Apple has announced a new, redesigned 24-inch iMac, featuring an M1 chip, a 4.5K display, and a range of color options, as well as an improved cooling system, front-facing camera, speaker system, microphones, power connector, and peripherals.
These look pretty good, but they come with the same limitations as all the other identical M1 Macs – 8 GB of RAM standard with a maximum of a mere 16 GB, lacklustre graphics chip, no high refresh rate displays (in 2021!), barely any ports, zero expandability, and Linux/BSD support will always remain problematic and years behind the curve.
Good processor, but at what cost?
I think it still remains to be seen whether it’s a good processor in the desktop space where power limits aren’t nearly as much as a concern. Last I heard Arm doesn’t scale up well, while x86 doesn’t scale down well; so we’ll see if M1 chips are all typical Apple marketing hype once reviewers get their hands on these.
“iMac offers two Thunderbolt ports” – somehow I can already guess that these don’t support egpus.
They are actually very competitive for content production tasks:
https://www.pcmag.com/news/first-tests-adobe-photoshop-for-m1-macs-is-quick-but-not-always-15x-faster
In the laptop market, the results are quite good. For the all in ones (like iMac) it should still be the same, since that sector uses mobile chips as well.
For high end, of course Threadripper will take the crown.
dark2,
It’ll be difficult to get better GPU performance due to the M1’s shared memory architecture. Unlike a system with a dedicated GPU, the higher resolution and FPS fundamentally competes with bandwidth needed for CPU tasks. To their credit, this chip is at the top end for an integrated iGPU, but integrated GPU perform worse in general and don’t think the M1 has really changed this outlook. Workstations with dedicated GPUs & vram still have the edge.
In terms of CPU performance, when it was revealed last year the M1 had outstanding single threaded performance, but disappointing multi-threaded performance. It was argued by many that multithreading performance would catch up in future models, we shall see if this is the case. Currently AMD is the one to beat for both single threaded and multithreaded workloads.
GPU use on Macs has long been an external option market.
With TB4 it’s possible that external GPUs could match internal ones in performance.
This is such a dumb statement it’s cringeworthy.
The 4core dekstop M1 has already been reviewed many times and it’s extremely fast for its power consumption and form factor.
E.g. check this: https://www.anandtech.com/show/16252/mac-mini-apple-m1-tested/7
I knew it wouldn’t be long before someone read only the benchmarks. Look up “M1 reality check” and you’ll see tiger lake already giving it a thorough beating (and amd laptop cpus were left out to begin with when they were already beating Intel), real world use showing the same Intel chips in Apple’s comparisons often beating the M1 when actually being used for work instead of benchmarks, etc. It’s good “for a 10w processor.” If you’re going beyond editing office documents and basic photoshop work, you’ll still need something much more powerful. There are even some reports of the limited RAM causing enough swap access to appear to cause the SSD to thermal throttle. No Apple has not proven the M1 is actually more powerful than x86 at all. It’s almost like they copied what the cell phone companies were doing a few years ago and upped the clock speed when benchmarking was detected. Until they release a true desktop level ARM processor with roughly the same performance and power requirements of a current 6 core x86 CPU, it’s all just typical Apple marketing hype to me.
Nah dude, for 99% of users who were on the Mac ecosystem, the M1 based macbooks seem to do the same or slightly better than the intel models they replaced, at a much longer battery life and nearly silent/cool. Which is the value added that most customers seem to prefer.
The amount of customers that use their laptops as mobile high performance rendering workstations or to do molecular modeling is minuscule. Heck, most of the wintel laptops that use high end chips are for high end gaming, which is something Apple has been traditionally not that interested in.
It’s a sing of the times, the mobile CPUs are now performing on par with the desktop micros. Just like desktop micros supplanted the RISC workstation/server CPUs, just like how the RISC micros supplanted the minis, etc, etc. Every generation thinks there’s some “special” sauce to their systems, when in reality they simply don’t understand how increased levels of integration work.
The M1, unlike AMD/Intel’s parts is a true SoC. It’s the next level of integration.
javiercero1,
That’s a clear advantage when it comes to portable laptops. ARM has had the energy advantage since forever and nobody is surprised that M1 is more efficient than x86. Data centers have been investing in ARM for the same reason.
