Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Interesting. according to the tech specs pace, the 5700 in the Mac Pro supports 3 Pro Display XDR whereas the Vega II (Single) only supports 2.

the Vega II 'Solo' should do 3. ( I suspect it doesn't handle it 'well enough' so Apple specs it to 2 ). But back I/O card port + TB bus 0 + TB bus 1 should work ( as 6 DisplayPort streams ).

The W5700X has a more modern display driving hardware and can do DisplayPort data compression. So sends less data to support the 3 large resolutions and has a better raster pipeline ( since more targets at higher frame graphics at more mainstream resolutions. )
 
Apple math .... $600 x 2 = $1600....

EDIT: .... you get a credit for the stock GPU.
$1000 x 2
...less $400 for the 580X 😉


Don't have any sense, cause Apple went away from Open GL a long tome ago, and created their own graphic system called Apple Metal that only works with AMS cards... Don't understand this Nvidia obsession in varios people around this forum....
I think it’s bc there is an incredible amount of software that only works on Nvidia
 
Finally. Now I have to wait for the Apple Store in Delaware to re open so I can save $700 in Maryland sales tax.
 

The buy button was greyed out on the Vega II models when the Mac Pro was initially released also. The initial supply of the cards will fill BTO orders. When start to have a surplus then will add standalone. May or may not take the month or so like timeline for that to happen if Mac Pro purchases are down and this isn't as popular as much of the discussions about the card.




That presumes that Apple/AMD are going to have Metal drivers for RDNA2 any time soon. Adapting to whatever software/hardware blend AMD didn't for real time ray tracing may/may not come to macOS quickly.

( AMD did a release in July 2019 for mainstream 5700 based cards. And it took into 2020 for the to arrive. There is also likely no big "RDNA2" product driver in the laptop space either where a MBP pulls along driver development. )

I can wait! Or at least I think I can! 😀

 
  • Like
Reactions: xnu
ADVICE NEEDED:

I want to upgrade the RAM and I know Apple does not recommend mixing RAM sizes and also recommends using either 6 or 12 slots for best performance - that being said:

Shouldn't need to worry about mixing RAM sizes, so long as you follow the configuration instructions here: https://support.apple.com/en-us/HT210103#mixed

I'd order 64GB (32GBx2 RDIMMs), currently $429.99 from OWC. That brings you up to 96GB, using the 6 DIMM configuration. Gives you room to expand later, and is cheaper than buying 96GB (16x6) $649-$100 trade-in.

Personally I like to keep the stock RAM handy even if I'm not using it, just in case 'weird' issues crop up, you can go back to the stock config and prove it's not a third party RAM issue (or possibly prove it is).
 
Interesting. according to the tech specs pace, the 5700 in the Mac Pro supports 3 Pro Display XDR whereas the Vega II (Single) only supports 2.

The Vega II is a legacy card. The future for AMD is Navi RDNA & RDNA 2 cards. The 5700X is not on the same level as the Vega II but it uses the RDNA architecture. I think the unannounced 5900 may be on the level of the old Vega II.
 
Don't have any sense, cause Apple went away from Open GL a long tome ago, and created their own graphic system called Apple Metal

Probably more about OpenCL than OpenGL. Nvidia did a bit of a "embrace , extend , extinguish" on OpenCL. They dragged their feet on support the versions. Whether they got off OpenCL before Apple did or Apple also switched to a path they didn't have to 'herd cats" on their own isn't 100% clear.

Nvidia uses CUDA as tool to control lock-in to their cards. Apple's isn't particularly interested in being dictated to by a component contractor. They both have lots of cash. So neither on is likely the blink any time soon. If Nvidia did the work to make Metal a first class citizen on their cards ( at least as good as CUDA or better on macOS in computation) then they might get signed macOS drivers. If intend to make Metal second class then Apple probably won't sign.
Similarly, there is likely intellectual property that Nvidia doesn't want to share with Apple unfettered (as Apple is in the GPU creation business now). [ At one point Nvidia was out to sue every major smartphone GPU vendor. That didn't put them on Apple's 'best friends' list. ] Both companies probably have a share in where there are no more drivers. Most folks on outside only point fingers one of the players ( which only serves to stall the issue further because that just enables the companies to engage in finger pointing at the 'other guy' ).

that only works with AMS cards..

Quite false. The most prevalent GPU in Macs is Intel's GPU; Metal works on Intel GPUs. Intel is a very significant GPU market player ( just not in discrete GPUs ... for now .. they are looking to change that over next 2 years. ). There is no AMD GPU exclusivity at all. The most prevalent GPU in iOS devices is not Intel or AMD GPUs. Metal worked on Imagination Tech's GPU. And it now works on Apple's pragmatic fork of that Imagination Tech GPU ( there is still some shared "front end" aspects even if underlying implementation has shifted. )

That is the other issue here. Nvidia posturing as though they are the only possible choice Apple has as an alternative to AMD. That is increasingly thinner and thinner ice as more players get into the field. (again probably a contributor to the issue is that one of those is Apple itself. )

. Don't understand this Nvidia obsession in varios people around this forum....

