Battery Life 14" 8 core vs 10 core?

[daroga]

Hi all!

Watching The Verge's video and some of their observations of the M1 Pro vs M1 Max battery life that the M1 Max with added GPU cores seems to have a noticeable impact on battery life. That got me thinking a little bit: have there been any comparisons between the 8CPU/14GPU M1 Pro and the 10CPU/16GPU M1 Pro in the 14" shell?

Obviously, you'll get more performance out of the one with more cores (if using something that can use it), but I wondered if there was any noticeable effect on battery life when some of those cores are hard-disabled.

 

[JohnnyGo]

Logic dictates that there will be some impact but the real question is if that is significant for the average user.

For power users, if your performance is up 20% and there is a 10% hit in battery life… they actually got more done with their battery!

 

[bill-p]

I don't have a 8-core machine to compare, but I don't think the saving will be significant. Unused cores won't impact efficiency much. It's what's in use that matters.

There's one fundamental difference between the M1, M1 Pro and the M1 Max: performance of the memory.

M1 has very low RAM bandwidth. It seems to top out around 65GB/s
M1 Pro has ample RAM bandwidth, at around 200GB/s
M1 Max has a lot of RAM bandwidth, at 400GB/s

And you cannot "turn off" unused RAM. So RAM should always be on. Faster RAM consumes more power, as that's how it goes.

Since the M1 Pro w/ 8 cores and M1 Pro w/ 10 cores seem to feature the same memory system, I doubt they'd differ much in normal power consumption.

 

[Malus120]

Hi all!

Watching The Verge's video and some of their observations of the M1 Pro vs M1 Max battery life that the M1 Max with added GPU cores seems to have a noticeable impact on battery life. That got me thinking a little bit: have there been any comparisons between the 8CPU/14GPU M1 Pro and the 10CPU/16GPU M1 Pro in the 14" shell?

Obviously, you'll get more performance out of the one with more cores (if using something that can use it), but I wondered if there was any noticeable effect on battery life when some of those cores are hard-disabled.

This is a good question that highlights a really problematic aspect of the way a lot of outlets have been covering the 14/16" MBP's in their reviews/previews. There are now so many different SKUs (8C.14C/10C.14C/10C.16C/10C.24C/10C.32C) that making a blanket statement about the battery life of either the 14" or 16" is really difficult.
While I certainly don't expect every outlet to test every variant, it would be nice if the outlets with means committed to testing at least 14 Pro / 14 Max / 16 Pro / 16 Max so we could have more like for like results to draw inference from in making our own purchasing decisions.

I'd also really like to see a major outlet measure these machines in low power mode. It's about a 40% drop in performance and while that is noticeable, it should still be noticeably faster than the M1. While it may or may not make sense for pro workloads, for office workloads where you're just trying to wring every last minute/hour out of the device's battery it could help cut the increased power draw and bring battery life back into line with the M1 machines...

There's one fundamental difference between the M1, M1 Pro and the M1 Max: performance of the memory.

M1 has very low RAM bandwidth. It seems to top out around 65GB/s
M1 Pro has ample RAM bandwidth, at around 200GB/s
M1 Max has a lot of RAM bandwidth, at 400GB/s

This is an interesting bit of conjecture. Apple's powermetrics command-line tool (see the link below for more info) should be able to tell us a lot about how much the extra memory bandwidth, cpu, and gpu cores each contribute to power draw. I'd love to see an outlet with access to multiple SKU's really geek out and run this on each one to give us a clear cut answer but if nothing else i've got a 10C/32C 14" coming next week and I'll try and compare figures with my M1 MBA.

Apple Silicon M1 Power Consumption Deep Dive Part 1: Safari vs Chrome

Apple M1 is an amazingly efficient piece of silicon. How much power do you think it consumes while streaming from YouTube or Netflix? Read on to find out.

singhkays.com

 

[JimBobHeller]

I have not been blown away by the 14's battery life. I had a zoom call running, and I played Civ 6 windowed at low-res (one up from min) and ultra graphic settings. I got about 3 hours. The system ran well the entire time though, and the fans stayed relatively quiet. The heat was not uncomfortable.

