The timing did look that way, but probably a bit less dramatic, since the roundtable happened totally out of the blue, instead of a meeting a hard schedule like a WWDC announcement on stage, so there really wasn't a pushing need for them to say anything at that specific time - they choose to say it there. I felt that the moment they must've already had at least chosen a prototype, have a pretty firm idea of what it should do and what it doesn't. The leading 2 years are for actual hardware engineering and logistics, especially considering if it will remain assembled in the US plant.
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.
Trends toward modular Mac Pro
- Thread starter MiJuConcept
- Start date
- Sort by reaction score
I don't buy that being anything other than a reactive crisis meeting, timed solely to try to head off the brand damage that was being caused by the near constant feed of "wtf is Apple doing in the pro markets" by even their closest media puppets, like Gruber, The Loop, Ritchie etc. I really don't buy the idea that they had anything remotely beyond "there will be a pro display, and a machine without a built-in display" as an idea on a post-it note, prior to deciding to hold that meeting.
Last edited:
Your take may be closer to the truth, to be honest. But then what I meant to convey is not an educated guess, but more like a wishful thinking, that "those guys can't possibly be this assbackwards to not even have a ****ing prototype at least floating around out of brainstorm sessions to dick around with". If the iMac Pro was seriously the answer that they thought was enough to satisfy, then they were literally just out of their minds.
A Self-proclaimed leaker I name "DarkNetGuy", stands the Modular Mac Pro to come next quarter, build on 4 Modules, CPU+Memory on the front module, and two side GPU module with integral independet cooling (a blower and heath-pipes), rear module for I/O and a single 1200W PSU using new smart power saving technology being a full order magnitude more efficient at idle load than the best equivalent PC power supply, Will be moreless shaped as a cube with vents on the top and bottom, FULLY SEALED (Apple will allow upgrades but will require to visit Apple centers for RAM, GPU, CPU, SSD upgrades), somehow will resemble both the cube mac and the trashcan, not departing from current Apple's Design Language,
Apple is testing prototypes with Intel Xeon-W and AMD ThreadRipper/Epyc
Didn't provide more details on specific GPU model or brands, but easy to guess GPU will be in proprietary form factor.
Consider it with a buldozer-sized load of salt.
Apple is testing prototypes with Intel Xeon-W and AMD ThreadRipper/Epyc
Didn't provide more details on specific GPU model or brands, but easy to guess GPU will be in proprietary form factor.
Consider it with a buldozer-sized load of salt.
It probably would have been easier to start the iMac Pro series in 2013... rather than spend time on the whole new trashcan Mac Pro (with all its problems)
At least by now it might have gotten a revision or two (unlike the tcMP)
This thread is an eye-opener. So what's the future of workstation Macs?
Since Apple is a one-stop shop... you rely on them (and only them) to provide the software and the hardware. But lately the hardware is an issue.
If it's true that Apple is only offering this supposed "modular Mac Pro" as a last-ditch effort... is this really a platform and company you can trust?
It's sad that there are so many "Waiting for...." threads on here.
Meanwhile... I think we can all assume that HP will update the Z-series next year... and the year after that. Or at least offer something comparable.
But with Apple... who knows! In a few years they might only sell laptops and all-in-ones.
Is MacOS so great that you're willing to put up with only those limited form-factors in the future?
Someone correct me, but I recall one of the in-depth looks at the iMac Pro more or less painted it as largely a re-organisation of the same basic internals as those in the 2013 - there's a lot less "new" development in that machine, and a lot more "build it from the parts bin" than is generally marketed.
For example, I'm a pissed off former / current Aperture user - my next photo migration will be to Capture One, or Affinity's DAM or Skylum's Luminar DAM, all of which are cross platform. Why would I buy a mac - literally the only thing I can think of is Airdrop from my iOS devices, but maybe I'd just get the cheapest mabook, or a mini for that, and just use it as a dumb airdrop receiver to a shared folder on my network, that could be accessible from a shared folder on a PC workstation.
There's a house of cards, insofar as unless you're using FCPX or Logic, most everything else is cross platform already, and probably in the Pro context, you're not interacting with the parts of the Mac experience that are the thing that makes the Mac "nicer"
This gets even more apparent when you get into VR where nothing of the user experience is furnished by the host OS, (which I know I've probably been an insufferable born-again about on here of late - hey, when you get the holy fire...) but which I honestly believe represents an existential threat to the future of the Mac as a creative desktop computing platform, because creative apps that use a flat 2d screen as a UI paradigm are going to be a legacy environment, one that iOS will eat. And until such time as GPUs can create a reality indistinguishable from the real world, rapid replacement of GPUs is going to be the defining characteristic of a non-legacy "pro" desktop system.
If it has standard retail-upgradable GPUs from both team red and team green, at least two so that for VR they can make a turnkey dedicated GPU per eye the differentiated factor, then yes.
