Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
John was introducing the UltraFusion process when he said that. That process may also allow them to connect two Ultras together, much like what you’re suggesting when you say they may use multiple M1 Ultras in the Mac Pro. It’s a plausible technical solution to the problem. That’s the whole point of the “chiplet” approach. I think you’re getting hung up on semantics, although I’ll concede that it was not a live event and everything said was carefully reviewed.
It isn’t hard to imagine how John would introduce the idea, “UltraFusion not only allows us to fuse two M1 Max together and create the M1 Ultra, but it also allows us to connect two M1 Ultras together …”
What does it take to make a motherboard compatible with PCIe? Just a bridge chip? Would Apple have to change or build new tech into a "M2" chip to make it compatible with a PCIe bus? Would the tech in UltraFusion be useful in creating a high speed interconnect in M2 chips to speed up PCIe interface?
My guess is that the Mac Pro will use the same M1 Ultra as the Mac Studio does. The difference will be in the system around the SoC. With a larger form factor, they have more cooling potential and could bump up the clock rates a little... but really, the M1 Ultra is a monster as it is (both in terms of size and performance). I would just take what Turnes said at face value, this is already the last of the M1 series. And I think we will see a Mac Pro that uses it.
So what could differentiate the Mac Pro? In a word: expandability.
1) PCIe slots. The M1 Ultra seems to have plenty of I/O potential, and a fast PCIe bridge chip would easily enable a lot of expansion potential.
2) Drive bays. The Mac Pro would have the same built-in super fast SSD, but in a large case a whole lot of additional storage can be accommodated.
3) RAM. This is where it gets tricky. The Apple Silicon approach is to use in-package memory, and there are real constraints on how much can be put into a single package. Some Pros just need more than can be fit into a single package, or more than is worth building in the TSMC production run. So conventional DIMMs are needed to supplement the super fast in-package memory. The question is, how does OSX use it? Apple seems to want to keep the programming model simple (i.e. CPU/GPU shared memory with a flat/uniform 64-bit virtual address space), so having some fast vs slow areas of memory doesn't seem like the direction they want to go in (although they could and just rely on the M1 Ultra's ENORMOUS caches). They are already doing virtual memory paging to flash, however... so why not do virtual memory paging to the DIMMs instead? Big DMA data transfers between in-package and on-DIMM memory across the very fast PCIe 5.0 lanes would ensure that the available bandwidth is used as efficiently as possible, and the latency is masked by the big (page-sized) transfers. A 128GB working memory (the in-package RAM) is huge, so doing VMM to get to the expanded pool is not as bad as you might think. Such a memory scheme may even just sit on PCIe cards so buyers only need to pay for the DIMM slots if they really need it. Such "RAM disk" cards have been around for ages, but are usually hampered by lack of direct OS support... and issue Apple could fix easily in their kernel.
One thing for sure, they won't include any drive bays for 5.25" devices or conventional hard drives. There is no advantage to having those devices inside of the case from a power, cooling or speed perspective. I also suspect the use of some kind of proprietary RAM expansion.
One thing for sure, they won't include any drive bays for 5.25" devices or conventional hard drives. There is no advantage to having those devices inside of the case from a power, cooling or speed perspective. I also suspect the use of some kind of proprietary RAM expansion.
That depends on what their big customers with lots of money want. The advantage is that its all in one box, rather than having external enclosures which can be problematic in some situations. I tend to agree that the age of 5.25" form factor drive bays is gone, but I'm also not the target audience for such a capability.
And an earlier reply to my comment:
zimmie said: On the topic of RAM, there's nothing inherent in the M1's design which precludes off-package RAM. They have a RAM controller on the chips, and the RAM controller is shared between CPU and GPU cores, but there's no fundamental reason the RAM couldn't be in DIMMs. Apple just hasn't chosen to do that. They might do so with the Mac Pro, or they might not. I don't see them doing a tiered memory structure, though. They just went to significant lengths to do away with NUMA concerns on the M1 Ultra. With the introduction of the Mac Studio, I wouldn't be surprised to see the Mac Pro go primarily rackmount. Very few people need more computing power than the Mac Studio offers at their desks. Almost everyone who does need more computing power is in an environment where they can rack the computer in a closet. For example, recording studios, film studios, scientific labs, and so on all have 19" rack space for other equipment, so putting specialist workstations in there isn't a stretch. That said, rackmount would mostly be relevant for a box with several full-height, full-length PCIe slots (e.g., to add hardwired audio and video inputs), and I'm not yet convinced Apple is interested in that at all. They might say the future is a rackmount interface which connects to the system via Thunderbolt. I'd be curious to know what they have seen the current rackmount Mac Pro doing.
The in-package memory has much tighter timing tolerances because the memory configuration is fixed, lower signal driver power levels, and at very short distances. I would imagine that their memory controller takes full advantage of those facts, and cuts a lot of the complicated corners that dealing with DIMM slots creates. So, I disagree: I do not think their memory controller could support out-of-package memory without some serious work, and it would represent a large power increase and performance impact.
