Discussion Apple Silicon SoC thread

Page 28 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Eug

Lifer
Mar 11, 2000
23,583
996
126
M1
5 nm
Unified memory architecture - LP-DDR4
16 billion transistors

8-core CPU

4 high-performance cores
192 KB instruction cache
128 KB data cache
Shared 12 MB L2 cache

4 high-efficiency cores
128 KB instruction cache
64 KB data cache
Shared 4 MB L2 cache
(Apple claims the 4 high-effiency cores alone perform like a dual-core Intel MacBook Air)

8-core iGPU (but there is a 7-core variant, likely with one inactive core)
128 execution units
Up to 24576 concurrent threads
2.6 Teraflops
82 Gigatexels/s
41 gigapixels/s

16-core neural engine
Secure Enclave
USB 4

Products:
$999 ($899 edu) 13" MacBook Air (fanless) - 18 hour video playback battery life
$699 Mac mini (with fan)
$1299 ($1199 edu) 13" MacBook Pro (with fan) - 20 hour video playback battery life

Memory options 8 GB and 16 GB. No 32 GB option (unless you go Intel).

It should be noted that the M1 chip in these three Macs is the same (aside from GPU core number). Basically, Apple is taking the same approach which these chips as they do the iPhones and iPads. Just one SKU (excluding the X variants), which is the same across all iDevices (aside from maybe slight clock speed differences occasionally).

EDIT:

Screen-Shot-2021-10-18-at-1.20.47-PM.jpg

M1 Pro 8-core CPU (6+2), 14-core GPU
M1 Pro 10-core CPU (8+2), 14-core GPU
M1 Pro 10-core CPU (8+2), 16-core GPU
M1 Max 10-core CPU (8+2), 24-core GPU
M1 Max 10-core CPU (8+2), 32-core GPU

M1 Pro and M1 Max discussion here:


M1 Ultra discussion here:


M2 discussion here:


Second Generation 5 nm
Unified memory architecture - LPDDR5, up to 24 GB and 100 GB/s
20 billion transistors

8-core CPU

4 high-performance cores
192 KB instruction cache
128 KB data cache
Shared 16 MB L2 cache

4 high-efficiency cores
128 KB instruction cache
64 KB data cache
Shared 4 MB L2 cache

10-core iGPU (but there is an 8-core variant)
3.6 Teraflops

16-core neural engine
Secure Enclave
USB 4

Hardware acceleration for 8K h.264, h.264, ProRes

M3 Family discussion here:

 
Last edited:

jeanlain

Member
Oct 26, 2020
149
122
86
Thank you (and Andrei of course for the follow-up). That puts the M1 core power consumption at 21 W fully loaded. I'm still thinking this is core only so add a couple of watts for uncore and then the Mac Mini system power numbers make perfect sense and this matches my earlier calculation.
One of the screenshots shows it's the total package power, inducing DRAM, during a POV-ray stress test. But it doesn't change much since the GPU is at 0 mW (lol) and the DRAM at just 60 mW.
It appears that POV-RAY is (in)famous for its power consumption.
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,629
136
One of the screenshots shows it's the total package power, inducing DRAM, during a POV-ray stress test. But it doesn't change much since the GPU is at 0 mW (lol) and the DRAM at just 60 mW.
It appears that POV-RAY is (in)famous for its power consumption.

I see at the bottom now. That's very impressive if true as that means total uncore power would only be 0.102 W. That seems almost too low to believe but I don't have anything else to go off of. How much IO does M1 have? Does it support any PCIe lanes, USB, etc?

Edit: Also, pov-ray isn't a power virus. It does hit the CPUs hard, but it is an actual renderer and has similar power consumption with something like Blender (another renderer). Something like p95 would have even higher power consumption (or force lower frequency).
 
Last edited:

name99

Senior member
Sep 11, 2010
404
303
136
My point was that even my iMac i5 in 2017 did very well with this app load test when it was brand new out of the box, so I think a lot of it has to do with the SSD speed.

My other point though was that this gets worse over time. It's great with a fresh install and newly installed apps, but with a machine that has been in active use for a couple of years, the load times clearly increased, at least on my machine.

IOW, it's true that M1 is fast, but I suspect a large of the great app loading performance in that video has to do with the fact that it's a brand-spankin' new machine with freshly installed apps, on a super fast SSD. Yes, the CPU matters, but that may not be the main bottleneck here.

