Battery life on M1X PRO vs M1X MAX?

[Aka757]

Thank you for being so thorough. Hopefully we can put to rest whether there is some kind of difference or not on this forum now. Really makes me curious on the pro 32gb pro vs 16gb pro though... As well as the 32gb Pro vs 32gb max with 24gpu

I am also curious about Pro32 vs Pro16; I have a Pro16 (base model) right now but was planning to get a Pro32 for peace of mind. If there’s a noticeable difference I’d happily keep the Pro16.

 

[hmorneau]

Please tell me if you want me to test anything else on them while I'm at it. I'll do it tomorrow at most, after that I'll pick a winner and stop doing tests

Can you open the Coconut battery on both and see the capacity? From my own experience sometimes when the battery is new the capacity is actually lower than after a few cycles (seems like the BMS needs a few cycles to understand the cells). Like mine was quite low when brand new, 98% of original capacity and now it's at 100%.

But honestly, if you have a 64GB in hand, for what 5 or 10% difference in battery life, keep the Max

I went with the Max 32GB 24c but kind of regret not spending the little extra to get 64gb. Even if 32gb should be good for the years to come for me.

 

[Charlesje]

@KirkDayne

Thanks for these results. Would you mind doing the cinebench test (cpu) while watching the power metrics? So we have numbers that can be compared between the two memory sizes of the Max. I’m still wondering how the memory bus and dram draw scales with relative cpu use (while not taxing the gpu). Between this, and power draw at idle use we could get a grasp of average use.

But I don’t expect the difference to be really big either. But it’s still a more or less open question.

 

[Fomalhaut]

@Fomalhaut
@Charlesje
@breath.by.breath
@obwohl
@hmorneau

Alright so as promised, I received the new mac and was able to make the tests I always wanted to do.

So first, which computers I am comparing?
1 - 14 inch, M1 Max 10c/32c/64GB -> referred bellow to as "Max64"
2 - 14 inch, M1 Pro 10c/16c/32GB -> referred bellow to as "Pro32"

First test "the idle battery drain test", aka reddit thread test
Methodology: please read the reddit post linked in earlier pages, I used the same.

I had already done the test once with Max64, and result was about 12/13% drained per hour.
This time, same test with Pro32 gave.... I guess similar results. I say "guess" cause I wish I did it with the same precision as Max64 but I forgot to note the percentage two times in a row, after what it weirdly seemed to have been drawn more than Max64, but it doesn't make sense.
In other words, I think it didn't have any impact, but I will retest that tomorrow just to be sure.

Second test "1080p youtube video playing in safari"
Methodology:
- disabled ability of screen to change brightness automatically
- nothing else besides safari is launched
- both computer are set to the exact same brightness level
- both play the same video
- 1080p video
- both computer charged to 100% before starting the test
- videos were set full screen

Limitations:
- At the very end I have had the doubt regarding if brightness was actually the same or not, because I figured out brightness change automatically when you plug-in the charger (it adds brightness), and when you unplug it (it lowers brightness). I know I had put the same amount of tick manually, but maybe that messed things up cause I dont remember if I controlled at beginning of the test. I guess it was same tho.
- I am unsure that both videos were to "1080p" maybe one of them was on "1080p (auto)" which means it theoretically could have changed during playback. I should have picked it manually for both.
- I forgot about ADs... Youtube launches ads that make video play unsynced, so conditions were not perfectly identical
- I messed up the time, I started at 36 min, and then for some reason I switched to 16min.

For those reasons I have made a third test where I controlled for all of these. But for now, here are the results.

Results:

Hour | Pro32 | Max64 | Difference
=======================
20h36 | 100% | 100% | 0%
21h36 | 96% | 94% | -2%
22h36 | 88% | 84% | -4%
23h18 | 81% | 77% | -4% (!warning: not one hour)
00h18 | 72% | 67% | -5% (!warning: one hour from last one, but only 1% diff increase)

Conclusion: I am not confident in this test and I decided to make another one with no flaws.

Third test "Two 1080p youtube videos playing simultaneously on different tabs, on Chrome"

Methodology:
- Browser is Chrome this time, not safari, because why not? Also I never use Safari as a developper so it better matches my use case
- Two videos playing at the same time on different tabs
- Audio disabled on video, computer audio disabled
- First video is playing full screen and other in the background
- Video playing fullscreen was a countdown timer so not very demanding in pixel color
- Both are set manually to 1080p
- Brightness have been double-checked to be exactly the same: 10 ticks.
- Start battery percentage was 66%, I wanted to try not starting from 100% and thats the first number I could manage to sync both machines on.
- I installed an ad blocker.