More than a few folks have software which is stuck in CUDA ecosystem. Feel they are not "stuck" because can move cards around but software is hooked to Nvidiai hardware. That plus substantive 'fan boy" rivalries. ( AMD vs. Nvidia , Apple vs Windows , etc. )
 
Shouldn't need to worry about mixing RAM sizes, so long as you follow the configuration instructions here: https://support.apple.com/en-us/HT210103#mixed

I'd order 64GB (32GBx2 RDIMMs), currently $429.99 from OWC. That brings you up to 96GB, using the 6 DIMM configuration. Gives you room to expand later, and is cheaper than buying 96GB (16x6) $649-$100 trade-in.

Personally I like to keep the stock RAM handy even if I'm not using it, just in case 'weird' issues crop up, you can go back to the stock config and prove it's not a third party RAM issue (or possibly prove it is).


OK , so mixing RAM sizes will not affect performance - I was merely going by what Apple wrote:

Vukulav.png
 
I can wait! Or at least I think I can! 😀


As mentioned in the video, there is no RDNA2 mobile coming until 2021. So Apple isn't likely going stumble over themselves.

In the Mac Pro / Workstation context, RDNA2 isn't what AMD has worked on. Arcturus likely is an equally 7nm optimized ( the way the Vega was optimized for mobile AMD Ryzen 4000 ). That's would likely be higher on Apple's Mac Pro driver update than something that is more video game focused.

If the RNDA2 lands in a iMac ( or iMac Pro ?) then perhaps will get a Mac Pro (and eGPU) knock-on with an off the shelf card with limited support. But probably not a "Mac Pro" focus.

If WWDC 2020 has huge session on extensive "real time ray tracing" updates to Metal API maybe. However, I wouldn't hold my breath on that.
 
Probably more about OpenCL than OpenGL. Nvidia did a bit of a "embrace , extend , extinguish" on OpenCL. They dragged their feet on support the versions. Whether they got off OpenCL before Apple did or Apple also switched to a path they didn't have to 'herd cats" on their own isn't 100% clear.

Nvidia uses CUDA as tool to control lock-in to their cards. Apple's isn't particularly interested in being dictated to by a component contractor. They both have lots of cash. So neither on is likely the blink any time soon. If Nvidia did the work to make Metal a first class citizen on their cards ( at least as good as CUDA or better on macOS in computation) then they might get signed macOS drivers. If intend to make Metal second class then Apple probably won't sign.
Similarly, there is likely intellectual property that Nvidia doesn't want to share with Apple unfettered (as Apple is in the GPU creation business now). [ At one point Nvidia was out to sue every major smartphone GPU vendor. That didn't put them on Apple's 'best friends' list. ] Both companies probably have a share in where there are no more drivers. Most folks on outside only point fingers one of the players ( which only serves to stall the issue further because that just enables the companies to engage in finger pointing at the 'other guy' ).



Quite false. The most prevalent GPU in Macs is Intel's GPU; Metal works on Intel GPUs. Intel is a very significant GPU market player ( just not in discrete GPUs ... for now .. they are looking to change that over next 2 years. ). There is no AMD GPU exclusivity at all. The most prevalent GPU in iOS devices is not Intel or AMD GPUs. Metal worked on Imagination Tech's GPU. And it now works on Apple's pragmatic fork of that Imagination Tech GPU ( there is still some shared "front end" aspects even if underlying implementation has shifted. )

That is the other issue here. Nvidia posturing as though they are the only possible choice Apple has as an alternative to AMD. That is increasingly thinner and thinner ice as more players get into the field. (again probably a contributor to the issue is that one of those is Apple itself. )



More than a few folks have software which is stuck in CUDA ecosystem. Feel they are not "stuck" because can move cards around but software is hooked to Nvidiai hardware. That plus substantive 'fan boy" rivalries. ( AMD vs. Nvidia , Apple vs Windows , etc. )

Honestly, my hope is that AMD and/or Intel are able to find a way to embrace, extend and extinguish NVIDIA, but I know that sheer fantasy on my part.
 
  • Like
  • Disagree
Reactions: xsmi123 and timduck
OK , so mixing RAM sizes will not affect performance - I was merely going by what Apple wrote:

  • Use the same size memory modules across all slots to maximize performance.

I hadn't seen that; had just checked out the configuration options that allow you to mix and match. I'd like to see some performance tests to verify, but my guess is that more RAM with mixed sizes will give you better performance than less RAM with all the same sized DIMMS. So until you're filling all the slots, you're still better to mix. Filling all the slots with 8GB DIMMS is more expensive, $597.99, and means you can't upgrade without replacing them all (unless you mix which puts you back at the same point).

Note: This is just an educated guess, I could be completely wrong. It's been years since I've put aftermarket RAM in a Mac. Would definitely like to see (or do when/if my Mac Pro arrives) some performance tests.
 
  • Like
Reactions: ChromeCrescendo
Anyone know if it’s possible to have both a Pro Vega II and 5700 in the same machine and receive the benefits of both? I’m interested in the hardware encoding the 5700 can do for streaming.
 
Interesting. according to the tech specs pace, the 5700 in the Mac Pro supports 3 Pro Display XDR whereas the Vega II (Single) only supports 2.

The W5700X runs on the newer Navi 10 / RDNA 1.0 gaming architecture while the Vega II runs on the Vega 20 / GCN 5.1 workstation architecture.
 
  • Like
Reactions: wilhoitm
Be aware: AMD will start using a whole new architecture for their GPU: RDNA and CDNA. RDNA will focuses on gaming while CDNA focuses on computing.

GPU_Fission.jpg


They have to create a whole new product with a new architecture. RX 5900 will be base on RDNA 2 which won't gonna be CDNA. Which means their strategy had been changed.

We probably need to expect CDNA based GPU, not RDNA based GPU like RX 5900 or 5800.
 
BOUGHT IT:

  • 3.3GHz 12‑core Intel Xeon W processor, Turbo Boost up to 4.4GHz​
  • 32GB (4x8GB) of DDR4 ECC memory​
  • Radeon Pro W5700X with 16GB of GDDR6 memory​
  • 2TB SSD storage​
  • Stainless steel frame with feet​
  • Magic Trackpad 2​
  • Magic Keyboard with Numeric Keypad - US English​
  • Accessory Kit​

ADVICE NEEDED:

I want to upgrade the RAM and I know Apple does not recommend mixing RAM sizes and also recommends using either 6 or 12 slots for best performance - that being said:

What would be the best configuration to upgrade to 96 GB RAM?

12 slots using 8GB R-DIMM, which would allow me to keep the 32 GB provided by Apple but would also take up all the available slots and, should I want to upgrade in the future, would cause me to have to reconfigure all the slots;

or

Trade-in the 32GB that comes with the 7,1 and just purchase a 16x6 kit (thus leaving me 6 open slots for future expansion)?


Finally, is the general consensus here to use OWC RAM or are there other sellers whom this forum trusts?

Thank you.

(I AM BEYOND EXCITED)
May I ask what your uses will be for this machine?
 
That presumes that Apple/AMD are going to have Metal drivers for RDNA2 any time soon. Adapting to whatever software/hardware blend AMD didn't for real time ray tracing may/may not come to macOS quickly.

( AMD did a release in July 2019 for mainstream 5700 based cards. And it took into 2020 for the to arrive. There is also likely no big "RDNA2" product driver in the laptop space either where a MBP pulls along driver development. )

Im not sure if RDNA 2 is useful on Mac system cause RDNA 2 is purely focused on gaming only. AMD changed their strategy by creating two new architectures instead of using one universal architecture: GCN. This means Radeon Pro version RX 5900 is not possible. They need to make different GPU for workstations.

But I'm not sure if AMD is going to do that soon or not.
 
Don't have any sense, cause Apple went away from Open GL a long tome ago, and created their own graphic system called Apple Metal that only works with AMS cards... Don't understand this Nvidia obsession in varios people around this forum....

This is wrong; my hackintosh has a nVidia 1080ti and it works with Metal just fine. In fact before Adobe CC updated I could choose in Premiere between Metal, OpenCL and CUDA.

I wish to hell I could put my 1080ti in my Mac Pro and have it work under OS X. The main reason I bought the Mac Pro is because my hackintosh is never going to be upgraded past High Sierra because of Apple's & nVidia's stubbornness with each other. It is just stupid that the rift between them is still going on. Like a bunch of school yard kids that can't swallow their pride a bit and let past things go.
 
BOUGHT IT:

  • 3.3GHz 12‑core Intel Xeon W processor, Turbo Boost up to 4.4GHz​
  • 32GB (4x8GB) of DDR4 ECC memory​
  • Radeon Pro W5700X with 16GB of GDDR6 memory​
  • 2TB SSD storage​
  • Stainless steel frame with feet​
  • Magic Trackpad 2​
  • Magic Keyboard with Numeric Keypad - US English​
  • Accessory Kit​

ADVICE NEEDED:

I want to upgrade the RAM and I know Apple does not recommend mixing RAM sizes and also recommends using either 6 or 12 slots for best performance - that being said:

What would be the best configuration to upgrade to 96 GB RAM?

12 slots using 8GB R-DIMM, which would allow me to keep the 32 GB provided by Apple but would also take up all the available slots and, should I want to upgrade in the future, would cause me to have to reconfigure all the slots;

or

Trade-in the 32GB that comes with the 7,1 and just purchase a 16x6 kit (thus leaving me 6 open slots for future expansion)?


Finally, is the general consensus here to use OWC RAM or are there other sellers whom this forum trusts?

Thank you.

(I AM BEYOND EXCITED)

What are you using the computer for? Is there a realistic chance that you'll ever surpass 96GB of ram? I think you need to weight whatever that likelihood is against the cost and uncertainty that comes with selling the factory installed RAM. I would personally go the route of leaving empty DIMM slots, but that decision may be inspired more by idealism and coolness factor rather than economical and practicality concerns
 
I have been fine with 32GB, but will update as soon as the **** storm is over.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.