Interesting. One of the Mac Pros I purchased (the one for me personally) has two Vega II Solo cards. Guess I'll see when it arrives (and on the dual Duo versions I picked up for others as well evidently). Thank you.
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.
MP 7,1 Waiting for Mac Pro 7,1 A1991 (no more)
- Thread starter Mago
- Start date
-
- Tags
- macpro 2016
- Sort by reaction score
Interesting. One of the Mac Pros I purchased (the one for me personally) has two Vega II Solo cards. Guess I'll see when it arrives (and on the dual Duo versions I picked up for others as well evidently). Thank you.
It _should_ come preinstalled.
The GPUs can work in tandem without the bridge, but it's going to all be much slower over the PCIe bus. And Metal deals with that situation quite differently, so apps might treat that situation differently or not support it.
Congrats on the dual Vega 2 purchase. Jealous.
Would love to have an Infinity Fabric setup. Looks real sweet.
I assume it’s a physical thing and uses the connector on the top of the cards shown here:
I'm curious why some are choosing two Vega II solos vs. one duo. What is the advantage? More ports? Something else?
I think it's really just about ports. If I had the money I would definitely go for one duo though. Just get a TB3 hub or the XDR and you'll have more than enough ports!
I'm doing this on my personal machine. Same price, I'm not putting enough cards in to use all the slots, more ports. For others I went with dual Duos for those who need the grunt and single Duo for those who don't need as much GPU.
... just get a device that does not, and cannot exist.
Sure ok, that sounds like a good plan.
They most likely meant TB3 backhaul which do exist, you can buy one on the Apple Store today.
... can you clarify what you mean by "TB3 backhaul" ? I've never heard that term used before, and can't find anything relating to it online.
Was talking about a docking station or the XDR display. Why you say it doesnt exist? Am I mistaken? maybe I didn‘t mean Tb3 but a Dock with several USB C ports. I think that should be fine in most cases, plus the ports that you already have with one Vega II... no?
If you buy two Vega II cards (over a single Vega II Duo) because you "want more ports" - you presumably want more TB3 ports.
No dock or display or any other kind of device will give you more than one daisy chain TB3 port. It can't.
So if by "TB3 hub" you mean, a USB hub, or a multi-port dock that connects upstream via TB3 - then yes of course those exists - but more USB-c ports is hardly a reason to buy two separate video cards. If it's a port issue, its because they want the TB3 ports, surely.
Yea you’re right.
can you elaborate for me how the ports On the Vega duo work? Like if I hook up two 5K displays to two ports, can I still use the other two for my TB 3 raids? Or would the bandwidth be too limited and I should rather use the ports on the “port card”
Take a look at the CalDigit TB3 Plus dock. It has a TB3 upstream connection to your Mac (and passthrough to a downstream device) and then has a variety of other ports on it (USB-C in 3.1 gen1 and gen2 variants, USB-A, memory cards, audio, etc...). "Backhaul" would be the upstream pipe on this hub, it is TB3 (as is the passthrough port) whereas the other ports are not but this is commonly known as a TB3 dock or hub.
Right, but my point was: if you want a bunch of USB or Ethernet or whatever ports, that a dock/hub type device like that can provide, that’s different than getting two graphics cards so you get more TB3 ports.
There’s an apple support doc specifically on which ports to use for best results in various scenarios, but generally it’s that multiple 5k/6k displays shouldn’t use ports right next to each other (so eg you could use ports 1+3 or 2+4 directly on the card, or one from the top of the case and one from the I/O card.
I’m not completely sure how the various display use affects the neighbouring ports for data transmission - at 4K I think it’s a non issue because the video ports use DP alt mode but for the higher res displays I think they operate as native TB3 for the extra bandwidth so a raid array right next to a 5k display may well be less than ideal.
Definitely check out the apple support doc about connecting displays, it has a heap of best use combinations and stuff to avoid.
[automerge]1576178389[/automerge]
Here’s the doc: https://support.apple.com/en-us/HT210228
Yea I saw that document but it doesnt say anything about data peripherals...
I’d suggest you perhaps need to ask Apple then. If the support staff can’t answer it directly, in my experience they’ll make a request to have engineering provide an answer within a few days.
Good grief for $2,800 minimal for a Vega II kit you'd think it would come in the box. ( and surprise surprise , it does on 'Solo' but doesn't on the even more pricer Duo. ). If Apple's hand waving was maybe someone wouldn't need one. Just how many is that? ( only 580X and W5700x folks who bought straight from the factory who go on to buy something up range). Sometimes Apple takes the whole green planet thing into goofy zone. ( but can see the tap dancing where folks are moving from one MPX to one Duo MPX where the need would disappear since buried inside the Duo. )
It doesn't show on accessory store search on "Infinity Fabric". I think they may need two different connectors because the Duo has two and Solo has one depending upon just how highly custom these are. It would make some sense though to see how many of each sell before start selling them loose if the types differ.
