I'm curious why that's the case. I would have though that the improved cooling (and lowered noise!) plus the ability to still plug in that eGPU would have slotted in nicely.
Peek Performance: March 8th Event [Event discussion begins on p6!]
- Thread starter Mhorydyn
- Start date
I'm curious why that's the case. I would have though that the improved cooling (and lowered noise!) plus the ability to still plug in that eGPU would have slotted in nicely.
Brad Oliver
Ars Tribunus Militum
Because I'd still want to plug a separate monitor into the eGPU for best performance, and I already had a dual-27" 1440p set.
Max Studio order placed. Yay! Now I've got to wait two months for delivery. Boo!
What are the odds they drop a $1,299 M1 Pro Mini before then?
What are the odds they drop a $1,299 M1 Pro Mini before then?
Has there been any official comment yet on the huge price leap to go from a Mac Studio with 48 core GPU to 64 GPU? Not that I can afford either, but it’s curious.
I think there will be more buyer’s regret than usual while the AS transition is still underway. And I also suspect that the M1 lineup of SoCs might not be indicative of Apple’s longer term plan. I wouldn’t be surprised if starting with M2, there will be only two basic Mac SoCs, rather than three. So the price/performance positioning of things like Minis and Studios might shift from where they are now.
I would imagine it's yield based. A high enough defect rate where you can get 48 good cores regularly, but 64 pristine cores more rarely, could lead to a big jump in price.
That plus even more margins is my best guess too. It is quite the jump -- 200$ gets you 8 extra GPU cores with the Max but it's 1000$ for 16 extra in the Ultra.
Going from 24 to 32 GPU cores on the Max version is only +$200, so you'd expect going from 48 to 64 on the Ultra version to be +$400, since it's two Maxes connected with an interposer. Instead it's +$1000. I wonder if they're just engaged in blatant price discrimination here, or if there's something more complicated going on, like the 64 core version actually has a higher bandwidth interposer, or they need to hit more aggressive power efficiency targets with a 32 core Max destined to be part of an Ultra in order to stay within the thermal budget of the cooling system.
I suspect some of both, but more that they're soaking professionals on the Ultra. A 48 GPU core Ultra is a $1200 jump over a 32 GPU Max, plus another $400 to go to the mandatory 64 GB of RAM. And really the better comparison for the 48 GPU Ultra is a 24 GPU Max, since it's two of those stuck together, and then it's an $1800 price jump (RAM included). When you consider that the entire computer costs $2000, then adding on another $1800 to exactly double the SoC definitely seems out of line. The Max is the upsell product, it's priced so that a home consumer can talk themselves into it as a nicer experience than a mini even if they'll never come close to using all that power. The Ultra is like the Mac Pro, if you don't know what you'll do with it you don't need it. It's priced for people who buy camera lenses and software packages that cost more than the machine and are going to write the entire thing off as a business expense anyway. Also it seems a little interesting that there's no 56 GPU core Ultra (pairing a 24 and a 32 GPU together). Wonder if that's for simplicity or if the fabric Apple's using to stick them together requires both chips to be identical.
That's part of what's keeping me from upgrading right now. Presumably the cadence will be more regular now, but it's still a matter of seeing what that cadence is. A-X chips took a while, like ~18 months? If the M2 comes out late this year that'll be like 2 years?
It'll be interesting to see what the A16 and presumed M2 look like, like if there's anything to indicate they're more amendable to faster scaling updates than before.
I would think a future M2 mini + display is more a 27" iMac replacement than the current product mix.
The latest Ars article on the Ultra reports that it's actually a single piece of silicon, not two, so making it really does require two adjacent Max chips with 32 working GPU cores. Which makes one wonder why it needs an interposer, but perhaps whatever method that allows them to cut Ultras in half to make Maxes doesn't allow the chips to completely touch, and the interposer is needed to bridge the gap.
This seems like it can't be right. The Apple event specifically referenced "physical limitations in creating a larger die than M1 Max" and described M1 Ultra as being enabled by "a groundbreaking die-to-die interconnect technology." This should cue up to the relevant bit (26:26). There's even an explicit reference to "this multi-die architecture."
It's still two Max dies, they just don't cut them up. There's usually a reticle limit in the ~800 mm² range, which is what prevents you from making an arbitrarily large die, which I'm guessing is what they meant. And to purely speculate on my part, it might not be possible to align the dies on the wafer to the nanometer level, so it's really not a single die, and could explain why an interposer is still needed to connect them.
