Battery life on M1X PRO vs M1X MAX?

[KirkDayne]

so Linus is getting 22-24hrs at idle for the 14"?!? (brightness is at 2/3, but still... haven't see any reports anywhere that come close to those times...)

Yeah I saw this test. Incredible.
Tho it was not really "idle" but playing a countdown youtube video.
Still it was a good test I think since all computers were in the same conditions.
But only M1 Pro versions were tested there and I wish they also included M1 MAX...!

 

[KirkDayne]

Doing the suggested discharge test with nothing running gave me some peace of mind actually.

Model Name: MacBook Pro
Model Identifier: MacBookPro18,4
Chip: Apple M1 Max
Total Number of Cores: 10 (8 performance and 2 efficiency)
Memory: 64 GB
Storage: 2TB

100% - 7:45am
98% - 8:45am = -2%
89% - 9:45am = -9%
79% - 10:45am = -10%
73% - 11:45am = -6%
66% - 12:45pm = -7%
57% - 1:45pm = -9%
48% - 2:45pm = -9%
40% - 3:45pm = -8%

And that's when I ran out of time. I'm going to do another test to see how much lower brightness helps.

Thank you for posting your results! Interesting.

Few questions:
- How many GPU cores?
- 14inch or 16inch? (I guess 14)
- Have you made sure that brightness were 100% and not auto-adjusting?
- How many days have you used it before doing the test? (to know if indexing might impact)

I wonder why your results are really different from those of the reddit thread OP despite it's the same machine...?

 

[Aka757]

I found an interesting video with a battery test for several MacBooks, including a 16" M1 Pro, 16" M1 Max, and 14" M1 Pro. The test itself isn't very long and it's focused on software development, but essentially the 16" M1 Pro and 16" M1 Max are neck and neck, with the 14" M1 Pro slightly less behind (73%, 74%, 70%). A M1 MBA was also tested and it was at 62% at the end of the test.

 

[mentalman1369]

Yes these are the right numbers. On a Max m1 with 32c gpu and 32gb ram I get

150 mW dram
20 mW cpu
1 mW gpu
about 500 mW package power

This idling test is meant to give you an idea of battery draw at light use (where cpu and gpu are not taxed). You can see the new m1 Max machines throttle back (and cut the power back) the cpu and gpu efficiently, but this is less the case for the memory that can’t be throttled back (nor memory bus), so the lower the load on the system, the higher the relative impact of the fast ddr5 ram (and large bandw of the bus) of the m1 Max. I would be most curious to see results from the m1 Pro with 32gb and the m1 Max with 64 gb to be able to compare them directly. This will at least give part of the answer of wat the subject of this thread was about.

any chance you could run powermetrics again but on a 1080p youtube video playing in safari with all other apps closed?
I've got a 32GB M1 Pro 14" and currently tempted to change up to 32GB M1 Max 24c if power consumption is negligibly different between the two.

Package power is around 700-800 mW on my M1 Pro, DRAM ~200-300 mW and CPU ~150 mW, GPU around 10 mW

 

[Fomalhaut]

I found an interesting video with a battery test for several MacBooks, including a 16" M1 Pro, 16" M1 Max, and 14" M1 Pro. The test itself isn't very long and it's focused on software development, but essentially the 16" M1 Pro and 16" M1 Max are neck and neck, with the 14" M1 Pro slightly less behind (73%, 74%, 70%). A M1 MBA was also tested and it was at 62% at the end of the test.

That was quite interesting, and at least shows the M1 Max having a tiny advantage, albeit running the test more slowly than the M1 Pro for some reason.

Unless some other downside is revealed in the next month, I think I'll be buying a 14" M1 Max.

 

[Fomalhaut]

any chance you could run powermetrics again but on a 1080p youtube video playing in safari with all other apps closed?
I've got a 32GB M1 Pro 14" and currently tempted to change up to 32GB M1 Max 24c if power consumption is negligibly different between the two.

Package power is around 700-800 mW on my M1 Pro, DRAM ~200-300 mW and CPU ~150 mW, GPU around 10 mW

I would love to see the results for this too. Is "powermetrics" known to be accurate BTW?

 

[crimlarks]

I wasn't sure about whether to order the M1Max or Pro, but after reading thru all these posts, I decided to go with the Max - ordered a 14" M1 Max with 10-core CPU, 24-core GPU, 32 GB memory and 2 TB drive. As James948 said, I'll rarely be in a place where I cannot plug it in for 30 minutes. Should arrive between 13-20 December which is perfect timing for me.