If you're using the 14 on your lap, the bottom feels cooler than the 16 (subjectively). You mainly feel the heat in the back center. Since I'm coming from a 2019 16 i9, this is perhaps the most important quality for me.

I think the high range in battery life that we're seeing has a lot to do with the high efficiency cores, and how much of the workload can be handled by them. Also, how much of the time the GPU is being challenged. This makes sense, but I think Apple's presentation presented a rosier portrait (surprise, surprise).

 

[smirking]

There's one fundamental difference between the M1, M1 Pro and the M1 Max: performance of the memory.

M1 has very low RAM bandwidth. It seems to top out around 65GB/s
M1 Pro has ample RAM bandwidth, at around 200GB/s
M1 Max has a lot of RAM bandwidth, at 400GB/s

Anyone have any thoughts on just how great of an extra draw on power we should see with memory at 400GB/s vs 200GB/s?

I'm starting to wonder if I should cancel my M1 Max and get an M1 Pro instead.

 

[Malus120]

Anyone have any thoughts on just how great of an extra draw on power we should see with memory at 400GB/s vs 200GB/s?

I'm starting to wonder if I should cancel my M1 Max and get an M1 Pro instead.

Given both the lead times on these things and Apple's generous return window I wouldn't cancel anything ATM.
IMHO wait for more benchmarks, and then test for yourself once it arrives. As strange as it is given that we're now a week out from the review embargo lifting, we're still only scratching the surface of what these machines can do and the differences between them.
I'll test dram power draw vs my M1 MBA once mine (14" M1 Max 32C) arrives later this week.

 

[bill-p]

This is an interesting bit of conjecture. Apple's powermetrics command-line tool (see the link below for more info) should be able to tell us a lot about how much the extra memory bandwidth, cpu, and gpu cores each contribute to power draw. I'd love to see an outlet with access to multiple SKU's really geek out and run this on each one to give us a clear cut answer but if nothing else i've got a 10C/32C 14" coming next week and I'll try and compare figures with my M1 MBA.

Apple Silicon M1 Power Consumption Deep Dive Part 1: Safari vs Chrome

Apple M1 is an amazingly efficient piece of silicon. How much power do you think it consumes while streaming from YouTube or Netflix? Read on to find out.

singhkays.com

I'm not the only one making this conjecture. Anandtech already alluded to this in their first-look article:

Apple's M1 Pro, M1 Max SoCs Investigated: New Performance and Efficiency Heights

www.anandtech.com

Relevant bit:

In multi-threaded scenarios, the package and wall power vary from 34-43W on package, and wall active power from 40 to 62W. 503.bwaves stands out as having a larger difference between wall power and reported package power – although Apple’s powermetrics showcases a “DRAM” power figure, I think this is just the memory controllers, and that the actual DRAM is not accounted for in the package power figure – the extra wattage that we’re measuring here, because it’s a massive DRAM workload, would be the memory of the M1 Max package.

So they did use Apple's powermetrics command and they're also not seeing the figure that would correspond with what they are measuring as power draw. Note that Anandtech also measured the M1 this way and the figures were much closer to what was reported, so the conjecture is that... Apple is basically just not reporting memory power draw, and that difference is pretty significant when memory usage is high.

And it does make sense that there has to be some kind of penalty for higher performance. It's not like we can just stack cores and get free performance out of them in a linear fashion.

 

[Malus120]

I'm not the only one making this conjecture. Anandtech already alluded to this in their first-look article:

So they did use Apple's powermetrics command and they're also not seeing the figure that would correspond with what they are measuring as power draw. Note that Anandtech also measured the M1 this way and the figures were much closer to what was reported, so the conjecture is that... Apple is basically just not reporting memory power draw, and that difference is pretty significant when memory usage is high.

Good catch. I'd forgotten about (ok, the first time my eyes kinda glazed over) that part of the Anandtech M1 Pro/Max overview. It does point to memory being a potential culprit but, as you'll notice, that part of the article doesn't include data for the M1 Pro (not to mention their M1 Pro was a 14" vs the Max being a 16") so again, I think we need more data to draw a definitive conclusion.

And it does make sense that there has to be some kind of penalty for higher performance. It's not like we can just stack cores and get free performance out of them in a linear fashion.

Obviously no one expected us to be able to get "free performance" (AKA higher performance at the same power draw,) on the M1 Max vs the M1 Pro (or on the M1 Pro vs the M1,) when the new chips are effectively just wider versions of the M1. However, if the M1 Max is getting significantly lower battery life without doing significantly more work and or at idle that isn't great design. Of course faster fabric and double the memory channels was never going going to be completely free in terms of power usage, but battery life in workloads that aren't stressing the gpu, memory or fabric really shouldn't be using massively more power...
(Also... If you look at the Anandtech article, the cores themselves do seem to be scaling in a fairly linear fashion power wise vs the M1)

 

[Pressure]

Anyone have any thoughts on just how great of an extra draw on power we should see with memory at 400GB/s vs 200GB/s?

I'm starting to wonder if I should cancel my M1 Max and get an M1 Pro instead.

Near zero. The RAM is clocked at exactly the same on the M1 Pro and M1 Max.

The difference between them is that the M1 Max has two extra 128-bit memory channels and two extra memory modules to populate them.

So the M1 Pro uses two LPDDR5 modules both connected to the SoC by a 128-bit memory channel giving you 256-bit of memory bandwidth.

The M1 Max used four LPDDR5 modules each connected to the SoC by a 128-bit memory channel giving you 512-bit of memory bandwidth.

The two extra 128-bit memory channels do consume some extra power though as well as the two extra memory modules themselves.

 

[smirking]

The two extra 128-bit memory channels do consume some extra power though as well as the two extra memory modules themselves.

So this means that we should expect to see a nominal increase in power usage of the M1 Max vs the Pro, but not enough for it to have any real world consequence?

 

[Pressure]

So this means that we should expect to see a nominal increase in power usage of the M1 Max vs the Pro, but not enough for it to have any real world consequence?

Well, it is not the power hungry GDDR6 or HBM like modern graphic cards use but it will still use power, obviously.

We don't know how well it is power gated and you also have twice the GPU cores, extra video encoders, neural engine, display engine and twice the amount of system level cache.

Depending on the workload the M1 Max SoC may be able to finish many tasks faster and thus going to idle quicker and conserve energy that way.

 

[JohnnyGo]

It’s well documented that RAM in mobile SOCs does affect power consumption/ battery life.

It stands to reason that higher memory versions of M1 PRO / Max would have lower battery life. The added CPU or GPU cores just further increase such energy consumption (when these extra cores are used).

And as another poster pointed out, the additional hardware/soc space used for more bandwidth controllers leads to an extra power draw (but much less than the ones mentioned above)

 

[smirking]

It stands to reason that higher memory versions of M1 PRO / Max would have lower battery life. The added CPU or GPU cores just further increase such energy consumption (when these extra cores are used).

Sigh, I guess we'll just go around in circles on this topic until we can get more specific benchmark tests.

 

[RamGuy]

LPDDR4X is typically 1.1V if they follow the standard PER module. Apple M1 Pro is running dual-channel, so you will have 2x 8GB LPDDR4X memory if you go with 16GB, and you will have 2x 16GB LPDDR4X if you go with 32GB.

Apple M1 Max is running quad-chanel, so you will have 4x 8GB LPDDR4X if you go with 32GB, and you will have 4x 16GB LPDDR4X if you go with 64GB.

