- dual slotted SATA III storage (RAID capable) and (16GB) ram - i7 2 (4) core processor - discrete internal GPU (no eGPU needed) - triple monitor support - list price under $1K
Is trying to turn the mini into a high end workstation a reasonable premise? How 'mini' is it then ? Is the concept more aligned to both entry level consumer and if dialed up a prosumer or office workstation at the lower middle performance category?
- dual slotted M2 or better storage (RAID capable) and (up to 32GB) ram - i7 6 (12) core (or i9) processor - discrete internal GPU (v20, 5500 or ?) - 5K (or 6K) monitor support - list price under $1K
"We'd like to see a computer like a complete NUC 9 kit, aimed hit the same market segment that the original $1599 Graphite G4 tower was for."
That's the market served by the standard 5K iMac. The apps that people bought a G4 tower to run back in the day will easily run on even the low end 5K iMac now.
The iMac is not suitable for a wide array of professional users. One big reason is the included screen. A second issue is the way Apple locks out the storage devices such that you are screwed if something goes wrong. I can almost understand Apples approach for the boot drive, however that just means they have to provide an open M.2 port so that a persons storage is not compromised.
I really don't get it when people offer up the iMac in this manner. The all in one platform simply isn't acceptable for many, especially one that goes out of the way to restrict access to the internals. If you should need to destroy the storage device on one of these machine you end up spending excessive time just to get to it. To put it mildly, for many professionals the storage devices is the life blood of the company and can't fall into the wrong hands. That include larger organizations that have massive server rooms.
So your main argument is you don't like the screen and you should be a able to rip open a box to throw a physical hard drive into a shredder. That's not very convincing. The vast majority of "professionals" are served just fine by both the very nice 5K screen and by FileVault 2 and a wipe. Your concerns are exceedingly niche.
Pretty sad day to see an apple fan site, ask,beg,plead Apple for a better computer. Don't get me wrong I absolutely agree with this post, it's just sad. Long gone are the days of Apple's superiority PPC RISC, special made macs. Now we just get macs from the ol parts bin. Apple has almost become all but irrelevant. That's the word in the Valley, not me saying it, just repeating/agreeing. SAD.
The xMac started in 1998 when the lineup was reduced to the quadrant after an existential crisis, and lived through PPC heyday as a Mac fandom dream. The iMac occupied the lower priced desktop tiers. People wanted a cheap expandable box at iMac prices. Apple back then gave up the low priced headless desktop and offered the iMac instead.
Only difference today is with the iMac now ranging past $5000, the high priced headless desktop Mac is pushed to $6000, the wish for an expandable desktop Mac now spans $1000 to $4000 or so. That range was narrower in the past, and the xMac was to fill it.
As has been said, the iMac is Apple's desktop machine. It currently offers a Core i9-9900K, which was top of the line as of 1 week ago, and with Comet Lake just announced, their availability will be low awhile for everyone for awhile. The iMac Pro comes with Xeon W 2100 CPUs and 16 GB GPUs has been available for people who want more cores. The Xeon W 2200 versus the 2100 is bit like Comet Lake is to Coffee Lake, a marginal 5% improvement. Apple isn't in a rush for that.
You can say Apple should use AMD, but they seem to have other plans with their own ARM designs in the near future. You can maybe argue that Apple would be more successful with a cheap expandable desktop, but given that they have no real desire to make gaming boxes and the rise of laptops, I think it would be tough to say a cheap desktop Mac would have made any difference to them. Some users would be happier, but Apple? Not much change.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
"We'd like to see a computer like a complete NUC 9 kit, aimed hit the same market segment that the original $1599 Graphite G4 tower was for."
That's the market served by the standard 5K iMac. The apps that people bought a G4 tower to run back in the day will easily run on even the low end 5K iMac now.