This all seems to be based on a ComputerWorld article (linked from the Ars article) that references "114 billion transistors on a 840mm squared die." Normally I'd say a die has to be a physically discrete piece of silicon. You're saying it could mean a logically discrete IC on the same piece of silicon. Sure, maybe. But neither of these is compatible with describing M1 Ultra as an "840mm die." That description just seems confused, so I don't think it's safe to draw any conclusions from it.
Also, the Apple event shows an animation of two physically discrete M1 Max chips moving toward each other and then "fusing" through the interposer. I suppose one could argue this is intended to be understood metaphorically, but, like, why? If M1 Ultra is one piece of silicon, that's a cool, notable thing. Why would Apple strongly imply otherwise, instead of just saying and illustrating that?
Anandtech describes it as bonding two dies together on one chip. It presents itself to the OS and software as a single chip rather than the classic multisocket multiprocessor machines, so maybe that's where the confusion is coming from?
Plus, the Studio Display doesn’t actually require a Mac Studio to use with it – you can hook it up to a regular Mac mini, too. So the entry price for a Mac and Studio Display is actually $2298 ($599 for the base model Mac mini, $1599 for the display), which is only $499 more than the old 27” iMac, and more powerful too just by virtue of the M1. I don’t think that really leaves any room for a separate 27” iMac.
And even without a 27” iMac, this is still the widest range of desktop Macs that modern Apple has ever offered: Mac mini, 24” iMac, Mac Studio, and Mac Pro, plus two different external displays. It’s hard to imagine there are people in the market for a desktop Mac whose needs aren’t covered by any combination of those products.
There is no good replacement for the 27" iMac in my case. I need at least 32GB of RAM and I'd rather not spend $3500+ on a Mac Studio + Studio display. The 24" iMac display is too small anyway.
Yeah, there's a big gap. If you don't need workstation power but you want an Apple-quality display and 32 GB RAM, you're out of luck. And I expect that describes a reasonable number of people. The secondhand market suggests that at least two-thirds of 27" iMac sales were base to midrange configs that were considerably less expensive to buy than a Studio + Studio Display. Some of those will do OK with a M1 mini + Studio Display but others will find it limiting, particularly with respect to RAM and dual-display capability..
I'm in a different boat: I'm willing to pay around the price for a M1 Max Mac Studio + Studio Display, but I feel disappointed that for all that money I wouldn't get a better display than I already have in my existing 2017 iMac Pro. The iMac Pro had meaningful spec upgrades to the display over the 2015 iMac 5K it replaced, but now it's several years later and there's no progress.
I'm in a different boat: I'm willing to pay around the price for a M1 Max Mac Studio + Studio Display, but I feel disappointed that for all that money I wouldn't get a better display than I already have in my existing 2017 iMac Pro. The iMac Pro had meaningful spec upgrades to the display over the 2015 iMac 5K it replaced, but now it's several years later and there's no progress.
Literally the only thing keeping me from ordering a mid-level Studio Max is whether work will pay for it. I expect they will, but I’m buying one regardless. This is exactly what I’ve been waiting for.
Ordered a Mac Studio Ultra. It will replace my 2019 Rack Mac 16-core. Or rather the 2019 will go in the server rack to replace the 2013, which gets donated.For the few things I still need intel for, I'll just RDP into the Rack mount. Everything is on 10GbE, so should still be plenty smooth. It is with the 2013 anyway.
Currently says early May delivery.
Currently says early May delivery.
ProMotion is interesting that way: I want it on my phone because it can reduce frame rate and increase battery life, but since I visually can't tell much of a difference, I don't care if my display has it. Other folks don't ever want to live without it ever again because of what they see.
The Studio Display looks close to what I want, but I'm not a launch day purchaser for three reasons:
1) Does it actually integrate better with the Mac environment than other displays? Specifically: If my laptop falls asleep and a pair of these are connected, and I wiggle my bluetooth mouse, can I expect that the monitors will wake up correctly? It's an Apple display, so I hope so, but I want to hear about people doing it.
2) Does the A-series chip under the hood change the way my laptop talks to my display? If tools like SwitchResX won't work and I'm stuck doing resolutions their way, I'm going to be unhappy. Initial impressions seem good here, too, but I'll wait 'til folks have 'em in their hands.
