Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M4 MacBook Pro Models Expected to Launch in Late 2024
- Thread starter MacRumors
- Start date
- Sort by reaction score
DisplayLink has worked quite well for me in the past. I always marvel at the derision it receives on these forums.
Some of us bought the Studio Max to save money. You save $1000 buying it over the Max MacBook Pro. For mobility, I can use an older Air or the junker the company-issued laptop. The Ultras have so far been horrible investments in every way possible. Becoming obsolete almost instantly. I am so glad I never pulled the trigger on one. My love for the Studio is solely for the Max version, the Ultras have never been worth the money. Not even close.
If the MP added more substantially volume , then the volume wouldn't be 'too low'. It is substantive part of the issue. But yes. Ultra's on every single M-series generation makes little economic sense.
it wasn't "M4 just around the corner" as likely they are probably going to fall into a update every n+2 generation pattern. TSMC's 'next' iterative iteration is 'just around the corner' on the road map going into the future. 2nd half 202x there is something new and shiny every year on the roadmap.
The 'volume' problem is also the MBP 14"/16" dropping out relatively quickly from Max consumption. That slows down switching 'half' of the Studio over to the next generation. Waiting for the Ultra to 'get to volume threshold' only adds to the problem. The Studio still would not likely go at the same time as the MBP 14"/16" even if they did split up the Max/Ultra variants. The Mini+Studio skipping M3 Pro/Max probably keeps them on M2 longer to recoup money to offset higher M3 Pro/Max costs that need to get paid for on lower volume.
Apple's SoC strategy has a substantive component of putting SoCs into multiple products to raise unit volume levels (to get higher economies of scale). There is always some product that gets the 'sop up extra volume' (or extra costs) duties.
Private Cloud Compute deployments will probably help keep the n+2 pace , but the bigger packages and/or dies are not going to churn at A-series rates.
That said the Mac Pro has a very narrow 'value add' over the Studio. The Studio moved to Ultra 4-5 months before the MP would likely erode that even higher. The spectre of an "Extreme SoC" differentiation probably does not help that problem much if it is double (or more) the cost of a Ultra. That would have even bigger volume problems. If the Ultra is deeply struggling for enough volume to update at same rate as rest of the series. An extreme would be in even worse shape ( if Apple keeps their non low cost chiplet approach.)
Well, if Apple Intelligence needs 2 GiB RAM, then virtual memory is going to use that up, as long as the process is running. Whether that memory is in actual physical memory, or partially or entirely in swap, or whether it is used by the CPU cores, the GPU cores, the NE cores, or a combination of them, doesn’t really change based on whether you use UMA or not.
Now, you’re right in the sense that, if there were a need to use it in a combination of cores, not having UMA would require more memory, and transferring between cores would come with significant overhead. But the original assertion was: if this is constantly running, then it’s constantly taking up RAM. Swapping portions of it, or only loading portions of the model, would be possible even without UMA.
That’s why I’m arguing that, for the given assertion, the presence of UMA is immaterial.
Yeah, the colleagues I’ve set that up for mostly don’t care. The permanent “screen sharing” warning is a bit of a bummer (I’m not sure if there are workarounds for that now).
Sure.
I find this world view so strange. As if the M2 Ultra’s performance were suddenly degraded by something even better existing.
'Obsolete' is a ridiculous word or connotation in this context. They are still supported. Still getting software updates. Still work just as fast as they were.
Paying more for an Ultra as 'future proofing' is just as flawed as buying Intel Mac Pro 2019 was for 'future proofing'. The Ultra gets more RAM capacity that the Max isn't going to cover. Likely more bandwidth that the Max isn't going to cover.
The Ultra is coupled. To get more CPU cores also buy more RAM and GPU cores. Same with the other permuations ( need more RAM , need to buy more cores , etc. ). It was never was a 'future proofing' thing. It was do you need (a requirement. not a 'desire' or 'I want to for bragging rights') these resources now and in the intermediate future thing to get paid work done.