There is no difference in power usage on 8GB vs 16GB modules. But as RAM is volatile meaning that they can't store any data unless they have power so RAM will always see constant power. The Apple M1 Pro will have 2.2V power to RAM when powered on, Apple M1 Max will have 4.4V power to RAM when powered on.

How much the difference between 2.2V and 4.4V is going to make depends on the battery. Obivosuly the Apple M1 Max will waste double amount of power running the quad-channel memory compared to the Apple M1 Pro running its dual-channel memory but how much of a difference this is going to result in at the end depends of how much the difference between 2.2V vs 4.4 is for battery life as a whole.

LPDDR4X do contain two different modes, so they do have one low-power mode reducing voltage under low loads but as this affects both Apple M1 Pro and Apple M1 Max the same way as they both feature the same kind of memory modules, it's just the M1 Max having two additional modules it won't make the difference in power draw between the Apple M1 Pro and Apple M1 Max any different.


This is why iPhone and iPad is still running single-channel memory. With their limited space for battery it doesn't make much sense for Apple to opt for dual-channel or quad-channel on iPhone or iPad as the performance gains won't be worth it compared to the additional power draw. Doesn't mean that Apple couldn't offer 16GB RAM on both iPhone and iPad by default as there wouldn't be any difference in power draw by going 16GB single-channel LPDDR4X compared to 8GB single-channel LPDDR4X or whatever.

From what I understand LPDDR5X is going to be under 1.0V but it's not produced in volume as of yet. It will also be even more dynamic featuring three different modes instead of only two on LPDDR5X. It also comes with the benefit of having on-board ECC capabilities so it will be much safer to run systems with 64GB RAM+ without running into memory corruption.

 

[kirk.vino]

LPDDR4X is typically 1.1V if they follow the standard PER module. Apple M1 Pro is running dual-channel, so you will have 2x 8GB LPDDR4X memory if you go with 16GB, and you will have 2x 16GB LPDDR4X if you go with 32GB.

Apple M1 Max is running quad-chanel, so you will have 4x 8GB LPDDR4X if you go with 32GB, and you will have 4x 16GB LPDDR4X if you go with 64GB.

There is no difference in power usage on 8GB vs 16GB modules. But as RAM is volatile meaning that they can't store any data unless they have power so RAM will always see constant power. The Apple M1 Pro will have 2.2V power to RAM when powered on, Apple M1 Max will have 4.4V power to RAM when powered on.

How much the difference between 2.2V and 4.4V is going to make depends on the battery. Obivosuly the Apple M1 Max will waste double amount of power running the quad-channel memory compared to the Apple M1 Pro running its dual-channel memory but how much of a difference this is going to result in at the end depends of how much the difference between 2.2V vs 4.4 is for battery life as a whole.

LPDDR4X do contain two different modes, so they do have one low-power mode reducing voltage under low loads but as this affects both Apple M1 Pro and Apple M1 Max the same way as they both feature the same kind of memory modules, it's just the M1 Max having two additional modules it won't make the difference in power draw between the Apple M1 Pro and Apple M1 Max any different.


This is why iPhone and iPad is still running single-channel memory. With their limited space for battery it doesn't make much sense for Apple to opt for dual-channel or quad-channel on iPhone or iPad as the performance gains won't be worth it compared to the additional power draw. Doesn't mean that Apple couldn't offer 16GB RAM on both iPhone and iPad by default as there wouldn't be any difference in power draw by going 16GB single-channel LPDDR4X compared to 8GB single-channel LPDDR4X or whatever.

From what I understand LPDDR5X is going to be under 1.0V but it's not produced in volume as of yet. It will also be even more dynamic featuring three different modes instead of only two on LPDDR5X. It also comes with the benefit of having on-board ECC capabilities so it will be much safer to run systems with 64GB RAM+ without running into memory corruption.

