If you want to discard both the Tim Cook emails and additional corroborating evidence, go ahead.
But unlike a lot of the rumors, there actually is corroborating information. Definitely more substance than usual for Apple rumors regarding the MP.
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.
What does 2013 mean?
- Thread starter yuma
- Start date
- Sort by reaction score
But unlike a lot of the rumors, there actually is corroborating information. Definitely more substance than usual for Apple rumors regarding the MP.
If they wait till next year to put Sandy Bridge in, ..... words fail me . ...
(but I don't put even that past Apple)
Here is the MacWorld article about the Tim Cook email to Franz. MacWorld states that Apple did confirm that this email was from Tim Cook.
http://www.macworld.com/article/1167247/cook_apple_planning_professional_mac_for_2013.html#lsrc.twt_jsnell
Whatever socket and chipset Intel ships with integrated Thunderbolt will be what they use if they continue to use Xeon at all. Everything is up in the air. Only thing we are fairly sure of is that a "Desktop Mac" will be announced that will not be an iMac. Meaning presumably "headless".
I think they're waiting for 10.9 (or 11), or whatever the next OS is called. They will do something with the software that will allow them "greater control" over the user experience and the user "less control", kind of like the iPhone and iPad. (edit: re control, not re how the OS operates, although I'm sure there will be more of that too).
Perhaps no more apps except through the app store as part of it?
Perhaps no more apps except through the app store as part of it?
Ivy Bridge EX is for 4 socket severs with more than 10 cores per package (around 15). The Xeon E7 line up. No, it won't be coming to the Mac Pro. There was no Sandy Bridge E7s because effectively they needed the process shrink associated with Ivy Bridge to crack 10 core economically. Westmere E7's already maxed out at 10 cores.
Even Ivy Bridge EP is not particularly likely. It may not show up until very late Q2 2013 or perhaps Q3. It won't "add" anything as the socket is exactly the same and the supporting chipset is exactly the same ( there is no "free" USB 3.0 upgrade to be had by waiting). Apple is already late. They'd likely be only even later if they wait on Ivy Bridge EP.
"Waiting for Ivy Bridge" isn't really some 'get out of jail free' card for the Mac Pro design. Likely looking at perhaps 10 core max (versus 8) in the E5 2600 v2 offerings and maybe 8 core max ( versus 6) in the E5 1600 v2 offerings. That's about going to be the major difference. No USB 3.0 magic . No Thunderbolt magic. Just a shrink and some minor architectural tweaks (slightly faster memory DDR3-1866 ).
----------
Thunderbolt is unlikely to be merged into Intel CPUs or core chipsets for a long time. Since the TB controller has placement constraintst to the physical port(s) of less than 2 inches. That rasies problematical issues because that will force the integrated functionality to also be moved. Either would have to move the CPU to the edge of the computer or the core chipset. Neither one of those is very desirable in space constrained laptop designs. It is also completely unnecssary integration for server/workstations which don't have tight motherboard space constraints.
As a discrete controller Thunderbolt is just for for the next 2-3 years. But the time it gets integrated the whole computer will likely be smaller so placement is not a issue. A whole System on a Chip (SoC) would just lead to a small box period.
Apple PR peeled the mini out of being jumbled in the Mac Pro 'hints' too.
The clarification was very clear about it being directed at the Mac Pro and only about the Mac Pro.
"Apple PR has reached out and clarified that only the Mac Pro is expected to be next updated in 2013 "
[emphasis added. ]
https://www.macrumors.com/2012/06/1...c-pro-and-imac-designs-likely-coming-in-2013/
I am unconvinced we will see the same box though because of such an insanely long product cycle. I would be quite disappointed actually. They could have done E5's sans TB if TB is never going to materialize on X based dual socket boards. SATAIII, USB 3.0 and PCI 3.0. Seems like a decent thing not to pass up if there was no TB plans for the future.