With that said, there’s a lot of pent up demand for more powerful macs from professionals and it is a fair question to ask how well these can handle the heavy workloads that many professionals need to work on. Last year many people were asserting that apple would easily be able to scale up performance and RAM to meet the demands of power users, but it hasn’t happened so far. Maybe scaling up has been more challenging for apple than was assumed, or maybe apple is not that bothered to offer higher end ARM workstations at this point in time. Either way, it’s likely going to leave a void where pros have to contemplate between buying hardware with lower specs than they wanted, or switching platforms. The 16GB limitation is quite disappointing. To an extent VM swapping will mitigate the effects of low ram conditions, but that brings up another problem that M1 owners are reporting:
https://www.phoneswiki.com/m1-macs-report-excessive-ssd/
(It’s been a couple months since I’ve read about this, if anyone has new news about it, I’d like a link!)
I hear you, but not everyone wants that. There’s actually a lot of us in the prosumer space who are really into ARM computers, I talk about this frequently on osnews. But many of us actually prefer having computers with components that can be swapped in/out for upgrades and replacement. We’re better served by a modular system board with upgradable components rather than having to replace the entire system.
I’m actually pleased that apple is pushing ARM technology. ARM needs that! But ultimately I’m more interested in a linux-friendly platform with extensible hardware.
People who need a mac workstation right now they can get a Mac Pro.
For the vast majority of Apple users, M1 seems to work just fine: It runs productivity apps at the same speed as the intel counterparts, it has the IP to accelerate video encoding for those doing light weight video editing, and the overall system is far more responsive than the intel counterpart due to the scale of integration.
Apple is doing a gradual replacement of Intel for ARM, so they’re doing it in a step fashion. And they’re targetting their products that lead their sales: small MacBooks, Mac Minis, and iPads. The large Macbook Pros and Mac Pros will come last at the end of the year.
So far, every M1 replacement performs better than the intel part they’re replacing. So there’s a clear trend that shows Apple knows what they are doing.
I think what you want is a powerful upgradable PC for a reasonable price, that is unfortunately not something you’re going to likely get from the Apple ecosystem or from ARM.
The market Apple is targeting is the consumer/prosumer/professional that sees their products as a turn key connected appliance. Which has been one of the holy grails of computing companies, interestingly.
javiercero1,
You keep saying intel, but that’s like competing for 2nd place. Even then the M1’s parallel performance is subpar. And the ram situation is limiting. Maybe you are happy with that, and maybe apple is happy with that, but obviously some mac users would like an M1 with better specs.
As for the x86 mac pro, I think it’s good hardware, but for better or worse apple has it at enterprise prices. $10k+ for decent specs is not very accessible and I would say that puts the modern mac pro out of reach for old-school fans of the mac pro. Moreover, since there’s no AMD option, the best CPUs are still out of reach.
You keep saying this as well, but the reality is that apple’s x86 products could not perform to spec because apple forced the CPUs to operate thermally throttled almost all the time. It’s a classic case of form over function. It’s well known that Intel processors use more power and are at a disadvantage in mobile applications. It’s easy for ARM chips to win this category. For people who want ultra thin/low power, this is great. I welcome ARM. But it’s not a good reason to just disregard power users IMHO.
No, the M1 performance is not “subpar.” Compared to the competing 4C AMD/Intel part, the M1 does as well or better in most benchmarks. It’s a well balanced system for the use cases it is intended for: media consumption, web browsing, productivity applications, and low/mid end creation. For the majority of Apple customers in that use case bracket, the M1 performs those tasks at the same or better level than the alternative Intel part, and it does it at a lower power which enables Apple the type of platforms they feel the majority of their customers appreciate: very responsive compact/mobile platforms with long lasting battery, and with little fan noise.
The Mac Pro is a workstation and it is priced as such. The mac pro is targeted for professional environments, this is; it’s a source of revenue. For the use cases the Mac Pro is used, the HW is not the main limiter to the cost per seat.
Apple has always seen the high end of the mac line as workstations, and they have never been accessible to home users when compared to a DYI PC alternative, period. Adjusted for inflation, a Mac II in the 80s, a tower Powermac in the 90s, and the cheese grater G5s in the 00s are all basically at the same cost bracket as the current Power Mac.
