Discussion Apple Silicon SoC thread

[Eug]

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:



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:

Page 78 - Discussion - Apple Silicon SoC thread

Page 78 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M1 Ultra discussion here:

Page 109 - Discussion - Apple Silicon SoC thread

Page 109 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M2 discussion here:

Page 127 - Discussion - Apple Silicon SoC thread

Page 127 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

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:

Page 215 - Discussion - Apple Silicon SoC thread

Page 215 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[Mopetar]

Alternately, Macs seem to represent ~15% of active laptops/desktops connecting to the internet:

Desktop Operating System Market Share Worldwide | Statcounter Global Stats

This graph shows the market share of desktop operating systems worldwide based on over 5 billion monthly page views.

gs.statcounter.com

10 years ago it was only about 5%, so significant growth.

There is also significant regional split.
Lower in Asia, Africa, South America, and India. (near 10%)
Higher in Oceana, and North America. (near 30%).
Europe in between. (near 20%)

Global PC shipments are down considerably over that span (2019 was the first year of overall market growth in nearly a decade), so it isn't so much that Apple has sold more as it may be that they're holding on better than other manufacturers. One quick result from a Google search shows they have 7.9% of the global PC market, at least for one particular quarter in 2019. I couldn't find a good chart with Mac shipments, but one that showed revenue looked fairly flat over the last decade (or at least the part of it the chart included), but it's hard to tell due to all of the other data being included. If we really wanted to know we could dig through financial reports I suppose since they have to report revenue at the very least.

The regional breakdown just shows that countries that are more well off financially buy more Macs. Not exactly rocket science, but unless Apple is willing to lower their prices (which is unlikely for a variety of reasons) there's always going to be a cap on the amount of hardware that they can sell. From the perspective of maximizing profit, the additional sales they would get from a price cut aren't likely to make up for the lost profit of the higher sale price. Never mind that they would likely need to invest some initial capital to increase their production capacity if they wanted to ramp up significantly.

 

[Heartbreaker]

One quick result from a Google search shows they have 7.9% of the global PC market, at least for one particular

I just wonder at the big mismatch between sales and usage numbers which are about double that. Half the Windows PC's running industrial equipment, and other use cases without internet access?

 

[scannall]

I just wonder at the big mismatch between sales and usage numbers which are about double that. Half the Windows PC's running industrial equipment, and other use cases without internet access?

Lots of PC's are stuck in offices and may not have net access.

 

[teejee]

I just wonder at the big mismatch between sales and usage numbers which are about double that. Half the Windows PC's running industrial equipment, and other use cases without internet access?

I think a normal Mac is used actively for more years than a windows PC (on average), that could explain this.

 

[shady28]

That seems like wishful thinking. How many people are willing to switch OS to get faster performance? You have to 1) NEED faster performance to the point you are willing to pay more to get it (so people who are just browsing the net and reading email are out) and 2) what you want/need to run is available on macOS (so most gamers and those who run some Windows software that doesn't have an identical version available on Mac are out) and 3) willing to potentially re-buy any commercial software they have (if the vendor doesn't offer some sort of free license transfer from Windows to macOS)

Pretty much no one, right? That's exactly the way I see it, and it goes double for businesses. If better hardware = adoption of the platform were true, Apple would not have been bleeding market share in the phone space to Android for the past 3 years.

Also, none of these initial devices can be configured for heavy duty work simply because their max RAM is 16GB. And for the MacBook Air, looks like that is 8GB RAM / 512GB storage for $1249 max config.

In the x86 world, those specs at that price point is a joke. It is good for nothing but facebook and twitter, web browsing and tinkering. If I saw an x86 laptop like that on Amazon, I would automatically assume it was a scammer selling old junk for inflated prices.

I mean you can get direct from Lenovo a Yoga 7i with TGL-U, 16GB RAM, 1TB SSD, 15 hour battery life, Thunderbolt 4, AX wireless, 500 nit 1080p HDR/Dolby Vision touch screen, for $150 less. And if AMD is your thing, they have 4650U versions with 6C/12T and AMD iGPU.

Real buyers are going to be looking at value of the package, not 'Look! The M1 is 2% faster at cinebench single core!'