The iMac is not suitable for a wide array of professional users. One big reason is the included screen. A second issue is the way Apple locks out the storage devices such that you are screwed if something goes wrong. I can almost understand Apples approach for the boot drive, however that just means they have to provide an open M.2 port so that a persons storage is not compromised.
I really don't get it when people offer up the iMac in this manner. The all in one platform simply isn't acceptable for many, especially one that goes out of the way to restrict access to the internals. If you should need to destroy the storage device on one of these machine you end up spending excessive time just to get to it. To put it mildly, for many professionals the storage devices is the life blood of the company and can't fall into the wrong hands. That include larger organizations that have massive server rooms.
So your main argument is you don't like the screen and you should be a able to rip open a box to throw a physical hard drive into a shredder. That's not very convincing. The vast majority of "professionals" are served just fine by both the very nice 5K screen and by FileVault 2 and a wipe. Your concerns are exceedingly niche.
I'm not that sure on this point.
It's impossible to know, but my personal thinking is that a G4 graphite style Mac which played to its strengths would cut a big hole into iMac and Mini sales. In fact, I sincerely believe that is why one doesn't exist now. Apple just doesn't want to see its margins hit.
After four years without an update, the Mini could have easily been considered a niche and the Mac Pro trash can and the new Mac Pro are also niche machines.
A low end tower or xMac or whatever you want to call it, would definitely sell more than the Pros - provided that - Apple made it competitive with the branded competition.
I definitely don't want an Apple screen bolted onto a desktop Mac after seeing a 27inch i7 slow cook itself to virtual death due to extremely poor cooling mechanisms. That paired with accessibility issues for even the most mundane tasks (like removing dust, cleaning inlets etc) and life extending or preference upgrades (RAM, upgrading screen etc).
At the end of the day it boils down to options and there is a huge hole where the option of this potential Mac should be.
The exact same needs catered to on the 'modular' Mac Pro are shared by a huge swathe of users but without the 'extreme' options. A Mac Pro for the rest of us as it were. Just like it was back in the day of the Sawtooth. No need for $700 wheels.
I’ve used NUCs for prototyping industrial applications that were later deployed to true, hardened industrial PCs. They are okay, but tended to be noisy and had issues with drivers that Intel was very slow to address. I’m not really sure where this “super NUC” fits into the general computing landscape but it could serve as a human machine interface (HMI) platform because it has 4 HDMI ports and it could serve as a high performance computing node for a in-process data acquisition & analytics system with its Xeon processors. Having two Ethernet ports is interesting if they can be configured for link redundancy. But as a general purpose desktop replacement, no way I’d buy one over a traditional form factor PC unless I was living in a tiny house. The NUC has too many compromises.
I agree with other folks who would love to see Apple build a mid range “Component Mac” that has replaceable components like memory, video, audio, and storage. Apple simply does not have anything in their current lineup that appeals to PC hardware enthusiasts, especially those who get all geeked up about benchmarks and having the latest still dripping wet chunk of silicon baking beneath a window air conditioner sized cooler in a gaudy box with blinking lights. If you’re into that sort of thing, you’re not going to find it at the Apple Store, and may never will. It’s not because Apple can’t build it, they can, but that type of machine is obviously in the category of things that Steve Jobs would have put in the “No, we’re not doing that” pile. Does that leave a hole in Apple’s lineup? Absolutely, but it’s a hole that Apple has purposely decided they aren’t jumping into, at least for the past 25 years and for the foreseeable future.
I think Apple builds machines that they feel serve the greatest number of customers for the types of jobs those customers must perform. I also believe Apple views the greatest differentiated value for the greatest number of customers coming from the software. The hardware is just there to support the software and the best way that it can do so is by providing a reliable, secure, and stable platform for the software to run on for as long as possible. Unless you’re a PC enthusiast or gamer the biggest investment in your time and billable hours is probably dominated by the software components and software related contributions that make up your production environment, not the hardware acquisition and upgrade costs. The hardware has to be good enough to not get in the way of the software and not to have recurring repair expenses or, worst yet, downtime.
