MVC's blog announced that Sierra natively supports eGPU, no hackery or tricks required.
http://www.macvidcards.com/blog/macos-sierra-had-native-egpu-support
http://www.macvidcards.com/blog/macos-sierra-had-native-egpu-support
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.
eGPU natively supported in macOS Sierra
- Thread starter Demigod Mac
- Start date
- Sort by reaction score
Hilarious... this is the nMP update everyone has been waiting for 
People with nntp, will have plenty of compute power. Also people with MP 6.1 will be able to most likely use the DeviceID's for GPUs from MP7.1.
Regardless if it will have Nvidia or AMD GPUs.
Yes, and more money for the Thunderbolt enclosure. And more wiring and boxes.
But on the plus side, it's still a dramatic improvement because now it's possible. And the old cards provide boot screens. And with Metal and DX12, they could in theory all contribute.
pci-e bandwidth and the TB to pci-e boxes cost as much as an good video card just for the box.
Empty external Thunderbolt expansion boxes are almost a nominal cost. ( Depending on how many slots. Single slot can be as cheap as $250.00 ) You are going to using your own video card anyway and most likely not both at the same time from the same manufacture.
$250/$199 > 1 ( a ratio greater than 1:1 )
The RX 480 is probably going to be a "good" video card.
http://www.anandtech.com/show/10389/amd-teases-radeon-rx-480-launching-june-29th-for-199
For cards in the $700-1000 range, yeah it would be more nominal percentage.
This payoff can be longer term rather than the immediate. One $250 chasis now and then amortize that over 6-8 year deployed lifetime ( similar to how folks are trying to stretch MacPr 1,1-1,2 boxes over extended lifetimes). $400/8 is $50/year or $4-5/mo. Folks pay more than that for Netflix/Amazon prime or a long list of other stuff.
The nnMP is likely going to be just the "Mac Pro". This whole slap a 'n' on the front is in part folks wishing the old one is going to come back somehow. If there is a "nMP 2.0" that extremely likely means that "return of the King" isn't going to happen; might as well just synch up with reality. What need is something like cMP "classic" or oMP ( old or original) as an abbreviation. (e.g., the MBP is the retina models at this point if talking about anything relatively current. )
Better cards doesn't necessarily mean going to get better, newer drivers/support.
For the deep bargain hunters, it isn't going to be so much the installed GPUs a not being able ti easily sell the "old" to defray the cost of the new.
In the Mac Pro space, seems likely that what is really being "replaced"/"upgraded" is more likely the 'Compute' GPU and not the 'Display' GPU. The 'Display' GPU is probably just as useful as it was ( all the more so if there are secondary displays to drive which can offload from the new one. )
eGPUs should make a bigger impact in "connect to display" usages more so in laptop ( and mini , iMac ) space more so than in Mac Pro one. By time Mac Pro GPU has fallen 'way behind the times' GPU wise most likely the CPU , RAM , TB ports, etc probably have also.
Very interesting that ASP implies that it supports automatic graphics switching. Wonder if that means it can draw to the internal display.
Umm... it will be very sad when (if) a 7,1 is released.. My prediction:
Everyone with a 6,1 + eGPU will have the latest GPU technology (they will be upgrading GPUs as available).
The 7,1 will be released by Apple, but the eGPU support will be purposefully removed so 7,1 buyers (suckers) will be forced to pay for Apple's highest GPU upgrade which will be already one (+) year old GPU technology.
You know my prediction will come true admit it. The only possible thing I may have wrong is that Apple will ever release a 7,1... 912 days and counting - certainly is a strong statement (of fail)...
Apologizes, I wasn't sure how to properly identify the tube v2. No, I don't think that Apple will bring back the tower again because Phil said the tube will be the computer for the next 10 years. I think they would cut their nose to spite their face on that.
Interesting. So if it is for compute, do you think that Apple would bypass the 2nd GPU and go directly to the eGPU for computer tasks? I don't quite see how they could run both the 2nd GPU and the eGPU for computing.
Do you think they will completely bypass treating both GPUs as a single unit?
Absolutely zero need to do that. 7,1 can easily be shipped with TBv3 a substantive differentiator. eGPU on TBv2 is hamstrung if want to push/pull data from it. It is nice if Apple back ported it to TBv2. However, long term it is a limited end. In short term, if the eGPU as primarily a 'Compute card' even more so.
7,1 will likely have better CPU , RAM , SSD. Folks who solely care about buying the "latest, greatest, discrete GPU card' every 12-14 months.. a high fraction of them never bought a 6,1. ( either kept dropping cards into pre-6,1 models or flipped to Windows). Far more likely will see a fraction of the pre-4,1 folks flipping over to used 6,1 as the used prices come down. Apple sees none of that whole "buy 2-3 generations back" market anyway so there is not much to motivate gimping the new model.
If Apple deploys an upgrade service the folks who want a more compact solution will still pay even if there is a eGPU capable path. As long as it is not ridiculously priced ( much higher than a external enclosure) some people will pay.
Doubt that, either way there is a script you can run that enables eGPU support fairly easily. I have an EVGA GTX 980 TI SC on my iMac prior to Sierra.
Sierra just seems to make it easier for EFI cards.
Update: It seems you still need the video card in the Thunderbolt enclosure with a modded EFI. MVC updated his post in the netkas forums.
As a collector of power bricks I'm always looking for a good reason to add more external boxes and cables to my desk. Hope they make the RAM external in the next round!