Limitations:
- I don't see any, tho I forgot to take note once at 18h00.

Results:

Hour | Pro32 | Max64 | Difference
=======================
12h00 | 66% | 66% | 0%
13h00 | 61% | 59% | -2%
14h00 | 55% | 53% | -2%
15h00 | 49% | 46% | -3%
16h00 | 43% | 39% | -4%
17h00 | 36% | 32% | -4%
18h00 |
19h00 | 24% | 19% | -5%
20h00 | 18% | 12% | -6%
21h00 | 13% | 6% | -7%

Conclusion:
I secretly hoped the Max64 to display the exact same battery performance, especially because the task was really easy for both computers.
Unfortunately, we can see that the Max64 has worse battery, even when not using its extra horse power.
We can note that battery drain speed seem to slightly increase the more we approach 0% left.
In the end, with that particular test, which is no near real use case where you could have many software opened at the same time, many tabs, virtual machines, etc, Pro32 seems to have at the very least 1 hour more battery.
In real world use, how much will that be? Not really sure, but its hard to do lighter use than this test, so I'd say at the VERY least, you loose 1h30 of battery life when you go for the Max64 compared to Pro32.
Maybe that difference can increase the more you push on it (with non-gpu tasks). EDIT: since I started at 66%, differences could have been broader if started from 100% and so it may be safe to assume around 2h difference.

What do you guys think?

Please tell me if you want me to test anything else on them while I'm at it. I'll do it tomorrow at most, after that I'll pick a winner and stop doing tests

Thanks for doing this!

These are quite encouraging results I think. You could have comfortably gone on for almost another hour with the Max64 and maybe 2 hours with the Pro32. A fairly consistent 6-7% battery usage per hour is better than I expected.

I still wonder whether the largest cause of the disparity between Max and Pro is actually the difference in RAM.

Could you please go and buy a 32GB M1 Max and repeat the experiment?

 

[Fomalhaut]

Thanks for testing!!! Though I don't agree with starting the third test at 66%, but I understand why it was done and based on the results / differences until the end it appears the max definitely uses more power regardless of that little nitpick.

I think starting at a lower percentage than 100% is actually a better test in some ways. It doesn't give "absolute" battery life from 100% of course but is arguably a more realistic test. We know that reported battery usage is highly non-linear near 100% (and I suspect the battery is actually charged to >100%, or it only reports any drop after capacity has dropped to well below 100%). You certainly can't extrapolate actual battery life from the time it takes to drop from a reported 100% to say 90%.

Starting at 2/3 battery means the battery drop is within the linear part of the curve, as demonstrated by the fairly consistent 6-7% drop per hour over the next 9 hours.

 

[ChpStcks]

Can you open the Coconut battery on both and see the capacity? From my own experience sometimes when the battery is new the capacity is actually lower than after a few cycles (seems like the BMS needs a few cycles to understand the cells). Like mine was quite low when brand new, 98% of original capacity and now it's at 100%.

But honestly, if you have a 64GB in hand, for what 5 or 10% difference in battery life, keep the Max

I went with the Max 32GB 24c but kind of regret not spending the little extra to get 64gb. Even if 32gb should be good for the years to come for me.

Coconut battery says my battery is 15,000 days old?

 

[ChpStcks]

I think starting at a lower percentage than 100% is actually a better test in some ways. It doesn't give "absolute" battery life from 100% of course but is arguably a more realistic test. We know that reported battery usage is highly non-linear near 100% (and I suspect the battery is actually charged to >100%, or it only reports any drop after capacity has dropped to well below 100%). You certainly can't extrapolate actual battery life from the time it takes to drop from a reported 100% to say 90%.

Starting at 2/3 battery means the battery drop is within the linear part of the curve, as demonstrated by the fairly consistent 6-7% drop per hour over the next 9 hours.

I understand where you're coming from with the absolute 100%, but I think the only way to test is starting both full charged over night at 100%. because if you're starting both at say 66%, one could've just hit 66% and one could be on it's way to 65%, that variance in 1% difference could easily be 6-10mins

 

[Fomalhaut]

I understand where you're coming from with the absolute 100%, but I think the only way to test is starting both full charged over night at 100%. because if you're starting both at say 66%, one could've just hit 66% and one could be on it's way to 65%, that variance in 1% difference could easily be 6-10mins

True...but I think 6-10 minutes difference would be just lost in the noise and variability of usage over an 8-12 hour period.