What I suggested isn't a software-visible tiered memory structure. It is just an alternative fast backing-store for the existing virtual memory system that all software currently works with (currently backed by flash memory). Implementing this would require no changes to the M1 architecture and very little OS change. The Mac Pro has a very small market (especially with the Mac Studio now taking a chunk of it), so custom work to support it doesn't make a lot of sense for Apple. That's a big reason why I think the M1 Ultra is what we will see in the Mac Pro. And probably just one of them as going multi-chip is a lot of specialized added hardware design work that I don't think they want to do.
The M1 Ultra has a pretty amazing amount of compute, after all. Bump the clock rate a little and you give it a small edge over the Mac Studio.
A rack mountable full-sized case (but still a desktop workstation) which can hold lots of extra drives, memory, and PCIe cards would differentiate it from the Mac Studio.
I doubt they will do this, but one thing they could do fairly easily in such a form factor is put the M1 Ultra motherboard itself on a PCIe card so one case could hold multiple of them (the case becomes just a PCIe backplane then). How such a machine would be used becomes more challenging though and would take them away from their preferred programming model of a single shared memory space for many CPU/GPU cores. Without a lot of OS work such a machine would look like several Macs on a high speed network... has some uses, but gets pretty obscure and way out of the consumer space. Then again, it is the "Mac Pro" so who knows?
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
So what do we make of the myriad rumors going way back of the Jade 4C-Die alongside those of the Jade 2C-Die which came to fruition with the M1 Ultra? It seems unlikely that was a very similar codename for a next gen 4x SoC design. The Ternus comment of course complicates all this, so did they maybe test that design at one point and decide to skip it this gen for some reason? Certainly possible, but it seems like it'd be a disappointment for many if the Mac Pro tops out with the same top end as the Studio. I feel like there's gotta be another trick up their sleeve.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
So what are we to make of the myriad rumors going way back of the Jade 4C-Die alongside those of the Jade 2C-Die which came to fruition with the M1 Ultra? It seems unlikely that was a very similar codename for a next gen 4x SoC design. The Ternus comments of course complicates all this, so did they maybe test that design at one point and decide to skip it this gen for some reason? Certainly possible, but it seems like it'd be a disappointment for many if the Mac Pro tops out with the same top end as the Studio. I feel like there's gotta be another trick up their sleeve. But, maybe it really is a case of them hitting a wall with production and pulling back a bit, much like it was rumored a larger iMac was going to come and they hit walls with that and reshuffled. The Ultra in a Mac Pro first gen that matches 28 core Xeon performance and most/all(?) of the current AMD GPU offerings isn't terrible, but it's not a great final act if the same power also comes in the little Studio box.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
So what are we to make of the myriad rumors going way back of the Jade 4C-Die alongside those of the Jade 2C-Die which came to fruition with the M1 Ultra? It seems unlikely that was a very similar codename for a next gen 4x SoC design. The Ternus comments of course complicates all this, so did they maybe test that design at one point and decide to skip it this gen for some reason? Certainly possible, but it seems like it'd be a disappointment for many if the Mac Pro tops out with the same top end as the Studio. I feel like there's gotta be another trick up their sleeve. But, maybe it really is a case of them hitting a wall with production and pulling back a bit, much like it was rumored a larger iMac was going to come and they hit walls with that and reshuffled. The Ultra in a Mac Pro first gen that matches 28 core Xeon performance and most/all(?) of the current AMD GPU offerings isn't terrible, but it's not a great final act if the same power also comes in the little Studio box.
No, the main sources behind the iMac rumor have said that everything they were interpreting as a new 27-inch iMac is accounted for by the Studio Display with the A13 + camera + microphones + sound system. It’s understandable they thought that = iMac.
Same deal for the rumored Mini Pro — Kuo and others have backed away, because everything they were seeing when they forecast that is accounted for by the Mac Studio. I don’t think we will see the Mini or iMac or Air get the M-series Pro, which is made from subpar Max chiplets that are cut down. It’s not a mass-production item, indeed the goal is to produce as few of them as possible!
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
So what do we make of the myriad rumors going way back of the Jade 4C-Die alongside those of the Jade 2C-Die which came to fruition with the M1 Ultra? It seems unlikely that was a very similar codename for a next gen 4x SoC design. The Ternus comment of course complicates all this, so did they maybe test that design at one point and decide to skip it this gen for some reason? Certainly possible, but it seems like it'd be a disappointment for many if the Mac Pro tops out with the same top end as the Studio. I feel like there's gotta be another trick up their sleeve.