 

[Heartbreaker]

I mean you can get direct from Lenovo a Yoga 7i with TGL-U, 16GB RAM, 1TB SSD, 15 hour battery life, Thunderbolt 4, AX wireless, 500 nit 1080p HDR/Dolby Vision touch screen, for $150 less. And if AMD is your thing, they have 4650U versions with 6C/12T and AMD iGPU.

Real buyers are going to be looking at value of the package, not 'Look! The M1 is 2% faster at cinebench single core!'

The whole package is what matters, but that isn't the spec sheet, check list that you think it is.

Real buyers are going to look how it blasts through real applications like Final Cut pro, or Xcode Compiles, while remaining ridiculously cool, and quiet, while just sipping battery life. That is the whole package that matters, not who has the biggest spec sheet check list.

 

[dmens]

In the x86 world, those specs at that price point is a joke. It is good for nothing but facebook and twitter, web browsing and tinkering

That is a damning indictment of the x86 ecosystem more than an attack on Apple "specs". Heh.

Here is a free pro-tip: the sole reason Firestorm is not outpacing x86 cores on single thread perf by enormous margins is because Apple does not bother playing stupid PCU games, like with Intel shoving 1.4 volts into a single core when it detects a single thread workload and pray the work is done within the turbo window before it has to throttle down and not blow through the thermal sink capacity.

 

[shady28]

The whole package is what matters, but that isn't the spec sheet, check list that you think it is.

Real buyers are going to look how it blasts through real applications like Final Cut pro, or Xcode Compiles, while remaining ridiculously cool, and quiet, while just sipping battery life. That is the whole package that matters, not who has the biggest spec sheet check list.

RAM > CPU if you are doing video editing and very large image processing, at least to a point. But 8GB ain't that point. 128GB might be.

Go ahead and try to edit a 100GB video on an 8GB machine.

 

[Roland00Address]

like with Intel shoving 1.4 volts into a single core when it detects a single thread workload and pray the work is done within the turbo window before it has to throttle down and not blow through the thermal sink capacity.

Agreed

(gets a little dramatic for a little while.)

1985 Doc: No no no this sucker's electrical, but I need a nuclear reaction to generate the ONE point TWENTY - ONE gigawatts of electricity-
1955 Doc: Doc: What did I just say?
1985 Doc: No no no this sucker's electrical, but I need a nuclear reaction to generate the ONE point TWENTY - ONE gigawatts of electricity that I need.
1955 Doc: ONE point TWENTY - ONE gigawatts. ONE point TWENTY - ONE gigawatts. Great Scott.
Marty: What the hell is a gigawatt?
1955 Doc: How could I have been so careless. ONE point TWENTY - ONE gigawatts. [Looks at a picture of Thomas Edison]
1955 Doc: Tom, how am I gonna generate that kind of power, it can't be done, it can't.

Back to earth, no more fun, time for math 😎


So if I understand the math correctly dynamic power consumption is this formula below

P=c⋅V2⋅f+PS

Note V2 means Voltage squared, aka Voltage * Voltage, and f is frequency.

In 2019 TSMC demonstrated the voltages of a "sample" arm chip at 7nm. Not the apple chip just a generic cortex a72

Voltage	2.8 GHz	3.0 GHz	3.5 GHz	4.0 GHz	4.2 GHz
	0.775 V	0.825 V	0.95 V	1.20 V	1.375 V

If we keep the cpu frequency the same but use half the voltage for a chip that means the dynamic use of power would be 1/4th as much. Thus 1.4 voltage for a chip at the same frequency as a 1.0v chip would use 1.96 the amount of power due to the higher voltage.

That sample arm a72 chip may do 50% more work (4.2 / 2.8 = 1.5) but it is using 4.72x power and thus it uses 3.15x the effective power to do the same task.

So it is a big deal like @dmens that we sometimes see insane voltages on intel chips. Often the extra performance is not worth it if you have to touch the voltage. Server chips and mobile+laptop chips know this but we have tolerated it for so long since the difference between a 65w vs 95w chip is negligible on a large form factor desktop.

If the voltage to ghz curve for Apple M1 silicon is similar to this sample 7nm basic a72 chip there is no reason to run anything over 3.5 ghz for anything over 0.95v is obscene. Even the 0.775v vs 0.95v is 50% more power draw for similar work (if I did the math correctly)

 

[Heartbreaker]

RAM > CPU if you are doing video editing and very large image processing, at least to a point. But 8GB ain't that point. 128GB might be.