In any case, the only tests that "count" are the ones that exactly mimic *your* real-world usage, which is probably impossible to reproduce with any consistency unless you do exactly the same thing day-in, day-out.

Apart from screen brightness, which I prefer at about 70% (on my MBP16) in normally lit office or domestic spaces, I think that the biggest battery hit is using multiple applications and changing between them frequently (inevitably using swap space unless you have a lot of excess memory, which in itself may consume more power).

These tests are a great baseline, but my assumption is that I will probably only get 60-70% of the time shown at best, knowing my own usage (lots of apps, 50+ browser tabs, bright screen, power-hungry apps). If the machine consistently gets 8 hours unplugged and means I don't have to worry about carrying the power supply, then I would be happy.

 

[hmorneau]

I just switched from the M1 pro 16GB to the M1 Max 32gb (24c) and to me it feels the same. Also, the SOC temp while doing my stuff is about the same. I would say maybe the Max is cooler a bit (45C) vs 50C for the Pro, but I probably have less stuff running since I just reinstalled everything. I will see how it goes, but I just spent 4 hours installing software and setting up everything, I went from 85% to 54%. While installing all my stuff I don't think I went above 50C... which is impressive.

So far I have the feeling that the Max runs slightly cooler than the Pro on light load (well with those SOC, reinstalling all your app it's considered light load). But it's not scientific at all...

But for those who are afraid that the Max will run warmer and drain your battery, don't be.

Edit: Just tested a game, and the fan run a 1500rpm while gaming, same as the pro.

 

[Serengeti1]

It can feel the same but it's not. As all the evidence has already suggested. The sacrifice of battery life may be worth it for your usage case as it's not a massive sacrifice, but it's there.

This guy has both the models you have had and ran a cpu test.

I think the better questions at this point are 16gb pro vs 32gb Pro. As well as 32gb Pro vs 32gb with 24gpu cores. I personally highly doubt the ram is as impactful as the gpu cores. If it were, the narrative wouldn't be so strong about the gpu cores being the main culprit. The laptops have been out for a significant amount of time now. If that were the case, there would be more evidence suggesting so.

 

[Fomalhaut]

This guy has both the models you have had and ran a cpu test.

I think the better questions at this point are 16gb pro vs 32gb Pro. As well as 32gb Pro vs 32gb with 24gpu cores. I personally highly doubt the ram is as impactful as the gpu cores. If it were, the narrative wouldn't be so strong about the gpu cores being the main culprit. The laptops have been out for a significant amount of time now. If that were the case, there would be more evidence suggesting so.

The reason I mentioned RAM power usage was from here: #263

3 - ~800 mW package + 400mW DRAM on a 10c/32c/32GB M1 Max (https://forums.macrumors.com/threads/battery-life-on-m1x-pro-vs-m1x-max.2317578/post-30622884) vs
4 - ~1500 mW package + 750mW DRAM on a 10c/32c/64GB M1 Max (https://forums.macrumors.com/thread...st-idle-load-power-draw.2320968/post-30551266)

....and this post, #244 (in the same thread) shows the comparative power consumption of DRAM, CPU, GPU & Package. The DRAM power consumption (in this example) is 200x the GPU power consumption. Not entirely sure if that's a single GPU core or all of them, but even so you can see that DRAM power usage can be much more significant that GPU usage, when the GPU is not being used of course!

"So what I did is I recorded the screen with my phone (not with the computer itself or it would impact the numbers), for 1 minute, the values update 1 time per second, so this is 60 values for each parameter, and here is the average value of all 60 data points for each parameter:

DRAM Power: 210 mW
CPU Power: 15.6 mW
GPU Power: 1 mW
Package Power: 387 mW"

 

[Serengeti1]

The reason I mentioned RAM power usage was from here: #263

3 - ~800 mW package + 400mW DRAM on a 10c/32c/32GB M1 Max (https://forums.macrumors.com/threads/battery-life-on-m1x-pro-vs-m1x-max.2317578/post-30622884) vs
4 - ~1500 mW package + 750mW DRAM on a 10c/32c/64GB M1 Max (https://forums.macrumors.com/thread...st-idle-load-power-draw.2320968/post-30551266)

....and this post, #244 (in the same thread) shows the comparative power consumption of DRAM, CPU, GPU & Package. The DRAM power consumption (in this example) is 200x the GPU power consumption. Not entirely sure if that's a single GPU core or all of them, but even so you can see that DRAM power usage can be much more significant that GPU usage, when the GPU is not being used of course!

