Battery life of the MacBook Pro M1 while rendering / processing video

[samotivad]

I thought you might all find this interesting. I did a test today where I charged my MacBook Pro M1 to 100%, then after disconnecting it I converted a bunch of videos using Handbreak (quite an intensive task). I left Handbreak converting videos until the battery was almost empty. It lasted more or less 2 hours.

Usually I get at least 16 hours out of the Macbook. That's with general usage like browsing the web, writing but also using Final Cut Pro (editing, not rendering). Processing video through Handbreak runs the processor pretty hard (reaches 95 degrees celcius by a minute or so in). So while it's quite alarming to go from 16 hours to 2 hours battery life, I'm still pretty impressed it can last that long while doing something so intensive. In those 2 hours it converted hours of video (I don't know exactly how much but it was a lot).

But yeah, 2 hours while processing videos with Handbreak. I suspect that rendering from a 3D program like Blender, Cinema 4D or Maya might eat the battery life more but I have no way to check myself.

I haven't seen so many reviewers on the internet doing tests with the battery life of these Macs while putting it under strain, I only saw temperature tests. If any of you have done similar tests it would be interesting to know your experiences below.

 

[MK500]

This is a very interesting point that people need to be aware of. The M1 is an incredibly fast processor that also happens to be able to run incredibly slow. This means the range of power consumption is MASSIVE. I can easily have tons of apps running for 12 hours on my Air doing real work without problems. BUT, I can also run down the entire battery in about 3 hours by playing a graphic intensive game that also requires Intel translation.

This is a GOOD thing. Apple gave the M1 the overhead to be able to give you incredibly performance when you want it. But you must be more aware of actual power consumption and manage your battery accordingly. I like using coconutBattery (in free mode) to show my current watts in use. So I can just glance at that in my menu bar and know if I am burning through my battery faster than I want to. And I am becoming more aware of how efficient various apps are.

The M1 is really a paradigm shift. With most of the Intel chips I have been using for years I would get what I would call moderate performance all the time, with moderate power usage. So battery life didn't vary that much.

"With great power comes great responsibility."

 

[ed.]

By sheer chance today I've made a test along the same lines, not as accurate, but with similar results. I've just got an M1/16GB MacBook Pro. I do motion graphics and use AE as my primary program, so I grabbed a project I'm working on and launched the render from AE to Adobe Media Encoder (basically it renders and compresses to h264 in a single step). Mind you, neither AE nor AME are M1 optimised*.

So, my control system (2017 27" iMac, 32GB, core i7 4,2Ghz), took 14'14" to complete the render.

The MacBook took 12'13", and the battery dropped from 81% to 74%, so 7 percentage points. At the same rate a fully charged battery would have lasted 2 hours and 53 minutes - which is in line with OP's findings, as AE/AME doesn't seem to push al the cores to the max (it's a long standing AE complaint, there doesn't seem to be a good way of using all the processor cores during render. Available RAM per core might be a limiting issue, but both the M1 and the iMac seemed not to exceed 400% of CPU usage - the iMac has hyper threading so it has 8 logical cores).
The fans never revved up (as opposed to the i7 iMac that basically tries to take off at the first whiff of any sort of effort, which seems to have become more frequent with Big Sur), and the chassis only got a bit hot towards the end, but only in the stripe between the screen and the Touch Bar.
I think the gain in render time is impressive by itself because of the difference in clock speeds (the M1 runs slower at 3.2 Ghz).
I still have to test Handbrake compressions, but having a portable machine that renders faster than my desktop is a good feeling. Can't wait to see what the next iMacs can do, even though for AE specifically it seems like we need to pray Adobe manages to innovate an old code base that seems to be struggling to keep up with hardware advances.
I also didn't try the same render with a power connection, I'm not sure whether this could have made any difference.