It's possible Apple has information we don't, and a GPU manufacturer is going to include Thunderbolt, or Intel is going to start doing integrated graphics in the higher end Xeons.
Or Apple stops using the Xeon entirely. No one really knows. Whatever is happening, I would be very surprised if it's a minor revision.
Or Apple stops using the Xeon entirely. No one really knows. Whatever is happening, I would be very surprised if it's a minor revision.
You are assuming that they were working something that requires an insanely long product cycle. It is extremely likely they were not.
First, one aspect of the long Westmere -> Sandy Bridge Xeon ( and -E ) cycle was Intel. That has nothing to do with the "complexity" of the products. SuperMicro and Intel had Xeon E5 motherboards ready back Aug-Sept 11 and yet didn't ship till March 2012.
Ivy Bridge Xeon ( and Ivy Bridge E ; essentially the same die with different settings and submodules removed )
Again a sweeping assumption that Thunderbolt is the blocking issue.
If apply Occam's Razor to the issue, Apple doesn't have something because they weren't working on something. It is the most likely. Given the protracted delay by Intel Apple could have started late and still finished before volume shipments started.
The dual socket aspect actually makes Thunderbolt a completely moot point since with 80 PCI-e lanes it is almost trivial to add embedded GPU to the motherboard. In fact, it would be one of the most productive ways to use the 80 lines if Apple stuck to their 4 slot set up ( 16x 16x 4x 4x ) as almost half of the lanes would be going unused.
It is the single CPU socket that is slightly more problematical configuration. This is situation where slightly different motherboards for the different variants would simplify things. ( e.g., single socket would have just 3 PCI-e sockets 16x 4x 4x or take a bandwidth hit 16x 8x 4x 4x ) . Even the proprietary PCI-e GPU cards don't take years to get out the door if Apple went with the "inflict an even higher Apple tax on Mac Pro" route. The MBA and new MBPr have proprietary SSDs and it didn't take years to get them out the door.
Which is all the more indicative that it was not a choice between TB vs. the rest but actually having any suitable design ready to ship.
Thunderbolt is not an viable excuse. It isn't that "great" of a feature to justify a whole year delay. It isn't really all that hard to implement either (iMac has it without some Rube Goldberg contraption to deploy it. )
I think several folks are just doing very bad expectation management. The notion that the "Mac Pro will be so 'different' that it took 2-3 years to design" is grossly flawed. It will likely look different but not because it took some "Manhattan Project" to put to the product together.
I guess a bigger question is if this is a good use of Apple's resources.
Apple knows most people are buying these machines to use full GPUs. As soon as you install a full GPU, you're bypassing the integrated chip. Then you're back to square one.
It puts Apple in the awkward position of most likely selling an add on card that outright breaks compatibility with their own peripherals if you actually use it.
I suppose my personal "Doomsday Scenario" is that Apple puts in some high end integrated GPU like what the iMac has, ends using discrete cards in the Pro line, and tells everyone to suck it up.
Well, if the updated Mac Pro design won't ship till midway through the second half of 2013 and Ivy Bridge Xeon E5's ship in volume at the beginning of Q3 2013 then it would make sense.
Honestly though, I'm not sure the Mac Pro would survive a greater than 12 month delay past this very small incremental bump in June 2012.
I suspect Apple will get something out the door before March 2013. They are going to do a mountain of damage to the Mac Pro if once again the whole Mac line up refreshes and the Mac Pro is stagnant resting on a relatively minor speed bump.
If Intel's Haswell mobile/general desktop line-up slides into 2nd half 2013 maybe. That would push the overall Mac refresh back. That wouldn't be surprising since both Sandy Bridge and Ivy Bridge slid on Intel's roadmaps over time.
They could wait for Ivy Bridge Xeon. But given how highly inaccurate Intel's roadmaps have been that seems like a highly risky move. If for some reason volume Ivy Bridge Xeon shipments slide to Oct-November 2013 that would mean the Mac Pro would "have to slide" to Oct-November. Apple would have to live for 16-17 months on these Nehalem/Westmere designs while their major workstation competitors were all 1-2 generations in advance with all Sandy Bridge line ups.
