Thanks for correcting my incorrect assumption that they could combine two Ultra chips into one. Do you think there will be a tier above the Ultra, or do you think that this will effectively be the next generation (i.e. M3 Ultra would be considered an M2 “Extreme”, and on and on)?
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.
Gurman: All-New Mac Pro Still in Testing, But 'M2 Extreme' Chip Likely Canceled
- Thread starter MacRumors
- Start date
- Sort by reaction score
What they’ll always get, though, regardless of what Intel is doing, is the fastest Mac money can buy. There are probably folks that, more than anything else, want some Intel related number to be lower than the equivalent Apple Silicon related number. Except for VERY specific circumstances related to the architecture of the SoC, that’s not going to happen. For those that want a peak performance Mac that’s 15 to 20% more performant than their prior peak performance Mac? That’s well within expectations.
When I say it doesn’t look good now, it’s because I’d tell anyone today if they needed raw performance (and not macOS compatibility at all), they should look into non-Mac solutions. That’s likely to get even worse as Intel/AMD/Nvidia pours more and more power into what they’re putting out. If someone wants Intel performance, the best way to get that now and in the future is with Intel systems, likely something custom built by someone that has studied the varied landscape of PC desktop parts and know how to pick precisely which parts mesh well with others to provide that performance.
A 48 core cpu with single thread perf of about 2000 and multi at 30000 is exactly what the high end mp should be. It will not beat a 64 or 96 core threadripper at some brute fore tasks but would be very competitive for everything else. If there is no cpu of that magnitude, the purpose of the mp is even more narrow than now. And gpu power that is at 1/4 of a current 2019 mp? This just seems implausible and would be a joke to the market. No way to spin this as a real pro machine except for some niche audiowork. I still think Apple have a solution planned that will be better than the 2019 model. They are just good at keeping it secret. I hope 😂
Come to think of another possible solution. “Continuity pro” where Apple basically makes a seamless clustering solution so that we add ultras and macbook and whatever to our compute pool. Some kind of solution that makes this invisible to the software? There would oc be latency and bandwidth issues but might be great for many tasks like rendering where there isn’t a realtime need.
Apple is about as likely to flip the DIMM die into 4th level cache as pushing the LPDDR5 die into that role.
Nominally the macOS on M-series optimized applications would expect the LPDRR5 RAM to be there Uniform and Unified. To make that disappear is decently likely to bust some built in assumptions in some applications.
Apple's memory management system already can compress and swap RAM pages into other locations ( and re-inflate them when applications touch them again) all transparently to applications. Very similar with the file system which will dynamic cache file data into RAM as it see's fit (again in a manner that is largely transparent to the applications. Some nitpicking apps like to do their own file caching but many do not. )
The DIMMs could get treated like a RAM SSD by the memory and file management systems there would not necessarily have much impact changes on the applications at all. There is also a "RAM disk" utility that again could be used to extend the standard interface around a RAM SSD. (instead of carving it out of system RAM ... use SSD targeted RAM. )
Open Activity Monitor on MacOS that has been running for long while with Apps reading/writing data and the "cached files" usage can get decently high. If push that out to optional DIMM RAM than all RAM can to allocated to the apps more direct RAM usage. That won't jump a 196G system into the 600GB range but would be susbtantive extra GBs of RAM.
Is macOS really the #1 go to operating system for 1.5TB data footprint EDA workloads? I would expect Linux to have a bigger share of that market.
Treating the LPDDR5 RAM purely as 'cache only' RAM is likely going to run into problems with the integrated GPU cores. Apple could layer some complicated cacheing set up which always put the GPU data in there but then back to copying data over at least some subset. ( the whole point of unified memory was to do away with copying). Cannot really set things up so that the GPU doesn't any caching. (going out to DDR5 would be 'too slow' ) . If make it so the GPU only using the LPDDR5 are non-cache RAM no have apps with non homogenous system RAM (and likely require application changes. Happens for Intel for Optane or HBM hybrid app set ups... more than likely macOS apps would not be immune either ).
Apple recently introduced an API so 'GPU' can do disk system reads. Same thing as with the file system where it should be straight forward to use RAM SSD to cache file data for lower latency access.
If there is 190GB cache with no ECC which data integrity sensitive scientific/engineering workloads are going there?
Apple's GPU don't believe 64-bit floats exist is another class of workloads that are probably not going there either.
Similar issue with the volume that Apple finds interesting enough. Trying to make the Mac Pro being an everything for everybody workstation is probably not where Apple is going long term.
Are they truly limited to using Max dies fused together to make an Ultra? Could they not use something more than a Max to start with?