"So what I did is I recorded the screen with my phone (not with the computer itself or it would impact the numbers), for 1 minute, the values update 1 time per second, so this is 60 values for each parameter, and here is the average value of all 60 data points for each parameter:

DRAM Power: 210 mW
CPU Power: 15.6 mW
GPU Power: 1 mW
Package Power: 387 mW"

Way out of my depth to be able to know how reliable this is at predicting the outcome + the other variables involved in the outcome etc

I'd be very, very surprised if the hit from 16 Pro vs 32 Pro is the same as 16gb Pro vs max of any kind. That's just going by everything I've seen and read outside of this forum. I just think there'd naturally be more evidence by now.

I definitely agree it's worth testing it though as I'm not sure I've seen a thorough, formal test on this. Just people saying the 32 Pro is the sweet spot for battery life and performance.

 

[hmorneau]

It can feel the same but it's not. As all the evidence has already suggested. The sacrifice of battery life may be worth it for your usage case as it's not a massive sacrifice, but it's there.

This guy has both the models you have had and ran a cpu test.

I think the better questions at this point are 16gb pro vs 32gb Pro. As well as 32gb Pro vs 32gb with 24gpu cores. I personally highly doubt the ram is as impactful as the gpu cores. If it were, the narrative wouldn't be so strong about the gpu cores being the main culprit. The laptops have been out for a significant amount of time now. If that were the case, there would be more evidence suggesting so.

No, it's not. More and more reports of the Max lasting longer than the Pro actually.

Apple downclock the GPU to single-digit Hz when not in use, so they don't use much power. That's why I upgraded to the M1 Max. The most you will see is a 5% difference, but now there are reports of the opposite, that the Max outlast the Pro.

I won't even test the Pro vs Max as I feel it's a waste of time. Get the one that you need, not the Pro because of battery life. I'm freezing my finger on my Max right now... Those chips don't use much power.

Anyway, that was my last post on this thread. My conclusion: there is no significant difference between Max and Pro under similar load. Get the Pro or the Max, as long you get at least 32GB ram you are good to go.

 

[DoorsFreak87]

Despite more than a month of lengthy discussions, we still are not able to answer simple questions such as, how much less battery life do you get for choosing a 64GB version over a 32GB version, all other specs being equal...

Of course if you get better performance, you also get less battery, so who wants to know that? We want to know the battery life cost of that added performance when we're not using it.

I wish we could have been more organized to share results of standardized tests with each other, to answer those questions. Some of you helped in that sense but really too few overall, and all we have is anecdotical evidence. I guess I'm too picky and I should just put aside my scientific mind for this one : P

I currently have a maxed-out 14inch and tomorrow I'll receive a M1 Pro 10c/16c/32GB, cause when I ordered I had no idea it could be about the RAM and assumed battery life would be more impacted by GPU cores.

So I'll be able to make some good standardized tests like discussed in previous pages of this thread and draw some good conclusions. Tests will include:
- How much % of battery is drained per hour, while computer fully idle (a.k.a reddit thread test)
- How much % of battery is drained per hour while computer displaying a 1080 video on safari and nothing else

Unfortunately I guess I'll never have the answer regarding if differences will be due to RAM or something else... So I'll just pick my winner from here : )

Agreed. There hasn’t been any ample evidence so far, especially in regards to the 16inch variant. I’ve heard it’s very minimal if anything. I was hoping a YouTuber would test the 16inch M1 max 32gb vs 64gb, but no one has. This is probably due to the 64gb component being backed up in apple shipments. I have a 16inch 64gb coming, but if it doesn’t ship out, I’m keeping my 32gb max. Battery life is a concern for me, especially since you can’t fast charge the 16inch with a power bank.

 

[breath.by.breath]

Dear all

I've got finally in front of me:
- MacBook Pro 14" PRO 8C, 14 GPU, 32 GB RAM, 8 TB SSD
- MacBook Pro 14" MAX 10C, 32 GPU, 64 GB RAM, 8 TB SSD

I've made sure the same files and software are on both computers, it's indexing since about two days, so I guess that should be soon over.

Anyone interested that I do some battery comparison testing? Personally I am interested more in the low-use case.
If yes, how shall I go about it without necessarily installing external software? What (if anything) would you be interested in?

Cheers -

 

[Serengeti1]

More data..

No real difference in battery between 24gpu and 32gpu shown here which is interesting. Not sure how much ram each model had in the test though.

