Pro having poor thermal solutions. 

I think the bigger issue with that benchmark is that it doesn't specify which Mac Pro they were testing against. 

The Mac Pro goes all the way from an 8 core CPU to a 24 core CPU. The test is probably the 8 core model at 3.5GHz.

 

I think there have been some systems where it isn’t bad and some where it is.  The 16” in at least a few models was good, and the Mac Pro in I think all models was good. Some weren’t though. Such an issue could only apply to the models that did have such an issue as not all did.  My memory is with the 16” the first 16” was good (2018?2019?) but following models had increasing problems. 

[randomhkkid]

31 minutes ago, Curufinwe_wins said:

It is impressive, but we really need to remember that Apple has been underdesigning their cooling solutions so badly for so many years now that performance of the same chip in their laptops is way worse than in anyone elses. Almost as if to intentionally make this comparison look stronger.... 

 

But change the order a bit, close the gaps, it's still impressive stuff. With or without the "win".

 

(I do think it's extremely important to note that Apple has literally been doing a disservice with how bad most of their chassis designs have been, and literally everyone in the industry knows it, so giving a free pass over that is not fair).

The MBP 16 actually has pretty good cooling as long as CPU and GPU aren't loaded up at the same time. It's still not adequate for combined CPU and GPU load (I have a whole reddit thread of a VRM mod I did) but in terms of CPU benchmarks it's pretty comparable to top of the line windows laptops. It's not a valid knock on Apple to say M1 vs intel is gimped.

 

Edit: To backup my claims you can see here I posted that the MBP 16 can maintain 3.3Ghz on all 16 threads for an unlimited amount of time. Then notebookcheck showing the Blade Pro 17" maintaining around 3.3 - 3.6GHz in a stress test using a newer 10th gen 8 core.

[Belgarathian]

42 minutes ago, Curufinwe_wins said:

It is impressive, but we really need to remember that Apple has been underdesigning their cooling solutions so badly for so many years now that performance of the same chip in their laptops is way worse than in anyone elses. Almost as if to intentionally make this comparison look stronger.... 

 

But change the order a bit, close the gaps, it's still impressive stuff. With or without the "win".

 

(I do think it's extremely important to note that Apple has literally been doing a disservice with how bad most of their chassis designs have been, and literally everyone in the industry knows it, so giving a free pass over that is not fair).

Or alternatively, Intel has been consistently letting Apple down on the development front so that their hardware doesn't have sufficient cooling capacity. Hence the M1 we have today.

[Belgarathian]

1 hour ago, LAwLz said:

Are you sure the cooling in the Mac Pro is bad? 

The cooling in the MacBook Pro was bad but I haven't heard complaints about the Mac Pro having poor thermal solutions. 

I think the bigger issue with that benchmark is that it doesn't specify which Mac Pro they were testing against. 

The Mac Pro goes all the way from an 8 core CPU to a 24 core CPU. The test is probably the 8 core model at 3.5GHz.

 

The latest 16" MBPs all benched fine for thermals and the Mac Pro is mint. No idea what the dude you're responding to is on about. 

 

Also it's well known that part of the reason for the M1 is because Intel has consistently under delivered on their roadmap, so it's entirely possible that part of the reason for the underwhelming thermal performance in previous generation Macs is at least partially because they were designed for a more efficient chip that never arrived in time for the launch.

[vetali]

Blizzard has native day 1 support of ARM Macs with World of Warcraft. Hopefully somebody runs it and posts results, because that game is really CPU intensive in certain areas. Saw one review running it but it was likely through Rosetta because the patch came out this morning.

 

https://us.forums.blizzard.com/en/wow/t/mac-support-update-november-16/722775