2013, year of “something really great” for Mac Pro users is here. Hopefully, within the next six months, Apple will finally announce what they plan to do with the Mac Pro line. Time to speculate on what Apple could do!
No new Pro hardware?
Let’s immediately get Apple’s first rumored option out of the way: It’s highly doubtful Apple is going to announce nothing and try to push pros to an existing product. If Apple was planning on discontinuing the Mac Pro with no successor, they could have done it already. There would be no reason to ship a non upgrade in 2012, and there would be no reason to continue shipping a machine that can’t even be sold in Europe currently. It’s clear Apple’s approach to the pro desktop market has been anything but organized, but if Apple had a death wish for pro desktop hardware they could have ended the Mac Pro already.
Apple is also in the middle of a pro software push. If Apple is serious about winning back pro users, they’re going to have to address the lack of pro desktop hardware.
Mac Prosumer?
There’s also the idea that Apple could announce a more prosumer friendly successor to the Mac Pro. Maybe something based on an iMac or Mac Mini design, powered by Intel’s consumer chips like the i7.
This too does not seem likely to me. The Mac Pro is a unique machine in Apple’s lineup, due to it’s use of ECC memory. ECC memory is more resistant to corruption issues, which makes it a necessity for mission critical systems such as Mac OS X file servers or directory servers. While Apple has pulled away from the server market, there still is a need for a Mac that uses ECC memory for existing Mac network deployments, and only Xeon chipsets typically have that support.
Apple could take the path of just abandoning the large organization IT market entirely to Microsoft and Linux, but I have a feeling they won’t take that path for several reasons.
In the past, Apple has had an advantage of selling organizations the entire package. With one bid, they could supply a large organization with all the servers, desktops, laptops, and tablets they required, all under one support plan. If new Apple devices require servers from HP, HP is going to push their own PCs and tablets along with their servers, and that’s not a position Apple is going to want to be in. I’ve seen this happen before previously, and typically Apple’s prices don’t come out on top. Apple has also made significant strides in iOS device management with OS X Server, and they aren’t going to want to hand that responsibility over to Microsoft.
If anything, my feeling is that Apple is going to want to ship new hardware that could also possibly fill the gap left by the XServe, and not have to ceed Apple client management to other companies.
What are we going to see in a new Mac Pro?
A full size, PCI Express x16 slot seems like a certainty. 10.8.4 added support for the NVidia GTX Titan (a high end desktop card), and the AMD Radeon 7970 is already supported under OS X, so both seem like reasonable guesses for a possible stock Mac Pro card. An eight pin power lead also seems like an obvious addition. The current Mac Pro only has six pin power leads which reduce compatibility with some graphics cards.
Another likely change is greater support for third party cards. Current Mac Pros require graphics cards that support a protocol called UGA (Universal Graphics Adapter), which is part of the EFI firmware on the machine. On the PC/Windows 8 side, another standard called GOP (Graphics Output Protocol) is used for their EFI implementations, and new PC graphics cards are beginning to ship with that protocol. This difference is the reason that vendors still have to create separate cards for Mac and PC.
Over the past few years, Apple has slowly been moving all their Macs to use GOP and GOP cards instead. Because the Mac Pro is the only Mac that has not been updated since, it is the only Mac still on UGA. It is very likely the next Mac Pro will support GOP graphics cards, and allow users to pick up newer PC cards off the shelf and put them into their Mac Pros with no issues.
It seems likely that the number of expansion slots will be reduced. This would reduce the case size and make it more server rack friendly, and free up more PCI lanes to add more Thunderbolt ports. Speaking of which…
Thunderbolt
Thunderbolt is likely to be a major sticking point for the Mac Pro.
The Thunderbolt specification requires on board graphics, and doesn’t support using PCI graphics cards with on board Thunderbolt ports. The Xeon E5 series processors the Mac Pro currently uses don’t include integrated graphics, meaning Apple would have to add a GPU on the Mac Pro’s board. Even this doesn’t entirely solve the problem, as a Thunderbolt display plugged into the Mac Pro would be driven from the Mac Pro’s onboard graphics, and not the faster PCI Express card. This seems like a major compromise in using Apple’s flagship display with their flagship desktop.
Gigabyte has come up with a novel approach to solving this problem for their main boards. They run virtual displays off of the PCI Express card, and then push the output image out over the on board graphics on the Thunderbolt ports, very similar to using VNC to route video card output to another machine on a network.
This seems like the right approach for Apple to use, and could explain why the Mac Pro has seen a significant delay (this solution would take a while to implement.) It would also further solve the problem of compatibility with PC cards. The PCI graphics card would be abstracted away entirely from the firmware.
CPUs
One obvious path for Apple to take is to continue using the same Xeon E5 CPUs they have been. These CPUs come in 4 and 6 core configurations, support dual processors, and will be coming in upgraded varieties including 8 core versions in Q3.
An alternative would be for Apple to adopt the Xeon E3. These CPUs are cheaper, and come with integrated graphics which dovetails nicely with the Thunderbolt problem. However, they have significant tradeoffs. The E3 does not support dual processor configurations, and comes with a maximum of four cores.
The processing power would be competitive with most of the current Mac Pro line, but would be a step back from a possible 2013 12 core or 16 core Xeon configuration. This wouldn’t be new territory for Apple though. The Power Mac G4 was always a competitor to other consumer Pentium 4 machines, and not Xeon workstations. Apple didn’t enter the high end workstation market until the G5 or the Mac Pro. This move would anger high end users, but might make the Mac Pro more affordable to mid end users again. I would guess Apple probably has a lot of internal data telling them that the 4 core and 6 core configurations are the most popular Mac Pros anyway. Apple’s hope may be enough workflows are becoming more dependent on GPUs than CPUs, and that dropping multi processor configurations won’t be noticed as much.
There has also been no announced Thunderbolt support for the Xeon E5 series, while the Xeon E3 series Thunderbolt support is a maybe.
Optical and hard drive bays
The optical bays are likely gone. I’d expect some or all of the hard drive bays to become 2.5″ bays as SSDs become more popular, which also reduces the case size and makes the case become more server rack friendly.
Conclusion
I think the Mac Pro is likely to undergo significant changes, probably some of which will make some people very happy and other’s unhappy.
The traditional thinking is that the Mac will eventually be replaced by iOS devices, and that machines like the Mac Pro are in a lot of danger. Cannibalism from iOS is going to come from the bottom of the Mac product chain, not the top. Devices like the Mac Mini, Macbook Air, and Macbook are the devices most likely to see their sales drop and become discontinued. Higher end devices like the Macbook Pro and the Mac Pro are the least susceptible to iOS cannibalism, and could continue to be great sellers as long as Apple continues to support them.