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 All Models Mac Pro X,Y…WHAT IF…
- Thread starter ZombiePhysicist
- Start date
- Sort by reaction score
I think there are parallels to what Disney has done to starwars and what Apple has told the think different/Mac pros. When you tell your core audience you don’t care about them, you blow up a bridge you may need to cross.
The Mac pro/enthusiast/think different crowd is what Apple leaned on to save itself when steve came back and Apple almost fell to bankruptcy.
Pretty decent chance that routinely covering 8K display is coupled to Apple rolling out Thunderbolt 5 ( and DisplayPort 2.x ) ... which is probably 1-2 generations forward. M4/M5. And when do some Studio configuration would probably cover 4 displays also. There isn't going to be a big gap there in the userbase.
[ Yet another reason why the Mac Pro trying to track ever M-series iteration doesn't make sense. The userbase is small so the incremental stuff isn't going to draw tons of 'takers'. Number of folks running just one 8K display isn't that large. 2x , 3x , 4x ,etc of those is just pointing at an increasingly smaller group of folks. ]
But they need to upgrade the abortion that is the Mac Pro right now. An M3 Ultra would actually be very competitive across the board except for the scant few die hard 3D GPU rendering guys who want better than RTX 4090 (A6000) performance.
I am pretty convinced that the M3 Ultra chip will pretty much be faster at everything compared to any 2019 Mac Pro config that is running OS X.
Of course the case for keeping on Intel is so that there is a viable dual boot windows mechanism. 🤷♂️
I am pretty convinced that the M3 Ultra chip will pretty much be faster at everything compared to any 2019 Mac Pro config that is running OS X.
Of course the case for keeping on Intel is so that there is a viable dual boot windows mechanism. 🤷♂️
There is a big gap between "they need to" and "they might like to" . Apple doesn't "need to". Folks posted 100's of messages to that same affect during the 2015-2018 time period and Apple delivered nothing. Primarily because they didn't "need to" ( it wasn't going to significantly impact the tactical or strategic direction of the Mac ecosystem at all. Apple grew revenues and unit volume during that period just fine. Hence no 'need' fiscal , strategic , or tactical 'need'. )
The M3 gen probably isn't going to 'cure' soldered RAM , iGPU only , and BTO pricing for upgrades. Very likely a large block of MP 2019 users will grumble at the M3 version also and simply buy used and/or 'bargain bin' old RAM , video cards to keep their current, Intel systems going for another couple of years. ( very much like what happened in 2015-2017 time frame when the MP 2013 was the 'abortion'. ). As long as macOS on Intel is getting active , robust support many of them are not going to move.
For the primarily 'hyper focused' GPU computational rendering folks, there isn't going to be a large gap between the MP and Mac Studio(MS) if there is MP is still only stuck with just one, shared with MS, SoC then there is really no hurry. If the GPU is the same and the GPU is the just about the only thing that matters, putting the M3 Ultra in a MP isn't really going to sell to that particular crowd. Bulk of them with now substantive additional I/O throughput and/or data storage capacity needs are going to take the relatively more affordable option.
Folks who have a higher priority focus on non GPU I/O throughput are still going to perfer the M2 Ultra MP to the M3 Ultra MS ( lack of access to the large PCI-e backhaul bandwidth would still beat the TB4 sockets on a M3 Ultra system. )
"But this will mean there are more Studio sales than MP sales ". That is already true. M3-gen very likely isn't going to change that. A mystical M3 Extreme very probably would not either if it is priced extremely high.
Furthermore if Apple skips the M3-gen then the MP could relatively easily jump onto the M4 generation sooner rather than later. If Apple 'saddles' the MP with paying off the M3-gen then it cannot jump to the M4 generation sooner ( it has to help pay off the M3 work. ). The MP coasting on M2 Ultra would extend the lifecycle of both the support to that SoC ( Apple's vintage/obsolete countdown clock starts when the stop selling something. Discontinuing the MP 2023 faster means it is desupported faster. After spending $7-10+ K a substantive fraction of the owner userbase aren't going to prefer than option. Higher prices typically lead to systems being pressed into longer service lifetime cycles. )
If Apple had waited another 3-4 months to release the MP 2023 they could have placed a M3 Ultra in it in the first place. Pragmatically they could delay longer because they needed to complete the transition and MP 2019 was so 'old'. That baseline of the MP refresh cycle be long very likely has not gone away.
