Designing the New Mac Pro: Technologies & Wishes

| John Martellaro's Blog

Tim Cook has promised us a new Mac Pro in 2013. What would a new Mac Pro would mean to us, what is the essence of the Mac Pro design and what technologies might it use? I've been pondering all this and also have my own personal wish list.


Last summer, Tim Cook responded to an email from customer who was disappointed about the lack of a new Mac Pro. Apple's CEO said, "Our pro customers are really important to us ... don’t worry as we’re working on something really great for later next year."

Let's take a closer look at that statement. First of all, a key phrase is "later next year." That fits in with Intel's CPU roadmap which I'll discuss below. Perhaps more importantly is the phrase "really great." That sounds a lot like Apple's normal marketing hype, but in the context of a personal email and a promise, I take that more as an affirmation that Apple will introduce an exciting new product that has been rethought, re-engineered and will please us greatly.

Current Mac Pro design started with Power Mac G5 in 2003

What Should a Mac Pro be?

Before we can talk about a wish list or potential technologies, I think it's valuable to set up some expectations. After all, if we expect to be pleased with Apple's new product, we should have some idea of what the product ought to be for its customers. Here's my list.

  1. The essence of a Mac Pro is that it is the fastest Mac a reasonable amount of money can buy -- faster by far than any other Mac. We expect to pay for this luxury.
  2. A Mac Pro uses the latest available hardware technology. No holding back.
  3. A Mac Pro favors functionality over technical trends or fashion.
  4. A Mac Pro must acknowledge the considerable hardware investment professionals already have in displays, storage and other peripherals.
  5. User expandability is job one. The user must be able to easily expand RAM and HDD/SSD storage with standard parts.
  6. There are no proprietary connectors.
  7. BTO options must represent the technical needs of customers, not Apple's own agenda. See #3 above,
  8. As a plus, it should look like like something recently removed from the engineering deck of the starship Enterprise.

Expected Technology

There are some reasonable expectation for this new Mac Pro based on known technology.

CPU - Intel's CPU roadmap points to a terrific new multi-core chip, the Ivy Bridge-E. This is a 22 nanometer (nm) process CPU, the i7-4900 series, that will use a 3rd gen High-K+ Metal Gate. It's supported by the Intel X79 chipset.

Image Credit: Intel

The memory access is at 1866 MHz and PCI 3.0 will be implemented The number of cores is uncertain from what I've read, but it appears that the "E" series will have 8 cores, but only 6 enabled for starters to control heat. There is also a big brother, the "EP" that sounds to me like a heavily cooled server class chip with 12 cores.

Even the 6 core Ivy Bridge-E will be hot with a Thermal Design Power (TDP) of 130 watts. That's serious heat, and if this is the CPU Apple intends to use, that rules out right away a significantly smaller enclosure. We're still looking at some serious aluminum and some fans.

The timeframe for this CPU is Q3 of 2013, which fits right in with Mr. Cook's comment about "later next year."

Image Credit: Intel

SATA - Starting with the 2011 iMac, Apple finally made the jump from 3.0 gigabit per second (Gbps) SATA to 6 Gbps. I would expect to see that in the new Mac Pro as well. A major reason for that is that as SSDs have evolved, the very best SSDs can now saturate 3.0 Gbps SATA. It's about time.

GPU - It doesn't make sense to me to offer, by default, a very high end GPU into this Mac and price it into dismay. There are plenty of decent GPUs that can support an entry level mode and allow advanced customers to go their own way with, say, the US$1700 NVIDIA Quadro K5000. So I wouldn't look for anything breathtaking there.

Two PCI slots would keep the size down but still allow professionals to add their own legacy video cards that have DVI ports. And some advanced GPUs require two slots, so that would be supported in the basic enclosure.

Enclosure - Given the anticipated CPU and heat that must be dissipated, my guess is that the TDP will dicate the size of the enclosure. In turn, that will dictate how many drive bays and PCI slots will be supported and still reduce the overall size.

One approach which Jeff Gamet and I expect will be to reduce the main enclosure size as much as possible, consistent with the thermal load, and then offer a similarly stylish matching external Thunderbolt enclosure. Mr. Gamet calls it the "Thunderbox," all the while, with a smile, recognizing the other meaning for the Aussies down under.

The advantage here is that the basic unit is usable out of the box, but is as small as practical with, say, two drive bays and two PCI slots. Then, matching, stackable Thunderbolt expansion chassis fits on top with additional PCI slots and ports. This has some considerable advantages. First, the base footprint remains the same. Second, advanced users will drool for the matching expansion chassis for the life of the product.

In the course of researching this article, I ran across exactly that concept by Peter Zigich. I think it's brilliant, and I hope Apple goes down this road.