Go ahead and try to edit a 100GB video on an 8GB machine.

We have seen multiple examples where the M1 machines are doing a better job video editing than machines with double or even quadruple the RAM.

The whole package doesn't mean the biggest number on a spec sheet. It all the parts working well together.

You seem to want a bigger number on a spec sheet, more than you want a machine that actually performs better on real applications.

 

[dmens]

Thus 1.4 voltage for a chip at the same frequency as a 1.0v chip would use 1.96 the amount of power due to the higher voltage.

The rule-of-thumb squared relation for power to voltage is idealistic and optimistic. The reality is quite a bit worse. Intel CPU turbo to what I like to call the "desperation region", in line with MOSFET terminology.

 

[shady28]

That is a damning indictment of the x86 ecosystem more than an attack on Apple "specs". Heh.

Here is a free pro-tip: the sole reason Firestorm is not outpacing x86 cores on single thread perf by enormous margins is because Apple does not bother playing stupid PCU games, like with Intel shoving 1.4 volts into a single core when it detects a single thread workload and pray the work is done within the turbo window before it has to throttle down and not blow through the thermal sink capacity.

You still don't get it, your head is stuck in the specs of the CPU.

Space matters, 512GB is barely enough these days. You going to do video editing on 100GB videos on that? Oh now you need external thunderbolt drives. Also how is that going to perform with 8GB RAM? All I am doing right now is Firefox and one spreadsheet open and I'm sitting at 7.4GB.

So I open up Corel Videostudio and look, I'm at 7.9-8GB. Open up one teensy 2.5GB 4K video and look, I'm at 8.5GB and I haven't even started doing anything.

So yes, that spec sheet matters, even and especially from a practical point of view. The CPU is not as important as you think.

 

[dmens]

You still don't get it, your head is stuck in the specs of the CPU.

Space matters, 512GB is barely enough these days. You going to do video editing on 100GB videos on that? Oh now you need external thunderbolt drives. Also how is that going to perform with 8GB RAM? All I am doing right now is Firefox and one spreadsheet open and I'm sitting at 7.4GB.

So I open up Corel Videostudio and look, I'm at 7.9-8GB. Open up one teensy 2.5GB 4K video and look, I'm at 8.5GB and I haven't even started doing anything.

So yes, that spec sheet matters, even and especially from a practical point of view. The CPU is not as important as you think.

Oh noes, there isn't enough space! I do enjoy seeing someone grasp at absolutely meaningless trivialities trying to argue a meaningless point. Yours is like trying to argue the Tesla 3 is a bad car because the Volkswagen IDR exists.

And like I said, the fact that Wintel is RAM hungry is an indictment of Wintel, nothing more.

Also are you trying to simultaneously argue that the spec sheet matters, but:

Real buyers are going to be looking at value of the package, not 'Look! The M1 is 2% faster at cinebench single core!'

Nice double standard there. Actually the real margin is quite a bit higher than 2%, but whatever.

 

[shady28]

We have seen multiple examples where the M1 machines are doing a better job video editing than machines with double or even quadruple the RAM.

The whole package doesn't mean the biggest number on a spec sheet. It all the parts working well together.

You seem to want a bigger number on a spec sheet, more than you want a machine that actually performs better on real applications.

Seems to me like you don't know what you're talking about.

See this video, go to 8:00 to see what I'm talking about. He can't even scrub through the video because it's out of memory. I can do that way smoother than him on my 7700HQ with 32GB. *WAY* smoother.

Edit: From the video "Checking CPU, CPU was fine this entire time, but the memory RAM is really suffering.."

 

[Roland00Address]

The rule-of-thumb squared relation for power to voltage is idealistic and optimistic. The reality is quite a bit worse. Intel CPU turbo to what I like to call the "desperation region", in line with MOSFET terminology.

Thanks for that knowledge 🤣😭

 

[amrnuke]

Where are they "so far behind"? Unless you are comparing them against PCs with 8 big cores (and probably SMT enabled as well) they aren't far behind.

