Last year at WWDC we got the M2 SoC which, if we were expecting a yearly cadence, should have meant the M3 would have arrived today. The lack of a yearly cadence is both disappointing and worrying. If the M3 launches after the A17 will it feature the new A17 high performance cores or will we be a year behind and use the A16 cores instead. If it is based on A16 it likely means we won’t be getting ray tracing with M3 … so do you feel hopeful or worried?
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.
M3 Missing in Action at WWDC
- Thread starter bcortens
- Start date
- Sort by reaction score
How could Apple add hardware-accelerated ray tracing - as a new accelerator like the neural engine or inside the GPU like ARM and Nvidia? If Apple includes it by a new accelerator, does it matter which core Apple uses?
What would be the use case for hardware-accelerated ray tracing? gaming? 3D rendering? Does it make sense for an 8 GB RAM M3 MBA to have hardware-accelerated ray tracing for gaming?
What would be the use case for hardware-accelerated ray tracing? gaming? 3D rendering? Does it make sense for an 8 GB RAM M3 MBA to have hardware-accelerated ray tracing for gaming?
No way M3 will launch before Vision Pro (with M2) launches. It would diminish that product and pricing. And That launches early next year.
Expect M3 to be a spring launch then in April 😱
Expect M3 to be a spring launch then in April 😱
I'm not worried at all. I'm still using a 2012 MacBook Pro.
People tend to judge these things by their own personal needs. I am waiting until they put an ethernet jack back on the MBPs before I upgrade. For all I know I will be enticed to some version of linux in the future because Apple can't figure out how to make an ethernet port look pretty next to a USB-C.
People tend to judge these things by their own personal needs. I am waiting until they put an ethernet jack back on the MBPs before I upgrade. For all I know I will be enticed to some version of linux in the future because Apple can't figure out how to make an ethernet port look pretty next to a USB-C.
The M2's are such screamers that I don't understand the push for the M3's in these forums.
Given the volume of Mac sales, an 18 month cadence was expected. Designing bleeding edge chips is extremely expensive, even for Apple. The numbers have to work out.
Since Apple is already testing M3 Pro chips today per Gurman, I think they're likely based on A16 IP. It's a less risky decision given N3E is already brand new.
Since Apple is already testing M3 Pro chips today per Gurman, I think they're likely based on A16 IP. It's a less risky decision given N3E is already brand new.
The problem with a schedule at 18 months is that they are allowing iPhone and Mac devices to fall out of sync, if they let this continue too long they are going to end up with the Mac chip either 2 generations behind or they will have to skip an iPhone chip generation. The M2 Pro and Max launched a year behind the CPU and GPU cores in the iPhone. While the A16 cores are slight improvements and the A16 was released almost 6 months before they were…
Most of the core architecture is shared across platforms - it worries me that they lack the engineering talent to keep the two lines in sync, as an aside we should also note how long it has been since the S series Watch core architecture has been updated. The Apple Silicon team seems understaffed to me considering how they just can’t seem to keep up.
Nobody upgrades their Mac as frequently as iPhone. So how could it make sense to update M-series chips as frequently as iPhone? That is the root cause of the problem. It’s easy to say keep it “in sync” but there’s a huge cost to that - measured in billions. Masks, simulations, validation, all of that costs money and time.
M1 Pro and M2 Pro each took a year after the base chips were launched. So it’s clearly not a walk in the park.
It will be part of the GPU, as RT is GPU functionality. A separate accelerator makes no sense as you need low-latency communication between RT units and shading units, ideally sharing same L2.
Ye, yes, and probably yes?
I get the impression that real-time ray tracing is an expensive computational task that requires a lot of VRAM, and only high-end GPUs with more than 16 VRAM are good for it. Could it be that Apple uses hardware-accelerated ray tracing as an exclusive feature for the M3 Pro/Max/Ultra?
We'll see it later this year or early next year.
I'm in no rush for new laptop hardware.. Especially with the 15" Air & Mac Pro announced we have pretty much every sector covered now.
I'm in no rush for new laptop hardware.. Especially with the 15" Air & Mac Pro announced we have pretty much every sector covered now.
Depends on what you want to do. If you want to raytrace a large and detailed scene, yeah, a slower GPU will need more time. But isn't this the case for any kind of complex computational work?
I think the question should be: can a compact GPU with ~5TFLOPs and limited memory bandwidth (like the M3) be useful for raytracing? I believe so, yes. There are different ways to approach the raytracing problem. Just because AMD brute-forces it, and Nvidia needs a lot of power for RT doesn't mean that RT is fundamentally useless in entry-level configurations. Apple has a few tricks up their sleeves to bring down the RT cost (I wrote about their patents multiple times).
I don't think this would make any sense. That means using different GPU core designs between the same family. Why go through that extra effort and expense? I could see it if there is some sort of technology transition, e.g. new GPU cores not being ready for M3 and only making a debut in later M3 Pro/Max...
Talking about M3, I don't really expect it until next year unfortunately. They just released the 15" MBA with the M3, I don't really see them updating it in fall. Also, Vision Pro shipping with M2 next year(!) is a big surprise, I would be certain that it would use a 3nm chip.
M1 > M2 had a cadence of 19.5 months.
If that holds from M2 > M3 then that 3nm SoC will be out Q1 2024.
This is a reason why iMac 24" M3 was being rumored.
If that holds from M2 > M3 then that 3nm SoC will be out Q1 2024.
This is a reason why iMac 24" M3 was being rumored.
Apple really didn't say what process the R1 is on.
Apple doesn't have to launch M3 in a laptop. Apple has left the iMac 24" in the Rip van Winkle slumber. Might be nice during the 25th anniversary of the iMac to give it some time in the spotlight as opposed to a forgotten , comatose product dangling on the side of the line up.
The 24" iMac all by itself isn't going to be a gigantic sale volume buster; even with a M3 versus the Mini M2 and Mini M2 Pro. ( The M2 Pro will hold its own against the M3. Especially if the number of display out is still substantially kneecapped on the M3. And the plain Mini will just be cheaper. )
For the MBA 13/15 and MBP 13 which have very substantially higher volume ... yeah they can wait to get out of the iPhone SoC demand bubble. The iMac probably would not be a problem run concurrent with the iPhone bubble. The R1 even less so.
This assumes that N3 volume is still extremely poor. Given that A14 and M1 launched within 2 months of each other on the then brand new N5 process I am doubtful of this claim. The N3 process has been (according to TSMC) in volume for almost 9 months now and if the A17 is on N3 then that will have been nearly a year from TSMS first high volume manufacturing claims. One would hope that Apple has the resources to get an A17 and an M3 out this year. The company is bigger than ever, spends more on R&D than ever, and yet seems to be slowing down in some areas. People often make excuses about how we don't need yearly updates, performance is good enough, but these are just excuses for poor execution out of Apple.
Yeah, except if the M2 in the VP was just a placeholder (I don't think so) then M3 probably won't be until next year. Though there's always the possibility that Apple just ignores the weirdness of releasing an M2-based product after the M3 is announced and just goes with M3 in October anyway.
MR had rumors of Apple allegedly wanting to align iPhone chip cores with Mac chip cores by shortening the Mac chip cycle from 19.5 months to 12 months of the iPhone.
If Apple were to do this it would be nice to have a predictable time table for Macs like what we enjoy with the iPhone.
I hope so.
The iPhone chip CPU cores were aligned initially and have drifted as the M series update cycle drifted slower and slower. (A14 and M1 had essentially the same CPU cores). I don't know if the GPU cores have aligned yet - I know the M1 has more capable GPU cores than the A14 but I don't know if that is true of the M2 and A15...