At this point, for a Mac Pro, if the M1ultra and M2ultra is what they've got, perhaps they could create a grid computer with up to four motherboards in one computer. And make a special version of Mac OS that handles this seamlessly. Using the grid power for render cues, and massive computational stuff. If they wanna leap ahead, I have a feeling maybe M2ultra alone might not be insane enough to compete head on with the competition when it comes to this level of maxed out dedicated work stations?
If they could join together two SoC’s, what’s to stop them designing a motherboard that could also be joined together? Apple engineering is some of the finest in the industry and if anyone could do this, it’s Apple.
At this point, for a Mac Pro, if the M1ultra and M2ultra is what they've got, perhaps they could create a grid computer with up to four motherboards in one computer. And make a special version of Mac OS that handles this seamlessly. Using the grid power for render cues, and massive computational stuff. If they wanna leap ahead, I have a feeling maybe M2ultra alone might not be insane enough to compete head on with the competition when it comes to this level of maxed out dedicated work stations?
If they could join together two SoC’s, what’s to stop them designing a motherboard that could also be joined together? Apple engineering is some of the finest in the industry and if anyone could do this, it’s Apple.
"A mother board that could also be joined together"?
You mean a single motherboard with, er, two CPUs on it? They actually built machines like that from 1999 until 2013. ;-)
The problem with doing that is the busses shuttling data back and forth. What took the industry by surprise about the M1 Ultra is that the interconnect runs at full speed, and because the distances are so short, CPU power actually scales almost exactly 2x.
Once data has to leave the SoC and travel through the logic board, distances decrease, and speeds plummet. They may go this route, but it won't scale nearly as well.
9secondkox2 said: The mobile snd desktop setups are completely different scsenarios due to batteries and thermals.
The mba is the low end.
The MacBook Pro is the midrange and high end depending on config.
In the desktop space, you had mini on low end, iMac in mid snd Mac Pro at the high end.
Being that the iMac isn’t out for a while and neither is the Mac Pro, the studio fits right in - and benefits from having no legacy to live up to, so far less criticism or disappointment concerning its shortcomings.
There really isn't much point in arguing product differentiation based on a hardware architecture that Apple left two years ago.
Intel's architecture differentiates between desktop and mobile PC platforms. Apple Silicon does not — up to the "Max" chips. The "mobile" versions of the Apple Silicon architecture are those used in iPhones etc.
So up to and including the "Max" chips, it looks like the desktop and mobile configurations are identical, differentiated only by form factor.
Above that ("Ultra" and whatever will power the Mac Pro), the chips won't fit in a mobile form factor and will remain desktop-only.
The 24" iMac IS the low end. Its architecture is completely identical to the M1 MacBook Air, 13" MacBook Pro, and Mac mini.
The Mac Studio comes with an M1 Max. This is identical to the M1 Max available in the 14" and 16" MacBooks Pro.
What's currently missing is a "Pro" chip desktop — this looks like a useful future upgrade option to the Mac mini. I assume that they haven't offered it yet because they need all the available chips to be able to ship their MacBooks Pro, plus the Mac Studio is pulling in a lot of customers who were hoping for a souped-up mini.
The packaging still matters. Quite a it in fact.
that’s why the MacBook Air throttles, the Mac mini had to be stretched and perforated to accommodate huge fans, and why the MacBook Pro doesn’t have the Ultra.
So let’s not go down that road. It doesn’t lead where you think it does.
It's weird, because it sounds like you think you're disagreeing with me.
I'm saying that from the regular "M" to the "Pro" SoC's, the exact same architecture is available in both desktop and laptop form factors.
Yeah, the Air throttles, but it's still the identical architecture. And guess what — the 13" Pro is the exact same machine, as the Air, the Mac mini, and the iMac, and it doesn't throttle. Literally the only differences are whether a machine has a battery, an in-built display, and ventilation.
The only exception is the M1 Pro, which for some reason is not available in the Mac mini or iMac.
Where we disagree is that you seem to be saying the "mini had to be stretched", as if the Studio is just a stacked mini.
But it's a totally different machine — its internal design is completely different. Virtually nothing about its internal design is similar to the Mac mini.
This machine is not a beefed-up mini. It's a completely new design.
And then..what I'm saying is that the Ultra won't fit in a laptop. You're saying that…it won't fit in a laptop, and it won't fit in a mini case, either. Well…yeah?
