Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
More than a few folks have said Apple should stick to their 2013 'script' ( have to dog and pony at WWDC , etc. ). It really isn't beneath them since they have done it before.




If Apple has been running a rigid "just in time" set up then that may not happen. if they have been inventorying ( or making/paying suppliers ) to inventory for a couple of months parts then maybe.

Even at this highly price there is quite likely going to be a demand bubble that Apple supply can't hit for the first several weeks. If there is only 1-2 standard configurations there is a pretty good chance with this market demographic that there will be a big gap between what Apple could have completely pre-built and what is ordered.


P.S. In the 2013 script the "old" Mac Pro had been turned off by this time. The counter on the "just around the corner" is the 'buy' button still being active on the 2013 Mac Pros.

https://www.apple.com/shop/buy-mac/mac-pro

if it was super close Apple could just turn that off. It is boggling why they would even want it on at this point.

If you need it, you need it. I can't imagine they've been selling many Mac Pros these past few years, but it's better than having no option at all in the space.
 
If you need it, you need it. I can't imagine they've been selling many Mac Pros these past few years, but it's better than having no option at all in the space.

The ship times on 2013s are also suspiciously long right now. Doesn't mean that they're going to delay and then cancel orders. But definitely gives the feeling of an assembly line in transition. It shouldn't take a week and a half for me to get a stock Mac Pro shipped to me domestically.
 
I think if Apple was really going to release in December they wouldn't have said fall. They would have just said end of 2019.

I've heard stories about the 2013 Mac Pro launch and why it dragged out so late. I'd be really surprised if the same factors were in play here (it had to do with the custom design, of course.)

It was the screws. https://arstechnica.com/information...moves-mac-pro-production-from-texas-to-china/

And it had to do with manufacturing it in the US, which they are doing again. So, might not be the exact same symptom, but it might have the same cause.
[automerge]1572553996[/automerge]
The ship times on 2013s are also suspiciously long right now. Doesn't mean that they're going to delay and then cancel orders. But definitely gives the feeling of an assembly line in transition. It shouldn't take a week and a half for me to get a stock Mac Pro shipped to me domestically.

How do you even get there on the site. Apple.com -> Mac -> Mac Pro = the presumably 2019 machine. Where is the 2013?

Edit, I see it now. If I search, I get an option between "new mac pro" and "mac pro". The "mac pro" takes me to https://www.apple.com/shop/buy-mac/mac-pro

Very hidden though.
 
The new Mac Pro is such a redesign, I would be scared of buying what is practically a first-gen product. Its becoming obvious, there is some delay in the development. We did see a working system at WWDC 2019, but, that's just maybe one of 20 prototypes they easily manufactured on campus. Either some component is proving difficult to mass-produce or software development bugs are holding it back (performance, stability). Does anyone know if the prototypes at WWDC 2019 were running a beta of Catalina or Mojave?

With the level of security restrictions in Catalina, it might be the cause of the delay.
 
The new Mac Pro is such a redesign, I would be scared of buying what is practically a first-gen product. Its becoming obvious, there is some delay in the development.

Apple historically has eaten up almost all their buffer any time they pre-announce a product. The Mac Pro has languished for so long, they are also tending to announce a little ahead of what they normally do.

But I remember old OS X launches like Leopard, where when Apple finally stopped delaying the release, and announced it would be October 2007, it was in stores on Oct 26th.

We did see a working system at WWDC 2019, but, that's just maybe one of 20 prototypes they easily manufactured on campus.

I'm honestly not sure even Apple would do a lot of manufacturing on campus. You'd want to flush out the manufacturing issues quickly, and it's easier to do that if your manufacturing partner is at least building your boards/etc for you from the start.

But having some equipment on campus makes sense, but I'd wager it is more case design tooling, and the like.

Either some component is proving difficult to mass-produce or software development bugs are holding it back (performance, stability). Does anyone know if the prototypes at WWDC 2019 were running a beta of Catalina or Mojave?

With the level of security restrictions in Catalina, it might be the cause of the delay.

The WWDC prototypes were running Mojave, but Catalina restrictions shouldn't be an issue with the Mac Pro. Apple isn't jumping through hoops for their own drivers, like 3rd parties are.

My personal bet is on logistics, like with the 2013, which got brought down because of the need to source screws that weren't easily available in a US supply chain. Assembling the Mac Pro in the US means ensuring your supply chain in the US can handle the parts that aren't coming over on the slow boat from the PCB manufacturers/etc in East Asia.

You also have the issue that you have to slow boat your parts over before you can start assembly, where with Foxconn, Apple can have the first orders shipping out straight from the local factory to customers via Air Freight, with no slow boat involved. That is going to add weeks if not months to a manufacturing schedule that Apple usually doesn't deal with, except on the 2013 Mac Pro.
 