The thing about UltraFusion is it’s not a new technique — AMD and Intel have been using it for years. That said, it sounds like Apple is really pushing the boundaries of the technology, so it’s promising but because they are breaking new ground it isn’t yet clear what the limitations and/or possibilities are. See the section titled “UltraFusion: Apple’s Take On 2.5D Chip Packaging,” here:
Seeing this makes me think about “persistent memory” — a technology Intel has invested a lot in (Optane DC), and while it’s currently aimed at servers, their road map for it has/had a second phase focused on workstations. There are people here who moan about the lack of additional user-replaceable internal storage, but it’s possible they are thinking too small, long term. User-replaceable persistent memory, supplementing both Unified Memory and internal storage, is more like it…
zimmie said: On the topic of RAM, there's nothing inherent in the M1's design which precludes off-package RAM. They have a RAM controller on the chips, and the RAM controller is shared between CPU and GPU cores, but there's no fundamental reason the RAM couldn't be in DIMMs. Apple just hasn't chosen to do that. They might do so with the Mac Pro, or they might not. I don't see them doing a tiered memory structure, though. They just went to significant lengths to do away with NUMA concerns on the M1 Ultra. With the introduction of the Mac Studio, I wouldn't be surprised to see the Mac Pro go primarily rackmount. Very few people need more computing power than the Mac Studio offers at their desks. Almost everyone who does need more computing power is in an environment where they can rack the computer in a closet. For example, recording studios, film studios, scientific labs, and so on all have 19" rack space for other equipment, so putting specialist workstations in there isn't a stretch. That said, rackmount would mostly be relevant for a box with several full-height, full-length PCIe slots (e.g., to add hardwired audio and video inputs), and I'm not yet convinced Apple is interested in that at all. They might say the future is a rackmount interface which connects to the system via Thunderbolt. I'd be curious to know what they have seen the current rackmount Mac Pro doing.
The in-package memory has much tighter timing tolerances because the memory configuration is fixed, lower signal driver power levels, and at very short distances. I would imagine that their memory controller takes full advantage of those facts, and cuts a lot of the complicated corners that dealing with DIMM slots creates. So, I disagree: I do not think their memory controller could support out-of-package memory without some serious work, and it would represent a large power increase and performance impact.
What I suggested isn't a software-visible tiered memory structure. It is just an alternative fast backing-store for the existing virtual memory system that all software currently works with (currently backed by flash memory). Implementing this would require no changes to the M1 architecture and very little OS change. The Mac Pro has a very small market (especially with the Mac Studio now taking a chunk of it), so custom work to support it doesn't make a lot of sense for Apple. That's a big reason why I think the M1 Ultra is what we will see in the Mac Pro. And probably just one of them as going multi-chip is a lot of specialized added hardware design work that I don't think they want to do.
The M1 Ultra has a pretty amazing amount of compute, after all. Bump the clock rate a little and you give it a small edge over the Mac Studio.
A rack mountable full-sized case (but still a desktop workstation) which can hold lots of extra drives, memory, and PCIe cards would differentiate it from the Mac Studio.
I doubt they will do this, but one thing they could do fairly easily in such a form factor is put the M1 Ultra motherboard itself on a PCIe card so one case could hold multiple of them (the case becomes just a PCIe backplane then). How such a machine would be used becomes more challenging though and would take them away from their preferred programming model of a single shared memory space for many CPU/GPU cores. Without a lot of OS work such a machine would look like several Macs on a high speed network... has some uses, but gets pretty obscure and way out of the consumer space. Then again, it is the "Mac Pro" so who knows?
They objectively do not have tighter timings. The on-package RAM on the base M1 is LPDDR4X. Exactly the same standard has been used for off-package RAM in the Intel MacBook Pro models for the last several years. The M1 Pro and Max use LPDDR5, which has slotted variants. HBM2 is enough to require the RAM be on-package, but they're not using that. Putting the LPDDR4X or LPDDR5 on-package saves a few microwatts from the shorter traces, but that's it.
Off-package RAM as a first-tier swap level could be done, but would give developers inconsistent memory performance. Again, Apple just went to ridiculous lengths to engineer away NUMA specifically because inconsistent memory performance isn't good enough. I don't see them adding it back in when they could instead just connect their memory controllers to DIMM slots. Sure, DDR5 DIMMs are rare right now, but that's not a limitation as far as Apple is concerned.
I wouldn't be surprised to see an M2 Ultra as the first M2-family chip, introduced in the Mac Pro at WWDC.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
John was introducing the UltraFusion process when he said that. That process may also allow them to connect two Ultras together, much like what you’re suggesting when you say they may use multiple M1 Ultras in the Mac Pro. It’s a plausible technical solution to the problem. That’s the whole point of the “chiplet” approach. I think you’re getting hung up on semantics, although I’ll concede that it was not a live event and everything said was carefully reviewed.