The M1 Pro and Pro Max have the LPDDR5 RAM

 

[Grohowiak]

Under moderate use I got 6.5h last night with the base 14".
8 safari windows with dynamic charts running at all times + spotify + some browsing and charting and all connected to external 38" with internal screen on.

 

[Malus120]

My 14" 32C M1 Max Macbook Pro arrived today so as promised here are some super preliminary results on power consumption compared to my (thermal modded) M1 Macbook Air.

First impressions/TLR - Just launching the powermetrics command line tool... holy crap the DRAM power usage is absolutely insane. I would not be surprised at all if this is our battery killing culprit... (although there does seem to be some other unknown factors that play a role as well)

Anyway, on to the results!

Idle:


These screenshots show the power usage of each machine Idling, with a third shot of the lowest score I could get on the MBP w/ Low Power Mode:

Next up we have both machines running Cinebench Multicore:

Finally here are both machines running the same Cinebench Multicore test, but in low power mode:

Here are some (very) preliminary observations and (somewhat dubious) "conclusions" we can draw or theorize about regarding power/performance and battery:

0. If you have a keen eye, you may have noticed that on both machines, Package Power is higher than the total for CPU/DRAM/GPU/ANE (the component parts,) but for the M1 Max it is significantly higher (more than 2x), whereas on the MBA its only about 10% higher on average.
a. This may be be due to the MBP having multiple CPU/GPU clusters but the math isn't clean so I can't say for sure.

1. The power usage of the memory subsystem appears to have increased by almost 10x (!) in the limited scenarios I have tested.
a. In idle/near idle workloads, its not uncommon for the DRAM alone to exceed the power draw of the entire M1 power package in the MBA.
b. Consequently the idle DRAM power usage, and thus overall idle power consumption is incredibly poor compared to the M1.
c. Conversely, when under 100% load, the DRAM can consume almost half as much power as the M1 power package in the MBA does when in low power mode at 100% load.

2. Low Power mode on the M1 Max only reduces the multicore CPU performance by around 20% (from ~12295 to 10,179) compared to the ~40% reduction seen on the M1 MBA (7601 vs 4379)
a. It manages to reduce power consumption by around 40% however, making it much more efficient, while still managing to be extremely fast. The M1 meanwhile, sees a more massive ~70% reduction in package power draw albeit with a consequently much larger drop in performance.
b. As noted above, M1 Pro/Max CPU multicore performance under Low Power Mode is still incredibly impressive (40%~50% faster than the M1 for 10C Pro/Max models, roughly equivalent to a desktop 8C16T Ryzen 3700X)
b. DRAM power usage however, appears to be almost completely unaffected by low power mode, so power usage at idle and with lighter workloads is improved but still fairly poor compared to the M1.

Taking the above into account I would tentatively say that (a lot more data is needed and) Low Power mode is going to be a godsend for this machine and is probably the optimal way to use the M1 Max if you need anything like the all day battery life the first M1 MBA/MBP could deliver.
However, it's not a silver bullet. The amount of power drawn by the DRAM (and package power draw being roughly 2x the component parts) even when idle and using low power mode means it likely can't compare to the power efficiency of the M1 MBA/MBP during casual use and therefore battery life will be noticeably worse despite having a 20% larger battery than the 13" M1 MBP and a 40% larger battery than the MBA (49.9 MBA, 58.2 13" MBP, 69.6 14" MBP

Anyway, there's still A LOT more to unpack here and I'll try and follow up with more testing in the coming days and report what I find.

 

[jabbr]

^ really appreciate the breakdown. Super interesting about idle power usage of the DRAM. Could the voltage be tweaked in a firmware update for a better low power mode or is this a fixed hardware value?

 

[Pressure]

Anyway, there's still A LOT more to unpack here and I'll try and follow up with more testing in the coming days and report what I find.

You also have the bigger and brighter display.