I think we already went through this; Apple has never considered AMD as a source for the CPUs, so Thread Ripper is out of the question. They stopped trying to compete on performance per dollar long time ago, they have focused, for a while now (almost 2 decades), on a different value proposition for their system. If the Apple ecosystem and a turn key solution based around it does nothing for you, you are not the target for their value proposition.
javiercero1,
You’re comparing the M1 to x86 chips that are themselves older subpar even for x86. That’s the problem.
Predictably enough, once again you go out of your way to miss the point.
It’s impossible to compare the M1 to a current x86, because neither Intel nor AMD have a similar product. We can compare on its discrete components like CPU and GPU, in which case the M1 either matches or surpasses the alternative x86 part at the same power level.
DYI geeks with high demands and low disposable income are not the target audience for Apple’s products. You’re basically a farmer asking for a cheap tractor at a mercedes leadership:
Does the tractor have more torque? Sure. Does the tractor carry more bushels of wheat? Absolutely. Can the mercedes sedan plow a field, not really. But see, the average upper/middle class urban driver doesn’t give a shit about any of that, they want a good looking luxurious and comfortable vehicle that fits with their lifestyle.
Similarly; do you care about the design of the product? Do you care about Apple’s ecosystem of media/apps/services? Do you see your computers/phones as accessories of a specific lifestyle or professional tool? If the answer is no to all those questions; you’re simply not Apple’s target audience.
.
javiercero1,
Oh don’t pretend I’m the one being dense. You know that apple’s line up has a hole in it for power users. Does apple intend to rectify this? Who knows. Either apple is having troubling scaling up specs, or apple is fine ignoring the needs of power users. Ether way it’s disappointing. Are you suggesting mac users who want a more powerful macs to leave the mac ecosystem for other vendors that offer higher specs? Surely even you should be able to understand why this is disappointing.
Are we pretending there are really that many Apple professional customers, which are mostly in the creative/video editing space, for which a desktop w 28 core, 2TB RAM, 4GPU with 64GB of VRAM, Video encoder accelerator, 10Gbps ethernet, loads of storage, and with the best monitor in the industry are somehow not powerful enough?
The mythical “power user” that has very little disposable income is not in Apple’s radar as a consumer. Or on any other workstation vendor for that matter.
If you want a many core machine with better specs, then you have to go thread ripper. But a 32-core from lenovo, for example, doesn’t end up being that much cheaper once you configure it with comparable specs.
So yeah, you’re just being dense for the sake of being dense. Like most arguments we end up having.
For the low/mid range, the M1 Macs seem to be doing just fine. The higher/premium tears are still serviced with intel parts, and since they are significantly smaller in market size, Apple will replace those last. As the high tier ARM chips are going to be more expensive given the economy of scale there. But the overall architecture of the core is there, within the M1, so Apple has already shown they have a competent microarchitecture.
javiercero1,
Not at all, but you are pretending that there’s no middle ground.
That an ignorant statement, there are lots of vendors who sell high end specs to power users without forcing us to buy enterprise grade gear. You think apple shouldn’t cater to the prosumer democraphic, fine but that’s your opinion only! There’s no reason for you to become so confrontational when people disagree with your opinion.
This was reported in november of last year…
https://www.macworld.com/article/234843/m1-macs-memory-isnt-what-it-used-to-be.html
There was an expectation that higher specs were coming. And now that the new 2021 imacs are out, being limited to the same specs is a bit of a letdown. One of these is logically true: apple is either unable or unwilling to deliver better M1 computers at this time. Either apple decided against giving M1 imac customers better specs, or apple is having difficulties increasing the specs of the M1 SoC. It’s not in character for apple to let down customer expectations for new products. I believe apple would have delivered higher specs if they could have, so I’m guessing they’re having trouble scaling. It’s possible they painted themselves into a tight corner with the integrated SoC design. By comparison discrete components are relatively easy to upgrade. But hopefully for the sake of progress and competition they get it sorted sooner rather than later. We shall see.
M1 mac minis have been out for a while, and they have been shown to perform just fine.
For the target audiences, these machines will perform well.
This is the next refresh in the scaling down, so now the mobile SoCs are taking over, just like the desktop microprocessors did a couple decades ago. ARM has now matched the microarchitecture performance of x86, so it’s going to be interesting times from now on. Intel and AMD are certainly worried.
Just 8GB of RAM and it’s not upgradeable, otherwise it’s a beast of a PC inside a … monitor.