I have no doubt that there are cases where having the ability to upgrade and reconfigure a machine post-purchase is a huge benefit for an individual or business, but for a large number of individuals and businesses having a reliable and stable hardware platform that just works with very little administrative overhead and serves the needs for several years keeps a lot of folks happy, and Apple’s sales numbers reflect this. For people who want to tinker and tweak their hardware the Windows and Linux world has plenty of opportunities. As an enthusiast you don’t have to choose one or the other, you can choose “all of the above” and be happier for having done so. Otherwise, it should be a data driven decision. No problem.
Big bulky inefficient hot chips - thanks Intel! Their days are numbered. Failed at making radio chips, can’t figure out how to shrink the die, a big monopolistic drag on the advancement of the chip industry.
"We'd like to see a computer like a complete NUC 9 kit, aimed hit the same market segment that the original $1599 Graphite G4 tower was for."
That's the market served by the standard 5K iMac. The apps that people bought a G4 tower to run back in the day will easily run on even the low end 5K iMac now.
Read one more paragraph below this one in the article.
And? What does predicting that a Mac Pro Mini or xMac would sell more than the Mac Pro have to do with what I posted? I'm saying Apple is already selling a product to that market: the standard 5K iMac. They don't need to create a new product. The type of legacy apps that people bought a G4 to run are going to run just fine on non-Pro hardware.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
I'm curious if you have ever looked at buying Mac minis, networking them, and doing compile jobs this way?
The tradeoff with getting a Mac Pro would be interesting because at some amount of dollars for a single user, spending the extra money for a single box would be better. Then again, you can get 6 6-core Mac minis for $6k. That's 36 cores to play with.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
I'm curious if you have ever looked at buying Mac minis, networking them, and doing compile jobs this way?
The tradeoff with getting a Mac Pro would be interesting because at some amount of dollars for a single user, spending the extra money for a single box would be better. Then again, you can get 6 6-core Mac minis for $6k. That's 36 cores to play with.
I have looked at it, but when making cross platform apps they also need to compile for the other platforms which are often done in virtual machines, which does not work with that distributed build approach.
We have replaced some of the 6/12 core/threads minis with trashcans refitted with close to new 12/24 core/threads Xeon cpus and they both do the job faster, are virtually silent and have graphics performance that is quite a few notches up from the minis (although quite mediocre when it really comes to it). These machines also run virtual machines pretty well.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
I'm curious if you have ever looked at buying Mac minis, networking them, and doing compile jobs this way?
The tradeoff with getting a Mac Pro would be interesting because at some amount of dollars for a single user, spending the extra money for a single box would be better. Then again, you can get 6 6-core Mac minis for $6k. That's 36 cores to play with.
I have looked at it, but when making cross platform apps they also need to compile for the other platforms which are often done in virtual machines, which does not work with that distributed build approach.
We have replaced some of the 6/12 core/threads minis with trashcans refitted with close to new 12/24 core/threads Xeon cpus and they both do the job faster, are virtually silent and have graphics performance that is quite a few notches up from the minis (although quite mediocre when it really comes to it). These machines also run virtual machines pretty well.
Can you distribute the compiles on a VM per networked machine basis? Ie, instead of using all the machines to compile an app for one platform, have each machine compile for different platforms concurrently?
Seems like your computational needs aren't that far away from needing to get some racks and a server room if your compute jobs are lasting on order hours. The longest thing I do are video transcodes on my 2013 iMac 27. 10 videos can go all night. A 2018 Mac mini would do my H.265 transcodes 10x faster, but I don't do it enough to really need to change anything.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
I'm curious if you have ever looked at buying Mac minis, networking them, and doing compile jobs this way?
The tradeoff with getting a Mac Pro would be interesting because at some amount of dollars for a single user, spending the extra money for a single box would be better. Then again, you can get 6 6-core Mac minis for $6k. That's 36 cores to play with.