I think the best example of this was when I was playing with some new OS installs. I ended up doing a clean install of Mojave and re-installing the apps, and all of a sudden, everything was screaming fast, including app loading. I then restored my original backup to the same computer, and the increased load times I was having before came back. So there was some sort of cruft on there that was slowing things down and I never figured out what it was. Note though the slower version of the install did have about 300 GB of photos and videos on it, so that couldn't have helped. (1 TB SSD.) The new clean OS installs were less than 30 GB even after all the applications were installed.

Oh, sure. SSDs make launching all apps faster. I wouldn't deny that at all.

But, along with all the other lunacy we are seeing today, there are claims that the reason M1 is doing so well in a variety of developer-centric tasks (like installing XCode, or large compiles) is all because the new machines have so much faster SSDs than their replacement machines...
 

name99

Senior member
Sep 11, 2010
404
303
136
I guess I should have been clearer in my question. How many PCIe lanes, how much USB, SATA, etc.? I didn't see that in the Anandtech article at least.

Why do you care?
There will be a bunch of internal lanes connected to WiFi and the SSD. There will be a bunch of internal USB connections to things like the keyboard/trackpad and the camera.
But none of that matters. What matters is there are two IO ports with the characteristics described.

More generally why are you so surprised by any of this? How do you think the SSD is connected to the SoC in an iPhone? It's been that way since, I think, the iPhone 6S. And my iPad Pro has a USB-C connector.
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
Oh, sure. SSDs make launching all apps faster. I wouldn't deny that at all.

But, along with all the other lunacy we are seeing today, there are claims that the reason M1 is doing so well in a variety of developer-centric tasks (like installing XCode, or large compiles) is all because the new machines have so much faster SSDs than their replacement machines...
The alternative explanation (that the M1 is really good) is getting pushback in some ways because it's kind of unbelievable. But some of the pushback comes from a different place: the results go against their worldview, so people look for chinks in the armor or ways the result could be invalid because of something else, or something not intrinsic to the M1 to criticize, like the ecosystem.
 

ultimatebob

Lifer
Jul 1, 2001
25,135
2,445
126
The alternative explanation (that the M1 is really good) is getting pushback in some ways because it's kind of unbelievable. But some of the pushback comes from a different place: the results go against their worldview, so people look for chinks in the armor or ways the result could be invalid because of something else, or something not intrinsic to the M1 to criticize, like the ecosystem.

I think that my one big problem with the M1 is that I still have a ton of old 32-bit Mac software that will not run on it.

I also know that this means that I will probably not get 7 years of Mac OS X upgrades on my Intel based 16" MacBook Pro like I did on my 2013 era 15" MacBook Pro. Apple will be looking to cut us off after 4 or 5 years, like they did on that old G4 iMac Mini that I got a long time ago.

The new systems actually look pretty good, but I wouldn't be trading my PC gaming rig for one anytime soon.
 
  • Like
Reactions: shady28

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,629
136
Why do you care?
There will be a bunch of internal lanes connected to WiFi and the SSD. There will be a bunch of internal USB connections to things like the keyboard/trackpad and the camera.
But none of that matters. What matters is there are two IO ports with the characteristics described.

More generally why are you so surprised by any of this? How do you think the SSD is connected to the SoC in an iPhone? It's been that way since, I think, the iPhone 6S. And my iPad Pro has a USB-C connector.

Why do you care why I care? If you don't have the answer, that's fine, you don't have to reply to my post.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
The A13 was budgeted at around $64 (again by those who model these things). If we bump it by 50% (this is considered aggressive when you look at the iPhone 12 BOM) we're at $90.
So we can take an M1 at maybe $135 or so?, about 1.5x (quibble about the price of the RAM if you insist on wasting time)

There is a HUGE range of numbers for BOM estimates for iPhones for the last few generations. SoC cost estimates have ranged from $30 to as high as $70. Obviously the people putting publicly available BOMs together either don't have access to TSMC's actual pricing on leading edge nodes, and/or have wildly different estimates for the yields achieved early in the node's life. I'm sure the more detailed information you can get if you're willing to pay $10,000 to one of the consulting firms that does this sort of thing as their job has better info - but unfortunately we don't get to see that info.

Everything I've read has the per wafer variable cost for succeeding TSMC processes increasing very little, definitely in the single digit range. The big increase for successive process generations are the fixed cost stuff like mask sets. But for someone with Apple's volumes that's almost irrelevant - if N5 mask sets cost $100 million that's only 50 cents per iPhone. $100 million for the A14X/M1 would be a little higher, maybe a couple bucks per iPad Pro/Mac.
 
  • Like
Reactions: Tlh97 and name99

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
more like 19% (0,84)

the only thing i don't like about this chip is the lack of i/o, it's a smartphone soc in disguise, no egpu support (no pcie lanes for it), just 1 external monitor. it has no point of differentiation from a possible a14x

Why would you expect eGPU and multi monitor in Apple's lowest end stuff? I doubt there will EVER be eGPU support for Apple Silicon Macs - all evidence is that Apple will be using their own GPUs across the entire line so why would NVidia or AMD write drivers for the handful of people who might want to use eGPU?

As for multi monitor, the laptops support only one external monitor but the Mini can do two - one up to 6K monitor via DP 1.4 (using the TB port) and one 4K monitor via HDMI 2.0.
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,685
136
Yes caching. Good point. But was that test cached? If so, then that totally negates the test. (I didn't check.)
I didn't check either, but it doesn't matter to me as it still shows a great level of finish on both OS and hardware side. CPU performance is best outlined in more intensive tests anyway.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
Looking at refresh timeframes on the Mac mini, MBA, and MBP - as well as their cadence on the iPad Pro and as such their X/Z chips, it looks like a 15-16 month cadence might be a starting point. However, the cadence on the iPad Pro muddies things. The device was refreshed in early 2020 but the chip is still using the same Vortex/Tempest 4+4 setup from the previous generation.

The iPad Pro was probably only refreshed because they wanted to add the ToF sensor so developers could start working on AR stuff for the iPhone 12 release. It didn't improve anything CPU wise other than the 8th GPU core being enabled.
 

Eug

Lifer
Mar 11, 2000
23,583
996
126
Under synthetic loads M1 can go up to 32W during combined stres with CPU @ 21W and GPU @ 10W, according to Andrei from Anandtech.

I don't get it. Is he saying that 21 Watts is max core power usage + max 10 Watts GPU = 32 Watts TDP?

That wouldn't be TDP though. That would be significantly higher than TDP unless 32 Watts isn't the maximum.
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,629
136
I don't get it. Is he saying that 21 Watts is max core power usage + max 10 Watts GPU = 32 Watts TDP?

That wouldn't be TDP though. That would be significantly higher than TDP unless 32 Watts isn't the maximum.

I think he's just saying that the full SOC is allowed to use up to 32 Watts if you stress both the CPU and GPU at the same time, there's no limiter preventing it from using that much power (unless you hit thermal throttling). I wonder if you could ever get that high (or close to it) with a real load though. It would need to be a game that can max out at least all 4 big cores and the GPU. Either that or have a game running that maxes out the GPU and doing something in the background that also used a lot of CPU. Would be an interesting test but probably more of an academic type experiment for these devices and the types of systems they are in.
 
  • Like
Reactions: Tlh97 and coercitiv

Eug

Lifer
Mar 11, 2000
23,583
996
126
I didn't check either, but it doesn't matter to me as it still shows a great level of finish on both OS and hardware side. CPU performance is best outlined in more intensive tests anyway.
Well, here is my lowly i5-7600 iMac with cached applications. I skipped the applications that had private information but launched the rest of the ones that were in the dock. The lag you see is due to the video recording. The launches themselves were actually quick, but the computer couldn't keep up with the frame rates for the screen recording.

I guess one big advantage of M1 besides overall speed is the fact it has enough cores including low power ones to keep the screen recording smooth while doing all these app loads.

 
  • Like
Reactions: Tlh97 and coercitiv

name99

Senior member
Sep 11, 2010
404
303
136
Why do you care why I care? If you don't have the answer, that's fine, you don't have to reply to my post.

I'm not trying to be sarcastic; I'm tryin to make you think through why you are asking these questions. If you insist on treating these machines as PCs, that you can "rank" by using PC-style numbers, you're going to be frustrated endlessly. You can continue banging your head against a brick wall, or you can listen to those of us with a lot of Apple experience.

The expansion capabilities are described by the two IO ports. That's it. If you explained why you need this info beyond knowing the IO capabilities are, I might be able to help further.
 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
I guess I should have been clearer in my question. How many PCIe lanes, how much USB, SATA, etc.? I didn't see that in the Anandtech article at least.
My impression so far is that M1 albeit expanded is still a mobile SoC at its heart with comparatively limited I/O.