3) I like having one cable going to my laptop, and my current dock requires one of my monitor outputs to be DisplayPort. Does that work? I don't mind losing camera/microphone on one of the two displays (having just one ought to be fine, right?), but if I lose anything else, I need to get my hands on a dock that has two Thunderbolt 4 outputs. (Looks like Caldigit's TS4 will do that, but it appears to have only been paper-launched at this point.)
If I can't get all of those three things, I'll probably grab a pair of second-hand LG 5Ks from an early Studio Display adopter.
1. It would be a huge misstep if the monitor *didn't* wake when the computer did. I'm 99.999999% certain it will.
2. From a video perspective my gut says that SwitchResX will still work, I think the A13 is really just there for the camera/speakers.
3. DisplayPort probably won't work, you'll need a Thunderbolt dock. Considering that Apple says the display is compatible with Macs that have Thunderbolt 3 ports, I don't think you'll need one that's TB4 specifically; an older TB3 dock with a passthrough port might be fine, and save some money.
The question I have about the display is what resolution of video it sends to the computer for the webcam. It's apparently a 12MP camera; can I get 4K video from it? Can I turn off the "follow me around" smarts? Or will it be a 1080p stream, and you just have to deal with what the monitor wants to do?
I'm right with you on both of those... but "monitors that don't wake from sleep correctly when the machine does" isn't just a Mac problem, it's a huge misstep across an entire industry. On this machine, monitors may not reliably wake, on my Windows machine, since they're not identical, they wake at different speeds, giving Windows time to decide, "Well, need to move everything to this monitor, since it's all I have now" and then "Oh, I have another monitor. Too bad I can't remember where everything goes in the two monitor config, let's move everything over to the new monitor."
So, sure, it'd be a huge misstep... but missteps are exactly what I expect from this segment of the industry. Can Apple fix it? Oh, probably. Have they? Remains to be seen, but I've been burnt on this one more than a few times. If they have fixed it, that alone's enough to make me pay the asking price for their Studio Displays. But if they haven't, what's the point?
This is actually one of the things that would immediately drive me to an Apple monitor if it behaved properly (and they made one in the size/resolution I want at a reasonable for Apple price). I have 3 identical LG 4Ks on my Mac Pro. It's a crap shoot which will wake first and what resolution I will end up with. Particularly for the DP monitor (1 via DP, 1 via TB3, 1 via HDMI). On the other hand, the 1080 espresso touchscreen also attached (via USB-C) - wakes instantly. Heck, about 50% of the time the LG via DP wakes and goes back to sleep, I have to switch from DP 1.4 to DP 1.2 back to DP 1.4 to get the computer to recognize it. Doesn't matter which monitor - they all behave the same if plugged in via DP.
Same issue with my Windows work laptop across multiple HP (at work) or LG (at home) monitors.
This is on systems that the system itself never goes to sleep, just the monitors.
I also will wait for first-party reports on the ASD, but the LG UltraFine 5K behaves absolutely perfectly in this respect, and I'm sure there's some Apple special sauce behind that that I'd guess will be in the ASD as well. It's the first monitor I've had that behaves 100% flawlessly since my 23" ACD back in the day.
When you're using a display that Apple considers "retina," neither Apple's method nor SwitchResX actually changes what the display sees. The scaling happens on the Mac end and the display always sees a 5120x2880 signal. I'd be very surprised if the ASD worked any differently.
LATE EDIT: The stricken section may not be correct. It depends on whether the ASD can accept a compressed DisplayPort stream. See below.
My impression is that some of the ambiguity comes from the fact that the interposer has a particularly high density of silicon, much closer to that of the M1 dies themselves (maybe it’s even manufactured 5N too?) than is usually the case for such interposers. So even if it is two or three separate parts that are then combined, the result can be more meaningfully described as a single chip than would be the case for two dies connected by a more traditional interposer that is a appreciably less sophisticated and lower bandwidth than the internal connections on the two dies.
Hm. Caldigit claims their TS4 can drive a 4, 5, or 6K display, or it can drive dual 4K or dual 6K displays at 60 hz... but strangely enough, doesn't say anything at all about dual 5K displays.
Given that they don't mention dual 5K displays, I don't know what would be weird about 5K where it wouldn't be supported but dual 6K would, but it may not be wise for me to assume that driving two 6K displays implies driving dual 5K displays.
...everything being subject to change when products actually ship...
