Apple Silicon is Overpriced and over valued in technology market

[bobcomer]

That does seem to be the case!

Yep! I definitely agree you need stronger ST performance.

I'm an IT Manager and I always have a lot of things going on, terminal sessions to our big server, remote desktops to other servers, web pages, an excel or two for quick calculations mainly, lots of monitoring stuff, VNC sessions. My main processing type tasks are done on the servers actually. You're focused, and I have to track a lot of different things. Choice of hardware to do it is good.

 

[theorist9]

Yep! I definitely agree you need stronger ST performance.

I'm an IT Manager and I always have a lot of things going on, terminal sessions to our big server, remote desktops to other servers, web pages, an excel or two for quick calculations mainly, lots of monitoring stuff, VNC sessions. My main processing type tasks are done on the servers actually. You're focused, and I have to track a lot of different things. Choice of hardware to do it is good.

The challenge for me is that it's easy to get more cores (well, easy if you're willing to spend the money). But faster ST speeds have to wait for technology to advance, and I suspect it's going to be a while before AS is enough faster than my iMac for it to make a practical difference. For instance, here's a benchmark I created for myself using symbolic calculations in Mathematica. I ran it on my i9 iMac, then asked someone I knew from another site to run it on their 16" M1 Max MBP. All these calculations are ST. The average improvement for each suite of tests was only 2% to 19%.

Now it's possible there are more optimizations Wolfram could make for Mathematica to run faster on AS—the first native build of Mathematica for AS was 12.3.1 (released July 2021), and this comparison was made with 13.1 (released June 2022), so they've only had a year to tweak it, but still...

 

[ahurst]

The challenge for me is that it's easy to get more cores (well, easy if you're willing to spend the money). But faster ST speeds have to wait for technology to advance, and I suspect it's going to be a while before AS is enough faster than my iMac for it to make a practical difference. For instance, here's a benchmark I created for myself using symbolic calculations in Mathematica. I ran it on my i9 iMac, then asked someone I knew from another site to run it on their 16" M1 Max MBP. All these calculations are ST. The average improvement for each suite of tests was only 2% to 19%.

Now it's possible there are more optimizations Wolfram could make for Mathematica to run faster on AS—the first native build of Mathematica for AS was 12.3.1 (released July 2021), and this comparison was made with 13.1 (released June 2022), so they've only had a year to tweak it, but still...
View attachment 2100014

In R and Numpy, you can get massive improvements in some mathematical operations (but not others) on Apple Silicon when the software is made to use Apple’s Accelerate.framework, which leverages the built-in matrix math hardware on the M1/M2 when possible. Never used Mathematica, but it sounds like the kind of thing that could benefit from that kind of optimization!

And I’m in the same boat with single-core performance: when I’m running a heavy number-crunching script (signal processing, statistical modeling) I’ll often have multiple cores at 99% for minutes to hours (depending on the amount of data and model complexity). Adding cores is always nice for multitasking, but I see the biggest gains from improved ST.

 

[theorist9]

In R and Numpy, you can get massive improvements in some mathematical operations (but not others) on Apple Silicon when the software is made to use Apple’s Accelerate.framework, which leverages the built-in matrix math hardware on the M1/M2 when possible.

The last I read about using Accelerate with R was this statement from CRAN. Is this still the official stance—which makes it sound like Accelerate hasn't yet been fully validated on R—or is this posting out of date?:

R for macOS FAQ

R for macOS FAQ

cran.r-project.org

Never used Mathematica, but it sounds like the kind of thing that could benefit from that kind of optimization!

I don't know how Mathematica operates internally, and thus don't know whether or not Accelerate would help the operations in my benchmark, which were entirely symbolic tasks (integrations, equation solving, expression simplification, etc.).

One of the other issues faced by Mathematica on AS its performance in numerical tasks. There the problem has been that the numerical libraries available on Intel (Intel MKL) are very fast, and what has been available on ARM hasn't been as performant--thus some numerical tasks can actually be slower on AS than on older Intel machines. Would Accelerate help with that?