  • Like
Reactions: OkiRun
Could anyone please elaborate what the practical difference between two Vega II vs one Vega II Duo would be? I don’t get it, except that obviously two Vega IIs would take up both MPX modules.
 
Could anyone please elaborate what the practical difference between two Vega II vs one Vega II Duo would be? I don’t get it, except that obviously two Vega IIs would take up both MPX modules.

I think it depends on what kind of applications you are using and how much those GPUs will be integrated or supported by that particular app that you are using. One good example is Davinci Resolve on the Mac platform. This app uses every GPU it can get and even the eGPU works great, - way above average. Davinci will profit form a Dual GPU for sure. I see the second GPU (or dual GPU) as a better integrated and faster eGPU on steroids, something of that nature.
But in order to make sure it works, one would need to wait till this whole shabang will be released and tested.
 
  • Like
Reactions: chfilm
I think it depends on what kind of applications you are using and how much those GPUs will be integrated or supported by that particular app that you are using. One good example is Davinci Resolve on the Mac platform. This app uses every GPU it can get and even the eGPU works great, - way above average. Davinci will profit form a Dual GPU for sure. I see the second GPU (or dual GPU) as a better integrated and faster eGPU on steroids, something of that nature.
But in order to make sure it works, one would need to wait till this whole shabang will be released and tested.
Hm, yea I get that, my question is if anyone understands if Apps will use the two GPU cores on one Vega II Duo as one GPU or will this still be treated like an eGPU for example, Or absolutely the same way like two single Vega IIs?
 
The MPX module will have an impact on this. A lot depends on how it actually functions and how the "infinity fabric link" really works. According to initial specs, two GPUs can be on one MPX module and four GPUs can be on two MPX modules. FCPX and C4D are clearly "advertised" in this section.
 
Lol, do people really argue about how long autumn lasts? It can‘t get any better 😂
i dont think there is anything else to talk about. Im so desperate for the release, ill start talking about the weather too...Just give me something apple, cmon!
 
Hm, yea I get that, my question is if anyone understands if Apps will use the two GPU cores on one Vega II Duo as one GPU or will this still be treated like an eGPU for example, Or absolutely the same way like two single Vega IIs?

They should appear to the systems as two GPUs. It isn't like an eGPU in a substantive way at all. If had two modern PCI-e v3 ( or v4 ) GPU cards stuffed into a legacy MP 2010-2012) they'd appear as two GPUs at reduced (from native) bandwidth; there is nothing particularly "external" about that at all.

Apps may introduce a layer that makes Inifinity Fabric connected GPUs present like a "virtual" GPU, but I don't think Apple's software is going to do all that 'work' for them automatically. ( it will be a non uniform access virtual GPU so need to know what you are doing anyway. ).

Two single Vega II has some upside in that each GPU package gets x16 PCI-e v3 worth of bandwidth to the host system. They also should be connected via Infinity Fabric so should have same "backend" bandwidth as a Duo card has between its packages. Apps the push the envelope in RAM to VRAM bandwidth requirements would do better. ( Apps that don't probably won't see much of a difference. ).
Sustained peak loading for long periods of time probably run better on two Vega II 'solo' as spreading the heat dispersal over a larger area. Also more TB/video out ports if need that kind of fan out.

If using the default Mac Pro host TB ports ( not the one on the video cards) two Vega II cards conceptually could remove the gap that a single Vega II 'solo' cards leaves. ( Depends if Apple provisioned two ways to fill one of those TB pair's DisplayPort feed. If out to trim some cost they may not have done that. The DP video feed pins out of MPX Bay 2 may be a dead end. ).

The Duo saves space (and access to two double slots for mainstream cards or internal 'real' hardware RAID on HDDs ). It fully provisions the host TB ports (if using those is a requirement). It should have more "backend" bandwidth ( two intercard infinity Fabric links). So if have a virtual GPU layer then can loop all four GPUs into that computational space. For example if the problem needs 100GB of HBM2 footprint then two Duos would work better. Slower VRAM fill time but if computation last much longer than fill then the bigger backend bandwidth will pay off.
 
There may have been several hundred (or more) 2019 Mac Pros already assembled in China. Apple then talked with U.S. regulators for import tax reductions, depending on if and when the final assembly has been re-located to Austin, Texas. Using the same identical parts as the China-made products, just the final assembly of those same parts now occurring in Texas. Whether any of those parts are actually fabricated in the Austin factory, such as the new Mac Pro case for example, seems somewhat unlikely.
 
....
The WWDC prototypes were running Mojave, but Catalina restrictions shouldn't be an issue with the Mac Pro. Apple isn't jumping through hoops for their own drivers, like 3rd parties are.

How does Apple demo new Catalina features on Mojave ? A WWDC convention to introduce a new OS and do substantive demos with the old , last years OS. Really?
 
....

I've heard stories about the 2013 Mac Pro launch and why it dragged out so late. I'd be really surprised if the same factors were in play here (it had to do with the custom design, of course.)

My guess is next week they'll start taking orders and this speculation is going to look really silly.

If it is next week how many sales of the old system would happen in a week? Again "on" at this point is boggling. What Apple is looped into in term of service lifetime support at this point is a bigger need to start the Vintage/Obsolete countdown clock as soon as they can. Apple will need parts , replacement systems for years for something that is already old and highly custom.

Unless Apple has screwed their resellers by packing their partners' inventories with a boatload of Mac Pro 2013s , it doesn't make much sense.


I've heard more than a few major orders have been placed for 2013s. Customers who have custom rack systems, or dependencies on things that T2 is incompatible with have been doing last minute ordering and stocking up. Plus Apple probably has inventory to clear anyway.

Yes 'placed' as in past tense. When the 'heads up" went out in June and folks are still sitting on the pot not doing something the third week or so in October .... when would they get something done? I'm not talking about Apple turning it off back in June, July, August , September, .... we are in November now. If going to ship in November why is it still on. XServe got a 4-5 month notice. The Mac Pro 2012 was axed from new order in October 2013 and didn't ship new until December

If using the same factory facilities it makes even LESS sense. The Mac Pro 2013 has been a zombie product for over two years. Apple did their first "dog ate my homework" session back in April 2017. They pragmatically buried the order link in June on the website. If the factory is deep on a working smoothly ramp MP 2019 now, then it would be time to call it quits.

If they were having problems yet again I can see holding off. Otherwise it is just highly dubious dogma and sad sales barker shill to still having it on.
 
  • Like
Reactions: ssgbryan
They should appear to the systems as two GPUs. It isn't like an eGPU in a substantive way at all. If had two modern PCI-e v3 ( or v4 ) GPU cards stuffed into a legacy MP 2010-2012) they'd appear as two GPUs at reduced (from native) bandwidth; there is nothing particularly "external" about that at all.

Apps may introduce a layer that makes Inifinity Fabric connected GPUs present like a "virtual" GPU, but I don't think Apple's software is going to do all that 'work' for them automatically. ( it will be a non uniform access virtual GPU so need to know what you are doing anyway. ).

Two single Vega II has some upside in that each GPU package gets x16 PCI-e v3 worth of bandwidth to the host system. They also should be connected via Infinity Fabric so should have same "backend" bandwidth as a Duo card has between its packages. Apps the push the envelope in RAM to VRAM bandwidth requirements would do better. ( Apps that don't probably won't see much of a difference. ).
Sustained peak loading for long periods of time probably run better on two Vega II 'solo' as spreading the heat dispersal over a larger area. Also more TB/video out ports if need that kind of fan out.

If using the default Mac Pro host TB ports ( not the one on the video cards) two Vega II cards conceptually could remove the gap that a single Vega II 'solo' cards leaves. ( Depends if Apple provisioned two ways to fill one of those TB pair's DisplayPort feed. If out to trim some cost they may not have done that. The DP video feed pins out of MPX Bay 2 may be a dead end. ).

The Duo saves space (and access to two double slots for mainstream cards or internal 'real' hardware RAID on HDDs ). It fully provisions the host TB ports (if using those is a requirement). It should have more "backend" bandwidth ( two intercard infinity Fabric links). So if have a virtual GPU layer then can loop all four GPUs into that computational space. For example if the problem needs 100GB of HBM2 footprint then two Duos would work better. Slower VRAM fill time but if computation last much longer than fill then the bigger backend bandwidth will pay off.
Great reply, thanks!
 
If it is next week how many sales of the old system would happen in a week? Again "on" at this point is boggling. What Apple is looped into in term of service lifetime support at this point is a bigger need to start the Vintage/Obsolete countdown clock as soon as they can. Apple will need parts , replacement systems for years for something that is already old and highly custom.

Unless Apple has screwed their resellers by packing their partners' inventories with a boatload of Mac Pro 2013s , it doesn't make much sense.

I dunno. If they're clearing existing inventory, and people are still ordering (which I know some people are), why take them down?

They're just going to have to move any remaining inventory to the refurbished/clearance side anyway, which has the same three year warranties.

They'll sell the old MacBook Pros right up to the moment they announce the new ones. I kind of feel like this is a non-issue. We're trying to use the 2013 Mac Pro as a benchmark when that was almost the only Apple product launch that worked that way.
 
Interesting possibility - although Apple's not a cosponsor of Adobe MAX - perhaps deliberately, if they're trying to be secretive like they usually are. Mac Pro, MBP or both could make sense at Adobe MAX (or a press release announcement during the conference).
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.