It isn’t hard to imagine how John would introduce the idea, “UltraFusion not only allows us to fuse two M1 Max together and create the M1 Ultra, but it also allows us to connect two M1 Ultras together …”
Look, he made it pretty clear that the Ultra was the last M1 chip. I don’t know why people insist that isn’t true. He didn’t say it had anything to do with Ultra /fusion, or anything else, just that the Ultra was the last. Earlier on, when they announced the Pro and Max versions, they could have said that too, and then popped out the Ultra with the UF connect, and acted as though it was just the same chip.
but they didn’t. And like it or not, that means something. What would have wrong with not saying anything? It’s not semantics. Semantics is something that’s interpretable. A definitive statement is just that.
the other thing thats] you guys are forgetting is that the cost of making these chips increases more than the added area because of increased defects and the risk of unusable chips. The greater percentage of wafer area a chip takes up, the more the cost. It’s a $1,000 upgrade to go from a 48 core Ultra to the 64 core version. And the cheapest Ultra costs more than twice what the Max version costs. So, even if they would do it, this new chip would likely cost at least three times as much. Maybe more. would that be worth it? I’m not so sure.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
So what do we make of the myriad rumors going way back of the Jade 4C-Die alongside those of the Jade 2C-Die which came to fruition with the M1 Ultra? It seems unlikely that was a very similar codename for a next gen 4x SoC design. The Ternus comment of course complicates all this, so did they maybe test that design at one point and decide to skip it this gen for some reason? Certainly possible, but it seems like it'd be a disappointment for many if the Mac Pro tops out with the same top end as the Studio. I feel like there's gotta be another trick up their sleeve.
Well, they could use two chips. People seem to think that Apple has come out with a more advanced connector than Intel’s, and AMD’s versions, and that’s it for Apple. They can’t possibly think of a faster way of connecting two chips together on a mobo. Why would people think that?
I just watched MacBreak weekly. Gurman was on it, and was talking about this very thing. He did mention a possibility of a four chip solution, but he thought it was likely to be an M2 thing - if Apple did it. That would be more likely, I think. It’s also possible, as you say, they tested it and it proved deficient, at least, for now. Not everything scales up linearly. Take multi cpu core chips. They never perform as all of the cores would indicate, including Apple’s chips.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
So what do we make of the myriad rumors going way back of the Jade 4C-Die alongside those of the Jade 2C-Die which came to fruition with the M1 Ultra? It seems unlikely that was a very similar codename for a next gen 4x SoC design. The Ternus comment of course complicates all this, so did they maybe test that design at one point and decide to skip it this gen for some reason? Certainly possible, but it seems like it'd be a disappointment for many if the Mac Pro tops out with the same top end as the Studio. I feel like there's gotta be another trick up their sleeve.
The thing about UltraFusion is it’s not a new technique — AMD and Intel have been using it for years. That said, it sounds like Apple is really pushing the boundaries of the technology, so it’s promising but because they are breaking new ground it isn’t yet clear what the limitations and/or possibilities are. See the section titled “UltraFusion: Apple’s Take On 2.5D Chip Packaging,” here:
Seeing this makes me think about “persistent memory” — a technology Intel has invested a lot in (Optane DC), and while it’s currently aimed at servers, their road map for it has/had a second phase focused on workstations. There are people here who moan about the lack of additional user-replaceable internal storage, but it’s possible they are thinking too small, long term. User-replaceable persistent memory, supplementing both Unified Memory and internal storage, is more like it…
The major difference is that nobody else, AMD, ATI, Intel or Nvidia have figured out how to make an interconnect that would allow two GPU chips to act as one. Only Apple has managed that here. It’s considered to be a major advance there, and the others will have to figure out a way to equal it without breaking Apple’s patents, which I would think they have.
zimmie said: On the topic of RAM, there's nothing inherent in the M1's design which precludes off-package RAM. They have a RAM controller on the chips, and the RAM controller is shared between CPU and GPU cores, but there's no fundamental reason the RAM couldn't be in DIMMs. Apple just hasn't chosen to do that. They might do so with the Mac Pro, or they might not. I don't see them doing a tiered memory structure, though. They just went to significant lengths to do away with NUMA concerns on the M1 Ultra. With the introduction of the Mac Studio, I wouldn't be surprised to see the Mac Pro go primarily rackmount. Very few people need more computing power than the Mac Studio offers at their desks. Almost everyone who does need more computing power is in an environment where they can rack the computer in a closet. For example, recording studios, film studios, scientific labs, and so on all have 19" rack space for other equipment, so putting specialist workstations in there isn't a stretch. That said, rackmount would mostly be relevant for a box with several full-height, full-length PCIe slots (e.g., to add hardwired audio and video inputs), and I'm not yet convinced Apple is interested in that at all. They might say the future is a rackmount interface which connects to the system via Thunderbolt. I'd be curious to know what they have seen the current rackmount Mac Pro doing.
The in-package memory has much tighter timing tolerances because the memory configuration is fixed, lower signal driver power levels, and at very short distances. I would imagine that their memory controller takes full advantage of those facts, and cuts a lot of the complicated corners that dealing with DIMM slots creates. So, I disagree: I do not think their memory controller could support out-of-package memory without some serious work, and it would represent a large power increase and performance impact.