 

[DoorsFreak87]

This YouTuber got 6 hrs using premiere, chrome, and Spotify with max brightness on her 64gb config. Not sure how that’s possible, since tech chap got less than 3 hrs. However, tech chap used blender which is not optimized for metal and therefore has to rely on cpu power too.

 

[Charlesje]

Blind spot indeed is the 32gb vs 64gb m1 Max comparison. At low everyday use battery life will certainly be strongly dependent on dram power draw. We now that the gpu is efficiently throttled down. But the new MacBooks having unseen memory + cache bus/bandwith infrastructure could surprise us regarding power draw differences relating to memory size. As there were early signs of bad battery draw on the 64gb m1 Max I was suspecting that memory could be a factor. But the notebookcheck reviews don’t mention big battery draw differences on their different tested models, so I don’t think there to be any. But It remains a blind spot to a certain extent. Still hoping people could repeat the idle and cinebench tests while checking (dram) power draw.

 

[KirkDayne]

Hi all!
Thanks for your comments! I'm glad you found my tests useful.

@Serengeti1, @Aka757 Pro32 vs Pro16 would be interesting for sure. With my tests we can't know for sure if difference comes from ram capacity. Like @Charlesje and @Fomalhaut, I just assume that because of the powermetrics shared earlier in this thread that pointed in that direction. Also, too bad linustech didn't include the RAM capacity of their computers... But they find 2 hours difference which roughly matches what I found as well.

@ChpStcks Yes low power mode might help, but then... battery life would be EVEN BETTER with M1 Pro on low power mode as well That's why I'm not too interested in testing it (battery testing takes time).

@hmorneau I have done what you said and installed "Coconut battery" and, tbh Pro32 battery was at lower capacity (95% vs 99%) so it cannot explain my results.

@breath.by.breath have you seen my test on page 14?

@Charlesje I did the cinebench test that you asked for. Using Cinebench r23 (download here), on multi-core CPU test. I ran the test on both computers under exact same conditions and screen brightness.
I found different interesting results!

First, I compared powermetrics, since that was what you were asking for. This time I didn't do an excel sheet for it, but just stared at the screen and I came up with:

Pro32
DRAM Power: 800-900 mW
CPU Power: 27000-28000 mW
GPU Power: around 8 mW
Package Power: around 32000 mW

Max64
DRAM Power: 1600-1800 mW
CPU Power: 26000-27000 mW
GPU Power: around 8 mW
Package Power: around 32000 mW

What do I see from it:
1. DRAM is always about 1.8 to 2 times more on the Max64. Which is also the case at idle btw.
2. CPU is always about 1k to 2k LOWER (wtf?) on the Max64 (read on to see my explanation)
3. Other parameters were roughly the same, just weird GPU Power was not 0 tho.

HOWEVER, I noticed Pro32 seemed to drain MORE battery. About 1%, or 2% more than the Max64. This made me want to dig up more. Was it because this CPU weirdly consumes more? But RAM consuming 2 times less would offset it no?hmm.
But then, I also noticed that I could hear the fans on Pro32, and could barely hear them on Max64...!

That was weird enough: I installed TG Pro (download here), to monitor fan speed and temperature during the test. And here's my findings:

1. Pro32 temperature reads increase faster than Max64's.
2. Pro32 fans kick in earlier, around the 80 degree read, they started at about 2300-2500 RPM, I cannot hear that (I can start hearing the fans only from 3000 RPM).
3. Max64 didn't kick in the fans even at 80 degree, I guess they kicked in from around 85-88 degrees.
4. Towards the end of the test, Pro32 had a temperature stabilized around 95 degree with fans around 3800 RPM, and Max64 has the same temperature, although with fans at only 3200 RPM. Overall slower fan speed on Max64 for all the duration of the test.
5. I touched the body near the end of test, both on the area between screen and keyboard, and the back of the computers, under the chassis. And despite both of them showing 95 degrees on TG Pro, the Max64 was wayyy hotter!

So at first I thought, "wow, Max64 is more efficient, it consumes less battery during the CPU test! and also fans don't kick in as much so its more silent! Awesome". But hold on a second here.

Let's try to put everything together... so Pro32's CPU consumes more on powermetrics but half as much of RAM. Its CPU temperatures on TG Pro increase faster. Fans kick in quickier. They also blow stronger, which results in louder fans (typical old macbook fan noise, not huge but audible), but way cooler (not like old macbooks when their fan is loud, they are very hot as well). On the other hand, Max64 seemed to be tuned to be less loud, but it doesn't mean its less hot, even tho TG Pro reads same temperatures, there is no debate when you touch it, very hot. Its quiet like new macbooks but hot like old ones. Pro32 consumed more CPU power on powermetrics, I think because of the increased fan speed which resulted in slighly faster battery drain.

