Most of what Apple has done since the M1 add more graphics core and juggle the processing core and bump clock and M3 is no different. Just listen to Apples presentation and how much they say graphic when talking about performance. The know they hit point of diminishing returns for the typical Apple user with the M1. So they are focused on graphics performance for multiple reasons. If the screen is faster user think the compute speed is faster, that's an old IBM trick from OS2 days. Next Mac's are used a lot in video and graphic with long render and processing time those people see reduction in long rendering and processing times. Last Apple is talking about games but real target is video performance for the Vision Pro that is going to require lots of graphic performance. So M3 no different focus is all on use for graphics. Even for people doing audio/music work the M1 was plenty fast and little to no boost with the M2 and suspect the M3 won't help either other than addition RAM size will help people doing film composing and using lots of large sample libraries.
Most run of the mill software productivity applications don’t benefit much from adding more cores beyond a certain point because their speed up is greatly hampered by the non parallelism in their code execution.
However, these apps still benefit from increasing the clock frequency or the number of instructions per clock cycle as long as everything else like memory and I/O doesn’t impose limitations and move the bottleneck. Reducing power consumption per clock cycle allows the clock frequency to be increased to higher rates before running into power and heat issues, which is seen in the increased single threaded scores.
Increasing the number of GPU cores and their efficiency greatly benefits applications that have a high degree of parallelism in their execution, but the improvements are not linear with the increase in core count because at some point all of the results of processing done in parallel need to be consolidated, synchronized, or coordinated with shared resources that can be blocked by other executing processes. Even small amounts of serialization cause a large reduction in the actual speed up.
As several others have said, buy whatever best fits your current and anticipated needs for however long you plan to keep the machine you’re buying. I can’t think of a time when an Apple product didn’t live up to my expectations, prompting me to purchase a newer model sooner than I anticipated. I’m sure it happens with some folks, especially those who are nearly maxing out their systems capabilities at the time of purchase.
But time is money, so there are inevitably some buyers who have no choice than to buy whatever machine maximizes their profitability, regardless of their expectations going into their previous purchase.
What's with all the weird ass comments saying that Apple should stop development of new chips?
Indeed - we don't want the history of 68k or PowerPC to be repeated. On both occasions Apple couldn't deliver. With PowerMac G4 we had the situation where all specs got downgraded after launch by 15% as they couldn't ship for months.
The benefit of x86 was "there is no difference". With M-series... Apple should deliver the difference. To see ray tracing as a base feature is great. But then we have Thunderbolt 3 from 2015, the 1 display limitation (better than iBook clamshell that had 0 ports), the M3 "no longer so" Pro, 8 GB, and the heavy Apple Tax on memory, the use of LPDDR5 instead of faster LPDDR5X, and the lack of upgrades to iMac (not even the camera got center stage, display is still 23.5", and accessories are not on USB C).
The presentation with "iMac display is huge" and the launch of old accessories indicates that this iMac is not going to last for long. Apple is aware that the display is small, and the accessories need to be replaced within 6-8 months. And the current production node at TSMC is simply too expensive. Not sure if a node change will result in M4 Pro going Pro again or memory prices dropping, but it should.
Black Friday might bring clearance sales on MBP 14 M2 Pro and iMac M1. They could both deliver great value compared to the M3 versions.
At this stage, I find 20% a speed increase meaningless and abstract.
To me what matters is: what does it enable me to do what I couldn’t do before?
That answer on a generation A to generation B basis usually leads to: not much. On a multi-generational skip that starts to answer that question: a much better battery life, exponentially faster video editing and exporting.
Even then there’s the question of when you really go ‘pedal to the metal’ and you find out these are usually short sprints. At this stage the baseline CPU usually delivers the best cost to performance ratio.
Nothing more although it helps slightly with running ML models, but it shows the general direction Apple is going in, and that from here on out we can expect small incremental improvements in performance for the next 5-6 years, meaning hey have a solid platform to build on. There is still a lot of work ahead for Apple.
At this stage, I find 20% a speed increase meaningless and abstract.
To me what matters is: what does it enable me to do what I couldn’t do before?
That answer on a generation A to generation B basis usually leads to: not much. On a multi-generational skip that starts to answer that question: a much better battery life, exponentially faster video editing and exporting.
Even then there’s the question of when you really go ‘pedal to the metal’ and you find out these are usually short sprints. At this stage the baseline CPU usually delivers the best cost to performance ratio.
If that’s the basis for not upgrading, then you wouldn’t EVER upgrade as rarely does any CPU/GPU/SOC offer something that you couldn’t do before. What it DOES offer is doing what you’ve done before much faster and more efficiently. So in essence, it actually does allow you to do what you couldn’t before: namely accomplish tasks at this level of speed while using less power. Add to that the fact that hardware ray tracing is here, and there’s plenty you can do that you couldn’t before.