What I suggested isn't a software-visible tiered memory structure. It is just an alternative fast backing-store for the existing virtual memory system that all software currently works with (currently backed by flash memory). Implementing this would require no changes to the M1 architecture and very little OS change. The Mac Pro has a very small market (especially with the Mac Studio now taking a chunk of it), so custom work to support it doesn't make a lot of sense for Apple. That's a big reason why I think the M1 Ultra is what we will see in the Mac Pro. And probably just one of them as going multi-chip is a lot of specialized added hardware design work that I don't think they want to do.
The M1 Ultra has a pretty amazing amount of compute, after all. Bump the clock rate a little and you give it a small edge over the Mac Studio.
A rack mountable full-sized case (but still a desktop workstation) which can hold lots of extra drives, memory, and PCIe cards would differentiate it from the Mac Studio.
I doubt they will do this, but one thing they could do fairly easily in such a form factor is put the M1 Ultra motherboard itself on a PCIe card so one case could hold multiple of them (the case becomes just a PCIe backplane then). How such a machine would be used becomes more challenging though and would take them away from their preferred programming model of a single shared memory space for many CPU/GPU cores. Without a lot of OS work such a machine would look like several Macs on a high speed network... has some uses, but gets pretty obscure and way out of the consumer space. Then again, it is the "Mac Pro" so who knows?
They objectively do not have tighter timings. The on-package RAM on the base M1 is LPDDR4X. Exactly the same standard has been used for off-package RAM in the Intel MacBook Pro models for the last several years. The M1 Pro and Max use LPDDR5, which has slotted variants. HBM2 is enough to require the RAM be on-package, but they're not using that. Putting the LPDDR4X or LPDDR5 on-package saves a few microwatts from the shorter traces, but that's it.
Off-package RAM as a first-tier swap level could be done, but would give developers inconsistent memory performance. Again, Apple just went to ridiculous lengths to engineer away NUMA specifically because inconsistent memory performance isn't good enough. I don't see them adding it back in when they could instead just connect their memory controllers to DIMM slots. Sure, DDR5 DIMMs are rare right now, but that's not a limitation as far as Apple is concerned.
I wouldn't be surprised to see an M2 Ultra as the first M2-family chip, introduced in the Mac Pro at WWDC.
I believe Apple isn’t using 4 for the m1 series. It’s 6,400 speed 5.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
John was introducing the UltraFusion process when he said that. That process may also allow them to connect two Ultras together, much like what you’re suggesting when you say they may use multiple M1 Ultras in the Mac Pro. It’s a plausible technical solution to the problem. That’s the whole point of the “chiplet” approach. I think you’re getting hung up on semantics, although I’ll concede that it was not a live event and everything said was carefully reviewed.
It isn’t hard to imagine how John would introduce the idea, “UltraFusion not only allows us to fuse two M1 Max together and create the M1 Ultra, but it also allows us to connect two M1 Ultras together …”
Look, he made it pretty clear that the Ultra was the last M1 chip. I don’t know why people insist that isn’t true. He didn’t say it had anything to do with Ultra /fusion, or anything else, just that the Ultra was the last. Earlier on, when they announced the Pro and Max versions, they could have said that too, and then popped out the Ultra with the UF connect, and acted as though it was just the same chip.
but they didn’t. And like it or not, that means something. What would have wrong with not saying anything? It’s not semantics. Semantics is something that’s interpretable. A definitive statement is just that.
the other thing thats] you guys are forgetting is that the cost of making these chips increases more than the added area because of increased defects and the risk of unusable chips. The greater percentage of wafer area a chip takes up, the more the cost. It’s a $1,000 upgrade to go from a 48 core Ultra to the 64 core version. And the cheapest Ultra costs more than twice what the Max version costs. So, even if they would do it, this new chip would likely cost at least three times as much. Maybe more. would that be worth it? I’m not so sure.
The good thing is we’ll know soon enough. There’s no way this uncertainty is prolonged past WWDC.
But reading your last paragraph here makes me think you haven’t looked at the mock-up I’ve been referring to, in conjunction with the patent about this packaging tech Apple filed in January. It’s two Ultras stacked on top of each other (back-to-back)—doing so doesn’t change anything about the wafer layout for making the Max/Ultra. That’s why, no matter how it works, it can’t be considered a new chip. Because it’s not. The Max Tech video that someone pointed to is goofy YouTube sensationalism, but the reading of the patent seems accurate.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
John was introducing the UltraFusion process when he said that. That process may also allow them to connect two Ultras together, much like what you’re suggesting when you say they may use multiple M1 Ultras in the Mac Pro. It’s a plausible technical solution to the problem. That’s the whole point of the “chiplet” approach. I think you’re getting hung up on semantics, although I’ll concede that it was not a live event and everything said was carefully reviewed.
