Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
The M3 Pro really got shafted. I just realized in the event they said it was 20% faster than M1 Pro, which is the same thing they said about the M2 Pro. Meaning that the M3 Pro is 0% faster than the M2 Pro....

I agree it would have been better to have the same number of CPU cores... but... if you want GPU performance, it's an easy choice over the base M3.
 
I was really excited about 3nm in Macs and phones, but, boy, what a letdown this last round of products was...
It seems to me that the remaining engineers at Apple are really being kept on the marketing/supply chain leash.
 
I wonder what this means, overall, in terms of performance for the M3 Pro.

Because it's hard to tell from the data we have.

What we know:

  • The video says p-cores are 15% faster compared to M2.
  • …and the e-cores 30% faster.
  • Memory bandwidth is down 25%. (Presumably, this is already factored into the above figures. I'm also presupposing that clock changes, if any, are already factored in.)
  • The M3 Pro has 6 p- and 6 e-cores, whereas the M2 Pro had 8/4.

So, it has 25% fewer p-cores, but each of them is 15% faster. And it has 33% more e-cores, and each is 30% faster.

My guess is this is a less than 10% performance improvement for most tasks, but a significant battery life improvement.

This also suggests — and Srouji’s artwork showing the three SoCs seems to confirm — that the M3 Pro, unlike the M1 Pro and M2 Pro, is no longer a chopped version of the Max. The non-suffix SoC was always a separate layout, but now the Pro is also separate from the Max. Interesting.

But overall, it sure reads like: the Pro’s performance barely changes at all; if you want that, you now have to upgrade all the way to the Max. And the Max, conversely, will have worse battery life due to fewer e-cores.
I’m disappointed to find out the decreased memory bandwidth across all M3 chips, the fact they lost a GPU core in some of the models, the fact the 12 core CPU is 6 performance + 6 efficiency cores as opposed to 8 + 4. I reserve judgement until we see real life comparisons but the M3 chips appear to be incremental at best, and certainly less of a leap than M1 -> M2, if though 3nm was touted to be game changing. We see the same thing with A16 -> A17 Pro.
 
  • Like
Reactions: VisceralRealist
Problem is most customers that are not a little tech savey just see a higher number and automatically assume it is better for them - not always the case.
I wasnt going to upgrade from my M1 Max / 64gb / 1TB anyways, and this bandwidth reduction has simply strengthened my case.
 
I’m disappointed to find out the decreased memory bandwidth across all M3 chips, the fact they lost a GPU core in some of the models, the fact the 12 core CPU is 6 performance + 6 efficiency cores as opposed to 8 + 4. I reserve judgement until we see real life comparisons but the M3 chips appear to be incremental at best, and certainly less of a leap than M1 -> M2, if though 3nm was touted to be game changing. We see the same thing with A16 -> A17 Pro.

I will reserve judgment as well, but it's absolutely true that the discourse leading up to M3/A17 was that 3nm was going to be game-changing and worth waiting for (how many times was it said that M2 was a stop-gap before the REAL performance increase with M3?) and so far it's been underwhelming. 🤷‍♂️
 
  • Like
Reactions: GuruZac
I’m disappointed to find out the decreased memory bandwidth across all M3 chips, the fact they lost a GPU core in some of the models, the fact the 12 core CPU is 6 performance + 6 efficiency cores as opposed to 8 + 4. I reserve judgement until we see real life comparisons but the M3 chips appear to be incremental at best, and certainly less of a leap than M1 -> M2, if though 3nm was touted to be game changing. We see the same thing with A16 -> A17 Pro.

I will reserve judgment as well, but it's absolutely true that the discourse leading up to M3/A17 was that 3nm was going to be game-changing and worth waiting for (how many times was it said that M2 was a stop-gap before the REAL performance increase with M3?) and so far it's been underwhelming. 🤷‍♂️

I suspect it's a mix of yield issues they ran into (so they're taking fewer risks with the higher-volume SoCs, and more risks in the high-volume Max), as well as different priorities: I personally would've liked more of a boost with the p-cores, and it seems they've focused, of late, on the e-cores and GPU cores.
 
I’m disappointed to find out the decreased memory bandwidth across all M3 chips, the fact they lost a GPU core in some of the models, the fact the 12 core CPU is 6 performance + 6 efficiency cores as opposed to 8 + 4. I reserve judgement until we see real life comparisons but the M3 chips appear to be incremental at best, and certainly less of a leap than M1 -> M2, if though 3nm was touted to be game changing. We see the same thing with A16 -> A17 Pro.
It is possible that the extra bandwidth and the loss of a GPU core in some models just didn't add any cost/benefit to the M3 SOC/SIP and it could just as likely they ran into global constraints or it is likely Apple were just being cheap asses. I guess we will find out.
 
  • Like
Reactions: GuruZac
What I meant is the speeds for SSDs, for example, M1Pros had higher speeds on 512gb storage than M2Pro speeds with the same storage. The read/write speeds have been halved on 512GB SSDs from M1Pro to M2Pro.
They really aren’t.
What I meant is the speeds for SSDs, for example, M1Pros had higher speeds on 512gb storage than M2Pro speeds with the same storage. The read/write speeds have been halved on 512GB SSDs from M1Pro to M2Pro.
It isn’t really a speed difference that will be noticeable outside of a benchmark test or a huge (think 10gb) write task that you do entirely on the SSD (think duplicating a bunch of files).

Modern integrated SSDs commonly utilize multiple NAND chips for storage; the 512GB of storage in the M1 MacBook Pro was distributed across four 128GB chips (previously confirmed in a teardown by hardware repair site iFixit), while it looks like the new model only has two NAND chips - presumably 256GB each.