M3 is finally here, it’s awesome, and it’s reason to upgrade for anyone still on m1 and Intel. M2 is less of a leap, but still a worthwhile upgrade if you’re performing mission critical, time-is-of-the-essence tasks.
What's with all the weird ass comments saying that Apple should stop development of new chips?
Seriously. Apple does something great and it’s as if the bleeding-edge-of-tech Apple user base has become a bunch of crusty luddites imported from the Windows world - if this forum is anything to go by.
Well… that and some disgruntled M2 customers that knew they should have waited. It’s not like they weren’t up to speed being active forum members…
One thing I can’t stand is the people that seemingly think Apple should not provide low-end options for the millions of people that only need one monitor and 8gb ram.
One thing I can’t stand is the people that seemingly think Apple should not provide low-end options for the millions of people that only need one monitor and 8gb ram.
Just another vector for negative nellies and juvenile trolls to complain their way to fame and fortune, not that any of them are in any way in the market for any of Apple's products anyway. Utterly ironic as well that Surface Pros and other competitors start their product lines the same way (w/8GB), so you know their whinging is nothing but childish, empty anti-Apple sentiment. Same boring cycle every announcement, leaving us to ignore, block and move on.
What's with all the weird ass comments saying that Apple should stop development of new chips?
Indeed - we don't want the history of 68k or PowerPC to be repeated. On both occasions Apple couldn't deliver. With PowerMac G4 we had the situation where all specs got downgraded after launch by 15% as they couldn't ship for months.
The benefit of x86 was "there is no difference". With M-series... Apple should deliver the difference. To see ray tracing as a base feature is great. But then we have Thunderbolt 3 from 2015, the 1 display limitation (better than iBook clamshell that had 0 ports), the M3 "no longer so" Pro, 8 GB, and the heavy Apple Tax on memory, the use of LPDDR5 instead of faster LPDDR5X, and the lack of upgrades to iMac (not even the camera got center stage, display is still 23.5", and accessories are not on USB C).
The presentation with "iMac display is huge" and the launch of old accessories indicates that this iMac is not going to last for long. Apple is aware that the display is small, and the accessories need to be replaced within 6-8 months. And the current production node at TSMC is simply too expensive. Not sure if a node change will result in M4 Pro going Pro again or memory prices dropping, but it should.
Black Friday might bring clearance sales on MBP 14 M2 Pro and iMac M1. They could both deliver great value compared to the M3 versions.
You can be sure the design engineers at Apple are frustrated by some of the shortcomings you identified. But Apple is driven by the Marketing Department and they like to slow and control the flow of new technologies. Maximizing returns on already paid for tech. They will never allow memory or storage upgrades, there’s too much profit in it.
I don't think it matters that the performance increase comes down to clock speed versus architectural improvements. That is, until we get to the next fabrication process. We're getting to point of diminishing returns on node reduction, and TSMC has already said its gate all around transistors are more difficult to fabricate. They may have won this round, but future rounds are going to be hard fought.
If they're not stacking the components (i.e. the transistors), then we approach the impasse. When I left the semiconductor world in 2009, that was the next "new technology" on the pipeline, but I stopped looking at that. I do remember the process engineers talking about the length of the interconnects and how the speed of light was a limiting factor, so going vertical on the chips can help reduce that. Instead of going up, over, then down (there are multiple layers of interconnects, so that path may be on multiple different metal layers), if they could just up and over to the next transistor, that would be huge.
Comments
The benefit of x86 was "there is no difference". With M-series... Apple should deliver the difference. To see ray tracing as a base feature is great. But then we have Thunderbolt 3 from 2015, the 1 display limitation (better than iBook clamshell that had 0 ports), the M3 "no longer so" Pro, 8 GB, and the heavy Apple Tax on memory, the use of LPDDR5 instead of faster LPDDR5X, and the lack of upgrades to iMac (not even the camera got center stage, display is still 23.5", and accessories are not on USB C).
The presentation with "iMac display is huge" and the launch of old accessories indicates that this iMac is not going to last for long. Apple is aware that the display is small, and the accessories need to be replaced within 6-8 months. And the current production node at TSMC is simply too expensive. Not sure if a node change will result in M4 Pro going Pro again or memory prices dropping, but it should.
Black Friday might bring clearance sales on MBP 14 M2 Pro and iMac M1. They could both deliver great value compared to the M3 versions.
Yes, for most people, the M2 chip is more than enough.
The Pro/Max variants exist for those millions of people in industries and environments where the base chips are a limiting factor for work.