It isn’t hard to imagine how John would introduce the idea, “UltraFusion not only allows us to fuse two M1 Max together and create the M1 Ultra, but it also allows us to connect two M1 Ultras together …”
Look, he made it pretty clear that the Ultra was the last M1 chip. I don’t know why people insist that isn’t true. He didn’t say it had anything to do with Ultra /fusion, or anything else, just that the Ultra was the last. Earlier on, when they announced the Pro and Max versions, they could have said that too, and then popped out the Ultra with the UF connect, and acted as though it was just the same chip.
but they didn’t. And like it or not, that means something. What would have wrong with not saying anything? It’s not semantics. Semantics is something that’s interpretable. A definitive statement is just that.
the other thing thats] you guys are forgetting is that the cost of making these chips increases more than the added area because of increased defects and the risk of unusable chips. The greater percentage of wafer area a chip takes up, the more the cost. It’s a $1,000 upgrade to go from a 48 core Ultra to the 64 core version. And the cheapest Ultra costs more than twice what the Max version costs. So, even if they would do it, this new chip would likely cost at least three times as much. Maybe more. would that be worth it? I’m not so sure.
The good thing is we’ll know soon enough. There’s no way this uncertainty is prolonged past WWDC.
But reading your last paragraph here makes me think you haven’t looked at the mock-up I’ve been referring to, in conjunction with the patent about this packaging tech Apple filed in January. It’s two Ultras stacked on top of each other (back-to-back)—doing so doesn’t change anything about the wafer layout for making the Max/Ultra. That’s why, no matter how it works, it can’t be considered a new chip. Because it’s not. The Max Tech video that someone pointed to is goofy YouTube sensationalism, but the reading of the patent seems accurate.
I would just add that TSMC and others have been talking about and carefully crafting the techniques of silicon-pass-through connectors which is the basis for Apples patent mentioned above. It is not bleeding edge, but just getting into the realm of full scale application. Apples patentappears to be just a cleaver use of the pass though technology.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
John was introducing the UltraFusion process when he said that. That process may also allow them to connect two Ultras together, much like what you’re suggesting when you say they may use multiple M1 Ultras in the Mac Pro. It’s a plausible technical solution to the problem. That’s the whole point of the “chiplet” approach. I think you’re getting hung up on semantics, although I’ll concede that it was not a live event and everything said was carefully reviewed.
It isn’t hard to imagine how John would introduce the idea, “UltraFusion not only allows us to fuse two M1 Max together and create the M1 Ultra, but it also allows us to connect two M1 Ultras together …”
Look, he made it pretty clear that the Ultra was the last M1 chip. I don’t know why people insist that isn’t true. He didn’t say it had anything to do with Ultra /fusion, or anything else, just that the Ultra was the last. Earlier on, when they announced the Pro and Max versions, they could have said that too, and then popped out the Ultra with the UF connect, and acted as though it was just the same chip.
but they didn’t. And like it or not, that means something. What would have wrong with not saying anything? It’s not semantics. Semantics is something that’s interpretable. A definitive statement is just that.
the other thing thats] you guys are forgetting is that the cost of making these chips increases more than the added area because of increased defects and the risk of unusable chips. The greater percentage of wafer area a chip takes up, the more the cost. It’s a $1,000 upgrade to go from a 48 core Ultra to the 64 core version. And the cheapest Ultra costs more than twice what the Max version costs. So, even if they would do it, this new chip would likely cost at least three times as much. Maybe more. would that be worth it? I’m not so sure.
The good thing is we’ll know soon enough. There’s no way this uncertainty is prolonged past WWDC.
But reading your last paragraph here makes me think you haven’t looked at the mock-up I’ve been referring to, in conjunction with the patent about this packaging tech Apple filed in January. It’s two Ultras stacked on top of each other (back-to-back)—doing so doesn’t change anything about the wafer layout for making the Max/Ultra. That’s why, no matter how it works, it can’t be considered a new chip. Because it’s not. The Max Tech video that someone pointed to is goofy YouTube sensationalism, but the reading of the patent seems accurate.
So, I don’t remember seeing the stacking that you’re referring to. But Apple’s interposer is part of the silicon and makes the two Max’s into an Ultra, one die chip. If this stacking isn’t doing that, but is instead just taking two Ultra’s and somehow connecting them together without turning them into one die, as with the Ultra itself, then that’s just two chips. It’s not another chip. I’ve been saying that Apple could do that in some way to offer more performance from the Mac Pro. Not double the performance, as in many tasks, as we’re seeing with the new ArsTechnica review that came up today, but pretty good.
I don’t really care about mock ups. They’re just guesses, and rarely, until just before release, often don’t mean much. After all, they’re called MOCKups for a good reason.
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
John was introducing the UltraFusion process when he said that. That process may also allow them to connect two Ultras together, much like what you’re suggesting when you say they may use multiple M1 Ultras in the Mac Pro. It’s a plausible technical solution to the problem. That’s the whole point of the “chiplet” approach. I think you’re getting hung up on semantics, although I’ll concede that it was not a live event and everything said was carefully reviewed.
