*corrects his original post above* š
That's right -- Do it right, or don't do it at all!
Hope they apple the same logic to the next Mac Pro. I really do.
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 Mac Pro 8.1 - WHAT IF...
- Thread starter maikerukun
- Start date
-
- Tags
- 7.1 mac pro next generation
- Sort by reaction score
That's right -- Do it right, or don't do it at all!
Hope they apple the same logic to the next Mac Pro. I really do.
I know it's tongue in cheek, but seriously, never apologise for putting your livelihood ahead of Apple's ecosystem. Unreal is going to be the universal engine for everything in a few years - especially in film, everything is going to be Unreal-centric, so whoever makes the best Unreal workstation is going to own the industry.
Frankly, at this stage, anyone reliant on using Apple gear in a professional context should have their bank manager questioning their risk profile.
I donāt disagree with your sentiment. But the home office basically almost mimics the studio now for me. And quite frankly, many folks I know who have upgraded their offices in the last few years have a very similar setup. Iāve just been reluctant to do it cuz I was holding out for the 8.1. Basically:
1. MacBook Pro for field work, travel editing
2. 2019 Mac Pro for sound/music production
3. Mac Mini/Mac Studio for office editing
4. PC workstation for 3D and animation
Now on the lot āparamount, Warner bros, Disney, Sony, etcā¦ā itās literally 100% MacBook Pros on the soundstages and in the offices. I mean everywhere. And the reason is itās the absolute most efficient plug and play solution for editing on the stage, on location, and sharing it with everyone who needs it all at once. Airplay and drive shuttling is a nonstop dance on set, there are an insane amount of custom apps specifically for iPads and MacBooks that were created just for set and on location scenarios. And quite frankly for the edit, MacBook Pros and Macs will remain king. I think color will remain 50/50 depending on the post production house you choose, and 3D animation will remain a 90/10 split on the side of PC.
Now video village is always going to be ran on PCās because they come with a prebuilt cart that runs the whole set but in terms of playback, file sharing, dailies, quick decisions, rough cuts, and previs, MacBooks will always be in like 50 peoples hands on any given set lol.
It really is a delicate and at this point streamlined dance of integration between both sides.
Iāve just been reluctant to add that other piece. But what you said is 100% correct. I basically bought this because of Unreal 5. Which the Beast (7.1) can run, BUTā¦UE5 is CPU intensive 90% of the time and your GPUS are far less important and unfortunately Intels Xeonās are absolute garbage for Unreal, while that 64 cores of CPU nonsense I just purchased will rip right through it. So, it was always a no brainer, but oddly enough, you just donāt want to do it if thereās any way around it ššš.
It sounds dumb but, my entire aesthetic is thrown off now and needs to be redone around this thing now to integrate it into the clean minimalistic industrial design of everything else I have. I also have to reconcile how to get this thing to run on my Apple Cinema Displays as efficiently as possible. I have to find an elegant solution to make the room symmetrical and clean.
Let alone the thread ripper, 2 etc 4090ās, and 2 w6800x duos are all going to be drawing powerā¦thatās literally 6 of the most power GPUās all doing there thing 24/7ā¦electricity bill about to vomit on my wallet Lolol.
And the idea is to move to a new studio where I have space to build out a far bigger virtual set and incorporate that within the next year or so once I have time.
Currently VFX Supervisor on a new film and Iāll be going into production on one of my own near end of July so earliest I can see having time for that step is probably spring next year.
But I do agree with you, and would never apologize in anything other than jest (with a small side of actual shame lol)
macOS is capped at 64 threads. If turn SMT off then have a decent match between OS and hardware. When TR 7000 series comes and over 64 cores, there is a mismatch even with SMT off. It is not a long term path for maximizing 'bang for bucks'.
[ edit: treads -> threads]
Last edited:
Yes, it actually does! That said, it's far more reliant on CPU than GPU, whereas programs like Cinema 4D, Maya, Blender, and After effects are all GPU based. Oddly, Cinema 4D's VIEWPORT is CPU based, but not many use the viewport as most tend to focus on Octane or Redshift as their render engine of choice "V-Ray, Arnold, and others are all options as well, but these are the primaries".
So it really comes down to what tasks you are going to be doing. Quite frankly, 64 core Threadripper CPU would see FAR MORE USAGE in science and huge data sets than it ever will anywhere else. And the 32 core Threadripper is the recommended one for Unreal UNLESS you plane on doing a lot of light baking "baking in lighting", in which case it will chew up all 64 of those cores.
Very interesting. And problematic. This 8.1 has a lot of holes it needs to fill...
Thx for the info.