 

[blaraka]

Would be great to see same data for M1Max 14, I'm on the fence between m1 pro 32 gb and m1 max 64gb because of battery life.

 

[archi penko]

Anyway, there's still A LOT more to unpack here and I'll try and follow up with more testing in the coming days and report what I find.

What did you find out?

 

[doublemycoresplz]

My 14" 32C M1 Max Macbook Pro arrived today so as promised here are some super preliminary results on power consumption compared to my (thermal modded) M1 Macbook Air.

First impressions/TLR - Just launching the powermetrics command line tool... holy crap the DRAM power usage is absolutely insane. I would not be surprised at all if this is our battery killing culprit... (although there does seem to be some other unknown factors that play a role as well)

Anyway, on to the results!

Idle:


These screenshots show the power usage of each machine Idling, with a third shot of the lowest score I could get on the MBP w/ Low Power Mode:
View attachment 1900566 View attachment 1900567 View attachment 1900568

Next up we have both machines running Cinebench Multicore:
View attachment 1900571 View attachment 1900662


Finally here are both machines running the same Cinebench Multicore test, but in low power mode:
View attachment 1900574 View attachment 1900575

Here are some (very) preliminary observations and (somewhat dubious) "conclusions" we can draw or theorize about regarding power/performance and battery:

0. If you have a keen eye, you may have noticed that on both machines, Package Power is higher than the total for CPU/DRAM/GPU/ANE (the component parts,) but for the M1 Max it is significantly higher (more than 2x), whereas on the MBA its only about 10% higher on average.
a. This may be be due to the MBP having multiple CPU/GPU clusters but the math isn't clean so I can't say for sure.

1. The power usage of the memory subsystem appears to have increased by almost 10x (!) in the limited scenarios I have tested.
a. In idle/near idle workloads, its not uncommon for the DRAM alone to exceed the power draw of the entire M1 power package in the MBA.
b. Consequently the idle DRAM power usage, and thus overall idle power consumption is incredibly poor compared to the M1.
c. Conversely, when under 100% load, the DRAM can consume almost half as much power as the M1 power package in the MBA does when in low power mode at 100% load.

2. Low Power mode on the M1 Max only reduces the multicore CPU performance by around 20% (from ~12295 to 10,179) compared to the ~40% reduction seen on the M1 MBA (7601 vs 4379)
a. It manages to reduce power consumption by around 40% however, making it much more efficient, while still managing to be extremely fast. The M1 meanwhile, sees a more massive ~70% reduction in package power draw albeit with a consequently much larger drop in performance.
b. As noted above, M1 Pro/Max CPU multicore performance under Low Power Mode is still incredibly impressive (40%~50% faster than the M1 for 10C Pro/Max models, roughly equivalent to a desktop 8C16T Ryzen 3700X)
b. DRAM power usage however, appears to be almost completely unaffected by low power mode, so power usage at idle and with lighter workloads is improved but still fairly poor compared to the M1.

Taking the above into account I would tentatively say that (a lot more data is needed and) Low Power mode is going to be a godsend for this machine and is probably the optimal way to use the M1 Max if you need anything like the all day battery life the first M1 MBA/MBP could deliver.
However, it's not a silver bullet. The amount of power drawn by the DRAM (and package power draw being roughly 2x the component parts) even when idle and using low power mode means it likely can't compare to the power efficiency of the M1 MBA/MBP during casual use and therefore battery life will be noticeably worse despite having a 20% larger battery than the 13" M1 MBP and a 40% larger battery than the MBA (49.9 MBA, 58.2 13" MBP, 69.6 14" MBP

Anyway, there's still A LOT more to unpack here and I'll try and follow up with more testing in the coming days and report what I find.

Absolutely stellar investigation. Thank you.

 

[Malus120]

Absolutely stellar investigation. Thank you.

Thanks, I was just doing the testing I wanted to see done

What did you find out?

I've unfortunately had a lot less time to be able to do this kind of testing this kind of testing than I hoped. Maybe I'll have more time this next week...