It isn’t hard to imagine how John would introduce the idea, “UltraFusion not only allows us to fuse two M1 Max together and create the M1 Ultra, but it also allows us to connect two M1 Ultras together …”
Look, he made it pretty clear that the Ultra was the last M1 chip. I don’t know why people insist that isn’t true. He didn’t say it had anything to do with Ultra /fusion, or anything else, just that the Ultra was the last. Earlier on, when they announced the Pro and Max versions, they could have said that too, and then popped out the Ultra with the UF connect, and acted as though it was just the same chip.
but they didn’t. And like it or not, that means something. What would have wrong with not saying anything? It’s not semantics. Semantics is something that’s interpretable. A definitive statement is just that.
the other thing thats] you guys are forgetting is that the cost of making these chips increases more than the added area because of increased defects and the risk of unusable chips. The greater percentage of wafer area a chip takes up, the more the cost. It’s a $1,000 upgrade to go from a 48 core Ultra to the 64 core version. And the cheapest Ultra costs more than twice what the Max version costs. So, even if they would do it, this new chip would likely cost at least three times as much. Maybe more. would that be worth it? I’m not so sure.
The good thing is we’ll know soon enough. There’s no way this uncertainty is prolonged past WWDC.
But reading your last paragraph here makes me think you haven’t looked at the mock-up I’ve been referring to, in conjunction with the patent about this packaging tech Apple filed in January. It’s two Ultras stacked on top of each other (back-to-back)—doing so doesn’t change anything about the wafer layout for making the Max/Ultra. That’s why, no matter how it works, it can’t be considered a new chip. Because it’s not. The Max Tech video that someone pointed to is goofy YouTube sensationalism, but the reading of the patent seems accurate.
I would just add that TSMC and others have been talking about and carefully crafting the techniques of silicon-pass-through connectors which is the basis for Apples patent mentioned above. It is not bleeding edge, but just getting into the realm of full scale application. Apples patent appears to be just a clever use of the pass though technology.
I’ll just put this here because I don’t think the Max Tech guy was the origin of the mock-up. Regardless, he has backed away from it. So now he’s on the “M2 Extreme” (4xMax) bandwagon, pretty much what Marvin posted above.
Editing to note that this isn’t a total fantasy, it’s pretty much exactly what some of that patent describes and illustrates (to be clear, though, this image is not from the patent)…
Allow me to add that I don’t quite get the argument that it has to be M2 — I gather there is a technical reason for it, but I think that’s hard to say without knowing a lot more than we do. I find the idea that Apple would design this entire M1 line but not account for the Mac Pro to be absurd.
In terms of naming, I don’t think they will call it the Ultra Pro or Ultra+, they will all be Ultra, just with different core counts. Basically an Ultra is 2 or 4 Max fused together.
It’s pretty clear that they are finished with the M1. Will people please stop making new M1 chips up?
They may use two or even four M1 Ultra chips. They may change their concept of introducing more powerful chips over the year, and have an M2 Ultra for the Mac Pro. We don’t know. But they won’t have a four chip Ultra. John made that pretty clear.
John was introducing the UltraFusion process when he said that. That process may also allow them to connect two Ultras together, much like what you’re suggesting when you say they may use multiple M1 Ultras in the Mac Pro. It’s a plausible technical solution to the problem. That’s the whole point of the “chiplet” approach. I think you’re getting hung up on semantics, although I’ll concede that it was not a live event and everything said was carefully reviewed.
It isn’t hard to imagine how John would introduce the idea, “UltraFusion not only allows us to fuse two M1 Max together and create the M1 Ultra, but it also allows us to connect two M1 Ultras together …”
Look, he made it pretty clear that the Ultra was the last M1 chip. I don’t know why people insist that isn’t true. He didn’t say it had anything to do with Ultra /fusion, or anything else, just that the Ultra was the last. Earlier on, when they announced the Pro and Max versions, they could have said that too, and then popped out the Ultra with the UF connect, and acted as though it was just the same chip.
but they didn’t. And like it or not, that means something. What would have wrong with not saying anything? It’s not semantics. Semantics is something that’s interpretable. A definitive statement is just that.
the other thing thats] you guys are forgetting is that the cost of making these chips increases more than the added area because of increased defects and the risk of unusable chips. The greater percentage of wafer area a chip takes up, the more the cost. It’s a $1,000 upgrade to go from a 48 core Ultra to the 64 core version. And the cheapest Ultra costs more than twice what the Max version costs. So, even if they would do it, this new chip would likely cost at least three times as much. Maybe more. would that be worth it? I’m not so sure.
The good thing is we’ll know soon enough. There’s no way this uncertainty is prolonged past WWDC.
But reading your last paragraph here makes me think you haven’t looked at the mock-up I’ve been referring to, in conjunction with the patent about this packaging tech Apple filed in January. It’s two Ultras stacked on top of each other (back-to-back)—doing so doesn’t change anything about the wafer layout for making the Max/Ultra. That’s why, no matter how it works, it can’t be considered a new chip. Because it’s not. The Max Tech video that someone pointed to is goofy YouTube sensationalism, but the reading of the patent seems accurate.