My personal conclusion
I think that Apple tweaked each machines based on their configuration in order to smooth-out the battery life differences that would otherwise occur between each model.
For example, as Max64 consumes more power due to its 64GB compared to Pro32 (even when doing nothing, see my previous test post), they tried to counter-balance that with a higher temperature threshold for kicking-in the fans. So it consumes less fan power to offset the battery hit of the 64GB RAM. Otherwise battery life differences would have been even bigger. Even though, I think it would have benefited from higher fan speed, cause that's how Apple chose to set up models less constrained by battery. That idea also matches with MaxTech's findings, that for a 14inch M1 Max 32 GPUs machine, GPUs are being throttled in power, again to ease the hit on battery.
That also explains why we hear some say that actually Max has better battery life... it's because it doesn't kick fans enough and gets hotter, basically. It's also more silent, but NOT as cool, so silence is the symptom of a temperature-tradeoff for battery life, rather than a feature here.
It's just my attempt at reverse-engineering Apple's thought process, but I think it makes sense overall.

What do you guys think?

 

[Charlesje]

Interesting indeed! There Seems to be a different cooling management between the Pro and Max. Did you try turning on / off high power mode on the Max?

 

[DoorsFreak87]

Hi all!
Thanks for your comments! I'm glad you found my tests useful.

@Serengeti1, @Aka757 Pro32 vs Pro16 would be interesting for sure. With my tests we can't know for sure if difference comes from ram capacity. Like @Charlesje and @Fomalhaut, I just assume that because of the powermetrics shared earlier in this thread that pointed in that direction. Also, too bad linustech didn't include the RAM capacity of their computers... But they find 2 hours difference which roughly matches what I found as well.

@ChpStcks Yes low power mode might help, but then... battery life would be EVEN BETTER with M1 Pro on low power mode as well That's why I'm not too interested in testing it (battery testing takes time).

@hmorneau I have done what you said and installed "Coconut battery" and, tbh Pro32 battery was at lower capacity (95% vs 99%) so it cannot explain my results.

@breath.by.breath have you seen my test on page 14?

@Charlesje I did the cinebench test that you asked for. Using Cinebench r23 (download here), on multi-core CPU test. I ran the test on both computers under exact same conditions and screen brightness.
I found different interesting results!

First, I compared powermetrics, since that was what you were asking for. This time I didn't do an excel sheet for it, but just stared at the screen and I came up with:

Pro32
DRAM Power: 800-900 mW
CPU Power: 27000-28000 mW
GPU Power: around 8 mW
Package Power: around 32000 mW

Max64
DRAM Power: 1600-1800 mW
CPU Power: 26000-27000 mW
GPU Power: around 8 mW
Package Power: around 32000 mW

What do I see from it:
1. DRAM is always about 1.8 to 2 times more on the Max64. Which is also the case at idle btw.
2. CPU is always about 1k to 2k LOWER (wtf?) on the Max64 (read on to see my explanation)
3. Other parameters were roughly the same, just weird GPU Power was not 0 tho.

HOWEVER, I noticed Pro32 seemed to drain MORE battery. About 1%, or 2% more than the Max64. This made me want to dig up more. Was it because this CPU weirdly consumes more? But RAM consuming 2 times less would offset it no?hmm.
But then, I also noticed that I could hear the fans on Pro32, and could barely hear them on Max64...!

That was weird enough: I installed TG Pro (download here), to monitor fan speed and temperature during the test. And here's my findings:

1. Pro32 temperature reads increase faster than Max64's.
2. Pro32 fans kick in earlier, around the 80 degree read, they started at about 2300-2500 RPM, I cannot hear that (I can start hearing the fans only from 3000 RPM).
3. Max64 didn't kick in the fans even at 80 degree, I guess they kicked in from around 85-88 degrees.
4. Towards the end of the test, Pro32 had a temperature stabilized around 95 degree with fans around 3800 RPM, and Max64 has the same temperature, although with fans at only 3200 RPM. Overall slower fan speed on Max64 for all the duration of the test.
5. I touched the body near the end of test, both on the area between screen and keyboard, and the back of the computers, under the chassis. And despite both of them showing 95 degrees on TG Pro, the Max64 was wayyy hotter!

So at first I thought, "wow, Max64 is more efficient, it consumes less battery during the CPU test! and also fans don't kick in as much so its more silent! Awesome". But hold on a second here.

