USB-C: A Mac Pro killer?
The Mac Pro hasn’t been updated for over 3 years. There is some discussion conjecturing that Apple may have abandoned the Mac Pro entirely. And judging by his twitter history the CEO of Apple does not appear to use the Mac as a platform much himself, or at least not to tweet from. It should be noted that marketing chief Phil Schiller has a profile that would be more expected from a high level Apple executive (scroll down to see the tweets sent from OS X).
But now that the MacBook Pro has fully embraced Thunderbolt 3, which supports a massive 40Gbps of data throughput with dedicated specifications for power, data transfer for storage, and even data transfer for graphics.
With the forthcoming external GPU’s it would be relatively easy for Apple to add full compatibility with any GPU on the market, which would mainly just require Nvidia drivers for macOS since the AMD GPU lineup is already supported. And with Thunderbolt 3 any card would get fully utilized since it’s part of the Thunderbolt 3 specification which Apple already fully supports, albeit with a small performance penalty for running through the cable instead of being directly connected. When it’s complete it’ll even be able to feed the data stream from the machine to the eGPU, then back down the cable again in the opposite direction to draw the graphics to the internal display.
Assuming this gets implemented the future is looking bright. The internal SSD’s on the 2016 MacBook Pro’s are already up there with the industry-best with read speeds of nearly 3000 megabytes / second, but considering the Thunderbolt 3 spec allows data transfer of 5000 megabytes / second it probably won’t be long before we start seeing external storage which exceeds the speed of the internal drive, especially considering that multiple disks could use RAID to achieve that. We’ll see a similar situation with graphics cards as they are released in the future.
With the bandwidth that Thunderbolt 3 affords it’s time to reconsider what a personal computer should look like. There is no significant advantage to having a large metal box which you place components into if you can have all of the same power by plugging a cable into your laptop and connecting it to large powerful components to give you desktop performance in a desktop environment, but with using a laptop as the central component.
Imagine you were to have a desk, and on it:
• A large external display
• A RAID configuration of Thunderbolt 3 drives for large storage & high performance
• An external GPU
If these are all Thunderbolt 3 powered then the only thing this setup is lacking is for you to plug your MacBook Pro into it. You can think of the whole setup as a sprawling inside-out computer. A box could indeed be designed to house these components if desired. The distinction is that it’s a network composed of standardized and fully optional components. This would be excellent for upgradability and compatibility with other hardware, and even other operating systems. The ability to disconnect the centre of the system and use it as a laptop is a bonus for people with desktop needs.
What do Mac Pro users need that cannot be provided using external Thunderbolt 3 components?
It’s limited to 16 GB for reasons which I previously wrote about in depth, and which Phil Schiller personally responded to. However it appears that later this year there may be a 32 GB MBP model released due to the LPDDR4 RAM specification getting finalized, which will make 32GB & up practical.
Granted, this may be a complaint for some users who do heavy video processing using some tools.
What do Mac Pro users need that can be provided using external Thunderbolt 3 components?
This could be really good. Besides potentially supporting things like the latest Nvidia 1080, if Thunderbolt 3 enclosure manufacturers decide to support it there is no limit on how much GPU processing power they could provide, all the way up to multiple chained up GPU’s in one enclosure, or daisy-chained GPU enclosures.
External drives have been hampered by the slow speed of USB-3 cables for far too long now. Thunderbolt 3 finally brings us to the moment where the cable specification is no longer the bottleneck. We will be seeing huge data transfer speeds to storage which will be enough to satisfy anyone with huge storage speed requirements.
• Future tech:
If this standard is widely embraced it makes the future of standardized compatible components look very interesting. Take the Tensor Processing Unit for example (a chip designed by google specifically for performing calculations for neural nets / AI quickly and with less power draw than a GPU, which is traditionally used). At present it is connected by a hard drive slot in a PC, but the point is is that something like this ever becomes a consumer product Thunderbolt 3 would be a suitable connection interface for it, so it could be compatible with any laptop which uses one. If we were still using USB 3 this would be unlikely to be a suitable connection for it, and it could only be used inside a PC – making the MacBook Pro future proof.
It’s not hard to envision other products that could be made with this bandwidth. How about a Thunderbolt 3 connection leading to a DIMM (RAM module), for some ultra fast ephemeral storage? If you were to use it for SWAP storage it’d effectively become a kind of RAM upgrade to your system that you can plug in at will (it would run nowhere near actual RAM stick speeds, but it’d still be a lot faster than using swap space on the internal SSD). The same goes for 3D XPoint (which is a distinct possibility given that it’s partially created by Intel, who have a finger in that pie as well as Thunderbolt 3).
And what of the future?
It’s hard to overstate how hampered electronics been by USB 3 and the lack of a universal connection standard for power and data for recent history. Its measly power delivery coupled with low data throughput has relegated it to a fraction of the potential potential of what a power and data connection standard could be. Now that we finally have a standard interface which has the capacity the possibilities are endless.
So which is it? A disinterested CEO, or a brilliant plan to embrace standardization and upgradability and shape the future of all electronics? Perhaps only time will tell. And it’s partially up to the industry as a whole to decide.