Back in December I pulled the trigger on upgrading my Intel 27″ iMac and pondered whether I’d made the right decision… With the recent reveal of the Apple Studio product line and the attendant withdrawal of the Intel iMacs, I am hugely relieved that I did.
Technical capability aside, an “Apples for Apples” [sic] comparison of the specs provides the single biggest reason to be relieved. In fact, many, many thousands of reasons.
To recap, my 27″ Intel iMac was configured as follows:
- 3.6Ghz 10-core Intel i9
- 8GB DDR4 RAM
- 2TB SSD
- 27″ 5K Retina Display (standard glass)
- 16GB Radeon Pro 5700XT
The 8GB RAM configuration was the minimum available from Apple, with an OWC upgrade kit purchased separately.
Taking that into account, the total cost of the machine, including the OWC RAM upgrade kit (and shipping etc) came to $7,500. That’s NZ $’s.
Not cheap by any stretch of the imagination, but a beast of a machine (and – IMHO – a good looking one to boot! ymmv).
This is no longer available, having been replaced by a Mac Studio and Studio Display combination.
So let’s configure equivalent specs…
The base configuration of the Mac Studio comes equipped with a 10/24/16 M1 Max. That’s 10 CPU cores, 24 GPU cores and 16-core Neural Engine. With only 32GB of RAM and a 512GB SSD, this base configuration still clocks in at $3600, already close to half the total cost of my iMac system.
Adding a 27″ Mac Studio display with basic, tilt-only stand, is another whopping $3,000. A height-adjustable stand is also an option, for a further $700, but the iMac isn’t height adjustable (though I gained all the height adjustment I needed by adding a TwelveSouth HiRise Pro at half the cost of a Studio height-adjustable stand, with additional storage features and nice desk presence).
Increasing the SSD capacity from 512GB to 2TB is the simplest next step, bringing the cost to $7,649.
And already this configuration has reached the total cost of my iMac and we have not yet addressed the difference in RAM.
To do so, things are a little more complicated.
First of all, unlike the Intel Macs, customers don’t have the option of fitting their own RAM in M1 hardware. You have to configure the RAM from the factory.
Secondly, you cannot simply choose to upgrade from the base 32GB to 128GB.
M1 Max processor options top out with a maximum of 64GB. If you need – or just want – 128GB of RAM, then you are forced to first upgrade the processor to an M1 Ultra. It must be mentioned that the lowest configurable M1 Ultra option (20/48/32) comes with 64GB of RAM to start with, so the 2,450 additional dollars account for more than just the CPU.
If you’re keeping track, that brings the cart up to $10,800
To finally achieve spec parity with the 128GB RAM requires shelling out a further $1,400 for the extra 64GB. As a reminder, that’s DOUBLE what I paid in total for the 128GB kit for my iMac.
Bringing the final system cost to an eye-watering… $12,200 New Zealand dollars.
Out of the gate, that’s 4,700 Reasons To Be Happy with my iMac purchase. And whilst the $3,000 display wouldn’t need to be replaced if/when the Mac Studio itself is ever upgraded, that still leaves a Mac Studio alone costing almost $2,000 more than the iMac with its integrated 27″ 5K Retina display.
Then there’s the question of form-factor.
Whilst there are pros and cons when it comes to all-in-ones, the overriding pro (for me) is that “all-in-one” nature, making managing an increasingly cluttered desktop that bit easier.
The severability advantage of not having to replace the screen when replacing the machine is somewhat undermined by the extraordinary price inflation on that machine.
“But It’s Not Just About The Specs…”
Much has been made of the advantages of the M1 architecture over Intel, but in more recent times the hyperbole has given way to a somewhat more muted appreciation. The power-per-watt advantage is most compelling in mobile form factors which don’t apply in my desktop use case, and when multi-thread work-loads have been more meaningfully measured overall performance advantage over comparable Intel chips has been shown to be less dramatic than initial impressions suggested.
In particular, you might point at the 10-core i9 and think that the 20-core M1 Ultra is self-evidently in a different league. And if you were running single-core tests you would come away feeling correct. But those 10 i9 cores are built on a foundation that relies on hyper-threading to provide 20 virtual cores, which single-core tests do not properly reflect. More importantly, I cannot think of any real-world scenario in modern computing where I am concerned with using each core in splendid isolation. Real software just doesn’t work like that.
Unless your day-to-day work consists solely of running single-core benchmark suites, of course. 🙂
In my initial post about my purchase, I mentioned not being too concerned with content and media production work, where the M1 is said to have the greatest advantage. Whilst that absence of concern has recently changed, I cannot say that my i9 equipped machine is struggling. Quite the opposite.
Perhaps an M1 Ultra would provide some superior performance in some areas but until I feel I need that, I’m very happy to have that $4,500 to spend on other things.
However, you should not forget the power savings over the runtime of the device compared to Intel chips and the M1. The M1 chip is extremely power-efficient compared to the Intel chip, so you can save a lot of money on electricity over the runtime of the device. You should not always only look at the hardware, but always add up the total costs over the runtime. Then it quickly becomes relative.
That’s a fair point but as you say, it is relative and the difference is only maximised when operating at peak power consumption.
Even assuming an unrealistic scenario where I was running the CPU at peak power consumption all of the time, relative to the total energy consumption in my workspace (never mind the household as a whole) the difference is still negligible and any savings will not make a significant dent in the $4,500 price premium over the lifetime of the product.
It would be interesting to see what difference it makes to the total power consumption of a fully equipped workstation, including display and any additional storage solutions. Curiously, the specs for the Apple Studio 27″ Display do not include power consumption (which will be more or less consistent when on, aside from ambient brightness variations, regardless of how much power the CPU is drawing).
Power per watt is also of significance in laptops and mobile devices due to the direct impact this has on battery life, both in terms of how long you can use the device away from an outlet and also the greater longevity of the battery itself as a result of fewer cycles over the life of the device.
On the desktop aside from potential energy savings, the only other real benefit of greater efficiency is the reduced need for active cooling, i.e. less noise from fans.
But that has subjective value, and in the real world is sensitive to other factors that vary from individual to individual.
For example, in my case, I have a 4-bay Thunderbolt enclosure directly behind my iMac (which acts as something of a “baffle” to reduce the sound of the enclosure fans). Although not excessively loud, these eliminate any benefit of “silent” operation of anything else on the desk.
Worth mentioning here is that the fans in the i9 are far less active than those on my previous 2017 i7 iMac and don’t kick in at all unless I’m transcoding media (not even whilst editing). Again, there is greater noise pollution from the fans on my Intel NUCs which kick in far more frequently even when “idle” (probably running Windows updates, though I can’t tell as they run “headless”).
So in my case, neither energy cost nor silent operation are significant factors. 🙂
ymmv.