Just starting learning Unreal, so a 7950X will have to suffice (along with a single 4090)
The whole M-series is already filling chunks of that 64 thread limit "gap". Apple is adding more specialized workload cores. There is no macOS cap on 'threads of work" on GPU cores , NPU cores , image processing cores, or AMX compute units work offloaded from CPU cores. Giving those other cores full access to the RAM means they are not "RAM capped" worse than the CPU cores.
It requires new , better software. And Apple making very good choices at to which specialized cores to allocate bigger transistor budgets too. However, It is not a bigger hammer to pound older legacy software libraries/applications with.
Chasing embarrassingly parallel workloads with big chunky CPU P cores is problematical also. AMD is about to release their new "cloud oriented" Bergamo Eypc CPU package that going to jump to 128 cores and be a big deal. Ampere Computing countered yesterday with AmpereOne that maxes out at 192 cores ( smaller , easier to scale cores)
Ampere Gets Out In Front Of X86 With 192-Core āSirynā AmpereOne
The largest clouds will always have to buy X86 processors from Intel or AMD so long as the enterprises of the world ā and the governments and educational
www.nextplatform.com
The > 64 CPU core workloads are probably splintering off further over time from what a single person workstation user is going to want. Those > 64 core processors are going to increasing shift to more optimally handling workloads of far greater than just one person (and largely just one single application).
Apple chasing that subset when they pragmatically only really sell macOS to individual single users doesn't make sense. No way Apple should put a design constraint on the Mac Pro SoC that it has to go chase the server package market like a dog chasing a car.
If the recommended TR in some cases is the 32 core model then that is under the same 64 thread cap that macOS has. Hence, not particularly a huge problem. There may not be an intermediate term solution in next 2-20 months on the hardware side that puts serious pressure on those limits , but macOS long term probably has little need for some major change there.
Last edited:
Yep, and it beats every intel processor with a 30% gain over 12th gen Intel CPUs.
I don't disagree with you, however it's less about does it need to and more about, CAN IT, for general audiences. A lot of people just want to know that their product CAN do a thing. I could've got 32 cores and I reckon I would've been just fine, but I chose 64 cores...just in case...that was a $6,000 just in case add-on and you know what? I slept better last night because of my decision lol.The whole M-series is already filling chunks of that 64 thread limit "gap". Apple is adding more specialized workload cores. There is no macOS cap on 'threads of work" on GPU cores , NPU cores , image processing cores, or AMX compute units work offloaded from CPU cores. Giving those other cores full access to the RAM means they are not "RAM capped" worse than the CPU cores.
It requires new , better software. And Apple making very good choices at to which specialized cores to allocate bigger transistor budgets too. However, It is not a bigger hammer to pound older legacy software libraries/applications with.
Chasing embarrassingly parallel workloads with big chunky CPU P cores is problematical also. AMD is about to release their new "cloud oriented" Bergamo Eypc CPU package that going to jump to 128 cores and be a big deal. Ampere Computing countered yesterday with AmpereOne that maxes out at 192 cores ( smaller , easier to scale cores)
Ampere Gets Out In Front Of X86 With 192-Core āSirynā AmpereOne
The largest clouds will always have to buy X86 processors from Intel or AMD so long as the enterprises of the world ā and the governments and educational
The > 64 CPU core workloads are probably splintering off further over time from what a single person workstation user is going to want. Those > 64 core processors are going to increasing shift to more optimally handling workloads of far greater than just one person (and largely just one single application).
Apple chasing that subset when they pragmatically only really sell macOS to individual single users doesn't make sense. No way Apple should put a design constraint on the Mac Pro SoC that it has to go chase the server package market like a dog chasing a car.
If the recommended TR in some cases is the 32 core model then that is under the same 64 thread cap that macOS has. Hence, not particularly a huge problem. There may not be an intermediate term solution in next 2-20 months on the hardware side that puts serious pressure on those limits , but macOS long term probably has little need for some major change there.
That's just the psychology of the consumer. I may not need it to be able to, but if I can future proof it, even if I never use the extra...then I'll have peace of mind.
With Apple they're already in a weird situation. This past year was the first time in ages that they sold LESS MacBook Pros, and it's because...they did TOO GOOD of a job with the original M1. Nobody that got an M1 was ever like "this thing is terrible, too slow, I need an upgrade NOW". And only the most affluent of us bothered on such an incremental jump to the M2 family to replace our M1's. Now the M3 will be a different story, and because there's been a 4 year GAP since the last Mac Pro, a lot of us are itching to replace that machine...but I'm willing to bet a lot of us that want to replace that machine are also wanting to replace it with something that is AT LEAST AS POWERFUL as what we previously purchased.
The overwhelming majority of the community got the very mid $12k - $16k range of the Mac Pro...GPU's nowhere near what the machine is actually capable of, a CPU half of what it's actually capable of, and so on...so it won't be difficult for the new Mac Pro to outperform what they previously had. BUT...then there are those of us who maxed the machine out and know what it's really capable of...the new Mac Pro becomes a much harder pill to swallow if it doesn't 2x our previous experience.