Interestingly, that's also an issue for Mathematica on AMD. The numerical libraries available on AMD are significantly slower than MKL. You used to be able to spoof MKL into thinking it was running on an AMD chip, but Intel killed that workaround.

 

[falainber]

Seriously? System responsiveness on Mac OS is dominated by single-threaded tasks which spin up the big cores when needed.

System resposiveness has been fine on all OSes/systems for many years/decades now. Even Windows 98 was very responsive. CPU performance becomes critical only when one runs the apps that do heavy processing.

 

[senttoschool]

System resposiveness has been fine on all OSes/systems for many years/decades now. Even Windows 98 was very responsive. CPU performance becomes critical only when one runs the apps that do heavy processing.

Please stop.

There is a threshold for multi-core performance. Once reached, multi-core does not matter. Even the base Apple Silicon reaches this threshold comfortably for 90% of users.

Single-core performance has a more significant impact on most users. Every improvement in single core speed is an improvement in responsiveness for all applications.

 

[falainber]

Please stop.

There is a threshold for multi-core performance. Once reached, multi-core does not matter. Even the base Apple Silicon reaches this threshold comfortably for 90% of users.

Single-core performance has a more significant impact on most users. Every improvement in single core speed is an improvement in responsiveness for all applications.

You are talking about smartphones (of several years ago) not computers. For computer OS, UI does not need a lot of computational power period. It will depend more on other factors such as RAM size/speed and, especially, SSD speed. The threshold you are talking about does not exist. It's true that some apps only use limited number of threads but those are usually not the most computationally challenging one (like games, which nobody plays on Macs anyways). Most heavily computational apps can use as many cores/threads as they can get.

 

[throAU]

You are talking about smartphones (of several years ago) not computers. For computer OS, UI does not need a lot of computational power period. It will depend more on other factors such as RAM size/speed and, especially, SSD speed. The threshold you are talking about does not exist. It's true that some apps only use limited number of threads but those are usually not the most computationally challenging one (like games, which nobody plays on Macs anyways). Most heavily computational apps can use as many cores/threads as they can get.

It's a good thing the M series chips are good at both:

• High core count in low power envelope
• high IPC

As far as thread scheduling goes, an app can hint to the OS whether it is a background task or a high priority interactive task (via grand central dispatch which Apple pushed out like 10 years ago) and get scheduled appropriately. Additionally, the OS will start scheduling things onto P cores if the E cores are busy.

I've noticed zero response issues with my m1 pro, despite most stuff running on the efficiency cores most of the time. If needed, the transition to performance cores is seamless.

Even running virtual machines, etc. in the background - you forget they're there. Do anything like that on an intel MacBook (either my 15" or 13" Pros from previous purchase, or the 2020 Air I have palmed off to the GF) and the fan ramps like crazy.

 

[senttoschool]

You are talking about smartphones (of several years ago) not computers. For computer OS, UI does not need a lot of computational power period. It will depend more on other factors such as RAM size/speed and, especially, SSD speed. The threshold you are talking about does not exist. It's true that some apps only use limited number of threads but those are usually not the most computationally challenging one (like games, which nobody plays on Macs anyways). Most heavily computational apps can use as many cores/threads as they can get.

Speaking as a software engineer, every improvement in ST will improve the responsive of the application in all areas. It's not just the UI. It's everything.

MT will only improve the performance if the threshold is not met, which the base Apple Silicon should meet for the vast majority of use cases.

Since you keep insisting that MT matters more than ST in modern CPUs, why don't you share with us a few of the applications where MT is bottlenecked for your usage?

 

[leman]

You are talking about smartphones (of several years ago) not computers. For computer OS, UI does not need a lot of computational power period. It will depend more on other factors such as RAM size/speed and, especially, SSD speed. The threshold you are talking about does not exist. It's true that some apps only use limited number of threads but those are usually not the most computationally challenging one (like games, which nobody plays on Macs anyways). Most heavily computational apps can use as many cores/threads as they can get.

