But if you had a 53k Mac Pro you probably wouldn’t be using it to develop VisionOS apps. Anyone who’s spending that kind of money is probably using it for some kind of specialised workflow which needs 1500gb of RAM and which probably won’t change much in the future, maybe something like processing 16k video, complex CG rendering, simulations etc
It still hurts the resale value of the thing, makes it less useful so more likely to end up junked, and even if the original owner “got their moneys worth” out of their purchase, this is still a relatively new computer sold to pro users who Apple should support better for that kind of price tag.
You can still ignore the fact it's a Mac and install Linux on it. With 1.5TB of RAM, it'll be really good to play with large-ish datasets or spin up tons of VMs up to the point storage becomes the bottleneck.
I just saw a listing on ebay (https://www.ebay.it/itm/115853434151) selling a machine with 1TB ram and 4x "Xeon Platinum 8160" (24c/48t each) for ~8'000 euros.
The motherboard should support up to 12TB ram.
And it's a plain regular HP ProLiant server, so it's not going to give you any hassle when running Linux. Heck, some distributions (likely RHEL or SuSe) might even be certified for use on that platform.
So to sum up: for people with money and wanting to play on large-ish datasets, a second hand mac pro is really a poor choice on how to spend their money.
Note: i didn't even look that hard for a better offer or a better-priced listing.
OTOH, the HPE is much noisier - you don't want it under your desk. And the Mac still looks cooler.
But yes - good luck selling a 1.5TB Mac to anyone not doing video production (and depending on the platform) for anything remotely close to retail prices.
Apple probably realized they only sold approximately three fully configured machines and decided 192GB should be enough for anyone.
I agree it's disingenuous to take a fully-spec'd system to make a stronger point here, it sounds silly on the surface, but as you said, those machines are purchased for one job and probably do that 24/7 for a few years until retired.
I think it's a much worse state of affairs that a $6000 base Mac Pro from a few months ago, which was marketed to developers, cannot do this development. Arguably the Mac Studio was a better targeted machine for those developers, but even that only takes us back a year or so. A $6000 machine for development should last a while – even for professional use where more expense can be tolerated, it should still have 3+ years of support.
A 4 year old device on a different CPU architecture. The Mac Pro wasn’t a device aimed at developers, it was aimed at creatives. This title is grandstanding.
Would be a good machine for data scientists as well, but those usually prefer cloud instances that reside near the data rather than bringing the data to the workstation.
Different archtiecture is utterly irrelevant. No one is developing watch apps on a watch, all development is cross develooment, even if the architectures are similar.
There is no actual technical limit here, only a typically inconsiderate policy one.
Any architecture may produce any collection of bits for any other, and even run them and test them. Every m1 user sees this every day when they continue to run intel apps.
The wording of the title is obvious flame bait as evidenced so far in the comments, while the core point “AppleSilicon machines required for VisionOS development” would be a much more productive thing to talk about. I wish OP would’ve had a bit more maturity, but hopefully @dang can help fix so we can get back on track.
But that does not highlight the problem, and makes the requirement/limitation seem reasonable.
The whole point is to distill the problem to it's essense and make it impossible to ignore.
Obviously there are various reasonable sounding reasons how such a state of affairs comes to be. The point was that regardless, at the end of the day, there is an absurdity here, and it is not "I assumed my 50k Apple computer would be usable for Apple development." There is no hard tecnical limit to excuse it. It's completely fair to call it out.
For me it's less surprise and more disgust. Well okay, they did say as much, so I'll give them that, but their perpetual march of obsolescence isn't something we should shrug away.
As someone in Windows world, that is extremely weak long term support. I would reasonably expect decade or two of software support. Like from any good OS.
I'm sorry for the click-bait title, comments are right, it's misleading to compare the abilities of that machine with the ones of M-series machines. They are different machines with different purposes.
30 comments
[ 3.0 ms ] story [ 76.9 ms ] threadOnly if more than a tiny sliver of the resale market for that product is for VisionOS development (a market segment that does not yet really exist).
The motherboard should support up to 12TB ram.
And it's a plain regular HP ProLiant server, so it's not going to give you any hassle when running Linux. Heck, some distributions (likely RHEL or SuSe) might even be certified for use on that platform.
So to sum up: for people with money and wanting to play on large-ish datasets, a second hand mac pro is really a poor choice on how to spend their money.
Note: i didn't even look that hard for a better offer or a better-priced listing.
But yes - good luck selling a 1.5TB Mac to anyone not doing video production (and depending on the platform) for anything remotely close to retail prices.
Apple probably realized they only sold approximately three fully configured machines and decided 192GB should be enough for anyone.
I think it's a much worse state of affairs that a $6000 base Mac Pro from a few months ago, which was marketed to developers, cannot do this development. Arguably the Mac Studio was a better targeted machine for those developers, but even that only takes us back a year or so. A $6000 machine for development should last a while – even for professional use where more expense can be tolerated, it should still have 3+ years of support.
Would be a good machine for data scientists as well, but those usually prefer cloud instances that reside near the data rather than bringing the data to the workstation.
There is no actual technical limit here, only a typically inconsiderate policy one.
Any architecture may produce any collection of bits for any other, and even run them and test them. Every m1 user sees this every day when they continue to run intel apps.
(yes I know this person in European and they your decimal point as thousand separator)
And was only kept around at all was to give more time for legacy users to migrate.
You can buy a Mac mini which is a great development machine for < $1000.
The whole point is to distill the problem to it's essense and make it impossible to ignore.
Obviously there are various reasonable sounding reasons how such a state of affairs comes to be. The point was that regardless, at the end of the day, there is an absurdity here, and it is not "I assumed my 50k Apple computer would be usable for Apple development." There is no hard tecnical limit to excuse it. It's completely fair to call it out.
If it looks like a duck, and quacks like a duck, it right well might just be a duck.
You need to write a medium.com rant and submit that :-)