 

[kgreen6901]

Thank you for posting your results! Interesting.

Few questions:
- How many GPU cores?
- 14inch or 16inch? (I guess 14)
- Have you made sure that brightness were 100% and not auto-adjusting?
- How many days have you used it before doing the test? (to know if indexing might impact)

I wonder why your results are really different from those of the reddit thread OP despite it's the same machine...?

- 32 GPU cores.
- 14"
- set brightness to 100% and turned off auto-adjusting
- I got it on November 5th, haven't done much but move my files over, general use since then

I also think it's suspicious that the discharge wasn't the average 10% seen in that reddit thread. Tonight I'm going to watch a movie on lower brightness with no other apps running and see how it does. Tomorrow I will try a zoom sessions since I have class. My portable battery arrived today too. Hoping that this all just pans out in the end.

 

[Charlesje]

any chance you could run powermetrics again but on a 1080p youtube video playing in safari with all other apps closed?
I've got a 32GB M1 Pro 14" and currently tempted to change up to 32GB M1 Max 24c if power consumption is negligibly different between the two.

Package power is around 700-800 mW on my M1 Pro, DRAM ~200-300 mW and CPU ~150 mW, GPU around 10 mW

Yes my results with a 1080 video covering the largest part of the screen gives 400 mw dram, 60mw cpu and 3 mw gpu... Could you do the proposed idling test in return (so after rebooting, doing nothing, wait until the values stabilize a bit)?

These values will be more comparable as the YouTube movie will make the values jump quite a bit. The idea of the idle test is to give you an idea of the base situation where the machine will throttle back to when they can. The M1, Max or pro machines excell in this kind of behaviour and have a great battery life, partly because of this throttling behaviour. The difference between the Pro (your machine) and Max with the same amount of memory will be interesting to see.

 

[mentalman1369]

Yes my results with a 1080 video covering the largest part of the screen gives 400 mw dram, 60mw cpu and 3 mw gpu... Could you do the proposed idling test in return (so after rebooting, doing nothing, wait until the values stabilize a bit)?

These values will be more comparable as the YouTube movie will make the values jump quite a bit. The idea of the idle test is to give you an idea of the base situation where the machine will throttle back to when they can. The M1, Max or pro machines excell in this kind of behaviour and have a great battery life, partly because of this throttling behaviour. The difference between the Pro (your machine) and Max with the same amount of memory will be interesting to see.

What was the reported package power for you?

When I do the idle test I get around 100-150 mW DRAM, 20-30 mW CPU 1-5 GPU and total package power 200-250 mW. It seems the DRAM value listed is separate to to total package power, as it doesn't seem to add together with CPU + GPU to total the package power. Some have suggested the DRAM that powermetrics lists is the DRAM controller; which means maybe included but not listed for package power is the actual RAM power draw. Also Anandtech's deep dive on the M1 Max seems to suggest their total package power was around 200 mW too when fully idle, whereas yours was reporting 500 mW which seems quite high.

 

[Charlesje]

What was the reported package power for you?

When I do the idle test I get around 100-150 mW DRAM, 20-30 mW CPU 1-5 GPU and total package power 200-250 mW. It seems the DRAM value listed is separate to to total package power, as it doesn't seem to add together with CPU + GPU to total the package power. Some have suggested the DRAM that powermetrics lists is the DRAM controller; which means maybe included but not listed for package power is the actual RAM power draw. Also Anandtech's deep dive on the M1 Max seems to suggest their total package power was around 200 mW too when fully idle, whereas yours was reporting 500 mW which seems quite high.

Thanks for testing this out! Package power, I estimate, is around 800 mW as well on my machine with the youtube vid playing, but it varies extremely.

I don’t understand how anandtech this low idle power figure, as it is consistently fluctuating around 500mw on my machine (sometimes it reaches 300 mW for short intervals).
I guess we need to calculate in the screen (brightness and background image) for the package power.

But I would say the dram power draw stands for the ram itself. It seems your dram power value is a bit lower but similar as mine, and alltogether lower then the 227mw value posted by another member of the forum with a 64gb memory Max. Still, a lot remains unclear. Best would be an indepth comparison of different memory sizes of the m1 Max, next to the comparison between the Pro and Max Models. Right now reviews are convoluting both.

 

[breath.by.breath]

Yeah I saw this test. Incredible.
Tho it was not really "idle" but playing a countdown youtube video.
Still it was a good test I think since all computers were in the same conditions.
But only M1 Pro versions were tested there and I wish they also included M1 MAX...!