I would just add that TSMC and others have been talking about and carefully crafting the techniques of silicon-pass-through connectors which is the basis for Apples patent mentioned above. It is not bleeding edge, but just getting into the realm of full scale application. Apples patent appears to be just a clever use of the pass though technology.
I’ll just put this here because I don’t think the Max Tech guy was the origin of the mock-up. Regardless, he has backed away from it. So now he’s on the “M2 Extreme” (4xMax) bandwagon, pretty much what Marvin posted above.
Editing to note that this isn’t a total fantasy, it’s pretty much exactly what some of that patent describes and illustrates (to be clear, though, this image is not from the patent)…
It's amazing that people don't think this is possible. Before the M1 Ultra was announced these same people would've said that it was impossible, too.
Comments
It isn’t hard to imagine how John would introduce the idea, “UltraFusion not only allows us to fuse two M1 Max together and create the M1 Ultra, but it also allows us to connect two M1 Ultras together …”
And an earlier reply to my comment:
The in-package memory has much tighter timing tolerances because the memory configuration is fixed, lower signal driver power levels, and at very short distances. I would imagine that their memory controller takes full advantage of those facts, and cuts a lot of the complicated corners that dealing with DIMM slots creates. So, I disagree: I do not think their memory controller could support out-of-package memory without some serious work, and it would represent a large power increase and performance impact.
What I suggested isn't a software-visible tiered memory structure. It is just an alternative fast backing-store for the existing virtual memory system that all software currently works with (currently backed by flash memory). Implementing this would require no changes to the M1 architecture and very little OS change. The Mac Pro has a very small market (especially with the Mac Studio now taking a chunk of it), so custom work to support it doesn't make a lot of sense for Apple. That's a big reason why I think the M1 Ultra is what we will see in the Mac Pro. And probably just one of them as going multi-chip is a lot of specialized added hardware design work that I don't think they want to do.
The M1 Ultra has a pretty amazing amount of compute, after all. Bump the clock rate a little and you give it a small edge over the Mac Studio.
A rack mountable full-sized case (but still a desktop workstation) which can hold lots of extra drives, memory, and PCIe cards would differentiate it from the Mac Studio.
I doubt they will do this, but one thing they could do fairly easily in such a form factor is put the M1 Ultra motherboard itself on a PCIe card so one case could hold multiple of them (the case becomes just a PCIe backplane then). How such a machine would be used becomes more challenging though and would take them away from their preferred programming model of a single shared memory space for many CPU/GPU cores. Without a lot of OS work such a machine would look like several Macs on a high speed network... has some uses, but gets pretty obscure and way out of the consumer space. Then again, it is the "Mac Pro" so who knows?
https://www.anandtech.com/show/17306/apple-announces-m1-ultra-combining-two-m1-maxes-for-even-more-performance
Seeing this makes me think about “persistent memory” — a technology Intel has invested a lot in (Optane DC), and while it’s currently aimed at servers, their road map for it has/had a second phase focused on workstations. There are people here who moan about the lack of additional user-replaceable internal storage, but it’s possible they are thinking too small, long term. User-replaceable persistent memory, supplementing both Unified Memory and internal storage, is more like it…
Off-package RAM as a first-tier swap level could be done, but would give developers inconsistent memory performance. Again, Apple just went to ridiculous lengths to engineer away NUMA specifically because inconsistent memory performance isn't good enough. I don't see them adding it back in when they could instead just connect their memory controllers to DIMM slots. Sure, DDR5 DIMMs are rare right now, but that's not a limitation as far as Apple is concerned.
I wouldn't be surprised to see an M2 Ultra as the first M2-family chip, introduced in the Mac Pro at WWDC.
but they didn’t. And like it or not, that means something. What would have wrong with not saying anything? It’s not semantics. Semantics is something that’s interpretable. A definitive statement is just that.
the other thing thats] you guys are forgetting is that the cost of making these chips increases more than the added area because of increased defects and the risk of unusable chips. The greater percentage of wafer area a chip takes up, the more the cost. It’s a $1,000 upgrade to go from a 48 core Ultra to the 64 core version. And the cheapest Ultra costs more than twice what the Max version costs. So, even if they would do it, this new chip would likely cost at least three times as much. Maybe more. would that be worth it? I’m not so sure.
But reading your last paragraph here makes me think you haven’t looked at the mock-up I’ve been referring to, in conjunction with the patent about this packaging tech Apple filed in January. It’s two Ultras stacked on top of each other (back-to-back)—doing so doesn’t change anything about the wafer layout for making the Max/Ultra. That’s why, no matter how it works, it can’t be considered a new chip. Because it’s not. The Max Tech video that someone pointed to is goofy YouTube sensationalism, but the reading of the patent seems accurate.
Editing to note that this isn’t a total fantasy, it’s pretty much exactly what some of that patent describes and illustrates (to be clear, though, this image is not from the patent)…