"6K" really means the Apple Pro Display XDR. It uses a single losslessly compressed (DSC) DisplayPort stream when in 6K mode. The 5K displays on the market pre-ASD use two uncompressed DisplayPort streams (MST), either muxed into a single TB3/4 signal (LG UltraFine 5K) or sent through two physical DisplayPort cables (all the third-party 5K displays). You only get a maximum of two DisplayPort streams per TB bus. We ran into this with the 2020 Intel iMac equipped with a 5700XT, which can drive two XDRs in addition to its own internal display but only one LG UltraFine 5K. It will also affect the TS4. The question is which technology the ASD will use. We know it can speak MST because Macs that can't speak DSC, such as the 2017 iMac Pro, can drive it. The question is whether there is a DSC mode for DSC-capable Macs. If so, the TS4 will be able to drive two ASDs when connected to such a Mac, although there may not be much bandwidth left for anything else. If not, you'll be limited to one.
...and now I know what would be weird. Given the pile of USB-C sockets on the back of the ASD, I could certainly see using compressed single stream to leave bandwidth for those ports, but until I see for sure that's what they're doing (since the current ones don't...), yet one more reason to hold off on spending the cash.
Yeah, it is weird that there’s no way to configure a desktop Mac with an M1 Pro – the Mac mini and 24” iMac only come with the regular M1, which is limited to 16 GB of RAM, while the Mac Studio can only be configured with the M1 Max or Ultra.
Notably, the high-end Intel Mac mini is still for sale, so I presume that means Apple intends to replace it with an M1 Pro model at some point to fill that gap. If you could get a Mac mini with an M1 Pro, that would give you a way to get 32 GB of RAM without having to get a Mac Studio.
I know that’s all hypothetical, but it seems like a piece of the puzzle is missing right now. It’s just odd for the laptops to have M1/M1 Pro/M1 Max while the desktops have M1/M1 Max/M1 Ultra and no M1 Pro.
Maybe it’s as simple as the Pro, being in the less expensive MBP model, is selling as many as they can make right now. The MBP is the more important line so why take supply-constrained Pro chips and put them in a Mini? Besides, they want to see how many people are going to step up to the Mac Studio. In a few months they drop an M2 Mini and see how the sales split between the M2 Mini and M1 Mac Studio. If it still looks like there’s a hole there then we get an M2 Pro Mini in spring of next yea with the Mac Studio revision. That would be the same time they potentially consider an M2 Pro iMac.
Yields may well be an issue, but also II did the math here and the other problem is that the mini as it currently stands is too expensive. A mini with an M1 Pro and all cores enabled with 32 GB of RAM would actually be more expensive than the entry-level Mac Studio. You'll need a redesign of the mini that gets the starting price down to about $550 before you can stick an Mn Pro into it and have it fit into the current pricing.
I think eventually Apple will have to drop the perfect price correspondence in all AS product lines. That would just result in too much padding in the prices of the high-end stuff. I agree we probably won't get a M1 Pro mini (if we were going to, we would have seen it Tuesday) but I think they will find a way to make the pricing work on some kind of intermediate product in the future. Whether M2 will have enough RAM, expansion, and GPU power to support that product or whether it will need to be M2 Pro… who knows.
Well Apple just stuck an M1 in a $600 iPad. I have to imagine that with the touchscreen, battery, camera and speakers the bill of materials for the iPad is much higher than the entry M1 Mac Mini. So there’s potentially room to lower the price point of the Mac Mini and maintain acceptable margins.
Given Apple’s penchant for keeping past year products in the lineup to hold down lower price points, I wonder if that will be their opportunity to adjust their pricing structure - keep the M1 Pro around as the entry level workhorse at say $200 less than it is now, price the entry level M2 Pro $100 higher than it is now compressing the pricing differential with the M2 Max slightly. Offer the M2 Mac mini as the entry to the Mini and the M1 Pro as the high end.
The average consumer now is a laptop/tablet/smartphone user.
The remaining desktop users are Technical and creative users. Mac Studio is going to take out a chunk of Macpro desktop uses.
My BTO Mac Studio has just switched to "preparing to ship" 
Maybe I'll get it sooner rather than later?
Maybe I'll get it sooner rather than later?
My BTO Mac Studio has just switched to "preparing to ship"
Maybe I'll get it sooner rather than later?
I have a hold on my credit card, but it's still at 'Processing'. No hold for the display yet though.