But again, I don't disagree with you.
Unless the next MBP offers 16TB of SSD or, dare I dream, they finally f'n add faceID (there is no more perfect an application for it than a laptop), I'll probably skip it too. No doubt it will be faster as well, but for right now, im not using the laptop that much so it's meh. It's weird, but really back to big iron the last few years. I guess with COVID it wasn't surprising shift.
Very true, and it's a shift I think obviously nobody was expecting. In a world that was full steam ahead on the mobile train, we all ended up back in the home office...interesting to see how people choose to "or not to" weave through it.
Paying $6K extra for 'piece of mind' is not the norm for the consumer psychology. At least those spending their own ( versus customer/clients) money.
As for selling less laptops, that has extremely little to do with the M2 vs. M1 and lots to do with tons of folks hyper buying during the pandemic. There is only a narrow subset of folks who went to "home office" and bought huge desktops. Most workers were sent home with laptops. So if have to come back in to 'hot desk' on internal office space 2-3 days a week they can bring the exact same computer they are using in both context to work with them.
And that buy every kid in school a laptop also blew up the laptop market.
The telling factor is that it isn't just Apple that is down on sales , it is everyone across the PC market. If run a sanity check to see of 2022-23 sales are higher than 2018-19 sales then there is no huge problem here. Remove the bubble outliers and not 'end of the world' drama here.
Apple doesn't need the M2 to detach folks from M1 systems. The M2 is far more likely aimed at either folks for which the M1 was 'bad timing' (not on their in progress refresh schedule) or were skittish about leaving Intel land. The vast majority of those folks were on either Intel Macs or Windows on Intel. The M1 really doesn't play a role.
M3 will be largely the same story. Mainly targeting folks being rotated off the Intel Macs. ( Intel Macs started off with 100M units. That isn't going to completely collapse in just 3 years. even if 20M a year left , then over three years there still would be 40M left. And the rate was probably no where near that high. And even more so if consider the continued inflow from Windows. )
There will be some substantive "ordered on day zero in 2020 to be 'first' " folks moving to M3 . But most of those folks likely were buying the newest shiny every 3 years back in the Intel era also.
For the most commonly sold Mac Pro components 16 core CPU and W5700X the M1 Ultra already covered those. So a M2 Ultra or M3 Ultra ....those have already got the "at least as powerful" already covered. You seem to be preoccupied with the way , way , way out fringe outliers ( most expensive config sold/bought). M3 generation isn't necessarily going to cover that.
Apple didn't 'cover' the upper end Mini Intel until 2023 ( after their two year transition window was over).
Furthermore the folks hyping the "I got dual duo W6800X " meme ... it hasn't been 4 years. Not even close. Those got introduced in August 2021. It will be about two more months before even get to the 2 year old stage ( 1/2 of that 4 year ). Apple already has had the users who might have bought in 2019 (or early 2020 ) spend an additional $2,800-10K on more stuff two years ago. ( and for the subset that bought off the shelf and didn't effectively pay for the driver R*D overhead ... Apple is going to be in a hurry to get undercut again how? )
Apple's statements says likely not. The more common ordered components where 16 cores and W5700 which is only about $8.5K. Boost to 96GB RAM + 4TB SSD range and at about $10.6K. Can start buying 'silly' amounts of RAM over 96GB directly from Apple but I doubt that is "most of the community". Same thing with storage.
The biggest hurdle that Apple has to get over for the M-series Mac Pro is putting more perceived user value in the sub $12K systems ; not the over $14K ones. If they don't, then it will likely fail.
Throwing in non-Apple parts could get into the $12K-16K range. Image capture, Audio DAW , M.2/U.2/SATA storage , 3rd party DIMMS, etc. But the baseline Mac Pro generating revenues for Apple; probably not .
There are not "lots of us who maxed the machine out". If there were, then the average CPU and GPU sold by Apple would be higher than a 16 Core CPU and a W5700X. if there were 10K who "maxed out" the MacPro and 95K systems sold then really only talking about around 10%. If look just at the 10K pool out of the context it looks like "a lot". But 10K people inside the Rose Bowl is relatively basically almost empty.
Where did I say "lots of us who maxed the machine out"?
And where did I say paying $6k more for peace of mind is the norm for consumer psychology?
Did I spend my own money or my client's money?
When did I say it's been 4 years since the w6800x duo's??
taxes and monitor figured, $10k+ Mac Studio exists...just sayin...
"Peace of Mind" is literally the entire basis for SUVs coming to dominate vehicle sales.
"Peace of Mind" is literally Apple's entire marketing message.