And the "massacre" would come when new generations of the M* come with more big cores, eventually scaling up to the Mac Pro in a couple years. I expect at least 32 big cores in the high end there, maybe more. You're going to have to be comparing with some awfully big (and expensive) x86 hardware to put that "far behind".

I'm not sure what he's getting at.

Compare GB5 ST -> MT scaling and CB23 ST -> MT scaling.

Let's assume worst case scenario - M1 is a full 8-core chip, and you classify the 5950X as a 16-core chip. So we're calling Icestorm a full core, and ignoring that the Zen3 chip uses SMT. The M1 gets 923.25 GB5 MT score per core (Firestorm + Icestorm), while the 5950X gets 1003.438 per core (and while SMT is turned on, we still just call it a 16-core chip). In CB23, the score is 1790 MT score per core for 5950X, and 979 MT score per core for M1.

If Apple decide to go 8+4 for an M1X, and scaling proceeds along the same path as Zen3 scaling as cores go up on GB5 and CB23 MT scoring (losing 40.42 GB5 MT score per core added, and 15.49 CB23 MT score per core added) then an 8+4 as a "12 core" chip should score:
GB5: 7386 + (7386/8 * 4) - (40.423 * 4) = 10917.3
CB23: 7833 + (7833/8 * 4) - (15.49 * 4) = 11687.54

In other words, an 8+4 M1X would be expected to perform a little better than a 5800X in GB5 MT and a little better than a 5600X in CB23 MT. And that's assuming that the 4 Icestorm cores are equally efficacious as the 4 Firestorm cores, which they're not, so we'd actually expect higher scores, but this is "worst case".

Power consumption figures should reveal that the M1 does so with better efficiency as well.

In other words, depending on task, at worst, we could expect the 8+4 either acts as good as an 8c Zen3 chip or a 12t Zen3 chip. Either way it appears that the M1 at 8+4 could be roughly equivalent to a 7-core Zen3 chip with SMT on, as a baseline. On mobile. Sipping power.

That means if Apple can (and that's the big question - with the Neural Engine and ML and fixed on-package RAM and the massive L2$ size, can this be scaled up easily?) put together a 16+8 M1XXX FTW GT3 Black Edition, we'd expect it to perform like a 14-core Zen3 chip with SMT on, so somewhere between a 5900X and 5950X, and closer to the 5950X.

And that's remarkable regardless of who's making the chip. But what will be even more remarkable is if they can do that at the same power draw per core, which I don't doubt.

 

[insertcarehere]

Seems to me like you don't know what you're talking about.

See this video, go to 8:00 to see what I'm talking about. He can't even scrub through the video because it's out of memory. I can do that way smoother than him on my 7700HQ with 32GB. *WAY* smoother.

Edit: From the video "Checking CPU, CPU was fine this entire time, but the memory RAM is really suffering.."

That would matter if 16GB M1 MBAs don't exist, but they do and according to people who do this for a living, the RAM seems to hold up just fine doing video editing. It's not like Windows Ultrabooks with 32gb RAM can be had for reasonable prices anyway.

 

[amrnuke]

That would matter if 16GB M1 MBAs don't exist, but they do and according to people who do this for a living, the RAM seems to hold up just fine doing video editing. It's not like Windows Ultrabooks with 32gb RAM can be had for reasonable prices anyway.

A ThinkPad P14s with a 4750U, 32GB 3200 RAM, 1TB SSD is $1565.62. The MBA with 16GB RAM and 1TB SSD is $1599. So, I mean, yeah, you can get a Windows Ultrabook with 32GB RAM for cheaper than a lower-spec MBA. But that's irrelevant if the MBA can edit video faster than the ThinkPad, and I think that's possibly the case, even for larger files, and especially if you find Final Cut to be better than Resolve or Premiere for your situation.

 

[mikegg]

That would matter if 16GB M1 MBAs don't exist, but they do and according to people who do this for a living, the RAM seems to hold up just fine doing video editing. It's not like Windows Ultrabooks with 32gb RAM can be had for reasonable prices anyway.

Yes, reviewers seem to agree that the MBA M1 is an amazing portable video editing machine.

Did people not see these 2 posts earlier?

M1 embarrassed a desktop Intel 8700 equivalent and trades blows when a 5700XT eGPU is attached to the Intel computer. How can anyone possibly think that they can get a better video editing laptop at $1000? Most Windows $1000 laptops still come with crappy screens with inaccurate colors.