I don't see Apple holding much credibility with the Mac Pro if the anniversary of this June's update rolls around and Apple has done nothing and keeps pointing at some nebulous 2013 "date".
If somehow Intel got things completely turned around an shipped Ivy Bridge Xeon in March-June 2013 maybe. Right now there are no roadmaps indicating they intend to do that. By Intel's Developers conference around September it should be much more clear what the Ivy Bridge E5 schedule is going to be. If first half 2013 then maybe Mac Pro will wait for them to leapfrog back into the game.
If targeted for the second half 2013 then it is a dubious move for Apple to make. They could ship Sandy Bridge versions in Jan-March time frame and then do a Ivy Bridge bump in November ( if Ivy Bridge managed to ship on time.) If Ivy Bridge slides then at least have a Sandy Bridge E5's for sale for most of 2013. If Apple updates twice in a year, there are upsides to restoring credibility and will help get them back in synch with Intel's release schedule. Haswell Xeon will likely be coming in 2014 around the half way mark.
Exactly my thought.
----------
That's generous of you to wait that long before deciding that Apple holds no credibility. I'll save you the wait, and say that APPLE HOLDS NO CREDIBILITY right now!
Yeah, I'm pessimistic.....
No you are not going back to square one. Most folks who want high powered GPUs want to hook them to "high powered" displays. Those displays have "mainstream" connectors on them that PCI-e GPU cards have. You hook the GPU card to the display and you are done.
Thunderbolt was, and is, 100% pointless to getting that job done.
There is nothing that obviates having two GPUs any more than having two CPU packages is "bad".
In fact, some high end set-ups are going this way. Nvidia's Maximus solution involves two GPUs based cards.
http://www.nvidia.com/object/maximus.html
One card oriented to GPGPU and another oriented to the visual side of the problem. An embedded iMac class GPU isn't going to equal a Tesla card but the structure of the solution is the same. In workstation class problems this will likely be an increasingly common configuration format. It wouldn't hurt for the Mac Pro to have it as a standard.
It only takes getting past the x86 cores are the "only thing that matters" viewpoint.
Assuming that the iMac gets one of the higher end Southern Island GPUs
http://en.wikipedia.org/wiki/Comparison_of_AMD_graphics_processing_units#Radeon_HD_7xxxM_Series
A Radeon HD 7870M has a 1024 GFLOPs rating (that's single but even at a quarter for double isn't too bad). That is large enough to be useful even if there is not display hooked to the Thunderbolt port.
With the video output hooked to a normal PCI-e GPU card and a 1024 GFLOP real time compute engine available the Mac Pro could do some intersting things.
Similarly a Mac Pro with no high end GPU PCI-e card and two 16x cards could also do very interesting work ( e.g., 1 very high speed LAN and 1 x8 (or x16) direct attached storage (DAS) card; 1 Tesla card and 1 x8 DAS card ; etc. ) The Mac Pro isn't necessarily about just one GPU card.
If doing digital audio , crunching non video data , and a long list of other things then top end 3D performance doesn't really make that much of a difference.
The notion of the Mac Pro as the primary vechicle of selling Thunderbolt Display docking stations is extremely dubious.
Even if want to go down that rat hole this is a cheaper way. Just have a BTO option with no GPU card. No you have a cheaper Mac Pro to sell so the user can more easily pay for their $999 docking statition with the entirely useless mag port connector.
Or Apple could make a somewhat more sensible move and .....
1. remove the power connector from the "regular" Cinema display.
(perhaps drop the web cam and maybe even the USB ports ). In fact, just remove the "attached fixed length" cable all together.
