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.
Waiting for modular Mac Pro 2019
- Thread starter Zarniwoop
- Start date
- Sort by reaction score
Not, not, and not.
Waiting for Navi doesn't buy that anything that makes up for expressing themselves in public with yet another year of "the dog ate my homework" pronouncement. Apple's major problem right now in the Mac Pro space is lack of trust. Period. The second is lack of access to a secondary open PCI-e slots in the high throughput/bandwidth group. Navi doesn't solve either one of those. Better tech spec porn is always going to be at the end of next year's rainbow no matter what year you are in.
The CPU is relevant. There is far more to the "pro" space than folks who have a mania for the latest big GPU. Audio isn't going to be driven by the latest top end GPU. Photography largely isn't solely driven by biggest GPUs. Folks running multiple VMs aren't. Code development overwhelming isn't.
One of the narrow updates that the Intel W/SP cores are getting is some ML learning instructions. Probably not the top performance possible options but baseline Assistant ( siri smarter PC) should be possible. In another iteration or so the core count is going to bump and in the interim dynamic clockspeed range expand.
As for lower end prices... in the Intel W space? You funny. That probably isn't going to change much. At least on the lower end. Intel's upper bleeding edge end may have gone too far, but game changing price drops? probably not. Some relatively marginal relief maybe.
[ AMD's Zen+ probably is incrementally better, but also not likely to cause a radical collapse in Intel pricing. ]
T3 objective isn't going to be "most i/o". The main points that the T1 and T2 focus on is security. Keeping the biometrics and master keys secure. The T2 expands that into securing the whole boot process and securing the default boot drive. There is a x4 PCI-e v3 link to the T2, but that is not real expansion past what the last iteration of Apple SSD controller PCI-e drives had.
PCI-e 4.0 isn't coming to the T-series any time soon. Its primary attachment point is the the PCH I/O controller which just got to PCI-e 3.0 after 3-4 years after PCI-e introduction. PCI-e 4.0 is not coming to the PCH chipsets any time soon at all (if ever to the fan out. PCI-e 4.0 between the CPU and PCH sure. That connection is already grossly oversubscribed. but fan out from the PCH .... to what. USB? SATA ? ). Besides the tie-in of PCI-e v4 to iOS devices is what?
TBv3 ... ties to iOS devices is what? Nothing. T-series handling Thunderbolt makes no sense at all. TB has placement issues. ( the TB switch needs within 1-2 inches of the port. ) Merging TB into the T-series makes placing the T-x needlessly more complicated. There isn't really a major security issue to solve either that T-series would be any better at.
Ditto for ETH for all of the TBv3 points.
T2 controls microphone and cameras because that has a tighter security tie in. Rogue apps doing peeping tom on mic and camera are a potential hole. The tie-in in with iOS devices as far as other function outside of security have much higher overlap. Taking (capture and baseline enhance) a picture/movie is the same whether on Mac or iOS. Audio capture. ... largely the same. Speakers coupled to audio is largely the same relationship.
T-series will likely subsume what the ARM SoC on the iOS devices does/will. Multiple USB ports ( other other I/O ports ) are not likely coming to iOS devices. The power amount and distribution is different ( iOS devices generally don't power other devices at 10's of watt levels and only have one port. ). The T-series is very unlikely to subsume the PCH chipsets at all.
intel stuffing the core functions of Wi-Fi into the PCH means those chipsets aren't going to remain a stationary target for Apple either. [ the core USB 3 functions of TB chipset might get moved into PCH and the location dependent switch become more a PHYS implementation 'helper' chip. ]
Last edited:
Yes, and this is going to be a key metric to see if Apple really did "Get It", or if they're going to just do whatever they want.
USB-A ports are an excellent 'Test Mule' to determine just if Apple really understands their customers' objective needs, since things like a wired keyboard (and even mouse) simply do not need (or benefit from) the bandwidth of USB-C (or TB for that matter): to have to allocate a high performance port to a low performance device is indicative of a very poor engineering decision in system optimization.
I'd personally expect to see some M.2-esque internal bays here as the scratch disk alternative, and I say "M.2-esque" because I suspect that Apple will use their proprietary version that they've had in their notebooks rather than using an existing industry standard (thanks guys). But fortunately, there is a viable 3rd Party supply for these to minimize the Apple Tax pain.