I have looked at it, but when making cross platform apps they also need to compile for the other platforms which are often done in virtual machines, which does not work with that distributed build approach.
We have replaced some of the 6/12 core/threads minis with trashcans refitted with close to new 12/24 core/threads Xeon cpus and they both do the job faster, are virtually silent and have graphics performance that is quite a few notches up from the minis (although quite mediocre when it really comes to it). These machines also run virtual machines pretty well.
Can you distribute the compiles on a VM per networked machine basis? Ie, instead of using all the machines to compile an app for one platform, have each machine compile for different platforms concurrently?
Seems like your computational needs aren't that far away from needing to get some racks and a server room if your compute jobs are lasting on order hours. The longest thing I do are video transcodes on my 2013 iMac 27. 10 videos can go all night. A 2018 Mac mini would do my H.265 transcodes 10x faster, but I don't do it enough to really need to change anything.
Xcode (re)builds the project with about 1.5 million lines of code in about 9 minutes on the trashcans. The Mini used 16 or so. Visual studio needs 2.5 hours for the Win version.
Xcode has a tendency to randomly rebuild the entire project, when xcodebuild (command line) would do an incremental build. But it can also go the other way.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
I'm curious if you have ever looked at buying Mac minis, networking them, and doing compile jobs this way?
The tradeoff with getting a Mac Pro would be interesting because at some amount of dollars for a single user, spending the extra money for a single box would be better. Then again, you can get 6 6-core Mac minis for $6k. That's 36 cores to play with.
I have looked at it, but when making cross platform apps they also need to compile for the other platforms which are often done in virtual machines, which does not work with that distributed build approach.
We have replaced some of the 6/12 core/threads minis with trashcans refitted with close to new 12/24 core/threads Xeon cpus and they both do the job faster, are virtually silent and have graphics performance that is quite a few notches up from the minis (although quite mediocre when it really comes to it). These machines also run virtual machines pretty well.
Can you distribute the compiles on a VM per networked machine basis? Ie, instead of using all the machines to compile an app for one platform, have each machine compile for different platforms concurrently?
Seems like your computational needs aren't that far away from needing to get some racks and a server room if your compute jobs are lasting on order hours. The longest thing I do are video transcodes on my 2013 iMac 27. 10 videos can go all night. A 2018 Mac mini would do my H.265 transcodes 10x faster, but I don't do it enough to really need to change anything.
Xcode (re)builds the project with about 1.5 million lines of code in about 9 minutes on the trashcans. The Mini used 16 or so. Visual studio needs 2.5 hours for the Win version.
Xcode has a tendency to randomly rebuild the entire project, when xcodebuild (command line) would do an incremental build. But it can also go the other way.
Visual Studio doesn't do incremental compiling? Sounds cheaper to just get a Windows PC as 2.5 hours is a pretty long time.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
I'm curious if you have ever looked at buying Mac minis, networking them, and doing compile jobs this way?
The tradeoff with getting a Mac Pro would be interesting because at some amount of dollars for a single user, spending the extra money for a single box would be better. Then again, you can get 6 6-core Mac minis for $6k. That's 36 cores to play with.
I have looked at it, but when making cross platform apps they also need to compile for the other platforms which are often done in virtual machines, which does not work with that distributed build approach.
We have replaced some of the 6/12 core/threads minis with trashcans refitted with close to new 12/24 core/threads Xeon cpus and they both do the job faster, are virtually silent and have graphics performance that is quite a few notches up from the minis (although quite mediocre when it really comes to it). These machines also run virtual machines pretty well.
Can you distribute the compiles on a VM per networked machine basis? Ie, instead of using all the machines to compile an app for one platform, have each machine compile for different platforms concurrently?
Seems like your computational needs aren't that far away from needing to get some racks and a server room if your compute jobs are lasting on order hours. The longest thing I do are video transcodes on my 2013 iMac 27. 10 videos can go all night. A 2018 Mac mini would do my H.265 transcodes 10x faster, but I don't do it enough to really need to change anything.