The initial release was coupled to the Studio's M2 transition , but that very likely is not necessary ( a requirement level 'need') over the long term. [ The Mini Pro didn't have to ship with the MBP 14" Pro, the iPad Pro didn't have to ship with the iMac M3, etc. etc. ]
The 4090 is on track to be superseded in late 2024 - early 2025 anyway. If Apple is going to try to 'chase' those class of cards they would be a system priced more like the studio to do the chasing , than one with higher upfront costs like the MP. Nvidia has already release a subset of the 4000-super offerings already. If M3 ultra is Q2-Q3 2024 Nvidia is likely to have moved the line anyway. There is always going to be a substantive gap between Apple and 2-3 other competitors who can release at different times spread out over the calendar year. In this space, Apple should deliver when they have very good , solid updates , than fixate on somebody else's calendar. ( AMD or Intel shoveling stuff out the door to beat Nvidia to some initial demo day fails at least as often as it suceeds. )
A likely significantly enough large fraction of configurations. However, folks with a 400-600GB (ECC) RAM data set requirements ... probably not. Folks with > 128GB VRAM resident problems probably yes.
Apple doesn't need something that covers every possible slot configuration of the MP 2019. They just need to cover enough so that they can get a decent ROI on doing a new MP. That is what is questionable. Just throwing 'specs' at the folks who didn't move with the MP 2023 isn't going to move most of them. It is 'form' (hyper modularity) not 'function' that a large fraction of the non-movers are hesitant about.
Apple should just put the M-series on a standard PCI-e card ( maybe squeeze M3 Pro onto a bus only powered card. Or bite-the-Aux-Power-bullet and do a M3 Max+ card. ). That way folks on legacy MP 2019 , Windows (or Linux) , and MP 2023 folks could throw a newer Mac into their systems. They wouldn't have to 'dual boot' again. They could have them both booted at the same time. Run a virtual Ethernet via the shared PCI-e bus so can mount shared disk and/or 'remote' screen on a private network.
If could put a "Mac inside Mac" that would be a bigger functionality gap between MP and MS. One of the current MP problems is Apple isn't doing much effective on differentiation. There is a small subgroup they are targeting that doesn't to be expanded to everything for everybody, but it could be bigger. Especially if willing to just run concurrently with something else ( like they do in many more places now because cannot dual boot).
I’m actually surprised Apple did not release some sort of MPX style card that was exactly what you mention a M series PCI card.
Maybe not enough Mac Pros to make it worth the engineering effort.
I’m pretty sure someone posted a breakdown of Mac model sales and the Studio was the lowest. I’m probably remembering wrong. I’ll research it.
I’m not holding my breath waiting for a new Mac Pro. Doubt it will happen.
Maybe not enough Mac Pros to make it worth the engineering effort.
I’m pretty sure someone posted a breakdown of Mac model sales and the Studio was the lowest. I’m probably remembering wrong. I’ll research it.
I’m not holding my breath waiting for a new Mac Pro. Doubt it will happen.
I’m probably dating myself, but this thread gives me xServe dejá vu…
Yeah I feel like "and the Mac Pro might not get updated" is more of a "so few people care about it that it's not really in the rumor sphere" rather than actual indications it's getting axed. At this point it's not a lot more effort than the Studio, and presumably it's got some crazy margins even if they only sell a couple thousand to the people who really need internal PCIe slots and storage.
(I don't think an Extreme is ever coming, though.)
I think the extreme may come. But it if it does, there is a decent chance it is offered on the studio as well.
I think apple knows the mac pro sucks as it was released in 2023. I'm not sure how much an M3 would improve many of the things that suck about the 2023, without an extreme version.
Doubt an Extreme could work with the Mac Studio cooling; Mn Max uses an aluminum heatsink, Mn Ultra uses one of the same size but made of copper...
Mn Extreme would most likely require a copper heatsink twice as large as the Mn Ultra in the Mac Studio, or one three or more times the size in aluminum...
Therefore, Mn Extreme would need a Mac Pro chassis; or, for those not needing PCIe slots, an all-new Mac Pro Cube chassis...?
This might be a bit off topic, but I have wondered if any enterprising individual has attempted to put the studio's guts into the G4 Cube's case.
You may be right. I was thinking with an extreme coming down to 3nm there would be enough of a reduction. But I haven't messed with the internals of the studio, so I defer to your comments.