So there aren’t as many channels which comes into play in a benchmark test or if you are activiely using swap memory while doing a massive write to the SSDs. If you are a YouTube content creator you can create the right situation to create an issue. But the individual SSDs are faster in the M2 and the SSD will be speedier in ordinary usage (like opening a large application).
 
They're not doing this to boost the M4: they're much shorter term than that. This is to push people to spend more than they would have on the M2 range. People who got enough performance from the M2 Pro are now "encouraged" to spring for the Max chip. People who were likely to go for the 13" MBP with M2 are now pushed towards the M3 MBP (or M3 Pro MBP) for "just a few $100 more" and even people who were looking at the 15" MBA are now being steered towards the M3 MBP to get sustained performance, promotion screen, multi-monitor support for just an inch less screen and 40 g more weight.
Performance gains over the M2 line are so modest, why would anyone who bought an M2-line model upgrade to an M3? Is Apple really expecting those users to buy a new computer every year?
 
Performance gains over the M2 line are so modest, why would anyone who bought an M2-line model upgrade to an M3? Is Apple really expecting those users to buy a new computer every year?
Can we get back to some sort of reality? The vast majority of people do not buy a new computer every year. Not even the vast majority of phone users buy new phones each year.
 
Performance gains over the M2 line are so modest, why would anyone who bought an M2-line model upgrade to an M3? Is Apple really expecting those users to buy a new computer every year?
I suspect even people with Intel machines, if they see a low performance upgrade from M2 to M3 could be more likely to seek an M2 system on clearence.
 
  • Like
Reactions: antonrg
I suspect even people with Intel machines, if they see a low performance upgrade from M2 to M3 could be more likely to seek an M2 system on clearence.
“People who got enough performance from the M2 Pro are now "encouraged" to spring for the Max chip.”

Why make this point?
 
“People who got enough performance from the M2 Pro are now "encouraged" to spring for the Max chip.”

Why make this point?
Because people who just needed CPU performance could get the Pro chips with M1/M2 without needing to spend extra for the additional GPU cores of the Max chip. Such a person would now have to get a Max chip (and the top spec Max chip at that) to get access to the full CPU performance, paying for GPU cores they don't really need.
 
I would say no, ARM is still the way to go for future processors. CISC was considered rather dead for a while with Intel wallowing in bad chips with mediocre performance gains year on year. Intel took some rather big performance jumps in the last two generations for two reasons: one is that they abandoned their own 10nm and 7nm processes and went with TSMC to make their chips, and two, they started adopting ARM approaches in their chip design, such as creating p-cores and e-cores. In prior designs, they used only what we would consider p-cores. Intel is moving towards ARM-like processes because it’s the only way they can find to increase performance. They were pretty much dead-ended otherwise.

Apple can only switch away from crap Intel processors once and get gigantic gains. People got spoiled by the switch from Intel processors to the M1 but have issues with 15-20% generational gains between the M-series chips. One thing Apple has yet to do is to fully go with ARM V9 architecture, though it can be argued that their own flavor of ARM takes a lot from ARM V9, so going fully to it won’t be as dramatic as some think. There’s nothing wrong with the architecture. People simply cannot expect 2x+ performance gains every generation.

The current quirkiness of having slower RAM bandwidth has more to do with binning and design decisions than a limitation of the architecture. I have a post a few above this one that explains more about that.
I hope to see ARMv9.2 from Apple soon!!
 
This is a move to push people towards the Max I guess
Or definitely a move to sour those looking and been waiting for a Black MacBook Pro with an under performing chip in comparison to last year on a few fronts.

* Like what was the technical reason for a slower memory bus? Its supposed to at least maintain or increase this: on the M3 Pro vs the M2 Pro.

Now I'm wondering why the M3 Ultra was not announced nor even hinted at. Either Apple engineers worked with TSMC and had to compromise on a few things via 3nm OR TSMC is having issues. Also seems to reason why so many other companies waited on first generation 3nm chips.

On a side note, related to 3NM chip fabrication with TSMC but isolated to mobile chips, Qualcomm could not order their full SnapDragon 8 Gen 3 chips from TSMC due to Apple boosting their orders (as we can see the M3 & M3 Pro series). They've worked with Samsung to produce the remaining orders they needed, which is VERY interesting as Samsung has in less than 1 year boosted performance of their Exynos chip (last years 2200) in the 2400 (which includes Graphics IP/tech from AMD). We're going to see a LOT more chip brands working with AMD and NVidia for their mobile chips and late 2024/early 2025 when AMD and NVidia release their Armv9 chips for Windows we're going to see some REAL competition for performance (relative of course as software refinement and real world use cases is key) for Apple's M3 Ultra and M4 series.

Lastly, TSMC and Samsung may have a serious competition as the current Lithography method for creating SoC's on die will have a competitor from Cannon. Cannon using a very old method that was left and ignored mostly due to a few issues that have now, according to Cannon been resolved.
 
Now I'm wondering why the M3 Ultra was not announced nor even hinted at. Either Apple engineers worked with TSMC and had to compromise on a few things via 3nm OR TSMC is having issues. Also seems to reason why so many other companies waited on first generation 3nm chips.
More likely because they didn't want to cannibalize sales of the relatively new M2 Ultra Mac Studio. The M3 Ultra Mac Studios will probably arrive late next January, or sometime next spring.
 
  • Like
Reactions: DeepIn2U
More likely because they didn't want to cannibalize sales of the relatively new M2 Ultra Mac Studio. The M3 Ultra Mac Studios will probably arrive late next January, or sometime next spring.
There's probably some of that, but even more likely that they don't have enough production capacity to even make M3 Ultra at this time. Which presumably gives them more time to tweak the design. Or perhaps forego M3 Ultra and move on to a complete M4 line.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.