Thunderbolt3/USB4 is 4x PCIe 3 lanes, all M1 Macs have two of them. The internal SSD likely is also 4x PCIe 3 at most. No SATA. The two USB A connections on the Mac mini are likely negligible.

But those are not necessarily adding to the power usage as long as they are not in use (one needs to test how much using both Thunderbolt connections at max bandwidth would add). The biggest power usage drivers are RAM/IMC and cache (which on M1 are still optimized for mobile), everything else of the uncore is comparatively easy to power down and gate off.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,670
1,250
136
M1 CPU Performance -> About what I expected.
M1 CPU Perf/W -> A bit lower than I expected.
M1 GPU Perf & Perf/W -> Substantially exceeded my expectations.
 
Apr 30, 2020
68
170
76
M1 CPU Performance -> About what I expected.
M1 CPU Perf/W -> A bit lower than I expected.
M1 GPU Perf & Perf/W -> Substantially exceeded my expectations.
I pretty much agree fully there. Very impressed with the GPU. I will also say I am extremely impressed with Apple's x86 emulator though, that's a FAST emulator. Clearly a tremendous amount of engineering resources went into that emulator. Clearly a tremendous amount of resources went into the initial software offerings for the Apple ARM platform in general. I do wonder if that same high quality and super-fast code will be the norm, though? Apple obviously had tremendous reason to make sure the debut software was tip-top and super optimized.

Really the only thing that "annoyed" me about the review was the Speedometer benchmark. Which browsers were they using? I feel like the other fair test would have been to use the same browser on every platform.
Why do you care?
There will be a bunch of internal lanes connected to WiFi and the SSD. There will be a bunch of internal USB connections to things like the keyboard/trackpad and the camera.
But none of that matters. What matters is there are two IO ports with the characteristics described.
The context of his post is in reference to the uncore power, not necessarily to the connectivity and I/O. The answer is, the M1 has very little I/O. No PCI-E lanes, No SATA. It's got a single display PHY (well, not counting the thunderbolt monitor muxing), a handful of USB/Thunderbolt PHYs, Ethernet, single audio port, and that's about it. Desktop SOC from Intel and AMD have 4x the I/O going on, which contributes heavily to "uncore" power consumption. Especially the memory controller needing to drive a fat 128-bit bus across a PCB. These things just cannot be overcome while still retaining the characteristics of a typical PC notebook.
 

Attachments

  • 113936160_954401865031178_8457159904871582057_n.png
    113936160_954401865031178_8457159904871582057_n.png
    1 MB · Views: 24

shady28

Platinum Member
Apr 11, 2004
2,520
397
126
For those complaining about Chrome on M1, the native version comes out tomorrow.

Apparently it was supposed to come out today actually, but sh!t happens...


MS Office beta is also out, looks like they were pretty well prepared. Chrome would be a seriously bad miss with average consumers.
 

Eug

Lifer
Mar 11, 2000
23,583
996
126
So, regarding Rosetta 2, I was just watching some reviews and it seems most of the photo apps run pretty well, but Pixelmator Pro doesn't run at all. However, the native version of Pixelmator Pro is coming this week, so no big deal. Lightroom is pretty good, and actually runs better than on many older Intel Macs, but does have some mild lag in some areas.

For video editing, some say Premiere Pro is usable but others say it totally sucks on M1. Davinci Resolve is native and is much smoother, but it's beta and somewhat buggy. Of course, the native Final Cut runs beautifully.

BTW, just about everyone says iOS apps on Mac is a total waste of time. Don't even bother, unless you really, really have to. For example, the HBO Max app works on the Mac, but it loads as a fixed size window and when you press the button to fill the screen, it does nothing.
 

Heartbreaker

Diamond Member
Apr 3, 2006
4,222
5,224
136
BTW, just about everyone says iOS apps on Mac is a total waste of time. Don't even bother, unless you really, really have to. For example, the HBO Max app works on the Mac, but it loads as a fixed size window and when you press the button to fill the screen, it does nothing.

Right from when ARM Macs were announced, iOS apps on Macs never made any sense, it seems like it would be the same kind of garbage experience Apple is famous for avoiding.

I wonder who, high up in Apple thought this was an idea worth pushing.