The point is that from the M1/M2 up to the M1 Max, the packaging is the ONLY thing that matters. There is no desktop or laptop version of the architecture; they are identical (except, of course, for the Air under sustained loads).
This hasn't been the case for intel systems in forever — every generation of architectures had separate desktop and laptop variants, with wildly differing specs.
The 2019 Mac Pro industrial design is very good imo. Just continue using it and only change the internal design in accordance to the expansion architecture they want to have. There are things they need to do to be competitive with a Mac Pro. Putatively, a product in the workstation market that can both go on top of desks and in racks, and would be able to address a wide range of workflows. So, >1 TB RAM capacity, large internal storage capacity, and flexibility to cover different types of work. I think it needs to continue to have 8 PCIe slots, and PCIe 4 is the minimum for a big box in 2023. How Apple enables large >1 TB memory and 40 to 60 PCIe lanes with their chip, memory and IO architecture has been a subject for debate for awhile.
Yeah, I'm not a potential customer for it, but I am really interested to see what they do with their "pro" machine. Honestly, the Mac Studio addresses like 98% of their "pro" market needs, so it's clear why they're not in a big hurry to get this to market. I suppose the biggest push right now is to just make good on their 2 year transition claims.
That said, yes, I agree that the true Mac Pro needs lots of memory... probably more than 1TB as you suggest. The existing pro can handle 1.5 TB, so this only makes sense. I think it needs PCIe slots of some sort, but I'm not clear on how much it needs. With the potential of 128 GPU cores on the SoC, I'm not sure if they need a dedicated card beyond that. They also don't need an afterburner card, etc. I'm thinking they just need something to handle some custom configurations like special purpose hardware controller cards, etc. The point being, fewer slots are needed.
I do agree with your point that the 2019 Pro is a great industrial design. That's likely the basis of the next design, but I wouldn't expect the same size and configuration for the Apple Silicon based Mac pro.
I hold out hope that the M2 Pro will find its way into the iMac and the new Mac Mini (6x6") as an option, along with the M2.
I was really hoping for a 27" iMac with an M1 Pro / M1 Max option. I've given up hope and just put in the order for an M1 Max Studio (4TB, 64GB) and Apple Studio display. That's probably what I wanted anyway, but I figured the all-in-one would be a better deal.
Apple right now has no real competition right now (it won’t last), Apple appears to sitting around dithering (they appear to be making marketing not tech decisions), overlap releases long term isn’t going to work, Apple didn’t take 13 years to replace Intel, because Intel, or those other sub standard companies like IBM, and Motorola before them did a good job. Apple’s cumulative experience with those chip makers let to the A and M series, and that is leading them to replace the Qualcomm modem in time.
The performance of the M series is too great not to put it into servers (give Qualcomm nothing) and that is no different than expanding the A series into mainstream computers which most people even on this site said they would not do. (Intel forever?).
Apple timing on the releases of the M series should be across the entire range of Mac’s at the same time, the M2 Max? in the next MacBook Pro probably stands a very good chance of out performing the current just released M1 Mac Studio. (Not good) but we shall see.
Apple’s hardware in current times has been all over in and out, Laptop keyboards, HomePods, MagSafe, Curated Monitors, Monitor Web Cam, Mac Pro’s, next up Apple router’s coming back? Pardon if there is some concern….
Again, I don't think you understand how this works. Apple doesn't just develop the M1 and then magically all of the other variants of the M1 are ready to go. They do much of this work in parallel. The core tech goes into the A series chips and then teams of engineers scale it up for the M series of chips. It takes time to do this. No amount of wishful thinking or even throwing money at the problem will change that.
Again, Apple is NOT in the server business. Nobody is denying that Apple Silicon couldn't be well adapted for servers. However, Apple's SoCs are NOT optimized for servers. They are optimized for mobile and desktop devices. Servers are almost exclusively about CPU processing power. Only a fraction of Apple's SoCs are dedicated to CPU. There is also the GPU, the NPU, media encoders/decoders, secure enclave, etc, etc. that have nothing to do with servers.
Finally, there are already other ARM based server providers that offer great performance and efficiency over Intel based chips. In fact, the most powerful supercomputer today is now ARM based. Yes, Apple could make a play here if they were actually competing in that market, but they are not.