For Firewire & similar, I expect the pragmatic solution to be a 3rd Party hub, particularly since Apple isn't bothering to even sell an OEM USB-C to FW dongle (the one that does exist is TB2-FW).
Probably, even though most Pros probably won't care ... and even for those that buy the matching keyboard & mouse on Day 1, in all likelihood these systems will no longer be aesthetically matched after only two years of use.
In any event, the focus on case color on a Pro system like this invokes in me personally the memory of the painfully expensive "20th Anniversary Mac", which was in several ways not much more than a repackaged PowerBook 3400c at ~twice the price.
Something was lost in translation, but what I meant, there's has not been and is not coming big changes in CPU world. 1% more power here, 5% more efficiency there, and thanks to AMD, the core wars have started. Still, nothing that wasn't already here five years ago.
If there's any hardware that Apple is waiting, it is not a CPU. Unless it is their own custom AMD APU.
Last edited:
I have doubts that is how Apple is approaching it. Type A is a easy way to distinguish to users that this is a "USB only" port. It is far more likely that Apple wants to keep Type C as a physical indicator that this is a multiple usage port ( TB + the 'normal' other alternative modes that come along with Thunderbolt ).
Yes the MacBook is the odd ball corner case to that. Probably the only reason it isn't Thunderbolt is that Apple shrank the case too small to fit TB into the solution. At some point they are probably going to correct that. If the system is going to be a one-port-wonder, then Thunderbolt is a better pick.
For the laptops you get into an either or because the edge space is limited in number and in height. The desktops (even the mini) don't really have those problems. So four Type A ports aren't a problem at all. They can all be provisioned out of the PCH chipset Apple had to buy anyway. So they just hook them up rather straightforwardly.
Technically Apple could slap limited range functionality on the PCH 'plain' USB ports. But that is likely going to cause user confusion. Lots of folks operate on the "if it fits it is suppose to work". Even if managed to sell a type-C-to-DisplayPort dongle is it really going to be worth it to field the support calls from the number of users to plugged that into their USB only Type-C port and it didn't work?
Apple would rather folks dump corded keyboards. I don't think they even sell them anymore. ( not even 3rd party ones). In Mac Pro space, software licensing dongles, affordable audio capture, pen tablets/screens , legacy external drives (flash thumb drivers) , and a few others would be better grounded drivers than keyboards and mice.
I would highly doubt that. Apple doesn't want to really sell loose, aftermarket SSD drives. They didn't for the MP 2013. They haven't for the iMac Pro. They haven't for the MBP that haven't soldered them to the motherboard.
More likely there would just be none ( like the MP 2103 and iMac Pro and the other laptops. ). If Apple is open to after-sale upgrades then M.2 makes sense. Apple doesn't have to sell them. Once over the other SSDs actually exist hurdle ( e.g. trim exists for other SSDs) then there is a clear established SSD sot standard. These wouldn't be default boot drives so deep integration with the security chip isn't necessary. ( or even going to work. since a 3rd party SSD controller. ). At point M.2 is relatively mature to be a viable ecosystem of both low and high quality parts. Apple can just test the high quality ones and offer those in BTO with appropriate mark up.
I think you are likely only going to get soldered onto the motherboard Apple SSD controllers going forward. Either the whole SSD is soldered down. Or the SSD controller (inside the T-series) is soldered down and the NAND daughter cards are movable. That's is probably the two options going forward. I don't see them as soldering down SSD controllers onto the future Mac Pro motherboard because maybe someone will plug in some NAND chips.
But even the 3rd party docks are drifting away from Firewire. As the TB dock market starts to be at least equally driven by the Windows PC market ( if not dominant in a year or two) the demand for FW is going to implode even faster.
USB 3.1 gen 2 with USAP (USB attached SCSI Protocol) vs FW isn't even a contest anymore. Older video cameras with FW are being displaced by smartphones with better video. There are some extremely narrow areas with sunk cost equipment that are hanging onto firewire in the computer space but they are stagnant and shrinking.
Without the mainstream demand the narrow corner case usage is going to get more expensive. Sonnet Technologies pitches their sole remaining FW card with optimized deployment in TB enclosure.