Absolutely, what we still need is the PRO-MAX comparison.

 

[kvlq]

Which one should i trust? Battery settings (sum all those blue bars) or activity monitor?
In battery settings from system preferences i always get about 8 and 8.5hrs.

PS: 16" M1 Pro / 8 bars brightness.

 

[mentalman1369]

View attachment 1912740
Which one should i trust? Battery settings (sum all those blue bars) or activity monitor?
In battery settings from system preferences i always get about 8 and 8.5hrs.

PS: 16" M1 Pro / 8 bars brightness.

I believe time on battery in activity monitor includes screen off time, e.g. time spent sleeping. Battery settings shows only screen on time, but is hard to add to a total sum.

 

[kvlq]

I believe time on battery in activity monitor includes screen off time, e.g. time spent sleeping. Battery settings shows only screen on time, but is hard to add to a total sum.

I’m asking because i dont know which one is taken into consideration from poeple who are posting here. Because i always get max 8hrs of screen usage and people here are getting 10-12+ hrs. Also Apple claimed more than 17hrs. I think everybody is reffering to that Activity Monitor time, right?

 

[applesed]

any chance you could run powermetrics again but on a 1080p youtube video playing in safari with all other apps closed?
I've got a 32GB M1 Pro 14" and currently tempted to change up to 32GB M1 Max 24c if power consumption is negligibly different between the two.

Package power is around 700-800 mW on my M1 Pro, DRAM ~200-300 mW and CPU ~150 mW, GPU around 10 mW

Any reason you’re targeting the 24c and not 32c?

 

[mentalman1369]

Any reason you’re targeting the 24c and not 32c?

24c much cheaper an upgrade than 32c, esp for what you get. It's $300 AUD to go from 16c Pro to 24c Max, for 1.5x the cores. For the next jump from 24c to 32c, it is $600 (edit - I mean $600 over the 16c, but is $300 more than 24c), which is only 1.33x more cores. Poorer value for money.

 

[mentalman1369]

Thanks for testing this out! Package power, I estimate, is around 800 mW as well on my machine with the youtube vid playing, but it varies extremely.

I don’t understand how anandtech this low idle power figure, as it is consistently fluctuating around 500mw on my machine (sometimes it reaches 300 mW for short intervals).
I guess we need to calculate in the screen (brightness and background image) for the package power.

But I would say the dram power draw stands for the ram itself. It seems your dram power value is a bit lower but similar as mine, and alltogether lower then the 227mw value posted by another member of the forum with a 64gb memory Max. Still, a lot remains unclear. Best would be an indepth comparison of different memory sizes of the m1 Max, next to the comparison between the Pro and Max Models. Right now reviews are convoluting both.

800 mW package power is pretty power-efficient then for the Max, since it is basically the same on my M1 Pro 32GB but with a smidge more DRAM power use. In any case, thanks to the extended holiday returns policy, I have just gone ahead and ordered the 24c 32GB Max and I will post some direct comparisons between it and my M1 Pro 32GB with both completely idle, youtube video power usage, general web browsing, and also readouts for total system from coconutBattery.

Current estimated delivery is 7-14th of Dec...

As for your readings of 500 mW when idling vs Anandtech's 200mW (which was on 64GB RAM for M1 Max)... I do wonder how they got that. Since another forum member posted here with their package reading around 470 mW at idle (for 64GB M1 Max):

Request: can some with M1 Max run this Powermetrics code to test idle + load power draw?

14" MBP, M1 Max 32C GPU, 64G RAM At 99.16% Idle right after a reboot and things have settled: DRAM Power: 227 mW CPU Power: 25 mW GPU Power: 3 mW Package Power: 470 mW Watching a YouTube video: DRAM Power: 746 mW CPU Power: 66 mW GPU Power: 35 mW Package Power: 1492 mW Anyone with a M1 Pro...

forums.macrumors.com

I suppose if you've checked Activity Monitor and closed every app and closed every user background program/process and there's no indexing etc happening, then it would be accurate. Only other thing is whether the frequency of powermetrics refreshing in Terminal is adding to the workload, perhaps needing to decrease from every 1000ms to 4000ms or something? Not sure...

 

[KirkDayne]

@Charlesje @mentalman1369

I received my 14inch M1 MAX / 32c GPU / 64GB RAM and ran the following command:
`sudo powermetrics -i 1000 -a --samplers cpu_power`

After booting, not logged in icloud, everything as default, no apps running except terminal, top right said spotlight was using significant energy so I waited for it to change to "No Apps Using Significant Energy".