But what do we know about Apple? (all together now)
Apple lies.
It was a very strange point for him to choose and argue lol. That's why I just left it alone. Some people seem like they're here specifically to fight or hear themselves talk lololol.
Totally unrelated and maybe impossible but I thought it would be cool to see a Mac Pro with a top of the line M chip but with the ability to add an entire Intel/AMD submachine into the chassis. The submachine would proxy through the main system running macOS and basically act as a psedohypervisor (I'm sure there's a proper word for this that I'm not familiar with). It would capture the audience of people like myself that use macOS as their primary machine but still want to use Windows WITH native hardware (such as Nvidia GPUs) but are forced to buy and maintain a separate machine.
If you want max performance from a macOS machine, you buy a Mac Studio. If you want max performance in macOS but want a Windows/Linux machine running alongside it, you buy the Mac Pro instead of a Mac AND an entirely separate tower for your Windows workflow.
It's unique because you only need one tower under your desk (great for portability also) and most importantly major software changes could integrate your submachine as a separate window in Mission Control along with the ability to mix audio and other I/O between the systems (imagine sliding over to your Windows machine in Mission Control but you can also play music/make calls on other screens through your Mac that sits on top).
Would never happen and probably a nightmare to engineer but that would be my dream machine: a true all in one that is macOS first but does not compromise on performance via virtual machines and macOS's lack of Nvidia GPU support.
Also if not a single tower all in one as described above, at least the ability to proxy a Windows tower into the Mac Pro so the software experience above is possible.....
If you want max performance from a macOS machine, you buy a Mac Studio. If you want max performance in macOS but want a Windows/Linux machine running alongside it, you buy the Mac Pro instead of a Mac AND an entirely separate tower for your Windows workflow.
It's unique because you only need one tower under your desk (great for portability also) and most importantly major software changes could integrate your submachine as a separate window in Mission Control along with the ability to mix audio and other I/O between the systems (imagine sliding over to your Windows machine in Mission Control but you can also play music/make calls on other screens through your Mac that sits on top).
Would never happen and probably a nightmare to engineer but that would be my dream machine: a true all in one that is macOS first but does not compromise on performance via virtual machines and macOS's lack of Nvidia GPU support.
Also if not a single tower all in one as described above, at least the ability to proxy a Windows tower into the Mac Pro so the software experience above is possible.....
I would imagine it would be easier to do the other way around - a Xeon machine, updating the 2019, with an M-series machine built in an MPX card - the volume of a Mac Studio, when it has the Mac Pro's large front fans to provide cooling, is much closer to a full-size MPX bay than a high performance Intel machine is.
The touch bar was kind of a. Mini version of this, no?
Yeah effectively, it was a small A(x) computer that had a secure connection (and handled overall security with touchID) to the macOS host.
Which lead to all sorts of comedy where the touchbar could crash, and require a system reboot. That's what everyone's looking forward to, right, GPUs with their own OS that can crash?
Man, now THIS is some out of the box thinking...like a hyperactive in-house BOOTABLE OS INSIDE AN OS!!! Like BootCamp on whatever Bruce Banner OD'd on LOL. I love it.
I guess you have seen this : https://www.pugetsystems.com/labs/articles/redshift-adds-amd-gpu-support/
Showing redshift perf on amd cards on windows. How does this compare to current 2019 mp with 1-4 6800 and the m1ultra? I get the impression that even if apple could make mpx module with dual 7900s it would not be much faster since it would have to be downclocked to meet power standards.
If I remember correctly a m1ultra is about 6:30 in this benchmark and a m2max 7:30, placing a potential m2ultra around the 4min mark (240s)
That is, similar perf as a single 6900 or 70% of the power of a 7900xtx
Clearly this will not be what powers a mac pro as the maxed out system.
A quad m2 max would still barely be on par with a maxed out mp2019.
Wouldnāt it make sense to be able to deliver at least 50% more perf as the high end? And making āthe common configā at 12k be about 2x perf?
Surely, apple want to impress. I bet we will be blown away š
Showing redshift perf on amd cards on windows. How does this compare to current 2019 mp with 1-4 6800 and the m1ultra? I get the impression that even if apple could make mpx module with dual 7900s it would not be much faster since it would have to be downclocked to meet power standards.
If I remember correctly a m1ultra is about 6:30 in this benchmark and a m2max 7:30, placing a potential m2ultra around the 4min mark (240s)
That is, similar perf as a single 6900 or 70% of the power of a 7900xtx
Clearly this will not be what powers a mac pro as the maxed out system.
A quad m2 max would still barely be on par with a maxed out mp2019.
Wouldnāt it make sense to be able to deliver at least 50% more perf as the high end? And making āthe common configā at 12k be about 2x perf?
Surely, apple want to impress. I bet we will be blown away š