M5 Pro and Max unveiled

RockRock8 said:
It is more efficient? P cores probably achieve most of what the S cores do but at a lower wattage. N1 is great, but I seriously doubt it the only reason for 2 hours longer battery life specifically on the Max model.
Click to expand...

I don’t quite follow the logic. E-cores are certainly more efficient than the P-cores, so I don’t see a valid argument for P-cores boosting battery life. Rather, changes to uncore, display optimization, and improved power gating would plays a key role. An improvement of 14 to 16 hours translates to a difference of just under 0.5 watt. I can totally see the network chip contributing a lion’s share of the reduction. Not to mention that Apple has been working on making the chip fabric more power-efficient.
 
leman said:
I don’t quite follow the logic. E-cores are certainly more efficient than the P-cores, so I don’t see a valid argument for P-cores boosting battery life. Rather, changes to uncore, display optimization, and improved power gating would plays a key role. An improvement of 14 to 16 hours translates to a difference of just under 0.5 watt. I can totally see the network chip contributing a lion’s share of the reduction. Not to mention that Apple has been working on making the chip fabric more power-efficient.
Click to expand...
The previous Max chip was 12/4, as in, the highest performing cores amount to 12, and the highest efficiency cores amounted to 4. The HP cores on the M4 Max draw more power than the HE cores on M4 Max. If you want to expand performance, while:

1. Exceeding battery life
2. Not continually pushing the E cores to be higher wattage gen over gen

Then you want to do what Apple did: design a new core built for high performing products like Mac notebooks and desktops.

You need to reason not just in the context of 1 generation, but multiple generations into the futture. Look at what happened with the M3 Pro with its core count and performance. I've seen multiple analyses note that E cores are increasing in wattage.

Solve both issues, boost performance by 20% for stuff like coding, and increase battery life by 2 hours by dropping the number of HP cores to 6 and create an all new core that offers most of the S core with way more efficiency and boost the number to 12. Now you've just upgraded to 18 cores across the board while maintaining thermal efficiency and boosting battery and not screwing yourself over for new designs for M chip cores. It's intelligent.
 
RockRock8 said:
The previous Max chip was 12/4, as in, the highest performing cores amount to 12, and the highest efficiency cores amounted to 4. The HP cores on the M4 Max draw more power than the HE cores on M4 Max. If you want to expand performance, while:

1. Exceeding battery life
2. Not continually pushing the E cores to be higher wattage gen over gen

Then you want to do what Apple did: design a new core built for high performing products like Mac notebooks and desktops.

You need to reason not just in the context of 1 generation, but multiple generations into the futture. Look at what happened with the M3 Pro with its core count and performance. I've seen multiple analyses note that E cores are increasing in wattage.

Solve both issues, boost performance by 20% for stuff like coding, and increase battery life by 2 hours by dropping the number of HP cores to 6 and create an all new core that offers most of the S core with way more efficiency and boost the number to 12. Now you've just upgraded to 18 cores across the board while maintaining thermal efficiency and boosting battery and not screwing yourself over for new designs for M chip cores. It's intelligent.
Click to expand...
Keep in mind, Apple's battery tests using streaming and web browsing are not typically not the intense all core workloads that would actually stress the new configuration - often they are even single threaded. So while it's good that Apple's new SOC design doesn't decrease efficiency in single/lightly threaded workloads, this isn't a sign, yet, of how the new cores will fare. For that, we'll have to wait for more intense tests done by third party reviewers measuring battery life or wall power directly. I do suspect the new design will indeed be more power efficient under load - hopefully we'll get die shots and we can also get a sense if it are more area efficient as well which I also suspect to be true.
 
dada_dave said:
Keep in mind, Apple's battery tests using streaming and web browsing are not typically not the intense all core workloads that would actually stress the new configuration - often they are even single threaded. So while it's good that Apple's new SOC design doesn't decrease efficiency in single/lightly threaded workloads, this isn't a sign, yet, of how the new cores will fare. For that, we'll have to wait for more intense tests done by third party reviewers measuring battery life or wall power directly. I do suspect the new design will indeed be more power efficient under load - hopefully we'll get die shots and we can also get a sense if it are more area efficient as well which I also suspect to be true.
Click to expand...
Keep in mind Max models suffered in battery life compared to Pro despite using E cores. Even just browsing the web Max models had less.

I think it is a sign of how the new cores will fare, but your opinion is your opinion and you're entitled to it. It doesn't mean that they didn't just boost 20% coding compilation performance while adding 2 extra hours of web browsing time and maintaining thermal efficiency despite also having more GPU cores, which does add to idle draw (which is in one reason in part why M5 Max gets 1 hour less battery life than the M5 Pro 16" model).

M3 Pro was a great chip but considered a "side grade" by many. I'm using this as evidence for what happens when you change the core configuration to be less HP and more HE
 
RockRock8 said:
Keep in mind Max models suffered in battery life compared to Pro despite using E cores. Even just browsing the web Max models had less.
Click to expand...