" ..
- Optimized for Maximum Performance Over Thunderbolt ..."
In that shrunken pool of hard core FW users, I suspect there is a much high ratio of those wanting > 1 FW port than the old FW market had. So these "one port" dongles and "maybe one port" docks aren't a good fit. That's why Apple isn't putting tons of effort into a single port Type-C to FW solution ( or pushing a 3rd party hard to come up with one to sell. )
If Apple forgoes M.2 completely then an open x4 slot is possible rather than Apple just dumping that bandwidth like they do on the iMac Pro. And open x16 and a open x4 slot would a compromise (relative to Apple's closing things up) that would get a decent number of folks on board with the Mac Pro.
I suspect they'd rather do a focused M.2 than an open generic x4 though.
I think it is highly doubtful Apple is going to paint the next Mac Pro. It is late as it is. Painting is only going to add more time (to finish out of Industrial design chokepoint) and expensive (product cost that adds zero value) .
The iMac Pro needs to be different from the iMac so there is a differentiation "value add" there. The differences between the iMac (and its Pro modifier version) and next Mac Pro will be obvious. Paint won't make any material difference in that perception dimension. It is simply a way to waste time and money.
Tx SoC of the future will take care of many other tasks. Security was a good start and reason to be (encryption plus touchID) but it takes or will take care of DSP, ISP, SIRI,FaceID,AI,keyboard,touchpad,iPAD APPS(UIKit),WIFI,LTE/5G, GPU for laptops (there's a tiny GPU in T1 already), video encoding/decoding and Thunderbolt 3 is very likely, especially if Apple ever starts to use AMD APU's. Sure it needs more I/O towards CPU, but there's going to be plenty left. The sky is the limit how Apple could use the future Tx chips and offload tasks from CPU.
Tx SoC's will eliminate the need to pay or rely on third party chip makers for I/O or networking services. There'd be one chip to rule them all...
Last edited:
Apple wasn't working on a new Mac Pro 5 years ago. Or 4 years ago. or 3 years ago. ( Even 2 years ago nothing along the lines they are doing now. ). What has/hasn't happened in CPU space over last 2-5 years has no material difference on the "go to sleep" approach Apple took and the "oh snap, maybe we should do something" change of mind they had after waking from their 'nap'.
If Apple had not decided to sleep away the future, they could have done a Xeon E5 v4 and AMD Polaris update to the baseline Mac Pro 2013 design. That would have solved all the problems but it also would have indicated that they were not completely sleeping on the job. If it takes them 3 years to gestate a "build from scratch" solution at least they would had something to ride out the longer than an elephant gestation process.
That there has be some huge big bang difference in CPU , GPU , and Thunderbolt for the Mac Pro to move forward is more a convenient middle management excuse to do a whole lot of nothing than a reasonable product strategy for the Mac Pro. At some point folks have turned the mantra of "the next Mac has to be best one of its type ever" into a nice excuse not to ship anything. Not that working on something and shipping when finished.... that it is a not work on anything across a number of products because it is a good enough milking cow. So can do less with less people and take home bonus for producing a higher $/headcount .
There was a straight forward socket change in the last 5 year they could have done while they figured out longer term where they wanted to go.
The core wars in the Xeon E5 space never left the core count war space. The only "core count war" new that AMD has brought recent is finally moving the mainstream desktop space past 4 and onto 6 (which they did in part by dumping the iGPU .. which makes them a no op for maintream Apple wins. ). Xeon E5 made progressions at the top end all along the way.
v2 12 core
v3 14 core (maybe 18)
v4 18 core (maybe 22)
[ maybe Primarily because those 'extra high' options were crazy expensive. Also because of the thermal corner the MP 2013 was in (GPUs main driver there though.) ]
The progression isn't anything new. If anything has flattened out as of late due to the delay in 10nm transition, but long term MP planning back in 2014-15 couldn't see that. Intel W line up starts off at 4 cores and probably still will with next minor bump update. There could be some small price changes toward the mid-bottom due to some competition with AMD but wouldn't hold may breath. Intel ins't going to dump the 4 cores, it just higher i/O that is more the selling point.