2. drop the price ( since it doesn't have to be a laptop docking station anymore ) or bump the backlighting and Display Port processing solution up so that is a full 10-bit color system.
and go back into making monitors. Not docking stations with monitors built in. The Thunderbolt and "general purpose" display could share major components ( LCD panel , major case elements , etc. ), but the internal elelectronics and connecting cable just differ. Apple could have two perhiperals just as they do now. Just two with better focus. The Display gets out of the docking station business and the Thunderbolt offering stops perpetrating that is designed for the Mac Pro (when it clearly isn't ).
There is zero reason to nuke the discrete cards entirely. If that is the philosophy then they should just clone the HP Z1 and merge the display in also.
Last edited:
optimistically, i could see Apple holding of on the major Mac Pro update to let the trends unfold
Possible but unlikely. Much of the roadmaps for the major components leak to blog/rumor sphere relatively quickly as Intel/AMD/Nvidia inform 100's of partners ... that just makes for loose lips. Apple may have several months lead, but not a whole years worth.
As pointed out above if the Ivy Bridge Xeon is going to come soon in 2013 it makes sense to wait and leapfrog. That schedule info will leak by the time Fall arrives.
That doesn't make sense. Enough information about Thunderbolt has been openly released now to make the "Thunderbolt on GPU card" a significant kludge.
Note that the TB controller also takes input to the switch.
http://www.tomshardware.com/reviews/thunderbolt-performance-z77a-gd80,3205-4.html
So it is not just PCI-e lanes and Display Port channel(s) coming in. There is also inputs that control the switch settings. This means that the GPU card still needs a proprietary connector to the base systems start-up firmware (EFI) to work. Hence the Rube Goldberg ASUS Thunderbolt expansion card with the extra wires going into the card and the Display Port input on the back so that can send it right back out again.
This whole "blame it on Thunderbolt" meme is weak.
\
That would be extremely unlikely for Ivy Bridge since it shares the same socket at Sandy Bridge which doesn't have it. Waiting for Haswell is almost suicidal from a Mac Pro perspective. It likely won't arrive before 2014. (and again incompatible. This time with Apple's 2013 proclamation. )
I thought there was a chance for Haswell bring in GPUs. But only the the E5 1600 descendant where could trade off x86 core and L3 space for GPU space. That seems less likely now with the early leaked Haswell slide-ware
http://vr-zone.com/articles/intel-haswell-ep-platform-detailed/16419.html
There will be socket changes for DDR4 but suspect they won't add pins for GPU out. Still stuck at 40 PCI-e lanes too in single CPU package configuration. Looks like the big deal in Haswell for Xeon will be the transactional memory instructions coupled to DDR4 .
That is a huge bandwidth hit. Haswell mobile / mainstream is still stuck at 16-20 PCI-e lanes out ( the socket pin count is actually going down). That like a single 16x and a TB controller unless jump into bandwidth over-subscription. Besides that would put the CPU offerings roughly on equal footing with the iMac's.
The subset of Core i7 that is -E class vs the E5 class ( -EP codename suffix ) is really only a matter of a several months when Intel isn't screwing with the schedule. It doesn't really buy much on cost either. And overclocking screams "not on Apple's top 10 wish list".
Not impossible, but unlikely.
"Minor" depends upon what the expectations are set at. If it is "suppose to be revolutionary" then moving to Xeon E5 and adding USB 3.0 and a TB controller to the motherboard would be "minor".
I was surprised this year that is was highly minor revision. Next year, that it will be what folks were expecting this year ... that won't be surprising.
The trend is already clearly present. The Mac Pro is the only OS X or iOS product that doesn't have an embedded GPU. Waiting another year isn't going to change that.
In the overall PC market, the bulk of AMD's and Intel CPU package offerings come with embedded GPUs. Intel already has E3 Xeons with GPUs. The majority of new, > $600 desktops sold from mid 2011- June 2012 had embedded GPUs. Every laptop, tablet and smartphone sold last, this , and next year have embedded GPUs
As for GPGPU being a major factor, Apple played a very key role in getting OpenCL off the ground. I seems unlikely that leveraging GPGPU capabilities is not already a strategic objective for them. Not sure it is going to get any more strategic next year after watching the overall market.