And the M3max laptops are getting hot enough that their fans are blowing full force, but the lackey mac/tech press is not reporting on it:
2023 MacBook Pro M3 Max: Intolerably Loud Fan Noise
macperformanceguide.com
Still a good chance I'd say . . .
Srouji gave this low key, 'more to come by the way' (for the M family) shortly after last year's very very disappointing WWDC. It's not a forgone conclusion, but if the M family of chips has another tier on the way, the extreme seems likely. My guess is this comment was Apple's attempt to fill the post WWDC 23 blues with a little ray of extreme hope!!
No. It just doesn't make any sense at all to proactively block the card from being placed in a non Mac Pro 2019. All that does is just reduce the places where it can be used. It is like only soldering a USA 220V cable to a Mac Pro and trying to sell it. Folks in a house with normal sockets cannot use it.
If the card is a full self contained Mac ( e.g., Ethernet , 1-2 TBv4 ports , HDMI (or Type-A ) socket ) then don't even have to implement the virtual Ethernet PCI-e to get some utility out of it. Conceptually someone with a TB PCI-e expansion box could use it with a Intel/M-series Mac also. The PCI-e bus is dominantly just a standard power plug (along with Aux-power if provision those also). The PCI-e data bus wouldn't be used for generic traffic. Nothing on the host side would need to interact with it. Just a driver for virtual Ethernet. The rest of the host system would just 'talk' to what it thinks is a Ethernet socket. That is it. ( If there is a physical Ethernet jack on the card then just skipping some extra wires and picking up some private networking speed. It is 'neater' and probably more 'snappy' GUI experience ; not critically necessary. )
Just make zero sense to narrow the addressable market more. Even if Apple wanted to be in the "windows/linux" hate mode , there are other Macs could make this work with relatively easily.
MPX was largely invented to solve the 'remote', discrete Thunderbolt controllers mounted on a removable card problem. The "Mac on a card" system doesn't have remote TB controller issue at all. A 100% complete GPU+PCIe+TB subsystem is right there on the card. MPX is a 'solution' in search of a problem in that context. Extremely likely why it is dropped in the MP 2023... it is largely pointless. The thunderbolt subsystem is elsewhere.
Yes, MPX has a 500W power provision component to it. But used at the extreme pragmatically also critically couples the card to the MP 2019 case specific properties. Also made into a 4x wide card that is defacto coupled/integrated into the MP 2019 specific fan layout and even more hobbled into deployment targets.
Finally, the MP 2019 is on an active Vintage/Obsolete countdown clock. Apple isn't going to put money into something desupported before the Mac sitting on the card is desupported.
If that is that CIRP survey from a year or so ago... that is extremely likely just an example of flawed survey sampling; not Studio or Mini sales. There is little corroboration of that data from independent sources at all. Large screen iMac and MP fans keep resurrecting that skewed survey from the dead to promote their causes in groupthink exercises.
A Mac Pro system that primarily worships at the alter of mega power draw discrete GPUs. Probably not.
A Mac Pro that provisions between than TBv5 bandwith and more than one internal drive. Probably yes. (how quickly iterate updates might be debatable). But Thunderbolt still has the same core issues now that it had in 2012-2017. (e.g., top SSDs about a generation ahead on bandwidth requirements. ) The height of the bandwidth hurdle to get over may have shifted, but general problem is still there. Multiple internal drives.... Apple's rigid pricing of $400/TB is doing less than nothing to solve that. New days ... same old problem.
It might collapse down to just the rack model chassis. ( and folks improvise deskside usage with a backet that tips it up on its side vertically. ). If the volume drops more making two different containers is probably a stretch.
[ Far more so 'XClient' as opposed to 'XServe'. So form factor somewhat similar but substantively different use case. ]
When TSMC gets to High-NA EUV and the max die size does down (and costs go up) there is a pretty good chance Apple is going to need to make a more deliberate chiplet design at the top end of even their laptop line up. Smilar with RAM densities going up and up... at some point some , perhaps half-hearted, ECC will likely kick in ( likely because the RAM vendor sticks it in; not major effort on Apple's part. Apple will buy ECC if they have no other choice. )
Little to do with the "Extreme". The M3 came after WWDC 2023 ...no extreme there either. At WWDC timeframe Apple had multiple working M3 chips that he had seen. That there will be further updates to the M-series family is like the 'sky is blue' . There is not deep , earth shattering , revelation there.
The M3 Max being able to get close to covering a M1 Ultra on some graphics tasks is a type of 'tier changing'. Apple didn't have to pour die area on like ketchup to get to a new performance level. They basically rebuilt the graphics hardware approach to something new.