Are you sure this makes sense to look at these numbers to draw comparisons? Numbers are jumping all the time. With some peak values in both directions...

I feel that if you wait long enough, you can have pretty much any readings.

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

I attached the table of data points.

Any chances you could perform the same test so we may get to something more comparable?

Also, are you sure the screen brightness doesn't impact? (I did it with 9 ticks of brightness btw).

 

[Fomalhaut]

@Charlesje @mentalman1369

I received my 14inch M1 MAX / 32c GPU / 64GB RAM and ran the following command:
`sudo powermetrics -i 1000 -a --samplers cpu_power`

After booting, not logged in icloud, everything as default, no apps running except terminal, top right said spotlight was using significant energy so I waited for it to change to "No Apps Using Significant Energy".

Are you sure this makes sense to look at these numbers to draw comparisons? Numbers are jumping all the time. With some peak values in both directions...

I feel that if you wait long enough, you can have pretty much any readings.

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

I attached the table of data points.

Any chances you could perform the same test so we may get to something more comparable?

Also, are you sure the screen brightness doesn't impact? (I did it with 9 ticks of brightness btw).

That's quite impressive - it looks like the average package power is well below 0.5W.

What is the actual power usage for the whole computer (e.g. measured using power meter when plugged in)? About 5W, giving maybe 14 hours with a 70Wh battery?

 

[Charlesje]

@Charlesje @mentalman1369

I received my 14inch M1 MAX / 32c GPU / 64GB RAM and ran the following command:
`sudo powermetrics -i 1000 -a --samplers cpu_power`

After booting, not logged in icloud, everything as default, no apps running except terminal, top right said spotlight was using significant energy so I waited for it to change to "No Apps Using Significant Energy".

Are you sure this makes sense to look at these numbers to draw comparisons? Numbers are jumping all the time. With some peak values in both directions...

I feel that if you wait long enough, you can have pretty much any readings.

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

I attached the table of data points.

Any chances you could perform the same test so we may get to something more comparable?

Also, are you sure the screen brightness doesn't impact? (I did it with 9 ticks of brightness btw).

Thanks for these results. I'm sorry but I don't have the time to do the 60 seconds, but I read out and wrote down my results for 10 (actually 11 ) seconds and averaged them. My values are jumping around but not so much.

Average numbers (for the 16'' m1 max, 32c gpu 32gb, see the printscreen below):

DRAM 140 mW
CPU 19 mW
GPU a small 1 mW
Package p. about 400 mW

I wouldn't concentrate to much on package power on idle use as I suspect it will depend also on screen size, brightness, background picture etc.

It's tentative to draw any conclusions, but the few data we have suggest:

-dram power draw is relatively big on these new MacBook Pros, which is not unexpected seen their memory structure. For reference see eg the 40 mw power draw of the Mac mini (also your results?) vs 200 mW of the m1 max when idling

-GPU Power draw at low use is (extremely) efficiently turned down, also for the 32 core GPU (thanks to the iPad/iPhone heritage of the m1...?) For reference 1mW with your m1 max vs 77 mW for the Mac mini

-there is upto 50 % extra power draw between 32 gb and 64 gb ram for these MacBook Pros at idle use, which might have consequences for battery life (in light load use). This, in my view, could help explain some preliminary reviews pointing out to the weaker battery life of the m1 max (albeit with the 64gb).

I suspect memory amount to be the most dividing factor after screen size for battery life at light (non gpu taxing) loads between the different configurations. For some people this trade off will probably be important to know. But again, this is tentative, and looks at initial conditions. We would need a real world test showing just how much battery life difference this amounts to.

 

[Sanpete]

-there is upto 50 % extra power draw between 32 gb and 64 gb ram for these MacBook Pros at idle use, which might have consequences for battery life (in light load use). This, in my view, could help explain some preliminary reviews pointing out to the weaker battery life of the m1 max (albeit with the 64gb).

A difference of .07W (.14 vs .21) would be .7W over 10 hours, so not a large difference.

 

[Charlesje]

A difference of .07W (.14 vs .21) would be .7W over 10 hours, so not a large di

No indeed not large, but it will depend on the scaling at bigger memory use.

 

[GeorgeVes]

CPU-bound task comparison from

And here's one where GPU is used :

Differences look negligible

 

[Sanpete]

No indeed not large, but it will depend on the scaling at bigger memory use.

Any significant additional power consumption should reflect additional work being done, though. Or that's what I'd expect, at least.