You actually bring up some of the reasons why I was questioning where the NUC 9 Pro and to some extend higher-end Mac Minis really fit into the computing space. They are quite pricey and seem more client app oriented than what I'd expect to see deployed into a server rack for doing automated builds/testing in a CI/CD environment. My desk can generally accommodate a tower/mini-tower machine very easily. If you're going to prune down the form factor and start to approach something loosely resembling a blade server, why include all the extra client side stuff like 4 HDMI ports that are not going to be doing much of anything in a rack based environment? This is the gist of the "compromises" issue I have with the NUC and the Mini. For sw dev work I have more than enough desk space with arm-mounted monitors and the ability to mount a tower on a rack under my standing desk. The compromises involved with stuffing all of the computer components into a tiny enclosure when they could could be living in a free range computing environment is lost on me. Sure, for embedded, rack based, or enclosure based deployments, size (and heat and power and headless and auto-restart and a plethora of other things) matter, but on my desktop the situation is different.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
I'm curious if you have ever looked at buying Mac minis, networking them, and doing compile jobs this way?
The tradeoff with getting a Mac Pro would be interesting because at some amount of dollars for a single user, spending the extra money for a single box would be better. Then again, you can get 6 6-core Mac minis for $6k. That's 36 cores to play with.
I have looked at it, but when making cross platform apps they also need to compile for the other platforms which are often done in virtual machines, which does not work with that distributed build approach.
We have replaced some of the 6/12 core/threads minis with trashcans refitted with close to new 12/24 core/threads Xeon cpus and they both do the job faster, are virtually silent and have graphics performance that is quite a few notches up from the minis (although quite mediocre when it really comes to it). These machines also run virtual machines pretty well.
Can you distribute the compiles on a VM per networked machine basis? Ie, instead of using all the machines to compile an app for one platform, have each machine compile for different platforms concurrently?
Seems like your computational needs aren't that far away from needing to get some racks and a server room if your compute jobs are lasting on order hours. The longest thing I do are video transcodes on my 2013 iMac 27. 10 videos can go all night. A 2018 Mac mini would do my H.265 transcodes 10x faster, but I don't do it enough to really need to change anything.
Xcode (re)builds the project with about 1.5 million lines of code in about 9 minutes on the trashcans. The Mini used 16 or so. Visual studio needs 2.5 hours for the Win version.
Xcode has a tendency to randomly rebuild the entire project, when xcodebuild (command line) would do an incremental build. But it can also go the other way.
Visual Studio doesn't do incremental compiling? Sounds cheaper to just get a Windows PC as 2.5 hours is a pretty long time.
Of course it does. I stated complete rebuild times. Giving VS a native PC (or rebooting the trashcan in bootcamp) only marginally improves build times. Versions before VS2019 also don't respond well to increasing number of processors over a certain level. It just spawns off more processes that fight with each other for command line builds.
great article, great discussion - AI definitely doing its part.
as subtly implied macOS definitely runs in NUCs (very well too), and eGPU noise levels are often too difficult without headphones. of course the base Mac Pro runs more silent, that seems to be the one advantage.. and sometimes i do wish i had all the money in the world so i didnt have to hear machines.
as many say the ARM-architecture transition may be the only explanation for the state of things.. unfortunately one thing we may deduce from that path is the likely complete disappearance of upgradeable or serviceable parts- the end of xMac possibility all-together. crossing fingers that like iphones and ipads, the new machines are so spectacular that such limitations arent felt anymore - as unless disposable income isnt an issue - they definitely are now.
I'm a software dev and push my machines fairly hard and have to say the 2018 Mini is pretty much perfect in every way. It's powerful, quiet, small, and cheap. And I 100x prefer using macOS over Linux on the desktop.
So for those thinking Apple is dead, supposedly in Silicon Valley even, I just have to roll my eyes.