He is referencing the M1 in your context quote. He also said "... you have seen the family..." (seen as in past tense) The M1 gen/family has multiple dies ( with cousins in other gen/family ) . The M2 gen/family has multiple dies ( cousins). M3 really didn't have to wait long at all to see multiple cousins all at the same time.
There is nothing here that Apple is going to continuously expand the breadth of the M series line up forever as move further into the future.
There have been a couple of indications that Apple attempted at "extreme" solution . What is blocking it largely isn't technical/tape-out reasons. It is who is going to buy it at the prices Apple wants to sell reasons. Sjroui also explicitly talked about the M-series shipping to millions of millions of users. The Extreme would not. It isn't going to get close to a million.
Hmm. The 16” one I have from work; haven’t heard the fans once. 🤷You may be right. I was thinking with an extreme coming down to 3nm there would be enough of a reduction. But I haven't messed with the internals of the studio, so I defer to your comments.
And the M3max laptops are getting hot enough that their fans are blowing full force, but the lackey mac/tech press is not reporting on it:
2023 MacBook Pro M3 Max: Intolerably Loud Fan Noise
macperformanceguide.com
I doubt Photoshop or whatever stresses the system more than After Effects. But I also ran the Cinebench and still didn't hear fans. Once native Lightwave is released for Apple Silicon, I have a very good scene setup to do a test render to see if that ramps them up.
What will the M3 Mac Studio look like? | Digital Trends
Apple's Mac Studio is its most attractive and accessible desktop PC, offering tremendous power in a refined package. What will Apple do to keep it on top?
www.digitaltrends.com
It's interesting, but the original title of this article has seemingly been changed at least in how it comes up in metadata from:
Not surprised, since "Mac desktops need a serious boost" is better as a clickbaiting title but "What will the M3 Mac Studio look like" is all about engineering something that will pop up in google searches as people wonder when the new Pro is coming.What will the M3 Mac Studio look like? | Digital Trends
Apple's Mac Studio is its most attractive and accessible desktop PC, offering tremendous power in a refined package. What will Apple do to keep it on top?
It's interesting, but the original title of this article has seemingly been changed at least in how it comes up in metadata from:
View attachment 2347125
It's kind of a shame that DigitalTrends is putting out a low-effort articles like this these days; obviously Apple is going to update the Studio, but they've always lagged behind the more important laptops the entire lifecycle of Apple Silicon, this isn't surprising. (I don't imagine they wanted a whole nine months between the M2 MBP and the M2 Mac Studios, but also I don't think they were trying to replace the M1 version in under a year, especially since there were issues with producing them fast enough for demand at the very beginning.) I'd expect the high-end M3 chips in desktops in the spring or WWDC timeframe.
(Now, does Apple keep selling the Pro, since I don't imagine they sold a lot of the "hey you can put slots in but you need to pay $1K over the last model for the privilege" model? That's a more interesting question. I don't think Apple particularly wants people to buy it.)
Welp, whether by design or accident, Apple has probably thrown the Mac Pro well and truly under the bus again. I’m leaning toward by design for numerous reasons. Don’t think there‘ll be a self flagellating apology tour like there was in 2017 after the glossy butt plug Mac Pro fiasco. Like the iMac Pro and 7.1 the 14,8 will quietly shuffle off to sad panda abandonware land. Pity.
www.lifewire.com
Why the Mac Pro Will Never Be the Ultimate Mac
Apple's high-end desktop computers, the Mac Pro, is due an upgrade, possibly to the M3 chip, but why would anyone buy them when a MacBook can do everything most people need?
In the mean time, E1.S drives are becoming an appliance thing, but Apple is nowhere:
They know Apple isn’t interested in third party, state of the art, high performance storage. That Apple has abandoned the HEDT market and their users after the brilliant, but abandoned 7.1, is terribly frustrating and unnecessary.
What might have been.
Imagine MacOS on one of these.
No need to reinvent to reinvent the wheel for best in class performance.
www.storagereview.com
Imagine MacOS on one of these.
No need to reinvent to reinvent the wheel for best in class performance.
Dell Precision 7875 Review: 96-Core Threadripper Pro and Dual RTX GPU Powerhouse
The Dell Precision 7875 is a highly impressive high-end workstation with an AMD Ryzen Threadripper Pro 7000WX series CPU.
www.storagereview.com