If they're going to do their usual strategy, which is to design things with a primary goal of reducing the risk that an older system could compete with a newer system (eg making eGPU a TB3-supported exclusive to age out the older systems), and make it a sealed box that has no ability for the customer to take it on a route that suits themselves, even if that's contrary to Apple's larger corporate goals, then no. They've burned up too much good will (eg Macbook Pro) by raising prices to give Pros things they don't want, while refusing to make the compromises (thicker body, for larger batteries, a more reliable keyboard and non-soldierd ram) to give them the things they do
The new z4 packs a config with a core i9 and dual PCI3x16 slots, plus a couple of x8 slots. That's a super, super sweet spot for a huge part of the creative market that doesn't need Xeons, assuming the power supply and space constraints can feed a pair of dual 8 pin GPUs.
I for one fanaticise about some bolshie European competition regulator handing out an edict that OS makers must be required to offer their operating systems, and the low level source for compatibility purposes, to any systems maker who wants it, under FRAND licencing terms.
Last edited:
The Z4 is available with 465, 750, or 1000 watt power supplies.
Like I was sayin'...
[doublepost=1520352704][/doublepost]
Depends on the design, but probably, yeah
[doublepost=1520352896][/doublepost]
Yeah, what you said.
LOL
Nah, same size or a little bigger if necessary for more PCIe slots would be just fine with me.
Last edited:
AMD Epyc vs Xeon-W Geekbench
Consider the Ryzen Threadripper 1950X has a Single core (stock clock) score of:4900pts, so an special Epyc SKU could appear with this clock as stock too. (Epyc ~== Threadripper)
Code:
Single Core Xeon W-2150B: 5310 pts (1 of 10 cores)
Multi - Core Xeon W-2195 : 48831 pts (18(36) cores)
Single Core Epyc 7351p : 3747 pts @2.9ghz (1 of 16 cores)
Multi - Core Epyc 7601 : 94358 pts @2.2ghz (32(64) cores)Threadripper and EPYC also have more pci-e lanes. Also with EPYC they can maybe Do an ADD an module CPU+RAM.
Say (MAIN BOX)-(pci-e BOX) or (MAIN)-(2th CPU)-(pci-e box)
Say (MAIN BOX)-(pci-e BOX) or (MAIN)-(2th CPU)-(pci-e box)
So, if your jobs need more than 36 threads, the AMD could be a win.
Let me catalogue all of the posts complaining that most software can't use more than a handful of threads.
In other words, better single thread performance can be much more useful for workstation loads than better multi-thread performance. This is especially true if your jobs can use a handful or two of threads. Running 5 to 10 threads at near single thread performance is a sweet spot for lots of uses.
But of course you know that. So why would you post a misleading message about AMD threading? Why?
Last edited:
After Apple's recent track record with top end professional hardware do you think that people who could afford to drop the thick end of 5 figures on Apple hardware would still choose to do so after the 2013 Mac Pro debacle, the axing of Aperture, and FCP (with FCPX taking years to get back to a semblance of what FCP was), the death of macOS server, and so on.
In a way, I think by allowing eGPU Apple are offering professionals the ability to upgrade something but surely as with their philosophy on Mac App store pricing (lower prices, more unit sales) the bottom of the range has to be carefully judged.
In a way, I think by allowing eGPU Apple are offering professionals the ability to upgrade something but surely as with their philosophy on Mac App store pricing (lower prices, more unit sales) the bottom of the range has to be carefully judged.
Its true most Apps need (if possible) to be re-written to make gains on more than 4 cores, but there are few apps updated to do so, as FCP.X which is now optimized for multi-thread processing (on both CPU and GPU offloading).So, if your jobs need more than 36 threads, the AMD could be a win.
Let me catalogue all of the posts complaining that most software can't use more than a handful of threads.
In other words, better single thread performance can be much more useful for workstation loads than better multi-thread performance. This is especially true if your jobs can use a handful or two of threads. Running 5 to 10 threads at near single thread performance is a sweet spot for lots of uses.
But of course you know that. So why would you post a misleading message about AMD threading? Why?
There are Applications like CAD where SMP dont have a chance (unless some one mabages to do an long due rewrite of the omnipresent CGAL library (a WIP).
But considers now languages as LaScala are focused on SMP-safe/optimal coding (functional programming),
A Cad User should choose (ok CAD has no big Mac user base) an iMac Pro, but those on ML an AMD based Mac with 32 cores should look more attractive.
Last edited:
And a Mac with tens of thousands of CUDA cores would be even more attractive.
Do you seriously push for CPU cores for ML. Seriously?
I know, but always there are some algorithms that also require some SMP CPU processing to support the Neural Networks.