It could be that Apple is waiting for T3 before they can release a modular Mac (Pro). Maybe T1 and T2 were built for closed systems, like laptops and iMac's. When they realised they need to design a modular Mac Pro as well, their roadmap for security enclave & co-processor SoC didn't work out. So now they need to work it out, by software or hardware, or implement new features to the next version of Tx SoC. And that takes time. Two years is little.
iPad apps and UIKit kit are completely at odds with security. The whole point of having a security enclave is that user apps/kext/whatever cannot run down there. It is random 3rd party cruft that is added to your security subsystem that typically completely hoses your security. If you run a small, very narrowly focus services run through a set of throughly testing suites focused on security, it is going to be more secure.
Random generic apps are at least as much a privacy bust as Facebook is. It is somewhat hilarious to see the histrionics going on about Facebook when many of the other apps on folks phones are at least as bad. You want to bring that to your Mac security processor.... not a good rational. That would be a sure sign the bozo clown car have taken over Apple. The Mac needs more and better Mac apps; not iOS apps run on some quirky side car screen/window.
The camera image and audio DSP are cross overs from the iOS world. So yeah they have been paid for ( and double paid for since PCH covers some (audio) of that too... but not as securely). But that isn't an expansion from the baseline of what iOS design already have that T-series are scaling down (to be cheaper and cost effective).
The touchbar stuff is merely a framebuffer copy. There is no UIKit stuff going on there. It is rendered on the Mac's x86 half and just the buffer is copied over. There is not a library to that.
People have brought up the notion of running iOS apps on the Mac but it is a waste of time to do that with a huge coprocessor. A framework that makes it easier to port apps between iOS and macOS would be far better solution. Several aspects to the new framework expected to be announced at WWDC point to that.
If the objective is to make the Mac more expensive that other Intel based solutions then having a whole bunch of 100% redundant functionality would make sense. The fundamental fact is that still have to buy the PCH if buying the Intel ( or AMD) CPU. Trying to 100% redundantly reproduce all of it is a good way to waste money and power consumption.
The GPU in the T1 runs a single iWatch sized screen. There is nothing in current or likely future iOS that can run multiple ( > 2 ) full sized screens at the same time. Apple is no where near demonstrating competence in that GPU space yet. Let alone putting it inexpensively into a sidecar chip in a x86 host system.
There is no where near enough volume in the Mac space to support a rogue A-series chip that is vastly different from the volumes that iOS needs and pays for.
Completely stationary Mac Pros are useful how with LTE ? Making every desktop Mac user pay for a LTE modem and licensing is good way to piss off a lot of desktop Mac users with added costs that add next to no added value.
You can arm flap as hard as you want, but TBv3 has no tie in with iOS. Hence unlikely to go into a T-series chip.
On something like a Mac Pro leaving the TB on the CPU PCI-e lanes is better balanced throughput design. The T-series isn't hooked to a CPU and likely won't be in the future.
That "very likely" has no rational design backing to it. Nothing Apple has said or is doing indicates that at all.
AMD APUs that don't have an integrated (either defacto or pragmatically) PCH ? Which ones are those?
The sky is not the limit. At some point Apple has to pay for these wierdo chips that are at a complete tangent from the iOS volumes. The Mac volumes aren't going to support some kitchen sink chip. MBP with touchbar adds Apple Watch costs to the system. That additional cost is a market problem at least as much as a market aid.
At best there may be a ARM SoC will a fair amount of this stuff that is the primary CPU in a "split market" mac that forks between ARM and x86. At a market size of 7% that "fork" strategy is about equally to shoot Apple in the foot as it would be to improve things. And liklely has next to nothing to do with the half that a Mac Pro would be in. The workstation CPUs in the x86 space are going to have their own PCH's you will pragmatically need to buy ( it not already attached to the package ... e.g., AMD ).
If it did that would be great. But pragmatically you have buy the PCH from AMD/Intel. The option of where there was choice by default is a last century notion.
Nor would that be a real or necessarily difference that Mac needs to have over the general Windows PC marketplace. Most likely it would only make Mac more gratuitously more expensive for no added real value.
T2's security is a difference and the security's value can be made more apparent over time. Meltdown and Spectre only highly that CPUs for random general workloads are more susceptible to issues due to the extremely wide vectors in which to dump augmenting software onto them.