For lengthy Xcode compiles (or something else in a VM) the mini is far from quiet. The fan noise gets pretty annoying over time when you listen to it for hours on end.
Agreed. The fans can get very noisy when you're pushing the Mini hard. That's just the reality of Intel CPUs.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
And yet I was able to get an app with fairly complex 3D models and simple editing of them (3 axis rotations, translation, scaling) working quite well on a 2014 Mac Mini running at 4K (around 30 fps). It was actually our 4K benchmark for machines with low end graphics hardware and it worked surprisingly well with only a few optimizations.
I don't think you know what optimizing your software for average consumer hardware actually means.
Comments
- dual slotted SATA III storage (RAID capable) and (16GB) ram
- i7 2 (4) core processor
- discrete internal GPU (no eGPU needed)
- triple monitor support
- list price under $1K
Is trying to turn the mini into a high end workstation a reasonable premise? How 'mini' is it then ? Is the concept more aligned to both entry level consumer and if dialed up a prosumer or office workstation at the lower middle performance category?
- dual slotted M2 or better storage (RAID capable) and (up to 32GB) ram
- i7 6 (12) core (or i9) processor
- discrete internal GPU (v20, 5500 or ?)
- 5K (or 6K) monitor support
- list price under $1K
Only difference today is with the iMac now ranging past $5000, the high priced headless desktop Mac is pushed to $6000, the wish for an expandable desktop Mac now spans $1000 to $4000 or so. That range was narrower in the past, and the xMac was to fill it.
As has been said, the iMac is Apple's desktop machine. It currently offers a Core i9-9900K, which was top of the line as of 1 week ago, and with Comet Lake just announced, their availability will be low awhile for everyone for awhile. The iMac Pro comes with Xeon W 2100 CPUs and 16 GB GPUs has been available for people who want more cores. The Xeon W 2200 versus the 2100 is bit like Comet Lake is to Coffee Lake, a marginal 5% improvement. Apple isn't in a rush for that.
You can say Apple should use AMD, but they seem to have other plans with their own ARM designs in the near future. You can maybe argue that Apple would be more successful with a cheap expandable desktop, but given that they have no real desire to make gaming boxes and the rise of laptops, I think it would be tough to say a cheap desktop Mac would have made any difference to them. Some users would be happier, but Apple? Not much change.
Also, the graphics performance makes it completely unsuitable to test anything 3D graphics for real. You can verify that it is working on Intel graphics, but you cannot test real performance that end users may want.
I don't think you know what pushing a mini hard for dev actually means.
It's impossible to know, but my personal thinking is that a G4 graphite style Mac which played to its strengths would cut a big hole into iMac and Mini sales. In fact, I sincerely believe that is why one doesn't exist now. Apple just doesn't want to see its margins hit.
After four years without an update, the Mini could have easily been considered a niche and the Mac Pro trash can and the new Mac Pro are also niche machines.
A low end tower or xMac or whatever you want to call it, would definitely sell more than the Pros - provided that - Apple made it competitive with the branded competition.
I definitely don't want an Apple screen bolted onto a desktop Mac after seeing a 27inch i7 slow cook itself to virtual death due to extremely poor cooling mechanisms. That paired with accessibility issues for even the most mundane tasks (like removing dust, cleaning inlets etc) and life extending or preference upgrades (RAM, upgrading screen etc).
At the end of the day it boils down to options and there is a huge hole where the option of this potential Mac should be.
The exact same needs catered to on the 'modular' Mac Pro are shared by a huge swathe of users but without the 'extreme' options. A Mac Pro for the rest of us as it were. Just like it was back in the day of the Sawtooth. No need for $700 wheels.