Supposing the M2 Max will be 12 core (8+4), Could they not do a 10+2 version as the basis for the Ultra's two Max parts then fuse them together? Giving 20 P cores and 4 E cores?
Or, extend that thought further, could they not push the Max die further and have more than 12 cores total per die? Other than die size, is there something in the current design paradigm that prevents them from having, say 16 cores (12+4) in each half of a new Ultra die?
Further, what is stopping them from ramping voltage on these M2 class chips for the Mac Pro, to push the frequency higher? Do these M2 class chips start to fail at faster clock speeds (like 4GHz?).
I ask these questions as they obviously have experience in adding more cores, as in the jump from M1 to M1 Pro. So could they not just add more cores once more and create a larger base die for the desktops, one that is both running at a higher voltage/frequency, and one that has more cores?
Supposing the M2 Max will be 12 core (8+4), Could they not do a 10+2 version as the basis for the Ultra's two Max parts then fuse them together? Giving 20 P cores and 4 E cores?
Or, extend that thought further, could they not push the Max die further and have more than 12 cores total per die? Other than die size, is there something in the current design paradigm that prevents them from having, say 16 cores (12+4) in each half of a new Ultra die?
Further, what is stopping them from ramping voltage on these M2 class chips for the Mac Pro, to push the frequency higher? Do these M2 class chips start to fail at faster clock speeds (like 4GHz?).
I ask these questions as they obviously have experience in adding more cores, as in the jump from M1 to M1 Pro. So could they not just add more cores once more and create a larger base die for the desktops, one that is both running at a higher voltage/frequency, and one that has more cores?
So the Mac Studio is going to be renamed to Mac Pro and the price will increase by $2000 for both CPU versions?
Got it.
This is dumb. Are they just going to keep the Intel Mac Pro around and upgrade that? I always wondered since they made such a big fuss about the Mac Pro being upgradeable how they would do that with Apple Silicon. Apparently the answer is there is no Mac Pro? None of this is making any sense.
Got it.
This is dumb. Are they just going to keep the Intel Mac Pro around and upgrade that? I always wondered since they made such a big fuss about the Mac Pro being upgradeable how they would do that with Apple Silicon. Apparently the answer is there is no Mac Pro? None of this is making any sense.
Several months before the Mac Studio was released I really needed a new computer… So i bought a maxed out 27”iMac with the thought that once the AS MP was announced I’d buy one of those as I certainly wasn’t goign to buy an Intel MP in late 2021…
Then the Mac Studio gets released and I think to myself I should buy one of those… I’ve had this thought many times over the better part of the last 12 months, especially as I render 8K footage to Pro Res 4444XQ… but as the year has rolled on and the promise (rumour) of the Mac Pro and the possibility of an Extreme SoC I thought, no I’ll wait… and when that was obviously not happening this year I was sad… but I thought… well at least it’ll be in the next few months of the new year… and at least if the AS MP is out of my budget well at least I can buy a M2Ultra Mac Studio instead and I’ll be good for many years…
Now it sounds like the AS MP might be… a bit ****… I really hope the M2 Mac Studio is going to be decent… hell it might not be updated in March at all… that would be a proper ballache….
Meanwhile the intel iMac is churning away quite happily…. slowly… but never missing a beat… 😂
Then the Mac Studio gets released and I think to myself I should buy one of those… I’ve had this thought many times over the better part of the last 12 months, especially as I render 8K footage to Pro Res 4444XQ… but as the year has rolled on and the promise (rumour) of the Mac Pro and the possibility of an Extreme SoC I thought, no I’ll wait… and when that was obviously not happening this year I was sad… but I thought… well at least it’ll be in the next few months of the new year… and at least if the AS MP is out of my budget well at least I can buy a M2Ultra Mac Studio instead and I’ll be good for many years…
Now it sounds like the AS MP might be… a bit ****… I really hope the M2 Mac Studio is going to be decent… hell it might not be updated in March at all… that would be a proper ballache….
Meanwhile the intel iMac is churning away quite happily…. slowly… but never missing a beat… 😂
There will never be modular Pro with current state of soc development. All of the speed gains come from soldered RAM and storage.
Hopefully not only IMAX or Dolby audio and video stuff but also 3D (near real-time rendering in Blender anyone?), scientific simulation etc. etc. — "pro" is quite a wide market.
So a dual M2 Ultra instead as an option?
Or a much faster-clocked M2 Ultra?
Or the main USP of Mac pro being upgradability/expandibility?
Or a much faster-clocked M2 Ultra?
Or the main USP of Mac pro being upgradability/expandibility?