FWIW, I think we're in alignment here, even though it doesn't sound that way. I agree that USB-C should be standardized on TB3...what I was pointing out is that old fashioned USB-A's are useful in (Pro) desktops (and that unlike a laptop, there's ample real estate to include them is a bonus), such as for peripherals which are low bandwidth.
Moving on,
Point noted, and I agree on what Apple wants, but this is simply another benchmark to see if Apple is actually listening to their customers or blowing them off.
For example, I personally have a strong preference for wired keyboards (and barely tolerate my wireless Apple Mouse) on a desktop for the profoundly simple reason that it is less maintenance for me: fewer batteries to manage charging is lower overhead which is value-added. And while I've read promises of long, long ("weeks") battery life, that's (a) when it is new, not 5 years old, and (b) doesn't apply when the cat slept next to your keyboard/mouse last night and wiggled stuff around enough to keep it awake & running all night, draining said 'long life' battery.
EDIT: plus it also depends on your work environment. One employer I've worked for had a strict IT Security policy of ZERO tolerance for wireless data communication. No WiFi. People even had to turn in wireless mice & keyboards in conference rooms, and company phones couldn't use Bluetooth synch on their automobiles, etc. In these Enterprises, a wired keyboard & mouse is a "Must Have" even if you don't like them.
(On M.2 or other blade SSDs)
Understood. What I was really alluding to here is exploring what solutions is Apple likely to consider for inclusion into a mMP to service their Pro customer desires for more local high speed scratch space that's not in a damn external box. To this end, an M.2-esque stick is an option that Apple would probably prefer over a 3.5" or 2.5" HDD bay because its a smaller footprint.
FWIW, another notional design solution here could be a SATA-3 "slot" where a hard drive could be plugged into the mMP, like an old 8-track tape (here's an illustration for kids under age 50)
(on Firewire)
Agreed, which is why I doubt that there's going to be anything FW specific on the mMP itself ... its going to have to be addressed in the (yes, shrinking) 3rd party market, even if that means daisy-chaining three $200 hubs together to get three FW ports. FWIW, I'm still a bit miffed that Apple never wrote FW drivers for OS X for anything other than external drives, as my FW/USB scanner from that era ended up taking a ~30% throughput performance hit when I had to switch the cable from FW400 back to USB2.
Agreed, that's another design option.
Yup, it pushes one of Apple's "Like" buttons which is to make Ive less grumpy by being an enabler for making the box smaller.
(on Space Grey and/or other colors)
Well, it will need to have some sort of finish (and the tcMP was black) and some approaches are fairly cheap/easy, such as plain aluminum of the G5 tower and cMP.
Given that the iMac Pro has already made matching Space Grey keyboard/mouse/etc, its a straight lift out of the parts bin to use those on a mMP (and raises their production quantities some), so that just leaves spec'cing the mMP case to be finished in Space Grey, which takes the Apple Engineers only five minutes to add to the drawing, since the actual hands-on implementation is the fabricator's problem, which if that's done out-of-house is not Apple's problem. I've not looked into any of the specifics of Apple's Space Grey finish, but to my eye, it don't look any more technically involved than a type of black(ish) anodized finish.
FWIW, one of the things that Apple probably is trying to roadblock themselves with to create a struggle is just how to integrate a touchbar onto this non-laptop Mac. Never mind that the iMac Pro didn't ship with one - - they'll probably try to use the overly long gestation period of the mMP as the way to develop a touchbar keyboard that they can employ as an upgrade on the iMac Pro.[/quote][/quote]
Last edited:
The primary objective of the T3 should be to be a cheaper, lowest common denominator solution for the Mac line than either the T1 or T2 were. Apple can't afford to throw Apple Watch cost increases across the whole Mac product line.
And Mac products with individual silo T-series processors is beyond goofy. That is only going to drive costs unnecessarily high.