I agree with other folks who would love to see Apple build a mid range “Component Mac” that has replaceable components like memory, video, audio, and storage. Apple simply does not have anything in their current lineup that appeals to PC hardware enthusiasts, especially those who get all geeked up about benchmarks and having the latest still dripping wet chunk of silicon baking beneath a window air conditioner sized cooler in a gaudy box with blinking lights. If you’re into that sort of thing, you’re not going to find it at the Apple Store, and may never will. It’s not because Apple can’t build it, they can, but that type of machine is obviously in the category of things that Steve Jobs would have put in the “No, we’re not doing that” pile. Does that leave a hole in Apple’s lineup? Absolutely, but it’s a hole that Apple has purposely decided they aren’t jumping into, at least for the past 25 years and for the foreseeable future.
I think Apple builds machines that they feel serve the greatest number of customers for the types of jobs those customers must perform. I also believe Apple views the greatest differentiated value for the greatest number of customers coming from the software. The hardware is just there to support the software and the best way that it can do so is by providing a reliable, secure, and stable platform for the software to run on for as long as possible. Unless you’re a PC enthusiast or gamer the biggest investment in your time and billable hours is probably dominated by the software components and software related contributions that make up your production environment, not the hardware acquisition and upgrade costs. The hardware has to be good enough to not get in the way of the software and not to have recurring repair expenses or, worst yet, downtime.
I have no doubt that there are cases where having the ability to upgrade and reconfigure a machine post-purchase is a huge benefit for an individual or business, but for a large number of individuals and businesses having a reliable and stable hardware platform that just works with very little administrative overhead and serves the needs for several years keeps a lot of folks happy, and Apple’s sales numbers reflect this. For people who want to tinker and tweak their hardware the Windows and Linux world has plenty of opportunities. As an enthusiast you don’t have to choose one or the other, you can choose “all of the above” and be happier for having done so. Otherwise, it should be a data driven decision. No problem.
The tradeoff with getting a Mac Pro would be interesting because at some amount of dollars for a single user, spending the extra money for a single box would be better. Then again, you can get 6 6-core Mac minis for $6k. That's 36 cores to play with.
We have replaced some of the 6/12 core/threads minis with trashcans refitted with close to new 12/24 core/threads Xeon cpus and they both do the job faster, are virtually silent and have graphics performance that is quite a few notches up from the minis (although quite mediocre when it really comes to it). These machines also run virtual machines pretty well.
Seems like your computational needs aren't that far away from needing to get some racks and a server room if your compute jobs are lasting on order hours. The longest thing I do are video transcodes on my 2013 iMac 27. 10 videos can go all night. A 2018 Mac mini would do my H.265 transcodes 10x faster, but I don't do it enough to really need to change anything.
Xcode (re)builds the project with about 1.5 million lines of code in about 9 minutes on the trashcans. The Mini used 16 or so. Visual studio needs 2.5 hours for the Win version.
Xcode has a tendency to randomly rebuild the entire project, when xcodebuild (command line) would do an incremental build. But it can also go the other way.
Giving VS a native PC (or rebooting the trashcan in bootcamp) only marginally improves build times. Versions before VS2019 also don't respond well to increasing number of processors over a certain level. It just spawns off more processes that fight with each other for command line builds.
as subtly implied macOS definitely runs in NUCs (very well too), and eGPU noise levels are often too difficult without headphones. of course the base Mac Pro runs more silent, that seems to be the one advantage.. and sometimes i do wish i had all the money in the world so i didnt have to hear machines.
as many say the ARM-architecture transition may be the only explanation for the state of things.. unfortunately one thing we may deduce from that path is the likely complete disappearance of upgradeable or serviceable parts- the end of xMac possibility all-together. crossing fingers that like iphones and ipads, the new machines are so spectacular that such limitations arent felt anymore - as unless disposable income isnt an issue - they definitely are now.
And yet I was able to get an app with fairly complex 3D models and simple editing of them (3 axis rotations, translation, scaling) working quite well on a 2014 Mac Mini running at 4K (around 30 fps). It was actually our 4K benchmark for machines with low end graphics hardware and it worked surprisingly well with only a few optimizations.
I don't think you know what optimizing your software for average consumer hardware actually means.