Apple would certainly never run an ad mocking a competitor.
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.
Apple's M1 Ultra Chip: Everything You Need to Know
- Thread starter MacRumors
- Start date
- Sort by reaction score
Eh, it’s fine.
And to be fair, Apple has kind of had their foot off the gas with Mail for a while.
I think Apple has the right idea making Mail/Calendar/Contacts/Notes/Reminders separate apps, unlike classic monolithic Outlook, but they haven’t iterated on several of them much. (Still love Notes, and Reminders has been evolving a lot, too. But the three others…)
It isn’t just petty advertising. It brings attention that when you have a need for general purpose HPC you are probably not going to find an Apple solution. If you are doing video work where the specific algorithms used are implemented directly in hardware in the M1 you are set but you shouldn’t expect that performance to translate to a broader more general set of problems such as those that rely on like using FDTD, FEM, etc.
There are cables you can buy for all those fantasyland display/connection options.
There are cables you can buy for all those fantasyland display/connection options.
Orrrrrrr the computer could simply, gasp, include an HDMI port, which is hugely convenient for almost any monitor, TV, or projector.
(But also, the point wasn’t even that. It was that Apple’s chips should implement HDMI 2.1 rather than 2.0A.)
When they list the following
- A display engine that drives external displays.
- Integrated Thunderbolt 4 controllers that offer more I/O bandwidth than before.
Also there is some discussion on Reddit that macOS itself doesn't not allow exceeding 4k 60hz via HDMI? link
The original post said: "hopefully the M2 family add HDMI 2.1."
I think that's a perfectly valid ask. Whether the SoC needs to implement it, or another controller in between, or the internal wiring isn't shielded enough, or whatever the limitation is.
Studio Display has one Thunderbolt 3 (USB-C) port, three USB-C ports... no HDMI. Now I have a HDMI port on my Mini that I will not be using. If it was a TB 3, it is usable for me. That to me is a waste.
The entry level Max is priced about $250 too high and the Ultra is priced about $1500 too high for the added utility they each provide. The M1 mini is far more compelling for most use cases. An expandable upcoming Mac Pro will steal from Ultra sales.
If the M2 mini gains an M2 Pro option, yep.
Not if pricing starts way above the Mac Studio, which it will. (Otherwise, why bother having both products?)
Thanks for that pointer to eclectic light. When I find time, I’ll probably skim through that in more detail, but found this interesting:
The hunt for the M1’s neural engine
How can you tell when software uses the Neural Engine in an M1 series Mac? How much power does it use, and what is Espresso? Mysteries unravelled.
eclecticlight.co
I think the conclusion is that the neural engines are under utilized, which probably means they will be doubly under utilized in an Ultra. Which is a shame, because there’s a lot of power there.
I agree with your assessment that this is still early days for such tools, and the point for now is to make the hardware available so people begin using it. The benefit of having the ANE cores in the M1 is likely less for the M1 users of today and more so Using them is mainstream by the M3, or whatever.
Based on what I read in the link above, it sounds like most third party AI work is still hitting the CPU/GPU Rather than the neural engines. If that’s true, then they are mostly being used for features in the OS and probably won’t benefit from the additional cores. The OS features are designed to run a common set of functionality fast enough to be transparent. Doubling the cores won’t help unless they’re going to make certain capabilities only available on the Ultra Studio, which they won’t.
One exception might be that someone commented that FinalCut does ML on input footage— that’s something that might benefit by running faster.
Same for the media engines— so far it doesn’t seem like tools like Handbrake are fully utilizing the media engines, but I’ll be interested to see if they eventually do.
Definitely. This will improve over time especially for first-party code, but third-party code will probably always underutilize it, unless Apple provides APIs that make it very similar to competing engines on other platforms, which I don't think they're interested in. As soon as you're a third-party dev who writes apps for multiple platforms, something as highly specialized as the Neural Engine that only exists on that one platform is relatively unappealing, no matter how fast it may be.
I believe Handbrake uses VideoToolbox where possible, and VideoToolbox in turn uses Apple's hardware encoder.
The M1 Max and Ultra are general purpose computers that are very fast at general purpose problems, not just things they were optimized for.
Surprising HPC results with M1 Max… | Apple Developer Forums
My worry is that’ll only close to 4x the performance but at what measure? The Ultra is not a double x jumó in performance in all measures, the majority but not all.
That word has a bad connotation with power-grabby supremacy, violence and dominance.
When you have to simulate a 3D structure (for instance an optical waveguide in my case but really any problem that requires solving differential equations with given boundary conditions) and you require a fine grid to get the solution to converge a 128GB of main memory will not get you far. If you don’t have enough main memory for the problem it isn’t that it just takes you longer to get the solution but that you crash the operating system and end up with no solution. I’ve seen this with both Windows and Macs. The only machines I’ve seen that can access more than 1 TB of main memory have Intel or IBM chips. I don’t think even AMD makes chips that can access that much memory. In addition companies that make enterprise grade software for electrical and optical engineering design and simulation develop their software to run on x86 under either Windows or Linux. Companies like Mentor Graphics, Keysight and Cadence are not going to develop their software to run natively on a Mac because the whole Mac market is too niche. So that means if you are an engineer that needs to perform that work you need a PC or the current Mac Pro.The M1 Max and Ultra are general purpose computers that are very fast at general purpose problems, not just things they were optimized for.
Surprising HPC results with M1 Max… | Apple Developer Forums
developer.apple.com
I like how Sony denominates their pro products. No BS.
Just call it M1001 M1002 M1003 M1004 and M1005.
Professionals know their business, no need for Michael Bay-esque marketing.