There is little about the T2 that needs to do with modular at all. There are parts that could be disused ( the camera) but no critical need for a design change.
this image was uploaded in the iMac forum "iMac Pro - First hands on Thread"
the purple SoC here is the T2. The T2 is hooked to
1. boot SSD flash storage. Not modular difference. ( e.g., MP 2013 had one. ... actually every Mac has one in some standard config. )
2. Audio codec . Not a modular difference ( headphone jack survives across all Macs so far. )
3. camera . Ok one possible modular difference. ( most Mac do though. current Mini and Pro really corner cases in Mac line up)
4. microphone. Not particularly a modular difference ( again current Mini and Pro corner cases )
5. speakers. Not a modular differnce ( all macs have at least one. )
6. Fans . Not a modular difference ( one fanless Mac which is a corner case).
7. "boot rom" Not a modular difference ( all macs have one.
)8. the more basic power management controller. ( SoC PMIC ).
9. the Intel PCH for x4 PCI-e lanes largely for the SSD.
A new Mac Pro probably won't have a camera but if leave that disconnected it isn't a problem. There is no fingerprint sensor either and that isn't the end of the world.
It would be relatively easy to slap a microphone on a base Mac Pro module if that was going to cause the end of world. ( probably won't though if left off).
A new Mac Pro probably would have more than two fans. But those additional fan(s) could be independently controlled by a controller in its own separate thermal zone ( or zone subset) that had temperature sensors and what not. The T2 controlling two or four ( two sets of slaved twins at the same speed. ) fans would work just fine in a new Mac Pro. [ Apple wanted to roll up a centralize thermal control that might be some additional software but that is probably unnecessarily complexity. local autonomic responses driven by local temp sensors are sufficient if not using some singluar cooling system ... which they likely aren't since that was one of the mentioned 'Problems". ]
There is no artificial show stopper here from just using the T2 , almost the same boot roms , Intel W , and GPUs from the iMac Pro in a new Mac Pro. It is 2017 parts that could easily go into a 2018 design ( i.e., finish engineering verification testing before end of 2018).
That is more like a "dog ate my homework" excuse.
That is a whole lot of hooey. The roadmap for the T-series likely included becoming the lowest common denominator chip across most of the line up from before it started. If look at the rest of Apple's usage of ARM SoCs they typically get volume by using the same chip across the broad range of the products targeted for.
iPhone 8 , iPhone 8Plus, iPhone X all use same SoC.
iPad Pro ( 10 , 12 ) all use same SoC.
latest Apple Watch Series all use same SoC.
the rest are a mishmash of older ones that have already gone through the above sequence for a launch. R&D paid for chips being used in other devices.
The T-series is probably going the same way. There will probably be one Apple puts in all newer mac designs that roll out. Perhaps older paid for designs in more trailing edge designs. Apple has firmly establish trend where they were using their own SSC controller and SSD design in Mac laptops. The T2 is completely in line with that and there is nothing about a new Mac Pro that would likely change that (at least for the boot SSD). Bypassing the PCH on audio and fans isn't that big of a deal. Neither is camera (specially if add in security aspects). Commonality across vast majority of Mac product line up is already in the T2.
"Modular" doesn't really throw in any loop at all. There might be stuff they don't hook up. ( e.g., no touchbar so don't hook that up. Their could be a GPU buried in the T2 ... don't really know on way or other. Wouldn't be surprsing since camera processing probably bleeds into GPU subsystem of Apple's SoC design. )
The next Mac Pro is late primarily because Apple wasn't working on it. So of course it didn't arrive earlier. The T-series can't be holding it back for any rational reason.
I understand there are those folks, but those aren't major market drivers. If the major drivers can get Apple to put 2-4 type-A sockets on the Mac Pro then they can be additionally used for other keyboards which don't align with which way they want to go.
The Mac Pro probably won't ship with a keyboard (or mouse). So hard to put it in the critical market driver category. It can be driving whether there are 4 instead of just 2 Type A sockets. However, the critical inertia to get to Type-A at all is probably other things.
Apple listening doesn't mean they are going to ship a "everything and kitchen sink" list of what was exhaustively enumerated by the group they listened to. The factors will be weighted by number (e.g. 75+ % are looking for this ) or outsized revenue (e.g., customer x buys $20+ M / year of Mac Pros wants Wifi-Bluetooth on a removable board or some firmware level disable. ) those are major drivers. There is difference between listening to the aggregate and listening and trying to chase the highest number of subsegments.