[automerge]1576183754[/automerge]
Perhaps later. But if have a Solo and buy another Solo in a kit you don't need to by it separate ( it is in the box ).
It is a physical product because it is a completely different physical bus.
Last edited:
They are implicitly coupled to each other. If Pair port 1 and port 2 means can only use Port 1 for very high video bandwidth data then using Port 2 for encoded PCI-e data traffic won't get entangled at all at the source points inside the Mac Pro. If there is "don't use for video" then the opposite is true is pulling from some other internal source.
I don't think the standard TBv3 ports on the Mac Pro are provisioned with baseline PCI-e lane bandwidth any better than the full sized MPX modules are. Each one of the full size modules blocks another PCI-e slot than the are using. Pretty good chance that blocked slot(s) bandwidth is shared with the MPX connector(s). The standard ports' feeds are likely sharing with other slots and/or devices that aren't necessarily blocked when a full sized MPX card is inserted.
Im Sorry maybe it’s because it’s not my native language but I don’t understand what you’re trying to say...
English is my native language and I don’t know what he’s trying to say.
Video data bandwidth comes from a different source than PCI-e data bandwidth. If Apple's guidelines tell to not use 2nd port in a pair for video for certain screens then implicitly that means that port is OK to use for some encoded data type other than video ( i.e. PCI-e ). If there are zero video consumers on the second port there there is going to be more bandwidth on the second port for other stuff. The bandwidth congestion problems only pop up when have too many devices collectively asking for "too much" data traffic. When 'ban' one type of data then that only opens up room for the other type. ( if ban trucks from a freeway you can put on more automobiles. )
Same is true if ban PCI-e data consumer/producer on a port that is going to host video. More bandwidth for the video. ( For the 6K display there should be nothing else on that daisy chain. Video or PCI-e data ).
So if Apple tells you what to do with one subtype on port pairing there more to it than just that. It is also implicitly telling you about info for the other port in that pairing.
Great, thanks for clarifying. So in my case it would look like this, according to Apple‘s documentation:
VEGA II Solo
HDMI Port (connected to AVX receiver for 5.1 and passthrough to HD projector)
BUS 0 connector 1: Thunderbolt 3 Raid.
BUS 0 connector 2: 5k Display via Tb3 to 2xDP adapter
BUS 1 connector 1: Another TB Raid
BUS 1 connector 2: 5k Display via TB3 to 2x DP Adapter.
Or do you think this is gonna be too much for the Vega? I can of course put the raids up on the E/A card. Right now on the Trashcan I have my 3 raids daisychained and connected to 2 TB2 ports, and the AVX receiver also in that daisychain, and the other 4TB ports are taken up by the 2 displays.
Errr... On the Mac Pro 2013 the HDMI connector was shared on one of the Thunderbolt Bus (TB controller). I suspect that is still true. Actually yes.
" ... If you connect a display to the HDMI port on your MPX module, the Thunderbolt 3 ports on Bus 0 can support one additional display at 4K or lower resolution. ..."
https://support.apple.com/en-us/HT210316
So presuming 6 DP streams max on Vega. In that case it would be the TB bus that is closest to the HDMI connector. [ the Radeon Pro WX9100 has a 6 min Displayport chips. It is Vega10 chip implementation but AMD didn't add anything there with Vega20 since it was mainly targeted for high compute products. I can't find evidence that Vega supports seven streams. ).
The Vega II solo has 2 DP ports going to each TB controller ( otherwise couldn't host a 6K on one of the part) and has two DP ports to the base TBv3 ports on the Mac pro ( one to each respective TB controller ... so they can't host the 5k-6k screens. ). That would leave zero leave for the HDMI port. (just like the Mac Pro 2013. ). So they 'steal' one from a TB controller for the HDMI port ( there is a DisplayPort switch upstream of the TB controller input and flip off the TB feed when the HDMI is active. ).
The current 5k monitors still soak up two DP streams.
If use the HDMI then loose one stream. A HDMI dongle of the I/O card's TB port would work. (there is a 1 stream being sent that way).
They put the same HDMI 'stealing one stream' restriction on the DUO too. I suspect just trying to maximize reuse on the edge port design work. Didn't "have to" do that with the other DP streams available. but it would mean trace rerouting on that part of the card so they probably don't want to. the "extra" two being feed into the MPX connector are on the other half of the card. That is much easier to do with little trace routing impact.
Last edited:
Or better to put one (or both) external storage on the IO Card? Being just an x4 device will that throttle full TB3?
@deconstruct60 do you have an opinion?
Last edited: