The Mac Pro’s goal is to replace Windows workstations. Not supplement them.
...what? This makes my brain hurt.
The Mac Pro was literally a Xeon class workstation that could run Windows as a secondary operating system. It was, tier for tier, component to component an equal to a PC workstation. And if you really weren't convinced you could load Windows onto the exact same hardware and run it as a Windows workstation.
Please note the above in bold. These are your original goalposts. These are the points that I've been debating here.
Again, you're taking a multi dimensional choice and folding it down to a single dimension: GPU benchmarks.
No. I'm saying that if GPU performance is the top priority (over the device in question being a Mac), then the Mac Pro isn't the best choice and never really was. If the device in question being a Mac is the top priority (over the device having performant graphics), then that's an entirely different point to be made. I don't understand what's so hard to grasp about this nor why you feel compelled to keep debating it like it isn't an obvious fact.
I only keep belaboring this because you keep fighting me on it. Your original point, again as quoted again above is that these Macs are tier for tier, component for component, comparable and competitive and they're not. Incidentally, this is a thread full of people realizing that Apple has abandoned them, neverminding the fact that the writing for this was on the wall for years. I'm not saying that makes it any better, nor that anyone in this boat doesn't have my sympathy. Y'all definitely have my sympathy for what little that's even worth. But acting like it's a surprise or like it couldn't have been predicted demonstrates a lack of attention to the obvious.
There are a lot of reasons people bought the 2019 Mac Pro for GPU work. Heck - Radeons even sell on the PC side. Saying the Mac Pro was a bad choice because it had Radeons continues to be nonsense.
Incidentally, I never said it was a bad choice. Just that there are substantially better choices IF MAXIMIZING GPU PERFORMANCE IS ONE'S TOP PRIORITY IN A WORKSTATION COMPUTER.
Are there reasons someone might choose a GeForce based computer? Yes. But that doesn't make the Mac Pro a bad choice.
You keep putting words in my mouth without buying me dinner first and I'm not very appreciative. I never said that it was a "bad" choice. I said that it's not a "competitive" or "optimal" choice. If you want to turn that into "bad", then that's on you.
You keep saying this - and as I keep pointing out - competitive does not mean it's the best in every performance axis.
Again, when you use verbiage like "tier for tier" and "component to component equal", that would directly imply that it's at least comparable in every performance axis.
There was even a lot of utility in having upgradable GPUs year over year.
Let's make one thing CRYSTAL clear: I'm not arguing against upgradability. What I am arguing is that the Mac Pro pales in comparison to pretty much any other workstation tower when it comes to upgradeability and that's far from limited to GPU selection.
A lot of users didn't care about raw performance but needed a machine where they could keep up with new versions of Metal.
...And if one NEEDS Metal, then fine. But, Metal still pales in comparison to CUDA which goes back to my original point that if raw GPU performance is more important than the device being a Mac, the Mac Pro is the wrong device. But if those priorities are flipped, that's irrelevant. A point that, again, I have no idea why you keep debating when it's otherwise obvious.
I'm kind of scratching my head here. That's both true and misses the point. A vast majority of CPUs sold are Intel. Does that mean AMD is non competitive?
AMD offers performance on par with (and in some cases exceeding) that of Intel. That makes them competitive.
AMD does not offer performance on par with NVIDIA and neither does Apple. That's the difference.
Hah. Yes. I'm intimately familiar with everything in that video.
Clearly not! There are things in that video that you are arguing against like that video wasn't straight from the horse's mouth! You are a Mac user, you are on a Mac forum. Like it or not, Apple makes the rules! I'm not saying I like it (for what it's worth, I don't like it any more than anyone else in this or any of the dozen of other anti-2023-Mac-Pro threads on here). But hell if I'm going to deny that when that's obviously how the Apple ecosystem works. They decide how things are going to go, and it's up to us to either take it or leave it. I'm not saying I like or condone it (again, I really don't), but denying that is to deny reality at this point.
Sure. You can go back to my posts discussing this on this forum in 2020 when that video was posted.
It's becoming hard enough to justify the expenditure of time to debate you here; I'm not exactly certain where there's added justification to dig up three year old posts of yours.
If not I'll re-summarize:
- That architecture technically does not rule out swappable GPUs. (Shared address space can work across removable GPUs.)
Technologically no. But Apple has stated that this is not what they're ever going to do. That is not their system architecture; they are not going to open things up to either (a) third party GPUs or (b) discrete GPUs. That is not how they've structured Apple Silicon as a hardware platform. That is deliberate. That is not going to change without a substantial announcement being made at a WWDC (and even that is an extremely unlikely course correction [seriously, you will see a flying pig outside your window before they change course on this]). So, your point of "well, they could do it" is utterly moot.
- That video does not rule out a competitive GPU (say like a rumored M2 Extreme that seems like it was cancelled.)
That video does not rule out a higher-end SoC. That much is true. You may very well get one in the M3 generation. Or, what's way more likely, the PCIe ecosystem on the Mac will be relegated to Thunderbolt breakout boxes, PCIe storage, and high-speed networking interfaces, while those for whom this Mac Pro (and M2 Ultra, by extension) doesn't suit (which, statistically is likely not even the vast majority of Mac Pro customers), will either gravitate toward the Mac Studio or some other workstation. I'm hoping against hope for the former, but I'm not fooling myself about how utterly unrealistic it is given Apple's obvious priorities and lack thereof in the workstation space.
I don't think your "facts" are quite what you think they are.
At this point, I have no choice but to take your word for it.
they could have worked on removable GPUs that still had a unified address space. They didn't.
Please explain how that's even remotely possible given Apple's stated design objectives. Because I'm fairly sure that it isn't, but I don't want to make the assumption that you don't have any sort of background in SoC design.
Apple Silicon is not incompatible with these things. Even within the guardrails of what they outlined in the 2020 video there are still options.
M2 Extreme is the only option you stated that was actually a POSSIBLE option, given Apple's stated goals and the laws of physics. And clearly, Apple made a calculation as to whether or not that would be worth doing and decided it wasn't. I'm not going to say that they made this calculation correctly (as I don't believe that, myself). But that's clearly how the cookie crumbled.
I don't think this goalpost has moved at all.
See above.
I mentioned that you could even put the same Nvidia cards in a Mac Pro and you freaked out.
Hyperbole much?
Apple even mentions installed Nvidia cards in a tech doc here:
Learn about the Apple MPX Modules and some of the third-party PCIe cards you can install in your Mac Pro.
support.apple.com
I actually didn't know that they supported that. That's actually very cool. Thanks for sharing that (seriously). Doesn't change the point I'm making nor my original counter argument to your "component to component" rhetoric. But it's rad that Apple would at least allow NVIDIA on there in that capacity.
You keep saying "optimal" without defining optimal.
"Bang for buck" or "Best tool for the job".
The 2019 Mac Pro was a good choice for a lot of GPU workloads.
If those workloads HAD to be Metal or the work HAD to be done on macOS, then sure. Otherwise, it's a terrible value proposition.
Was it a good choice for bitcoin mining or AI training? No. But there's a bigger world beyond that.
If that's all you think an NVIDIA card or Windows Workstation is good for then this debate is truly pointless.
I'm saying Apple's 2019 top end can still clobber Apple's 2023 top end. That's a problem. The 2023 Mac Pro got slower - not faster.
It sounds like this is not the Mac for you, then. Maybe buy something else? A 2019 Mac Pro, if that one suited your needs so well. Or, dare I say it, a different workstation that isn't hindered by Apple Silicon?
Incidentally, the comparison I was replying to was that of four AMD GPUs compared to a single NVIDIA GPU, not M2 Ultra compared to anything.
It's not really that the top end config got cheaper. It's that Apple cut the top end config. Literally. The top end config was suppose to be M2 Extreme which never shipped. The top end config is missing.
If M2 Extreme had shipped, I think we'd be having a different version of this conversation - maybe around upgradability. But looking at the numbers they would have had top tier GPU performance.
I get that. It's undeniable that upgradeability and customization got shafted with the 2023 Mac Pro and that the higher-end configurations were ultimately eliminated. It is undeniably a bummer. Truly. But this is a point that is now two months old. It is no longer novel. And there's nothing any of us can do about it other than figure out what to buy instead. The OP started this thread on the pretense of the 2023 Mac Pro being the end of the line. Belaboring what we've already established in the other dozen threads is a waste of time and energy.