They could wait to see if Thunderbolt was really going to become a contributing factor for the Mac Pro, but it isn't necessary for the core workstation mission. [ There are no other workstation vendors bending over backwards to put Thunderbolt on there offerings this year. There may be a few players next year but it isn't a game changer right now. ]
There is no Mac Pro update now because Apple screwed up. Why they dropped the ball I don't know exactly. But this missed update as being part of a "plan" to gain insight; I don't see it.
If anything it clutters insight into the viability of the Mac Pro because it isn't as competitive with other workstations as it should be. So it is hard to get a good read as to its long term growth potential. It is only after an update to bring it into a competitive stance against viable alternatives that can see if it has traction with the targeted market.
Delaying to 2013 only means the picture is muddled until 2013. The primary decision Apple had to accept was that it was OK to wander the wilderness for another year, before getting data with a much better signal to noise ratio.
did not know desktops were going the way of integrated gpu.... great - just one more thing that makes my MP archaic.
And Apple just screwing up is a stronger argument? Seems apologist and anemic to me.
Last edited:
Deconstruct...
Thanks for your input! It's been interesting.
So if Ivy Bridge is out and also assuming that waiting a full year to update would be suicide for Apple's pro line, what is holding up the Sandy Bridge launch?
I mean, what could be so revolutionary ("big things" are coming) that is within in the next six months?
This whole thing is so murky!
Thanks for your input! It's been interesting.
So if Ivy Bridge is out and also assuming that waiting a full year to update would be suicide for Apple's pro line, what is holding up the Sandy Bridge launch?
I mean, what could be so revolutionary ("big things" are coming) that is within in the next six months?
This whole thing is so murky!
Pro might be replaced by Apple web apps. I know the bandwidth isn't there yet, but Apple is always rolling out new stuff before it's ready.
There is likely nothing revolutionary coming. That's just normal Apple "our next product is better than sex" hyperbole.
At this point, "big" would be just delivering something roughly comparable to HP's Z820 and Dell's T7600 . Apple's "big" is most likely relative to other Apple products. After the Mac Pro has coasted so long on the same design, just a major refresh will bring it up to date.
My suspicion is that the only "big" transformation in the concept will be aligning it with the rest of the Mac line up and removing some aspects of modularity ( e.g., embedded GPU but optionally add another ) to the Mac Pro. Likewise with Apple jihad against ODDs and slight slower campaign to de-emphasize HDDs.
None of that really requires waiting till 2013. The Mac Pro is in 2013 most likely because Apple didn't start working on what they are going to deliver in 2013 until either late 2011 or early 2012. In other words a relatively normal gestation period, just didn't start (for whatever reason) until late.
I do think that high "horsepower" GPGPU card will be more mainstream in the workstation market in 2013. Once Nvidia, AMD , and Intel starting battling directly ( with Tesla K20 , FirePro W9000 , and Xeon Phi respectively ) it should open up more mainstream workstation apps leveraging these types of resources. It is an open question whether Apple will re-orient the Mac Pro to be a better host for these kinds of cards.
It is suppose to be murky. Whatever evolutains refinements the iMac gets should be another clue where at least some aspects of the Mac Pro 2013 will take on.
How are apps going to replace hardware? It is in a completely different class.
Like the Mac Pro ? Apple's web services are typically rolled out a bit prematurely ( unstable and bit flakey under the initial waves but stabilize over time. ). But that is actually somewhat opposite of their approach to hardware. It is far more the exceptional case where Apple makes some sort of hardware leap before others have taken stabs at it before.
Processing as fast as you want to pay for: Upload, render / transcode, download, pay Apple.
Perhaps "web apps" isn't accurate, but they may implement that too.