Adding to the drawing doesn't scope out the time and resource hit. Apple's Design tends to be OCD based. So a color means chasing down every minor detail all the way to mass production. The paint has to be sprayed on uniformly and color accurate both inside and others of the entire case. Apple is gong to look in the corner crevice of the case interior a look for imperfections. Working those out means iterating with the contractor.
Apple knew some folks would run out and buy Space Grey keyboard/mouse even though working 'regular' ones. Color is something that Apple tends to use when they are trying to juice more out of a relatively mature design and need some "easier than new design" solution. If the design is done then spending 3 months on an OCD color update is easier with limited resources than trying to do something new that takes 2-3x as long in time and resources.
The next Mac Pro is probably is at least 3 years late. Giving the OCD design folks more vectors to drift off into is only going to extend that delay longer. At some point Apple is simply going to loose trust with a critical mass of the Mac Pro market. "We spent 8 extra months of OCD on this design" isn't going to buy them a "get out jail free" card.
As OCD and myopic as Apple can be at times I don't think they are log jammed on a touchbar keyboard at all. They are far more customers logjammed on that than Apple is. It won't be cost competitive as long as "touchbar" is basically equivalent to Apple Watch in cost. Apple isn't completely blind and the keyboard market has lots price insensitivity at the range of that kind of relative cost increase. The others are real technical issues in that "touchbar" would basically mean doing a 180 on corded keyboards ( due to bandwidth demand (by the rapid framebuffer data pulls) among other factors. ). As a "back door" way of getting Apple back into corded keyboards I wouldn't hold my breath on that working. Even less on it sabotaging the Mac Pro replacement roll out.
It is in the ballpark of trying to saddle the base Mac Pro module that is targeted to sit desk-side with a camera and microphone.
Just doing a resized and readjusted case to the new Mac Pro internals would probably be a 6-10 month ordeal at Apple. Throw on top the limited design resources ( Apple can only work on a couple of products at a time) and that's probably pretty close to a 12-16 month wall clock window even with an approval into the design queue around mid-to-late in 2017. There aren't tons of open design windows and the shared resources team is already stretched thin on current Mac line up.
Saw this interesting info from that thread (referring to the Aquantia AQC107 10G NIC):
https://www.tonymacx86.com/threads/...d-extended-guide.229353/page-691#post-1728489
"I'm very pleased with the ASRock x299 Professional Gaming. It has 10Gbit/s nic (in addition to the intel i211 and i219). There are caveats in getting the AQC107 to work, you *must* go through 10.13.3. There was a bug in 10.13.3 that reflashed the AQC107 into an Apple branded card (the same as the iMac Pro). Once the AQC107 got reflashed, it would then work perfectly in macOS. That bug got corrected in 10.13.4, and the card no longer work. If it has been reflashed in 10.13.3 however, it will work."
Note: an ASUS branded AQC107 PCIe NIC card can be found for <$100 on Amazon, if the motherboard doesn't include one of those.
Hey maybe they surprise us with a silent TB3 upgrade meanwhile ? They should rename it Force iMac Pro
They should rename it Force iMac Pro 
Hey maybe they surprise us with a silent TB3 upgrade meanwhile ?
Jokes aside, it will be interesting to see what the update schedule for the iMac Pro is, since that's probably a good indicator of how committed Apple will be to updates for the Mac Pro too (especially if they're using the same -W chips.)
Are we even getting Kaby Lake -Ws or have we already seen all the Xeons that family's getting?
I know the argument has been made over and over again - I still believe Apple needs to stop being smart about the next MP's configuration and expandibility , and try their best to please everyone and serve every possible user case - within reason .
Kitchen sink with integrated bottle opener and corkscrew .
IMO, they need to return to a 'dumb' design just like the cMP , and on top of that work on third party compatibility .
That's what I'd consider 'listening' , and what well might be a major factor in winning back customers, and reassure existing ones .
Any shortcomings in usability, any propriatery solutions are likely to be percieved as further proof of Apple's inability to cater to the actual needs of their customers .
Apple can only win with a highly flexible workstation design, but can not gain anything with any limitations in a future MP .