Yes, that's why as a rough measure we target around 300 CUDA cores per CPU thread. (600 CUDA cores per CPU core)
I agree. To this I will add the following, there are only 4 computer designs.
1/ Modular / scalable / cluster
2/ Highly mobile / portable / companion devices / iPhones / laptop
3/ All in one / TV / Compact screen / Touch device / Tablet / iMac
4/ Unique purpose / Raspberry Pi / Universal gadget / Programmable / Apple TV / Homepod / Workstation
[doublepost=1520393204][/doublepost]
Apple took this approach in 2011. It was delayed a year by the fan.
But it should be seen as a reflection of American design and engineering ability. Why they stopped developing this platform is only something Apple will know. I dont think there will be one critic if the machine can support Xeon Gold CPU and Vega GPU.
[doublepost=1520393447][/doublepost]
And a Mac with tens of thousands of CUDA cores would be even more attractive.
Do you seriously push for CPU cores for ML. Seriously?
Take into consideration that some highly valued numerical analysis tools are still reliant on cores not GPU. The GPU side appears to favour fuzzy logic but not always hard coded data extraction.
Last edited:
you're seriously suggesting the design of the fan was why the 2013 was actually supposed to be a 2011 release? let me get this straight - a whole product is able to be held up by effectively 3 years by a fan (2014 wide availability), and that wasn't cause to bin the entire thing as a bad idea? That's supposed to be a positive?
Again, if it's not patently obvious - the problem with the 2013 for the entire segment of customers who don't want an iMac Pro, wasn't that Apple didn't update the parts every year, it's that Apple customers couldn't update the parts every year after purchasing. A 2013 with Xeon Gold and Vega, is still a "Nuts & Gum together at last" disposable appliance at workstation prices. Noone else in the workstation market is doing that as their product philosophy, for good reason.
In Wikipedia it states very clearly the SR71 replacement does not fly. However, it was doing some amazing Mach 3 or Mach 4 trials over Sydney Australia several times a month till it was mentioned in online media and then it stopped. Having seen this non-flying plane it started a design inspiration.
SR-71 inspired Jet Engine design was first offered to Shuttle Corporation. They held a design office in Germany. Shuttle were so certain no "commoner" could overcome the heat, size, build limitations that they flatly refused to consider or discuss it. Even the offer to drive 200km for a five minute meeting during lunch was not enough to provoke a meeting. So in reply the designer said he would throw it to Apple to prove someone could build it. Depending on how you look at it, Shuttle dodged a bullet or they missed the bus.
In the original design brief it was mentioned very clearly the fan is an enormous issue. That was mentioned on the first day. Apple finished all aspects of the Machine except the fan. I wish I can find the journalism that pointed to this fact. I know it was written in media. Apple did not offer to collaborate or ask for any assistance on the fan. Of course there was a fan design waiting for Apple but they never asked.
Anyhow it still has poetry. The delays coincided with 50th year (49th anniversary) of the first flight of SR-71. Media samples went out on Dec 19 and first shipments arrived on Dec 21 and 22 as per the historical homage - I'm sure it was accidental. The jet engine design won global regard for design purity. Red Dot and induction into MOMA NYC. This is part of technology culture.
This only proves one thing. Apple will listen to "commoner" and many other companies will laugh at you. If you will take the time to design things for them, Apple will take the time to review and consider. The 2013 MP experiment has helped Apple to "loosen up" a little and think more openly.
I will not be surprised if the modular Mac pays enormous respect to NEXT Cube and Steve Jobs' vision.
Last edited:
No dude, if you "design" things for Apple, they will run as far away from them as they can, to avoid accusations of theft, to avoid being seen as so weak they have to crib ideas from the internet, and to maintain the mystique that they're some font of originality.
Apple is a performance art project, that lives for the surprise and reveal - everything they do is secondary to, and for the purposes of, achieving that goal.
The only reason they're giving the impression of being "open" or "listening" is that they were facing an existential threat to what was once a core market, and for which their pride was at stake.
You really deserve a beer or a wine for that insight. Be sure to claim it. I don't agree with all the points raised but there is also some truth in how Apple is now compared to 25 years ago. I really don't see anyone casually dropping in to visit them now.
Apple now is almost exactly like Apple 25 years ago - when Spindler was running the company, and half-baked products were being thrown out for markets that didn't want them, core software was laughably fragile, and the creative segment was transitioning away from MacOS.
We went through a nice phase where NeXT was effectively wandering around in Apple's skinned corpse, building the best places to use standards, and now we're returning to the worst of old NIH-averse Apple.
So your leak is suggesting a machine with anti-user locking to prevent self-servicing, with prospects of component upgrades dictated by Apple not by open market, has inherit limit of utilizing any offshelf GPU which is the number one demand in multiple professional fronts, which also happens to be shaped as a monument that reminds us the company's past design failures?