Page 48 - Discussion - Apple Silicon SoC thread

Page 48 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

Page 48 - Discussion - Apple Silicon SoC thread

Page 48 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

Better yet, watch this:

 

[mikegg]

A ThinkPad P14s with a 4750U, 32GB 3200 RAM, 1TB SSD is $1565.62. The MBA with 16GB RAM and 1TB SSD is $1599. So, I mean, yeah, you can get a Windows Ultrabook with 32GB RAM for cheaper than a lower-spec MBA. But that's irrelevant if the MBA can edit video faster than the ThinkPad, and I think that's possibly the case, even for larger files, and especially if you find Final Cut to be better than Resolve or Premiere for your situation.

It's not just that the MBA is significantly faster than a desktop Intel 8700 equivalent at video editing and trades blows with the Intel computer when it's attached with an 5700XT eGPU. It's also the fact that the M1 has built-in accelerators that make editing certain high-end video files possible without major stutters experienced by Intel/AMD processors.

But let's be honest, no one is editing the next blockbuster movie on a Macbook Air. Most people buying these machines are just editing for Youtube and Instagram and the Macbook Air M1 is probably overkill for those workloads.

If you're a real professional who makes a living editing videos, you'd want to wait for the Macbook Pro 16", iMac, or even the Mac Pro.

 

[mikegg]

That seems like wishful thinking. How many people are willing to switch OS to get faster performance? You have to 1) NEED faster performance to the point you are willing to pay more to get it (so people who are just browsing the net and reading email are out) and 2) what you want/need to run is available on macOS (so most gamers and those who run some Windows software that doesn't have an identical version available on Mac are out) and 3) willing to potentially re-buy any commercial software they have (if the vendor doesn't offer some sort of free license transfer from Windows to macOS)

I agree, 50% within 5 years is an extremely bold claim. Maybe within 10 years.

This comes with the caveat that Apple will make an iPhone SE/Watch SE Macbook equivalent in the future, which I predict will happen. Basically a $700 Macbook 12" or Macbook Air.

I do want to address your numbered points.

1. Apple Silicon is much more than better performance. It's battery life, quiet and cool, machine learning, video editing and enhancement, image enhancement. All of these things Apple Silicon does significantly better than Windows laptops. We don't know what Apple will do for iMacs and Mac Pros but we can assume that it will be impressive. Regardless, most of the volume is on laptops and Apple's efficient SoC reigns supreme here.

2. PC gamers can belong to the other 50% who won't get a Mac. However, I do expect Macs to become a bigger player in AAA games in the future due to this ARM transition. I can explain more if you need me to.

3. Actually, this point is in favor of Apple. 50% of the U.S. market uses iPhones. 65% uses iPads. They can reuse the software they bought on those devices on their Mac. They can use their Apple One subscription on their laptops. For those who own thousands of dollars of Windows software, maybe they're stuck on Windows. Apple just needs to recapture their own customers.

For some reason, people here seem to think making the transition from Windows to Mac is a big jump. It could be for people on this forum, but not normal people. A huge portion of the U.S. population already uses Apple devices and they're already in the Apple ecosystem. And software is written for browsers nowadays and many of them are written with cross-compatibility in mind, ala Electron framework. It's less of a jump than you think.

I myself have a Windows desktop (I don't really use anymore), and a Macbook Pro for work, iPhone XR, and an iPad Pro. The only software that I run on Windows that I can't run on my Macbook Pro is games. Everything else I use is on both platforms or on the browser.

 

[amrnuke]

It's not just that the MBA is significantly faster than a desktop Intel 8700 equivalent at video editing and trades blows with the Intel computer when it's attached with an 5700XT eGPU. It's also the fact that the M1 has built-in accelerators that make editing certain high-end video files possible without major stutters experienced by Intel/AMD processors.

But let's be honest, no one is editing the next blockbuster movie on a Macbook Air. Most people buying these machines are just editing for Youtube and Instagram and the Macbook Air M1 is probably overkill for those workloads.

If you're a real professional who makes a living editing videos, you'd want to wait for the Macbook Pro 16", iMac, or even the Mac Pro.