Apart from a smaller footprint and a few design awards , which obviously isn't working for everybody ...
i couldnt agree more!
the worst apple has done to the pro community are those senseless innovations nobody wants.
as cg professional i want a standard tower using standard components, configurable with a single or double cpu with enough room for cooling to get all performances, being able to use the GPU of MY CHOICE, having multiple SATA and memory slots.
this is what we had 2012 and this is what we need since they came up with that off-way nMP.
problem is that i dont think apple gets it.
they will still try to invent some nonsens, and this will be either proprietary crap and/or second worse a physical module system which will just raise the costs and doesnt bring any advantages for the user. with such a system the sales would be low due to the costs and the future would remain uncertain as apple might stop selling pros because of the low revenue. a standard tower might be boring to some but it is the best solution to fit all needs and to keep costs low for apple and the users.
Last edited:
You could make the Mac Pro of yesteryear, updated for the modern era, and it'd still be a smaller footprint than its predecessor. You don't need 5.25" bays, you wouldn't have as many 3.5" bays, and the largest component that needs a minimum width is gone (the optical drive.) Hell, before the requirements for massive cooling with the PowerMac G5, the PowerMac G4s fit the same power and expansion into a chassis that was mathematically 10% smaller in volume, and also felt a lot smaller.
I know some people's response to that would be "just cram more in", but then we're getting back into the fact that Apple has not and never will make full-height, high-end workstations. The 1,1 Mac Pro wasn't that, the mMP probably won't either.*
As for "kitchen sink with bottle opener", if Apple includes standard PCIe slots than they've already provided a route for extensibility even if they don't offer it themselves. Want more flash modules than are on the mono? Throw in the cards. More USB ports? Cards. Dual graphics? Cards. Assuming we get a workable TB3 routing with the new machine too that'd give you a product with greater flexibility than the old Mac Pro because breakout boxes with far less performance penalties than the old days would be possible, and it'd require less mucking about with power re-routing on the interior.
I get people freaking out that the delay means that Apple is going to "overthink" things, but I think even if they weren't/aren't, it makes sense it would take the better part of two years to design a tower from scratch. I know the glib comments are that people would take the cheese grater in a heartbeat again, but that to me seems myopic. The Mac Pro was and is still a great design, but things have advanced since then too. Why should I need a screwdriver for any major component swapping, for instance? Or, (to vent about the competition), why do I need flimsy plastic shrouds ion my way that feel like I'm going to snap them off every time I open the thing?
I think the (understandable) frustration that Apple has let certain pro users languish without updated hardware masks the fact that it's always taken Apple a long time to build new designs. It's just that if you're not distracting people with upgraded tech the wait seems exponentially longer. To take an example from the iOS space, they've said that the iPhone X was being designed for three years before it actually came out. The iMac Pro is "just the same case" and that didn't come out overnight either (exactly how long it took will probably have to wait for the tell-all memoirs, along with when exactly Apple decided to go back to the drawing board with the Mac Pro.)
*Also from a practical standpoint I think the power that would go to those old internal connections is better routed to graphics anyhow.
I fully agree .
I used the cMP just an example of a general aproach I consider useful ; a new MP should certainly be smaller and lighter than that .
Personally I think two full sized drive bays for 3.5" HDDs - alternativelly usable for 2x 2.5" HDDs/ SSDs in each bay would help a lot with basic storage needs , save the cost, hassle and noise of an external enclosure for that particular purpose .
Plus 2 slots for fast m.2 PCIe SSDs for system and scratch etc. drives ; add 4 PCIe slots at ( 2x4 and 2x16 ) for GPUs and possible expansion cards .
Quient vents and proper cooling management, nice case design, , big flap to open the thing up .
A bunch of USB A ports, and TB ports , Ethernet, Wifi/BT, Audio ports, etc . - presto , done .
No chance IMO. I don't even see how that request makes any sense for a 2019/2020 Apple desktop.
I think 10gbit ethernet is the modern equivalent of internal 3.5" drive bays. Data on spinning rust doesn't need to be internal any more. Put it in the server closet where you don't have to hear it, keep it cool, or power it with the workstation power supply. Even a giant multi-platter NAS isn't going saturate the network.
Multiple M.2 slots (and enough PCI lanes to drive them) is the new target for workstations.
Multiple M.2 slots (and enough PCI lanes to drive them) is the new target for workstations.
Last edited: