If Apple decided that the MacPro gets an M2 chip, they may have just removed it from the 2 year goal. The MacStudio Ultra is powerful enough to where it probably doesn't matter if the MacPro takes an extra year to be released.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Do y'all think apple will meet their 2 year AS transition deadline?
- Thread starter mactinkerlover
- Start date
- Sort by reaction score
Am I remembering right that you're in bioinformatics of some sort? All I remember for sure is that you work with RStan models with your job, since it's not all that often I run into someone else (online or otherwise) that does Bayesian modelling. Now that Stan supports multi-threading chains I can only imagine how fast you could get a big model to run on a top-teir supercomputer.
I'm in cognitive science research so I'm mostly modelling behavioural reaction time/accuracy data with GLMMs. For heftier brms models my 14" M1 Pro MBP outperforms our lab's ~4 year old dual Xeon server by a clean factor of two, but hey, the RAM can't be replaced so it's not suitable for workstation use
If they do release a Pro, I suspect it will be a 1U packaged M2 Ultra (Quad option for +$3K) with two lateral card slots such that cards can be installed horizontally on either side, to fit the enclosure. Professionals will have to find some other device with which to grate their cheddar.
Am I remembering right that you're in bioinformatics of some sort? All I remember for sure is that you work with RStan models with your job, since it's not all that often I run into someone else (online or otherwise) that does Bayesian modelling. Now that Stan supports multi-threading chains I can only imagine how fast you could get a big model to run on a top-teir supercomputer.
I'm in cognitive science research so I'm mostly modelling behavioural reaction time/accuracy data with GLMMs. For heftier brms models my 14" M1 Pro MBP outperforms our lab's ~4 year old dual Xeon server by a clean factor of two, but hey, the RAM can't be replaced so it's not suitable for workstation use
I’m in linguistics and I don’t work much with Stan directly (but I do rely on models developed by my colleagues)
Most of the stuff I do is on much lower level using hand-developed algorithms dealing with pattern discovery and brute-force tree walks. But yeah, Apple Silicon has been fantastic for data processing, the performance is stellar. I have to chuckle a bit every time someone mentions Cinebench as example of Apple underperforming when my data processing pipelines on M1 are 2-3 times faster than an Intel based desktop.Most people don't use dGPU's that expensive. I don't even have a medium tier dGPU in my windows machine and the machine supports 11 monitors. (3 from the processor, 8 from the dGPU).
I would say so, I would venture to guess the Mac Pro's design and chips were done a year ago, they are just releasing in order of importance to their overall business and what lines needed it the most.
Err....I think you should check out @leman's past posts. He seems to know a hell of lot more than you appear to....
You don't sound like someone who is actually an IT professional, but a hobbyist. Nothing wrong with that, and it was my interest in computing as a hobby that led me to study computer science at university and have a 30 year career building enterprise software and hardware solutions. Feel free to correct me if I'm wrong, and post your credentials for us to inspect.
I encourage your interest in computer hardware, but you do need to realise that the choice of hardware and platform is only a small part of the overall considerations and cost of system implementation. The cost of software and people to write it, use it & maintain it is a far greater concern in most cases.
The choice of computing platform depends on the task and users. End users, developers and system compute/network/storage platforms are all likely to be different. e.g. an inventory control app running on phones & tablets, with software developed on Linux workstations with specialist PCIe cards, data entry and management done on iMacs, and running the back-end on hybrid public/private cloud infrastructure combining AWS/GCP/Azure services joined to customer data-center legacy systems.
In other words, you choose the tools appropriate for the job!
Last edited:
Onprem = on site, not remote hosted
AD server = Active Directory server. A Microsoft server tech that lets admins manage which devices are in the network an which are not.
Not sure how that is relevant to this discussion, though.
I believe Apple will release a M2 variant maybe dual this Fall for Mac Pro replacement! So be patient young Palawan!
And it's not necessary for that function, but it is fairly common. I prefer to have my firewall control what should be on the network or not.
Not really since you could fit a 27” iMac for the same as a Apple Studio Display with a bit more cash since it’s a 2 in 1.
AD is more about user identity management than device management. It identifies who can log on to devices on your network and what they have access to.
On prem typically means in a companies own data center on servers managed by that companies staff i.e. not hosted by a third party in the cloud. Microsoft licenses AD for on prem use but would prefer you to use Azure AD for your identity management.
Funny you mention that:
All this has happened before, and it will happen again...
Or, just a slotless M? Ultra in a 1U rack mount. Add a rack mount TB4-to-PCIe backplane if you want slots.
Question 1 is, will Apple ever support regular GPUs on Apple Silicon? If not, then Thunderbolt can probably cope with the lower bandwidth requirements of other types of PCIe card (and Afterburners are probably redundant with Apple Silicon). Question 2 is whether there's enough demand for >512GB RAM (the likely limit of a future "Ultra" with 8 LPDDR5x chips) or whether unified RAM and very fast swap would suffice.
This seems to be a more appropriate use of the Apple Silicon building blocks we've seen so far - the alternatives would be (a) for Apple to make some sort of ARM-based Xeon-W-a-like with extreme RAM and PCIe capacity that could drive a 2019 Mac Pro like machine, (b) go down the NUMA/clustering route (adding a suitable interconnect to a future M? chip) or (c) if it's just the RAM issue, maybe add some external RAM as ultra-fast swap space.
Trouble is, although the demand for a Mac Pro-like machine may be there, Apple has had a patchy track record, since about 2010, of keeping it up to date, and has shown signs of much preferring the "appliance" concept: we've had the initial neglect of the original cheesegrater & discontinuation of the XServe, the trashcan concept, the lack of a trashcan replacement for years, the iMac Pro (which I strongly suspect was intended to be the new 'Mac Pro'), the very belated 2019 Mac Pro - followed 6 months later by the announced abandonment of Intel, then the Mac Studio which is very much the Trashcan Mk2 (except maybe done better). Now, Mac Pro users are in limbo again with no idea whether the future holds an ARM-powered Big Box 'o' Slots or something more "courageous" (which would also be a better showcase for Apple Silicon).
I suspect, though, that is currently uneconomical (for Apple) since they're the only people using 5k@27" displays. It was always a bargain to get such a good display in a lower-end <= $2000 iMac and I guess it couldn't last.
The Mac Studio + Studio Display is in the same price/performance ballpark as the top-end 5k iMacs and the iMac Pro - and an even better deal if you prefer a cheaper (4k) or more expensive (Pro XDR) display. The 24" iMac will also have taken a bite out of the market for low-end 5k iMacs.
Wow, interesting. I still doubt Apple will do it because they want to be done with Intel, but I think it'd be a truly genius "pro" move.
Yeah, the Intel boat has sailed along with everyone that REQUIRES Intel to do their work.
Yeah, unfortunately Apple doesn't care. They'd rather lose customers than have to keep supporting Intel for longer. A Mac Pro with an Intel chip would mean macOS would have to support Intel for another decade and there is no chance that's happening.
They’ve been slowly whittling away the users that don’t want what Apple’s selling for quite awhile. The next Mac Pro and the current Mac Pro really only have to have 1 thing in common, execute macOS and macOS application faster than any currently selling Mac. Everything else is up for Apple to (re)define.
No, they will not make it within the original 2 years time frame. However, I expect the final announcements by end of this year:
- MacPro with an M3 UltraMax processor to be announced in Nov, available next year - they need the TSMC 3 nm process to get more CPUs / GPUs on the dies
- MacMini to be replaced by an M2Pro or M2Max processor, mainly to plug in more monitors and to some extend for better performance
- iMac 27+ to be announced later - I still expect an all-in-one design (at least I like my old iMac 27'', however nobody else apparently since the ebay auction is stuck...)
I’d agree expect your gonna see M2 Pro and M2 Max/Ultra/XXX naming. Apple has done this the A8X was a smaller process then the A8 in the iPhone if I recall correctly, I could easily see apple just being like yes our most powerful M2 chips are 3nm allowing for “insert marketing lingo” over M2 which is absolutely still a fine chip on the 5nm node.