If the MBA/MBP13 isn't for the "real professional who makes a living editing videos" but is overkill for Youtube and Instagram editing... why have you linked this video of all videos.

 

[jeanlain]

2. PC gamers can belong to the other 50% who won't get a Mac. However, I do expect Macs to become a bigger player in AAA games in the future due to this ARM transition. I can explain more if you need me to.

Please do. I've been discussing this topic a lot in other fora. My conclusion, which is shared by many others, is that the Mac hardware will not make a sizeable difference. Even doubling the user base won't be enough to entice game studios. Apple doesn't care about AAA gaming, and AAA game studios don't care about the Mac. The latter is not going to change all by itself. And there's not evidence that Apple is changing. We've seem some games mentioned at the last Apple event, but that's nothing more than usual.

 

[Mopetar]

I just wonder at the big mismatch between sales and usage numbers which are about double that. Half the Windows PC's running industrial equipment, and other use cases without internet access?

Without knowing how that site gets their results I honestly can't tell you, but a few others did point out valid reasons for Macs to be overrepresented in that metric. The first being office machines with no internet access, but the second (and what I consider a far bigger factor) being that Apple hardware likely has a longer life span so Mac users don't replace it as often.

I only upgraded to a new iPhone after 5 years and was considering holding out even longer. While I was in the store a woman next to me was upgrading from a 5C and wanted to get an XR. I'm also still using an iPad Air 2 that I got when it first came out and don't intend to upgrade that anytime soon. My brother had one of the earliest quad-core MacBook Pros that he kept for over 8 years and only replaced this summer. My mother had a MacBook Air the she had for 6-7 years.

How long do you think any of the ~$400 Windows computers that every company sells tend to last? Dell has a 15.6" (1334x768) 2 core Celeron (1.8 GHz) with 4 GB of RAM and a 128 GB SSD for $320. Coincidentally the specs on this actually line up nicely with my five year old iPhone 6S which had 2 ARM cores at 1.86 GHz, a 1334x750 pixel display, and 128 GB of solid state memory. It only had 2 GB of RAM though, but I suspect after accounting for operating system usage, the amount left for the user would have been similar.

Quite frankly, I wouldn't have accepted a laptop like that 5 years ago and if I had bought one with those specs I very much doubt it would have lasted me 5 years. The worst desktop I can get from Dell is a quad core i3 with a spinning hard disk and 4 GB of RAM. It's $380. That isn't lasting me 5 years either.

Not to many people here would buy anything like that from Dell or anyone else, but there are a few posters that build low-end PCs that they either give to family/friends or try to sling as a way to making a bit of money. There's a 6-year old AT article about building a budget PC that puts together a PC that costs $400 that's in some ways better than the $380 Dell model since it has 8 GB of RAM and includes an SSD for the OS. The RAM and SSD prices have probably come down enough to allow for an entry-level APU quad core APU.

However, the average price needs to account for the expensive high end computers that are being sold and if there are enough of those and at a significant price difference, it can distort the average. If you had a market where 9 people bought a $600 computer and then add 1 person who buys a $1,200 computer, the average sales price goes up by 10% (to $660) just because one person bought a computer that was twice as expensive as the old average. However, the more interesting effect is when you keep the average sales price fixed at $600. In that situation those other 9 individuals would have to have bought $533.33 computers to make it average out. One person essentially eats $66.66 of value away from each of the 9 on the other side of the mean.

Now that's a simplistic calculation for the point of illustration, but it's not too far off of reality. I found two stats from Statista (full details require membership) that put the average personal computer (I believe this includes laptops) sales price in 2019 at $632 and the average desktop sales price in 2015 at $544. That alone doesn't tell us much, but we can make a reasonable estimate about the average sales price for Apple's computers.

If we use the 7.9% market share from the previous article I linked that comes to about 5.388 million Macs sold in that quarter based on the estimated global PC shipments. Apple's financials for that quarter (note that for their company the fiscal fourth quarter corresponds to the third quarter of the actual year) report the Mac division revenue as $6.991 billion dollars. That's an average sales price of roughly $1,300 for Apple's computers. For convenience let's assume 2.1% of the market are PC sales that have a similar average sales price. This means with 10% of the market at $1,300 ASP, the other 90% needs to buy $557.77 computers to make the average work out to $632.

Of course the bottom 90% of the market doesn't buy just $550 computers. There's some number that are buying the $800 computers that Dell and HP sell. If that's 10% of the market, the remaining 80% needs to be buying $527 computers. If it were instead 20%, the remaining 70% has to buy $488 computers to balance it all out. The exact mix of sales isn't important, but we could estimate that. However, the $500 Dell computers aren't all that much better either and still include a spinning disk. Backblaze put out some information on their website which indicated that 22% of their drives failed within the first four years. While it's true they're probably using the drives more, they probably don't get quite the same level of awful that goes into these OEM machines either.

Suppose we have a hypothetical spread (which actually winds up looking close to a normal distribution which the actual market likely resembles for reasons I won't get into, so this is a decent approximation) where the top 10% buys at $1,300, the next 10% at $900, 20% at $700, another 20% at $600, and then 20% at $450. That still leaves 20% of the market that needs to buy for an average of $310 to balance the numbers. That's below the minimum you pay on Dell's website, but that part of the market exists and may be much larger than just 20%. Some people aren't just buying the current garbage, but last (or perhaps the one before that) year's bottom basement closeout deals that retailers want to get off the shelves.

Walmart has some utterly ancient hardware (based on what's there I suspect a Core 2 Duo is the minimum CPU you can get to run Windows 10) on their website (though a lot of it from third parties selling through Walmart) for $200 or less. There's one that's being sold refurbished for $88 which includes a 13 year old Conroe CPU. The product description lists one of the key features and benefits as being able to store 22,000 songs.

I don't know how much of that is original parts or cobbled together from dozens of salvaged boxes considering how hard it would be to buy some of the components. All but one of the 8GB DDR2 kits on Newegg run more than the $88 for the computer. Frankly though, it doesn't matter if the HDD or whatever shoddy power supply is in that thing craps the bed after only two years, or if the amount of malware (pre-installed or otherwise) has made it unusable before then. Just throw it out and get a new one, or well a new ancient pile of junk at any rate.

The cheapest thing I could find on Walmart's website that's actually sold by Walmart and not just through them is a $169.00 laptop with a 4-core Atom CPU (coincidentally one of the last made for consumer products) and enough flash storage to hold the operating system and quite possibly five high scores in Solitaire. It's got 4 GB of RAM, but the Intel spec sheet says it's going to be a single stick of DDR3-1600. I don't know what kind of battery it has and the only information I can find just says 9000 mAh without any mention of the chemistry. I'm thinking lead-acid.

I can't imagine trying to use it and it took some digging to even find relevant benchmarks for it and something to compare it to. Notebookcheck says it gets 31/101 points in Cinebench R15 single / multicore. AT's CPU Bench only has a single entry even close to that low which is a Conroe chip (curiously one just slightly older and .2 GHz slower than the one in the $88 Walmart special) that gets 54/102 points in the respective benchmarks. I would imagine that these fantastic kinds of system hang for at least 15 seconds out of every minute of use.

If you were to use one of these systems for two hours a day and you earn federal minimum wage it only takes four months before the time one of the computers has taken off of your life waiting on it to do something could have been spent working to buy the $400 budget PC from 6 years ago described above that can get 110 points in the Cinebench R15 single thread test! There's a more modern take on a $400 build that includes a Ryzen R3 1200 (135/482 Cinebench R15 scores) and GTX 1030. Obviously if we dropped the gaming focus (and discrete graphics card) and swapped in an APU it'd be easy to get an SSD for the OS.

I've wasted far too much time on this post and researching all of the various parts of it, but it was certainly a fascinating rabbit hole to go down.

TLDR: There are a lot of trash heaps being sold in the Windows PC market. They probably get replaced every 2 years with another steaming turd out of necessity. The people who buy such awful machines may repeat doing so 2 - 4 times before a Mac user gets a new machine.

 

[mikegg]

TLDR: There are a lot of trash heaps being sold in the Windows PC market. They probably get replaced every 2 years with another steaming turd out of necessity. The people who buy such awful machines may repeat doing so 2 - 4 times before a Mac user gets a new machine.

I didn’t read your long post but I read this and I came to the same conclusion. My primary Mac is 5 years old and is still going strong. I can resell it for $700-800 even today. $400-700 PC laptops are pure garbage in terms of longevity.