Credit: Peter Zigich

SuperDrive - I am somewhat up in the air about this. On one hand, I am prepared to argue that we have not yet ripped all the CDs and DVDs that we have on hand. Yes, physical rotating plastic is dead, but a Mac Pro is a blend of technologies -- advanced hardware balanced by the needs of advanced users and professionals. Until all that plastic is gone from our libraries and studios, a new Mac Pro may have to concede the point.

On the other hand, in order to reduce the weight and size, Apple will very likely go with the external SuperDrive here as well. Yes it clutters the desk. So put it in a drawer when not in use. It's too big a compromise on internal space. I'll declare it gone.

My Wish List

Now that we've more or less sized up a possible Mac Pro and have a feel for the hardware, there are some personal things I'd like to see.

  1. A beautiful Mac, still made of aluminum, but lighter and smaller than the current Mac Pro. No more razor sharp edges.
  2. A Mac, designed by Mr. Ive, that looks as if it's been brought back from the future. Simple and clean is nice, but carrying that too far hides essence of the Mac Pro. The design should also suggest speed, power and Apple class.
  3. A maximum usable, physical memory of at least 128 GB.
  4. Multiple Thunderbolt and USB 3 ports.
  5. A modular design so that legacy Mac Pro customers can get in on the base model and won't feel priced out of the market.
  6. Pleasant storage options. A base model with a 2 TB HDD keeps the initial cost down and lets us add our own, second bay, SSD. Later, even that 2 TB HDD can be replaced.
  7. No lock-in. Apple isn't going to sell as many of these Macs as, say, MacBooks. So don't try to lock us into Apple technologies for the sake of a minor, last minute panicked money grab.
  8. Remain true to the flagship concept. Create this Mac with a mind more towards honoring the tradition of the high-end, awesome, desirable speed-demon that researchers, scientists and developers will love. Make it a work of art that we can adjust to our own style and be proud to display it on our desks as Apple's statement that extreme power in the hands of the creator can lead to wonderful things. As you did in the past, Apple, make us drool, not grumble.

It should look right at home here. (Credit: Paramount)

That's my own list of the Mac Pro concept, candidate technologies and a wish list. Tell me what you have in mind.

Popular TMO Stories



I like the concept of a modular design. Buy what you need now and expand later. I’d look at offloading all of the PCI slots and drive slots from the CPU as well. Let the core hold the processors, RAM (at least 6 and better yet 8 slots) and one SSD for boot. Then add “thunderboxes” as you need for drives, PCI cards, and other expansions. Imagine if you could run parallel Thunderbolt lines to the external cases. Serious speed.


I think Thunderbolt is the wildcard on these new systems.

Right now, the i7 doesn’t allow multiple processors on a single board, and the Xeon doesn’t support Thunderbolt, likely because it has no onboard GPU to route the frame buffer through. Unless there’s a major change the Ivy Bridge-E chips, Apple may have to choose between Thunderbolt and multi-core on the next Mac Pro.


Thunderbolt technology would lend itself quite well to a rack mounted complex, or maybe something like the B25 system that Burroughs once sold, which was based on Convergent Technology CTOS bookshelf systems.


What geoduck wrote triggered an idea for me… one module could be a standalone thunderbolt/USB/lightning hub with a small power supply? Throw in the ability to pair a couple 2.5” drives and you have (well, a Mini sans the motherboard) an interesting adjunct to the laptop/mobile world…


While I like your ideas about what the future Mac Pro might look like, mine is more radical: Why doesn’t Apple make a “pro” version of Mac OS that can run on standard PC hardware, (possibly with a restrictions regarding supported CPUs and video cards)? They could exit the pro-hardware market completely, since that’s clearly not a large profit center for them. They could charge a hefty premium for the pro OS version (since they wouldn’t make any money from hardware sales), but given how cheap PC hardware is, the per-seat cost to the end user might wind up almost the same. The higher cost would prevent a mass exodus from Mac hardware in the consumer space, and it would give pros many more options.


Apple is not going to abandon hardware sales in favour of software at any price point. Pirating software is significantly easier and lower risk than stealing hardware.

If Intel doesn’t get Thunderbolt into their Xeon platform then Apple is basically screwed. You can’t charge Mac Pro prices for the same i7, generic RAM, PCI slots and drive bays that are in every $800 PC. Of course Apple could respond with a $1200 desktop Mac and sell optional PCI and drive bay expansion modules, but I don’t see that as very likely.

Even if Intel does get Thunderbolt into their premium hardware there’s the concern that inexpensive Haswell based machines might match Ivy Bridge E in raw performance. Right now Intel is saying that most of the Haswell innovations are around GPU performance and power saving and that performance increases on the CPU side will be minimal. That may or may not turn out to be true.

My department at work has two VM servers. One is an 8-core Mac Pro with 32GB of RAM and an internal drive array. The other is a 2012 Mac Mini with just 16GB of RAM and a bunch of USB 3.0 hard drives stacked on top. The mini with the newer architecture is both faster and more stable than the old Pro.