Due to the timeline of how long Apple said they will replace Macs with ASi versions, I don’t think M1 Pro version was really designed. The mini will stay as the consumer computer, the Studio will stay as the prosumer low to medium pro computer, and the Pro will be the high end. The Studio will not cut it for someone who needs 1.5tb of memory and lots of processing power. Yes the Sudio beats some configurations of the Mac Pro. The Studio wasn’t designed in a short time. The Studio took awhile to design and announced when they were ready.
I’m skeptical of this.
The Studio is basically a stretch Mini. They just had to take the Mini CAD files, edit the vertical dimensions, add perforations and port cutouts, and attach a simple tapered and perforated cylinder to the bottom.
The whole thing could have been designed, tested, and machined in a very short period of time, including the big honking fan assembly - which probably explains the numerous fan issues in the first run.
Another person that doesn't understand how long it actually takes to go from an idea to an actual shipping product. Hint: It's at least a 2 year process. Suggesting that someone just edits their Mini CAD file and just pumps out a new product at the last minute is pretty comical. I get why you might think that's actually possible, but reality is a bit different than that.
Comments
You mean a single motherboard with, er, two CPUs on it? They actually built machines like that from 1999 until 2013. ;-)
The problem with doing that is the busses shuttling data back and forth. What took the industry by surprise about the M1 Ultra is that the interconnect runs at full speed, and because the distances are so short, CPU power actually scales almost exactly 2x.
Once data has to leave the SoC and travel through the logic board, distances decrease, and speeds plummet.
They may go this route, but it won't scale nearly as well.
I'm saying that from the regular "M" to the "Pro" SoC's, the exact same architecture is available in both desktop and laptop form factors.
Yeah, the Air throttles, but it's still the identical architecture. And guess what — the 13" Pro is the exact same machine, as the Air, the Mac mini, and the iMac, and it doesn't throttle. Literally the only differences are whether a machine has a battery, an in-built display, and ventilation.
The only exception is the M1 Pro, which for some reason is not available in the Mac mini or iMac.
Where we disagree is that you seem to be saying the "mini had to be stretched", as if the Studio is just a stacked mini.
This machine is not a beefed-up mini. It's a completely new design.
And then..what I'm saying is that the Ultra won't fit in a laptop. You're saying that…it won't fit in a laptop, and it won't fit in a mini case, either. Well…yeah?
The point is that from the M1/M2 up to the M1 Max, the packaging is the ONLY thing that matters. There is no desktop or laptop version of the architecture; they are identical (except, of course, for the Air under sustained loads).
That said, yes, I agree that the true Mac Pro needs lots of memory... probably more than 1TB as you suggest. The existing pro can handle 1.5 TB, so this only makes sense. I think it needs PCIe slots of some sort, but I'm not clear on how much it needs. With the potential of 128 GPU cores on the SoC, I'm not sure if they need a dedicated card beyond that. They also don't need an afterburner card, etc. I'm thinking they just need something to handle some custom configurations like special purpose hardware controller cards, etc. The point being, fewer slots are needed.
I do agree with your point that the 2019 Pro is a great industrial design. That's likely the basis of the next design, but I wouldn't expect the same size and configuration for the Apple Silicon based Mac pro.
Again, I don't think you understand how this works. Apple doesn't just develop the M1 and then magically all of the other variants of the M1 are ready to go. They do much of this work in parallel. The core tech goes into the A series chips and then teams of engineers scale it up for the M series of chips. It takes time to do this. No amount of wishful thinking or even throwing money at the problem will change that.
Again, Apple is NOT in the server business. Nobody is denying that Apple Silicon couldn't be well adapted for servers. However, Apple's SoCs are NOT optimized for servers. They are optimized for mobile and desktop devices. Servers are almost exclusively about CPU processing power. Only a fraction of Apple's SoCs are dedicated to CPU. There is also the GPU, the NPU, media encoders/decoders, secure enclave, etc, etc. that have nothing to do with servers.
Finally, there are already other ARM based server providers that offer great performance and efficiency over Intel based chips. In fact, the most powerful supercomputer today is now ARM based. Yes, Apple could make a play here if they were actually competing in that market, but they are not.
Another person that doesn't understand how long it actually takes to go from an idea to an actual shipping product. Hint: It's at least a 2 year process. Suggesting that someone just edits their Mini CAD file and just pumps out a new product at the last minute is pretty comical. I get why you might think that's actually possible, but reality is a bit different than that.