Shame on Intel and AMD.
(There is 16GB option, but not on the website yet: https://9to5mac.com/2021/04/20/m1-imac-tidbits-storage-ram-more/ )
To be fair, mainstream CPU vendors (Intel, AMD) cannot be expected to compete with this design. This is a SoC with a very specific target computer in mind. GPU, AI and even RAM is on the same chip. A more apt comparison could be with other system vendors, like the Xbox Series X or PS5, that has a higher performance than comparably spec’ed PCs.
We could still make an argument for AMD/Intel making specialized SoCs and forgo PCIe and backwards compatibility. And use this to heavily optimize both the hardware and specialized versions of Windows / Linux. But I don’t think there is going to be a huge market for that.
8 or 16, some of it is eaten by GPU (e.g. modern web browsers can easily consume a gig of VRAM which is …RAM in this case and some workflows are even heavier), so in reality it’s even less than that.
In a SoC there’s reduced need for replicate system ram elements into the VRAM for the GPU, unlike discrete GPUs which need to move a lot of elements from system RAM locally into their VRAM to overcome the bus bottleneck.
This is, in a discrete GPU, the VRAM operates as both; display RAM, as well as a cache for texture/geometry/etc.
Obviously, the GPU in the M1 is not going to compete with a med/high end discrete GPU from NVIDIA/AMD. But for the majority of the uses cases these machines are targeted for they are more than adequate.
I think the level of integration in SoCs confuse a lot of people architecturally, specially when it comes to memory accesses.
It’s an iMac, so I am willing to give them the benefit of the doubt that they examined the use cases of most iMac users and found the M1 to be adequate (though having to pay extra for 16GB is ridiculous… at least it exists as an option).
Let wait until they reveal the replacement of the Mac Pro. That’s how we’ll determine whether they still consider the Mac to be a platform also targeting professionals or they only want to make bigger iPads from now on. Ok, bigger iPads with sideloading and self-hosted Xcode, but still.
> Good processor, but at what cost?
$1.299 USD
But anyway, it’s meant to fill in the low end iMacs which haven’t seen a refresh in ages. They’ll follow up later with their M1X or whatever it gets called.
I do wish the system, despite being user friendly, weren’t so user hostile.
I have my hopes up about it being far easier to repair than in previous models. The logic board looks like it should pop out without having to take half the bloody thing apart.
> Good processor, but at what cost?
We already knew it, of course, but now the external design also makes blatantly obvious which direction the os is moving towards.
White bezels??? Monster chin?? No M1x?
Thanos Mac is a pure disappointment.
The most interesting things announced was the use of the M1 chip in the new iPad Pro, plus the increased memory and storage options. The iPad Pro is now, in terms of hardware, a Mac with a touch screen and lots of extra sensors. I wonder if they will announce OS changes that make the iPad Pro dual booting at WWDC? Apple clearly thinks blending touch with a WIMP interface is sub-optimal, which I agree with, but if you could boot between iOS and MacOS with file compatibility that would be a very, very interesting option.
Personally, seeing it failed with Linux and after again with Microsoft, i am interested in seeing if Apple, being a closed ecosystem, can succeed with ARM on desktop. In lets say next 5 to 10 years. My prediction is Apple will have a hard time in doing that.
The iMac now uses a power brick instead of an internal PSU.
“ Are we pretending there are really that many Apple professional customers, which are mostly in the creative/video editing space, for which a desktop w…”
Actually the M1 decimates AMD and Intel chips in workstation graphics application. Only the top two ThreadRipper parts beat it.
Take a look at native M1 AV1 test results.
For the “normal” video workflow 265/AACHE encodes are lightning fast. 8:1 speed over real-time with 1080p/60.
Before the M1 was released I was extremely vocal about how this was a bad idea. Apple proved me wrong. Sometimes people need to just admit their error and move on. I called the move stupid. I was wrong.
Ported BSD workstation apps run better on M1 with Rosetta 2 translation. Load and function speeds are considerably faster. File compression, like video compression, is another real world benchmark. And again the M1 excels! Ported Intel apps are just as good.
And Apple’s ecosystem is such that venders will move to Risc native in time.
The only people today clinging to the claim that M1 performance is synthetic are those unwilling to admit they made a mistake in earlier discussion!