Let's try to put everything together... so Pro32's CPU consumes more on powermetrics but half as much of RAM. Its CPU temperatures on TG Pro increase faster. Fans kick in quickier. They also blow stronger, which results in louder fans (typical old macbook fan noise, not huge but audible), but way cooler (not like old macbooks when their fan is loud, they are very hot as well). On the other hand, Max64 seemed to be tuned to be less loud, but it doesn't mean its less hot, even tho TG Pro reads same temperatures, there is no debate when you touch it, very hot. Its quiet like new macbooks but hot like old ones. Pro32 consumed more CPU power on powermetrics, I think because of the increased fan speed which resulted in slighly faster battery drain.

My personal conclusion
I think that Apple tweaked each machines based on their configuration in order to smooth-out the battery life differences that would otherwise occur between each model.
For example, as Max64 consumes more power due to its 64GB compared to Pro32 (even when doing nothing, see my previous test post), they tried to counter-balance that with a higher temperature threshold for kicking-in the fans. So it consumes less fan power to offset the battery hit of the 64GB RAM. Otherwise battery life differences would have been even bigger. Even though, I think it would have benefited from higher fan speed, cause that's how Apple chose to set up models less constrained by battery. That idea also matches with MaxTech's findings, that for a 14inch M1 Max 32 GPUs machine, GPUs are being throttled in power, again to ease the hit on battery.
That also explains why we hear some say that actually Max has better battery life... it's because it doesn't kick fans enough and gets hotter, basically. It's also more silent, but NOT as cool, so silence is the symptom of a temperature-tradeoff for battery life, rather than a feature here.
It's just my attempt at reverse-engineering Apple's thought process, but I think it makes sense overall.

What do you guys think?

I’m curious what the results of this test would be if you substituted the 32 pro with the 32 max. The memory is much faster on the max vs the pro and of course there’s the higher core count.

 

[Sanpete]

My personal conclusion
I think that Apple tweaked each machines based on their configuration in order to smooth-out the battery life differences that would otherwise occur between each model.
For example, as Max64 consumes more power due to its 64GB compared to Pro32 (even when doing nothing, see my previous test post), they tried to counter-balance that with a higher temperature threshold for kicking-in the fans. So it consumes less fan power to offset the battery hit of the 64GB RAM.

The Max has a larger heatsink, so it may remove heat more efficiently with lower fan speeds. That might not keep the body from getting hotter, though.

 

[ChpStcks]

Dear all

I've got finally in front of me:
- MacBook Pro 14" PRO 8C, 14 GPU, 32 GB RAM, 8 TB SSD
- MacBook Pro 14" MAX 10C, 32 GPU, 64 GB RAM, 8 TB SSD

I've made sure the same files and software are on both computers, it's indexing since about two days, so I guess that should be soon over.

Anyone interested that I do some battery comparison testing? Personally I am interested more in the low-use case.
If yes, how shall I go about it without necessarily installing external software? What (if anything) would you be interested in?

Cheers -

One test that I'm really curious about and haven't see any to date is how the MAX64 battery lasts on low power mode vs PRO32 on regular battery mode... because if you can get similar or better battery life on low power I would say personally the MAX64 is justifiable, because if you ever need that extra ooof in performance you'd either plug it in or turn off low power mode and accept the additional power consumption...

my understanding for light daily tasks (browser, vdi, music, word processing etc [non GPU intensive, photo/video editing) the MAX on low power is pretty much same or similar to the PRO (on regular battery) in regards to performance.

 

[ChpStcks]

Hi all!
Thanks for your comments! I'm glad you found my tests useful.

@Serengeti1, @Aka757 Pro32 vs Pro16 would be interesting for sure. With my tests we can't know for sure if difference comes from ram capacity. Like @Charlesje and @Fomalhaut, I just assume that because of the powermetrics shared earlier in this thread that pointed in that direction. Also, too bad linustech didn't include the RAM capacity of their computers... But they find 2 hours difference which roughly matches what I found as well.

@ChpStcks Yes low power mode might help, but then... battery life would be EVEN BETTER with M1 Pro on low power mode as well That's why I'm not too interested in testing it (battery testing takes time).

@hmorneau I have done what you said and installed "Coconut battery" and, tbh Pro32 battery was at lower capacity (95% vs 99%) so it cannot explain my results.

@breath.by.breath have you seen my test on page 14?