Investment? It isn't going to appreciate in value. It is a capital cost that is largely intended to be spent in the pursuit of making money. Not that going to sell it and make more money.
If the Ultra is meeting business requirements then it isn't 'obsolete' in any significant sense at all.
It is ephemeral as a bragging rights system, but that really isn't what Apple is trying to sell.
Apple is covering two subgroups of user bases with the Studio. However, the Studio is physically speced for the Ultra. The Max underpowers the ports ( loose TB ports). The height increase is to cover the Ultra ( not the Max).
Same issue with the Mini and "Mini Pro". Same chassis two groups. The plain Mn buyers probably wouldn't buy Mn Pro at the higher price. The Mn Pro buyers probably won't buy the plain Mn version either (because didn't meet their requirements). Neither one is a 'future proof'/obsolete thing.
Ditto for Mn , Mn Pro , Mn Max Macbook Pro 14". Three subclasses , same base chassis.
Last edited:
Allocated and duplicated are two different things. Allocated is handing out chunks of memory. In UMA, memory is still allocated ( assigned to different processes for storage). Areas of memory are reserved/protected. You can't allocate the same memory to two different processes for two entirely different purposes and get any coherent results out of that (even more so if have multiple cores doing entirely concurrent work. ). In very real sense allocation is a reservation because the memory is assigned.
The secure enclave is getting a subset that nobody else gets access to. Kernel memory doesn't belong to any user process address space. Processes should not be able to poke and prod anywhere they wish to. That isn't secure.
Reducing duplication of data is different. To be shared the memory is still allocated ( assigned to some shared task and security parameters set). It is just coming out of one resource 'pool'. It will still be duplicated at closer cache levels to the core, but at the basic working space RAM storage level it isn't. Critical bits will still need locked (reserved) access. ( can't have two cores making changes at the same time for data that needs to coherently transfer between users.)
Last edited:
Good comment. Following that logic it would make sense for Apple to give us a roadmap of Studio upgrading on an n+2 upgrade basis. Alas, roadmaps are not Apple's style. <sigh>
When you can get reasonably comparable M series chip performance from a far less expensive newer chip, from a value perspective, it certainly "feels" obsolete. So perhaps not LITERALLY obsolete, but I get what you are saying and appreciate your perspective.
The Ultras have always been a horrible value proposition. Sure an M1 or M2 Ultra still run perfectly well, but buying one just a year or two later has always been kinda crazy as a value/performance proposition.
When the Ultras first came out I bought a Windows machine for heavy workloads instead, and it still outperforms the Ultras at a fraction of the cost.
True that for most of us core users, including me, "the Ultras have never been worth the money." But there is a subset of other folks for whom the Ultras are worth the money. Even though I will never personally need a box stronger than a [loaded] Studio Max, I will be very interested in seeing what happens with MP and with Studio Ultra.
Last edited:
Ultimately, the value proposition is “need this amount of performance today? Here it is”.
(For me, personally, neither the Max nor the Ultra are that interesting, because I have no use for their many GPU cores. Give me a Pro with more cores and/or more RAM.)
But three crucial points:
- consumers are buying less desktops today than five years ago—so no reason to think Mac mini and Mac Studio are a boon with consumers
- And to the few who were buying desktops five years ago, the MacBooks are even more viable as desktop replacements, so I'm sure many have just followed the trend of converting to MacBooks
- And Apple didn't replace the 27" iMac with the Mac Studio—they replaced the 27" iMac with a 24" iMac.
But looking at it from a sales perspective, iMacs started at $1,299, and the 27" started at $1,799—yet a Mac Studio + Studio Display starts at $3600—so thats not at all a viable alternative when the majority of iMac sales are base model buyers who loved both the relatively low price (considering the monitor is included) and the elegance of all-in-one.