That said, I have had the opportunity to use the machine on battery a bit lately so here are some much less granular findings (so... YMMV.)

First, 13-14 hours for "wireless web" AKA a mix of Safari/Mail/Office Apps/Music/SNS seems to be about what I'm averaging when out in "regular mode" (haven't managed to actually run the battery all the way down yet), and this seems like it can be extended by another 10~15% using low power mode, and possibly even more with some additional tweaking (FYI, when out I run the screen in the 500 Nits mode as I've heard that saves on battery, maybe I'll try it in "1600 nits for HDR" mode next time)

Second, regarding more moderate/demanding loads, I was processing 24MP RAW photos (Fuji RAF) in Capture One Pro 21 today at Starbucks with a good amount of masking/local editing and chewed through about 11% in 30 minutes in low power mode... It's not as bad as it sounds though, here is some more information:
1. I was in low power mode because, aside from the initial loading/caching, which does seem to take a tad longer, performance even in low power is great and I'll take more battery life when out.
2. I was working off an external USB 3.1 SSD (so that's going to take a good chunk more power than using the internal drive.)
3. I was connected to a VPN for security, which obviously creates a low, but constant and not non existent background load. I also had Safari/Mail/Messages/Office/etc open in the background, so I definitely wasn't trying for maximum battery life.
4. Screen Brightness was 7 clicks from the lowest.
5. The machine chewed through a variety of tasks I gave it over 30 minutes... while also chewing through 11% of the battery.... So basic math suggests that under these conditions the battery could be expected to last ~4.5 hours.

My take (ATM):
The IMHO massive upside is that under light, office or web loads the machine can achieve very good battery life. It's not quite as good as an M1 MBA, and certainly can't match the M1 MBP, but we're still well north of 10 hours and thus talking about being able to go a full day at work (with some overtime) or an entire transpacific flight, and still having a few hours left to spare.
The downside is that for more demanding loads like photo editing, the battery life while more than acceptable given the performance, is not really comparable to the M1, at least not without a lot of tweaking/optimization. That said, it's still good enough to get you through more than half a standard 8 hour workday so, particularly for people like myself, who do mixed workloads it should still last all day.
The real question I have yet to answer, but I'm sure someone on Youtube already has is, how long can the machine run on battery with a heavy/constant CPU, GPU or combined load?

Some more thoughts... in a lot of ways, while it doesn't quite deliver "magical" battery life for demanding tasks, this is really only when compared the M1 which isn't really fair. The M1 Max is in an entirely different performance class. This is a no compromise CPU/GPU compute monster in a relatively small and light 14" chassis. Honestly I'm often surprised how light it is. It is heavier than my Air but honestly not by THAT much. If you consider that you're getting one of the fastest 8(10) core CPU's on the market, and (for productivity) the equivalent of a Desktop RX 6800/RTX 3070 performance, it's absolutely wild that it can even run at full tilt in this chassis at all let alone for hours on end.

That said, it is expensive, and as I contemplate whether I really need this or not, one thing I've realized is... as gaming is unfortunately, not a thing on the Mac at the moment, I'm beginning to feel like dropping down to the 24C GPU Core M1 Max and getting 64GB of Ram instead may be a better use of funds. Although 32GB is the sweet spot for me (where I don't run into memory issues like with 16GB), 64GB would open the door to a lot of new possibilities and use cases in a way that 8 more GPU cores probably doesn't...

Anyway lots of rambling but I hope it helps someone. I love this machine so far! If anyone has a 24C model I'd love to hear about performance, battery life or links to benchmarks!

 

[doublemycoresplz]

First, 13-14 hours for "wireless web" AKA a mix of Safari/Mail/Office Apps/Music/SNS seems to be about what I'm averaging when out in "regular mode."

I'll be stoked if the 32 core Max 16' gets that much battery life for these tasks.