@Charlesje I did the cinebench test that you asked for. Using Cinebench r23 (download here), on multi-core CPU test. I ran the test on both computers under exact same conditions and screen brightness.
I found different interesting results!

First, I compared powermetrics, since that was what you were asking for. This time I didn't do an excel sheet for it, but just stared at the screen and I came up with:

Pro32
DRAM Power: 800-900 mW
CPU Power: 27000-28000 mW
GPU Power: around 8 mW
Package Power: around 32000 mW

Max64
DRAM Power: 1600-1800 mW
CPU Power: 26000-27000 mW
GPU Power: around 8 mW
Package Power: around 32000 mW

What do I see from it:
1. DRAM is always about 1.8 to 2 times more on the Max64. Which is also the case at idle btw.
2. CPU is always about 1k to 2k LOWER (wtf?) on the Max64 (read on to see my explanation)
3. Other parameters were roughly the same, just weird GPU Power was not 0 tho.

HOWEVER, I noticed Pro32 seemed to drain MORE battery. About 1%, or 2% more than the Max64. This made me want to dig up more. Was it because this CPU weirdly consumes more? But RAM consuming 2 times less would offset it no?hmm.
But then, I also noticed that I could hear the fans on Pro32, and could barely hear them on Max64...!

That was weird enough: I installed TG Pro (download here), to monitor fan speed and temperature during the test. And here's my findings:

1. Pro32 temperature reads increase faster than Max64's.
2. Pro32 fans kick in earlier, around the 80 degree read, they started at about 2300-2500 RPM, I cannot hear that (I can start hearing the fans only from 3000 RPM).
3. Max64 didn't kick in the fans even at 80 degree, I guess they kicked in from around 85-88 degrees.
4. Towards the end of the test, Pro32 had a temperature stabilized around 95 degree with fans around 3800 RPM, and Max64 has the same temperature, although with fans at only 3200 RPM. Overall slower fan speed on Max64 for all the duration of the test.
5. I touched the body near the end of test, both on the area between screen and keyboard, and the back of the computers, under the chassis. And despite both of them showing 95 degrees on TG Pro, the Max64 was wayyy hotter!

So at first I thought, "wow, Max64 is more efficient, it consumes less battery during the CPU test! and also fans don't kick in as much so its more silent! Awesome". But hold on a second here.

Let's try to put everything together... so Pro32's CPU consumes more on powermetrics but half as much of RAM. Its CPU temperatures on TG Pro increase faster. Fans kick in quickier. They also blow stronger, which results in louder fans (typical old macbook fan noise, not huge but audible), but way cooler (not like old macbooks when their fan is loud, they are very hot as well). On the other hand, Max64 seemed to be tuned to be less loud, but it doesn't mean its less hot, even tho TG Pro reads same temperatures, there is no debate when you touch it, very hot. Its quiet like new macbooks but hot like old ones. Pro32 consumed more CPU power on powermetrics, I think because of the increased fan speed which resulted in slighly faster battery drain.

My personal conclusion
I think that Apple tweaked each machines based on their configuration in order to smooth-out the battery life differences that would otherwise occur between each model.
For example, as Max64 consumes more power due to its 64GB compared to Pro32 (even when doing nothing, see my previous test post), they tried to counter-balance that with a higher temperature threshold for kicking-in the fans. So it consumes less fan power to offset the battery hit of the 64GB RAM. Otherwise battery life differences would have been even bigger. Even though, I think it would have benefited from higher fan speed, cause that's how Apple chose to set up models less constrained by battery. That idea also matches with MaxTech's findings, that for a 14inch M1 Max 32 GPUs machine, GPUs are being throttled in power, again to ease the hit on battery.
That also explains why we hear some say that actually Max has better battery life... it's because it doesn't kick fans enough and gets hotter, basically. It's also more silent, but NOT as cool, so silence is the symptom of a temperature-tradeoff for battery life, rather than a feature here.
It's just my attempt at reverse-engineering Apple's thought process, but I think it makes sense overall.

What do you guys think?

yeah I understand, I guess my reasoning was if the max can have similar battery life to the pro (low power vs normal) then that's good enough for me.. but if you're losing 2-3 hours on a usual 10 hour normal day the pro has for the max that 20-30% reduction wouldn't be worth it

 

[ChpStcks]

This YouTuber got 6 hrs using premiere, chrome, and Spotify with max brightness on her 64gb config. Not sure how that’s possible, since tech chap got less than 3 hrs. However, tech chap used blender which is not optimized for metal and therefore has to rely on cpu power too.

I thought the 6hr was what is expected from the 16" at full throttle?