Which is precisely the evidence that the cores have little to do with it.

I agree that replacing E-cores with new cores optimized for throughput/area|watt is a good choice for professional workloads (essentially what you state in #82). This also has been mentioned earlier, both by myself and others. However, this has little to do with the battery life.
 
1772642070019.png


I’m telling you, it’s confusing. Paul ain’t exactly a newb.
 
Theory: We will look back on M5 as an outlier (sort of like M3), and M6 Max/Pro coming later this year will look very different in terms of core arrangement.
 
RockRock8 said:
Keep in mind Max models suffered in battery life compared to Pro despite using E cores. Even just browsing the web Max models had less.

I think it is a sign of how the new cores will fare, but your opinion is your opinion and you're entitled to it. It doesn't mean that they didn't just boost 20% coding compilation performance while adding 2 extra hours of web browsing time and maintaining thermal efficiency despite also having more GPU cores, which does add to idle draw (which is in one reason in part why M5 Max gets 1 hour less battery life than the M5 Pro 16" model).

M3 Pro was a great chip but considered a "side grade" by many. I'm using this as evidence for what happens when you change the core configuration to be less HP and more HE
Click to expand...
To build off of what @leman said here's a chart:

Screenshot 2026-03-04 at 9.10.45 AM.png


The differences in ST efficiency between the various M4 models aren't due to the cores that aren't being used, but rather the differences in "un-core" - the larger fabric, the caches, the larger RAM/memory bus, etc ... (and remember I subtract out idle in my efficiency analysis, this purely load efficiency). Even the Apple M5 pictured might have slightly lower efficiency to the M4 not because of the core changes but because it was a 32GB model and the M4 was a 16GB model.

Cmaier said:
Theory: We will look back on M5 as an outlier (sort of like M3), and M6 Max/Pro coming later this year will look very different in terms of core arrangement.
Click to expand...

Oh? You think they'll make big changes again in the M6 that will be the pattern going forward?

Yoused said:
Could the M5 Pro and Max already be on N2? It has been in production for several months.
Click to expand...

Extremely unlikely - I would say somewhere between true 0 and epsilon :). N2 is probably only starting true mass production around about now (maybe a touch earlier) and I suspect is focusing on iPhone and AI chips. Plus there's no reason to believe that Apple's switched.
 
Last edited:
dada_dave said:
Oh? You think they'll make big changes again in the M6 that will be the pattern going forward?
Click to expand...
Just a hunch. It’s odd enough to have two MBP releases in the same year (lately, at least), and I suspect M6 development went on in parallel with M5.
 
Again, this is why the new names are just stupid.
 
If M5 or M6 is on a multi-die package, I could see Ultra maybe looking very different. If they put the "S" cluster on the same die as the GPU and P on the I/O die, they might be able to make it "top-heavy", say 12P/18S with more GPU power.
 
Yoused said:
If M5 or M6 is on a multi-die package, I could see Ultra maybe looking very different. If they put the "S" cluster on the same die as the GPU and P on the I/O die, they might be able to make it "top-heavy", say 12P/18S with more GPU power.
Click to expand...
From my sources, I believe the CPU cores are all on the same die, and the GPU is on the other die.
 
RockRock8 said:
Keep in mind Max models suffered in battery life compared to Pro despite using E cores. Even just browsing the web Max models had less.

I think it is a sign of how the new cores will fare, but your opinion is your opinion and you're entitled to it. It doesn't mean that they didn't just boost 20% coding compilation performance while adding 2 extra hours of web browsing time and maintaining thermal efficiency despite also having more GPU cores, which does add to idle draw (which is in one reason in part why M5 Max gets 1 hour less battery life than the M5 Pro 16" model).

M3 Pro was a great chip but considered a "side grade" by many. I'm using this as evidence for what happens when you change the core configuration to be less HP and more HE
Click to expand...

leman said:
Which is precisely the evidence that the cores have little to do with it.

I agree that replacing E-cores with new cores optimized for throughput/area|watt is a good choice for professional workloads (essentially what you state in #82). This also has been mentioned earlier, both by myself and others. However, this has little to do with the battery life.
Click to expand...

Despite my earlier statements, there is one way where the new SOC design might make a difference and that's if Apple shoves the streaming/web browsing thread onto the M-"Performance" cores instead of the P-"Super" cores on battery - Intel does this to save battery life. I don't think Apple will do that given their P-cores are extremely efficient compared to Intel's P-core and Apple tends to run full throttle on battery vs plugged-in while Intel very much doesn't do so, but Apple might do this. That could potentially result in the increased battery life and it be from the change in core design.
 
leman said:
Which is precisely the evidence that the cores have little to do with it.
Click to expand...
I don't follow the logic.

They increased performance by 20% in code compilation, they boosted core count to 18, they dropped HP cores to 6, and boosted HE cores to 12 with a brand new core. All of this, and it gets 2 hours longer in battery M5 Max to M4 Max.

It's directly tied to the arch change
 
Back
Top