Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

ScrappyCoco

macrumors newbie
Original poster
Aug 27, 2013
28
0
san mateo
Let's say in the future if manufactures like nvidia ever make cards for the mac are we going to have to buy 2 cards? Like lets say they make a titan 3 for nMP or something am i going to have to buy two of them? That would be sick but expensive.
 
The benefit of more cards means the extra one can be dedicated to OpenCL/Cuda processing without slowing down the user interface which is why the nMP has a minimum of 2. One is for the GUI and general OS and the secondary is for OpenCL processing or CrossFire if in Windows. Anything past 2 x in the future should be an optional add-on such as in 3 x GPU's or 4 x GPU's if they go that route.

Just an opinion.
 
Let's say in the future if manufactures like nvidia ever make cards for the mac are we going to have to buy 2 cards?

Doesn't have to particularly need to go to 3rd party cards. Apple doesn't sell D300/D700 or D300/D500 combos either. Is the symmetric GPU hard coded configuration or just more profitable ( both in component and support costs ) to sell only in pairs? [ support in costs in that always dealing in wired pairs for Crossfire and the 'compute' GPU is balanced twin. ]

The power on firmware may fail if the Mac Pro isn't fully assembled correctly. ( wouldn't want a Mac Pro with a GPU not seated properly to go out the door ).


Like lets say they make a titan 3 for nMP or something am i going to have to buy two of them? That would be sick but expensive.

Since this is default hardwired for Crossfire, unless there is tons of flexibility in the connector and traces between the connectors present, another GPU vendor isn't likely. If primarily tip-toeing around that issue by just having one card, it would not be particularly likely to want to scaffold that Frankenstien project in the firmware.
 
Since I got my nMP and looked at it from the beautiful inside... I'm having a really hard time imagining the gpus are user replacable in any way..

since you don't even see "the gpu". It's just like a mainboard with stuff on it and some of those parts belong to the graphics unit...

I'd say in 3-4 years, sell the nMP, make maybe 1000$ loss and get a new one.
 
I'm guessing the cards will need to be matching, especially in since there is a hardwired bridge.
 
Since I got my nMP and looked at it from the beautiful inside... I'm having a really hard time imagining the gpus are user replacable in any way..

The GPU cards pop right out after having been unscrewed. It takes a little time and patience, but it's not difficult to do. There have already been a few How-To vids and web pages that describe it fully.

IF new cards are ever released by Apple and are purchasable by end users, it won't be too hard to do the swap.
 
I wonder what happens when you put two gpu cards in there that both have the SSD connector. Wil it work, will you be able to put 2 SSD's in there?

I highly doubt it.

I think that the primary GPU board carries both the 6 sets of display port channels and the PCIe storage lanes, while the other card has neither.

A redundant set of PCIe lanes for the storage does not exist on the motherboard so they couldn't work even if a second set of chips were present.
 
I wonder what happens when you put two gpu cards in there that both have the SSD connector. Wil it work, will you be able to put 2 SSD's in there?

An early thread on this same question noted that the power connectors on the two cards are mirror-images - so it's impossible to put the SSD graphics card in the non-SSD slot.

Try the little search box at the upper right - you'll find your answer faster.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.