I would imagine less iMacs are sold today than 5-years ago, but still with such a relatively low price at $1299, screen included, I can't help but believe a 24" iMac double sales over the Mac Studio and Mac mini, combined. That doesn't seem crazy to me.
(I think we have to remember that we Mac enthusiasts are not representative of majority markets. We can rave and celebrate the Mac mini for its high performance to cost ratio, on a forum like this, where we think with our heads, but 99% of consumers aren't even considering it because its the least sexy product at an Apple Store and it doesn't fit modern lifestyles. I know exactly one person with a Mac mini and they were a senior citizen that just needed the cheapest thing to check email that could connect to the monitor they already owned in their computer room, and that was more than 20 years ago—No one in my extended social circle (in the last 15 years) has bought a Mac mini—1% seems generous to me—but thats just my anecdotal addition to the data)
If you can find a source on Apple saying that, then I will concede the point.
The 15" MBA is the only 15" laptop Apple sells, so how could it not be?
Last edited:
1. Higher throughput for the HDMI port (up to 8K 60Hz or 144HZ 4K)
2. Wi-Fi 6E
3. Bluetooth 5.3
4. Slight changes to Speaker quality (tested)
5. Space Grey finish for MagSafe cable.
I keep my eye on the Ultra and hoping the value proposition for it improves because it would be nice not having to buy a separate Windows machine for heavy workloads. I cannot add an Nvidia card to a Studio nor can they offer the performance of one, so an Ultra has never been an option for me. Hopefully, that changes one day.
Next year Nvidia will be offering its own consumer ARM SOC, so it will be very interesting what added value they can offer to consumer ARM chips with their experience with GPUs.
Functionally yes but physically the chips are a different size with a different pin out, so the logic board requires a respin too
Last edited:
For what’s worth, Apple claimed the OLED on the iPad Pro required a more advanced/sophisticated graphics controller than what was available on the M3 chip. And the M4 chip was capable of satisfying that requirement. The MacBook Air did not have an equivalent requirement.
Im confusing about Apple strategy with the Mac Studio.
MSMax looks like a bad purchase as it’s being released at half cicle, thought is way cheaper than the MBMax, it is also half cycle old. And the ultra is, in paper, double the better, but in real world it isnt but in Apple’s apps, so aims to a very small niche.
Both MS Max/Ultra are going to be CPU surpassed by the next Pro/Max in 6 months, not so in GPU, but newer SOCs will be very close. This makes the Ultra, again in theory, only worth for Apple apps users that need the best performance money can buy at THAT release date (if you wait six months to buy the Ultra, then you can buy the next MBP Max better specs, for the same price)
And then we have the Mac Studio M3 Mas/Uktra case…
The latest PR from Apple claims their AI stuff requires hardware with 8GB of RAM. Maybe that is true, maybe it’s not. But it is an open point of contention.
Apple Explains iPhone 15 Pro Requirement for Apple Intelligence
With iOS 18, iPadOS 18, and macOS Sequoia, Apple is introducing a new personalized AI experience called Apple Intelligence that uses on-device, generative large-language models to enhance the user experience across iPhone, iPad, and Mac. These new AI features require Apple's latest iPhone 15...
forums.macrumors.com
The latest PR from Apple claims their AI stuff requires hardware with 8GB of RAM. Maybe that is true, maybe it’s not. But it is an open point of contention.
Apple Explains iPhone 15 Pro Requirement for Apple Intelligence
With iOS 18, iPadOS 18, and macOS Sequoia, Apple is introducing a new personalized AI experience called Apple Intelligence that uses on-device, generative large-language models to enhance the user experience across iPhone, iPad, and Mac. These new AI features require Apple's latest iPhone 15...
But the iPhone has a very aggressive multitasking model. The Mac does not. 8 GiB RAM in an OS that’s mostly the frontmost app is a very different thing than the same amount in an OS that heavily emphasizes windows.
Just saying 8GB for Apple AI is still an open issue, it is not verifiable DOA.