They have pulled a u turn. It’s called the Mac Studio.Apple kinda settled this discussion in April 2017
Transcript: Phil Schiller, Craig Federighi and John Ternus on the state of Apple's pro Macs | TechCrunch
You have already read the news. But we thought we would also use this opportunity to share a transcript of the interview with Phil Schiller, Senior Vicetechcrunch.com
Unless they’re doing a U-turn once again.
I love Macs, but for $10,000 I can have one beast of a PC. 🤷♂️
I don’t believe that you recognize what is driving people insane. It’s not that they eliminated the chip that didn’t exist, but it’s what it says about what this product is really for. And more importantly, who it is for. Apples about to do it again.. they’re gonna offer a pro, that’s only for one kind of pro. And they’re gonna leave everybody else out to the windows PC’s. People who want to use Mac’s, don’t want to see this.
Last edited:
Ohhh another product that Apple fans are in love with that is never gonna come out.
I hate to break it to you, but Apple is never building a Mac that takes external video cards ever again.Apple has priced the MPX W6900 ($6K ) higher than a MPX W6800 Duo ($5K). Both use the same fundamental die from AMD. So two dies are less than one (albeit better binned) die.
A 7900 replacement at the same price point would have a very similar ( so expensive ... how many folks are really going to buy it ) problem as the $10K M2 Extreme SoC.
Apple and AMD collaboratively worked on AMD GPU generational drivers when Apple both 2-3M AMD GPUs for iMac and MBP 15" per year. That volume to amortize development costs over is gone. That was being used to help defray the costs of the $2-6K GPUs for the 2019 Mac Pro driver development. When the Mac Pro has to pay for it all with only maybe 3 MPX modules the costs are going to go down? Probably not.
If the vast majority of end users bypass the MPX W7900 for the low cost, off-the-shelf version there is substantially even less money to pay for the driver work.
This story about "too few users to sell to leads to Apple not interested in making a product" impacts the high end GPU space also.
With the used GPU glut sales of W6800 and W6900 are likely falling pretty fast. That really doesn't do much to 'enable' a next generation for Apple. The MP 2019 user base likely isn't getting substantively bigger. (if leave the core system with a 3 year old CPU ... how is that competitive with 2023 workstation offerings coming from Intel/AMD? )
The Mac Pro was far more dependent upon the rest of the Mac market to help manage kernel/driver/software costs than most folks want to admit. The 580X , W5700X , 5500X all had placements in other Macs (iMac , MBP dGPU).
Not sure why AMD would be jumping up and down at opportunity window of selling another 10-30K units when they are shooting for 100's of thousands per quarter.
AMD Will Reportedly Have 200,000 RX 7900-Series Cards for Sale
The more the merrier!www.tomshardware.com
The price points that Apple is selling at and the numbers likely to drive via Mac Pro sales might make some noticable impact on MI210 or future MI300 series cards. AMD position in the GPU market is not what it was back in 2016-2018. They don't really need for Apple to step in and sell GPU chips they can't sell elsewhere. Even more so to an Apple that would be buying in sub 25K/quarter volume rates. Buying another 20K/quarter MI210 might.
Pretty good chance that folks with a MP 2019 system that want a 7900 will do the same thing that the folks who wanted to run a 3090 30-70% of the time did. Boot the Intel system over into Windows and use the drivers there. Costs AMD/Nvidia/Apple an additional approximately nothing. Certainly what the hackintosh folks are doing.
No worries mate if your audio needs are already met with Appolo X racks. Simply stay with that.
There are plenty of other disciplines beyond audio where special purpose high-speed hardware processors/accelerators would be useful - especially if the next Mac Pro has a rack mount version similar to what the current Mac Pro has.
Where the hell is the new Intel Mac Pro that Tim Cook said they'd release when they announced the M1 processor transition? Did everyone just forgot he said there'd be one more Intel Mac?
Putting aside expandability, what are the use cases for something with more ooomph than a fully kitted out Mac Studio?
I’m genuinely curious - I’m sure there are plenty, but it’s outside my sphere of knowledge.
I’m genuinely curious - I’m sure there are plenty, but it’s outside my sphere of knowledge.
There was a fairly major 27" iMac revision in 2020 (same external design - significantly new innards with T2 chip etc.) as well as new GPU options for the Intel MBP.
(I'm sure he didn't promise a new Mac Pro - just that there were new Intel Macs in the pipeline.)
Another explanation is Apple’s Frankenstein assembly of CPU chips just doesn’t scale beyond two. Routing thousands of high-bandwidth interconnects is beyond the current state of magic.