Mr. Zigich seems to have a spelling problem.


Fascinating article & discussion.  As a MacPro user who is think of upgrading this year I’m watching this closely. I love the idea of the modular units, which would enable me to keep adding to the base unit. I would still like to keep my 30 inch cinema display going for a while yet, & so would like to see some way to plug this in either natively or via an adaptor.


What KitsuneStudios mentioned is also tangentally related to #1 in the list of “What Should a Mac Pro be?”  That is, it isn’t the fastest Mac money can buy, when you compare the low-end Mac Pro to the higest-end iMac.  Apple might be able to sell more low-end Mac Pros if they would use an i7 instead of Xeon and lower the price, but I guess they want to keep the Xeon consistent across all Mac Pro models.

Also, as far as beautiul design goes, the seams that come with a modular computer case are anathema to that.  I suppose it’s worth the trade-off to many people, but I don’t have much to complain about regarding the current form factor, aside from the lack of USB3.0/Thunderbolt/eSATA*, and better cooling (HDs tend to run hotter than I like).  Maybe an option to fit more drive bays inside the same case if one chooses to go entirely SSD.  Of course the dream is to be able to buy new motherboards (as long as whatever form factor that is current remains current), so that we don’t have to keep wasting money on new cases (and/or other parts we don’t need to upgrade), but that will never happen.

*Not saying I want all of them, just that different people have different needs.


Great analysis John.  I love how you break the problem down into smaller components and address them individually instead of just assuming that what you want for your needs are the same as everyone else’s. 

That said, I wonder what the market is for people who need a powerful desktop tower are anymore vs. the market of people who need a powerful home or office server.  I still feel like the void created by Apple’s cancellation of the XServe is greater and growing than the one that would be created by the cancellation of the Mac Pro.  More and more I feel like a wider audience of users would be better served by a centralized server than by multiple desktop or laptop systems.  Apple is already well positioned in this space through its iOS devices, laptops - particularly those with limited-storage SSD drives - and iCloud.  Maybe the XServe was too limited because it was rack-mountable.  Perhaps a modular Mac Pro could fill the in-between void?


I hear what you’re saying, and I can see why Apple would think making a version of Mac OS that runs on generic PC hardware would be (marginally) bad for them. However, the other thing I see, from this very discussion in fact, is why it would be very good for consumers. Don’t need a super fast machine, but need PCI slots? No problem. Need the highest performance possible, but don’t care about the case? Fine. Want a DVD drive? Can do. Want a compact machine with a few PCI slots? Yep. The range and number of configuration options available for generic PC hardware is fantastic. Companies are supposed to prosper by satisfying their customers. In this case Apple has done anything but. My opinion which, granted, Apple is unlikely to follow, is that they should simply admit that this isn’t a market they care about, and open it up to smaller companies that would be more than happy to compete for our dollars.


So, just like a RiscPC then? (Google it.) A great idea.


In this case Apple has done anything but. My opinion which, granted, Apple is unlikely to follow, is that they should simply admit that this isn’t a market they care about, and open it up to smaller companies that would be more than happy to compete for our dollars.

In my fantasy world, Apple releases a Mac Motherboard bundled with the OS.  It has a basic 1 year parts and labor warranty, no Applecare.
Onboard Audio, Ethernet, Wireless, Bluetooth, USB 3, Thunderbolt, and PCI expansion. Allows a standard-socket Intel i3 to i7 CPU.

This allows Apple to create a standardized board with all the major components, to keep OS system updates simple. Most of the added parts (drives, CPU, RAM, ATX Cases, Power Supplies) are pretty standardized, and thus are less likely to cause support issues that a licensed design might. The few areas that aren’t standardized, like Graphics cards, could see help from a wider market for the product.

By standardizing to ATX, Apple reduces the potential for sales cannibalization, since you cannot create a laptop, Mac Mini, or iMac from an ATX board, not build a Xeon Mac Pro. There will likely be some cannibalization from Mac Mini and iMac sales, but that should also allow Apple to refine offerings and further tighten supply chains. It also means recapturing the Hacktintosh market.

By selling the board with the OS, Apple creates a basic licensing system without the issues they had in the clone days.  Third parties could then build standardized Apple towers, and offer their own warrantees. This could revitalize the small Apple authorized resellers/Technician shops, which would be free to sell their own warrantees instead of Applecare.

It’ll probably never happen, since Apple hasn’t sold a motherboard since the Apple I. I think Apple could massively expand their desktop market that way, though, and would make me an insanely happy camper.


That would be a great idea, although if they are going to sell motherboards, I’d love it if they’d avoid ATX and stick with their superior (IMHO) form factor, or something like it.  Of course, that would require them to sell cases as well, and commit to staying with that case through multiple generations of motherboards.  That’s even less likely to happen.

Log in to comment (TMO, Twitter or Facebook) or Register for a TMO account