This is an example of discussion where the term "performance" itself becomes limited. I think it's much more useful to distinguish between latency and throughput. Latency-sensitive operations are there might be not so much work but you want to see the result quickly, throughput-sensitive operations is where you want to have a lot of work done quickly, but you don't care that much when the first result comes in. These are two different optimisation points. Consumer hardware often optimises for latency (quick bursts of activity) with fast power changes and high boost clocks while server hardware often optimises for throughout (many slower clocked cores).

It seems to me that you talk about workloads that rely on throughput. But UI is a prime example of latency-sensitive work. When a user does something, you want instantly to put your core into maximal performance state and burst through, so that the response is provided as soon as possible. High single threaded performance is absolutely vital here, but it's only pat of the equation. The other part is the ability to do very fast power state changes. Luckily, Apple Silicon provides both.

 

[yellowhelicopter]

Well, I've been playing Shadow of the Tomb Raider on Mac Mini M1, and I would say the graphics and speed are about the same as on PS4. But... it's small, light, almost totally silent, just slightly warm, and runs x86 code via emulator in this case. If it's not a revolutionary processor, I don't know what could be called so.

 

[leman]

Well, I've been playing Shadow of the Tomb Raider on Mac Mini M1, and I would say the graphics and speed are about the same as on PS4. But... it's small, light, almost totally silent, just slightly warm, and runs x86 code via emulator in this case. If it's not a revolutionary processor, I don't know what could be called so.

Nominally M1 is about twice as fast as the PS4 in the GPU department, and could be somewhat faster still after optimisations. But you know...

 

[cp1160]

It is overpriced in comparison to past and previous counterparts in todays market

What we are seeing is Apple taking over the market making prices with SoC's

Everything SoC is no where to be found we will now see minor improvements/designs/context with what little has changed now that SoC is in the market

ps I have been optimistic in previous posts I am speaking from a person point of view on Apple's game

Cite these views, and provide comparative evidence of Apple vs. AMD vs. Intel vs. ARM and others. It is ok to say I think it is overpriced because you want prices lower, but these are non-compelling evidentiary arguments.

I know I am way late to this party....It just bothered me, and I came back to it.

 

[MauiPa]

Cite these views, and provide comparative evidence of Apple vs. AMD vs. Intel vs. ARM and others. It is ok to say I think it is overpriced because you want prices lower, but these are non-compelling evidentiary arguments.

I know I am way late to this party....It just bothered me, and I came back to it.

I would have to agree with you, when I look up competing products with Intel and AMD chips with a comparable build quality, display quality, SSD speed (just forget Dell then), and performance, the other laptops are usually much higher in cost, much heavier, throttle a lot or have big honking fans (I said heavier) and lets not even get into battery life, they are mostly a joke. So Apple silicon is overpriced? Not even close.

Now I did just read about the Samsung Chromebook (Chromebook seriously? wonder if people can install Windows, Ok maybe ChromeOS has gotten better since I last used it, I have no idea). It did seem reasonable priced and had close to the performance of the new M2 MacBook Pros, and 17 hours battery life (LOL before the disclaimer - your usage may vary, you may get only 4 hours when actually using it, or maybe 2 - I guess we will actually have to see them in use). If true, there would be a new competitive standard for others to follow.

 

[falainber]

I would have to agree with you, when I look up competing products with Intel and AMD chips with a comparable build quality, display quality, SSD speed (just forget Dell then), and performance, the other laptops are usually much higher in cost, much heavier, throttle a lot or have big honking fans (I said heavier) and lets not even get into battery life, they are mostly a joke. So Apple silicon is overpriced? Not even close.

Now I did just read about the Samsung Chromebook (Chromebook seriously? wonder if people can install Windows, Ok maybe ChromeOS has gotten better since I last used it, I have no idea). It did seem reasonable priced and had close to the performance of the new M2 MacBook Pros, and 17 hours battery life (LOL before the disclaimer - your usage may vary, you may get only 4 hours when actually using it, or maybe 2 - I guess we will actually have to see them in use). If true, there would be a new competitive standard for others to follow.

Samsung sells great Windows laptops too. Here is the new model: Galaxy Book 3 Ultra (https://www.theverge.com/23580552/samsung-galaxy-book-3-ultra-laptop-price-hands-on). This one has OLED screen and might be faster than M2 MBP (definitely faster in graphics). It's also thinner and lighter than Apple counterpart.

 

[sam_dean]

Cite these views, and provide comparative evidence of Apple vs. AMD vs. Intel vs. ARM and others. It is ok to say I think it is overpriced because you want prices lower, but these are non-compelling evidentiary arguments.

I know I am way late to this party....It just bothered me, and I came back to it.

I'm not bothered by the MSRPs of each binned SoC.

8GB RAM & 256GB SSD for base models started in 2012. It's time for an upgrade 11 years later.

Below is the specs of standard Mac SKUs. I doubled the RAM & SSD but kept the same price, SoC, CPU cores & GPU cores.

Mac model	MSRP	Chip	CPU (Core)	GPU (Core)	RAM (GB)	SSD (TB)
iMac 24"	$1,699	M1	8	8	16	1
iMac 24"	$1,499	M1	8	8	16	0.5
iMac 24"	$1,299	M1	8	7	16	0.5
Mac mini*	$1,299	M2 Pro	10	16	32	1
Mac mini	$799	M2	8	10	16	1
Mac mini	$599	M2	8	10	16	0.5
Mac Studio	$3,999	M1 Ultra	20	48	128	2
Mac Studio	$1,999	M1 Max	10	24	64	1
Mac Studio**	$3,999	M2 Ultra	24	60	128	2
Mac Studio**	$1,999	M2 Max	12	30	64	1
MBA	$1,499	M2	8	10	16	1
MBA	$1,199	M2	8	8	16	0.5
MBA	$999	M1	8	7	16	0.5
MBP 13"	$1,499	M2	8	10	16	1
MBP 13"	$1,299	M2	8	10	16	0.5
MBP 14"	$3,099	M2 Max	12	30	64	2
MBP 14"	$2,499	M2 Pro	12	19	32	2
MBP 14"	$1,999	M2 Pro	10	16	32	1
MBP 16"	$3,499	M2 Max	12	38	64	2
MBP 16"	$2,699	M2 Pro	12	19	32	2
MBP 16"*	$2,499	M2 Pro	12	19	32	1
MB 12"***	$699	A16 Bionic	6	5	8	256GB
Mac nano***	$299	A16 Bionic	6	5	8	256GB

Note:

*If RAM & SSD were increased my choice would be the $1299 Mac mini M2 Pro if there was no iMac 27" replacement & the $2499 MBP 16" M2 Pro. Both of which would have 32GB RAM & 1TB SSD.

**My guess on the CPU core & GPU core count of the future 2023 Mac Studio M2 Max & M2 Ultra SKUs.

***iPhone chip-based Mac. If M1 & M2 can be used in an iPad Pro & iPad Air why not use iPhone chip in a cheap laptop & desktop? "Mac nano" uses the 2022 Apple TV 4K enclosure as to make it 0.27L instead of 1.39L of the Mac mini. This reduces shipping cost as you can pack in more "Mac nano" per shipping pallet.

 

[tevion5]

Samsung sells great Windows laptops too. Here is the new model: Galaxy Book 3 Ultra (https://www.theverge.com/23580552/samsung-galaxy-book-3-ultra-laptop-price-hands-on). This one has OLED screen and might be faster than M2 MBP (definitely faster in graphics). It's also thinner and lighter than Apple counterpart.

That seems like a very good high end Windows laptops but I'd be surprised if it managed even 3-4 hours of battery life Vs the MBPs 20-22 hours. The screen resolution is also lower, it has much more limited I/O, and unclear that the version with an RTX 4050 outperforms the M2 Max.

In terms of PC gaming it sounds great, but it's hard to argue that's the only criteria to judge laptops on when comparing them imo. For many people (I suspect far more) the MBP is the better buy.

 

[dmccloud]

That seems like a very good high end Windows laptops but I'd be surprised if it managed even 3-4 hours of battery life Vs the MBPs 20-22 hours. The screen resolution is also lower, it has much more limited I/O, and unclear that the version with an RTX 4050 outperforms the M2 Max.

In terms of PC gaming it sounds great, but it's hard to argue that's the only criteria to judge laptops on when comparing them imo. For many people (I suspect far more) the MBP is the better buy.

I haven't seen the Galaxy Book Ultra 3 in person, but even the Ultra 2 had an amazing OLED display. It was the closest thing I have seen to the Retina displays being used by Apple. IIRC, 8-10 hours was the rated life of the Ultra 2, but if they're dropping a 40xx series laptop dGPU in the Ultra 3, that would likely drop to around 6 hours, simply because even those laptop SKUs of the 40 series are power hungry.

 

[jdb8167]

Samsung sells great Windows laptops too. Here is the new model: Galaxy Book 3 Ultra (https://www.theverge.com/23580552/samsung-galaxy-book-3-ultra-laptop-price-hands-on). This one has OLED screen and might be faster than M2 MBP (definitely faster in graphics). It's also thinner and lighter than Apple counterpart.

I'd wait for an actual review unit to be released before heaping on the praise. Most high end windows laptops don't get anything close to their top speed unless they are plugged in. The battery life on this laptop is not going to be good. It is only 76 Wh compared to the 99 Wh in the 16-inch MacBook Pro. Let's just leave the fan noise out of this comparison too.

 

[Aquaporin]

Apple Silicon is fast but I am sad that many of my older apps, specifically games, that were native to Intel cannot run on Rosetta 2. I was similarly sad when Apple went from PPC to Intel but stopped supported PPC and my PPC apps were no longer usable.

 

[falainber]

I'd wait for an actual review unit to be released before heaping on the praise. Most high end windows laptops don't get anything close to their top speed unless they are plugged in. The battery life on this laptop is not going to be good. It is only 76 Wh compared to the 99 Wh in the 16-inch MacBook Pro. Let's just leave the fan noise out of this comparison too.

Not that many people care about laptop battery life (at least not beyond 5 hours).

 

[NT1440]

Not that many people care about laptop battery life (at least not beyond 5 hours).

Good luck getting 5 out of it in the first place, at least if you’re expecting the power of the i9.

 

[Gudi]

Samsung sells great Windows laptops too. Here is the new model: Galaxy Book 3 Ultra. This one has OLED screen and might be faster than M2 MBP (definitely faster in graphics). It's also thinner and lighter than Apple counterpart.

The Galaxy Book3 Ultra is Samsung’s shot at the MacBook Pro
It’ll start at $2,399.99 ($100 cheaper than the base MacBook Pro 16"), with a release date still to be announced.

Yeah, it better be faster than an M2 at that price. Like most users, I wouldn't even pay that much for an Apple laptop. For a Wintel laptop to even try to justify such a price is ridiculous to me.

 

[NT1440]

The Galaxy Book3 Ultra is Samsung’s shot at the MacBook Pro
It’ll start at $2,399.99 ($100 cheaper than the base MacBook Pro 16"), with a release date still to be announced.

Yeah, it better be faster than an M2 at that price. Like most users, I wouldn't even pay that much for an Apple laptop. For a Wintel laptop to even try to justify such a price is ridiculous to me.

Take a look at the i9/32GB RAM price 😉

 

[Gudi]

Take a look at the i9/32GB RAM price 😉

"Any customer can have a car laptop painted any color that he wants, so long as it is black graphite." − Henry Ford