As for actual AE usage, for what it's worth, in my very brief testing with a non M1 optimised AE, the interface seemed to be snappier than my 5k iMac.
This totally anecdotal and I have to look further into this, but the project I used as a test has a section with 3d camera with depth of field. It's the hardest to render and both machine slowed down significantly during the render progress of that section. While one is working in After Effects, there are ways to make the WYSIWYG preview of the project quicker: you can lower the resolution, use wirframes and so on. One option is 'adaptive resolution', which basically tries lo lower the resolution of the previewed frame dynamically as you scrub the timeline, to give you an updated viewport as quickly as possible, and then refining the frame at a higher resolution when one stops on a specific frame.
While I was setting up the project for rendering, I had the impression that the M1 MacBook struggled a little bit with this setting, and then I turned it off completely, just using the full resolution preview. To my surprise, the scrubbing suddenly was much smoother than the 'adaptive' one, which is supposed to be quicker. Maybe this is related to the new architecture, I have no idea really - it might just have been a random occurrence.
But if the M1 architecture with the new unified memory can bring this level of smoothness at high quality, this can really be a game changer for the (admittedly) very niche AE user base.



* although I suspect that since Rosetta is not emulating but translating, the actual gains whenever they will be made native will not be very impressive.

 

[xraydoc]

The fans never revved up (as opposed to the i7 iMac that basically tries to take off at the first whiff of any sort of effort, which seems to have become more frequent with Big Sur), and the chassis only got a bit hot towards the end, but only in the stripe between the screen and the Touch Bar.

You should hear my PC's air-cooled i7 9700K. Pretty damn loud at full-tilt, though it does move a lot of air across a simply massive heatsink and the processor never drops below 3.8Ghz even with all 8 cores cranking at 100% for extended periods. Dell actually built a working cooling system (albeit a noisy one).

The amazing thing is my little fanless MacBook Air M1 has slightly higher CPU benchmarks (though can't beat the graphics scores).

 

[ed.]

You should hear my PC's air-cooled i7 9700K

I realised too late that the i7 vs i5 would mean a lot more fan noise in exchange for a moderate speed gain (maybe around 10% or so?). It didn't really bother me at first, as it seemed to kick in only when under stress, but for the past year or so I've noticed the fans revving up more frequently even on seemingly non intensive tasks, so much that I started to actively desire again 'a silent computer' for the health of my workplace environment. Apple silicon Macs can't come soon enough.

 

[samotivad]

Great to hear of other people's experiences with the battery while running the computer on intensive tasks.

I just did another round in Handbreak, converting video on a constant basis, running the processor hard. This time it took 1h45m to run the MacBook Pro M1 down from 100% to 10%. I guess that 10% would have taken me to roughly two hours again.

Just as a reminder to anyone who might stumble across this, remember I am doing something untypical with the computer. You're not really meant to keep a small laptop like this converting video footage for hours and hours. Unless your job requires it or you're editing home movies etc then you probably don't need to do what I'm doing either. I'm doing this simply for testing purposes. 2 hours is pretty impressive for what I've been making it do.

 

[MBAir2010]

Every GPU and other abbreviated PU benchmark publishers on PC /windows sites always perform handbrake test in their studies. the new M1 MacBooks are the same specs and battery usage as the other hated manufactures here.
ahh handbrake!

 

[Slartibart]

Every GPU and other abbreviated PU benchmark publishers on PC /windows sites always perform handbrake test in their studies. the new M1 MacBooks are the same specs and battery usage as the other hated manufactures here.
ahh handbrake!

But they do this mostly in relation to video encoding times, not to run down the battery. Battery times are usually measured with video playback or some scripted mix of different apps mimicking some kind of “average user” (which none of us have ever met or know ??).

 

[ed.]

some scripted mix of different apps mimicking some kind of “average user” (which none of us have ever met or know

Yeah I think most people underestimate how much of the daily usage of a real person is just "standing around". Even typing this message I'm pausing every once in a while to think about what I'm writing.
If I analyse my actual work (motion graphics), a lot of time in front of After Effects is actually me trying to think about how to solve a problem. If the machine can detect that and basically shut down all inessential power consuming elements (and turn them on instantly when they're needed), I can see where "real life" long battery times come from.

 

[Toutou]

If the machine can detect that and basically shut down all inessential power consuming elements (and turn them on instantly when they're needed), I can see where "real life" long battery times come from.

That's exactly where the long battery life of today's laptops comes from. Many laptops are able to burn through their battery in two hours if pushed hard, just like 15 years ago. The real magic is in the various tricks employed when nothing is happening.

 

[MK500]

The thing is, you can do a lot of actual work without encoding video or doing cryptographically intensive chores. You can be writing code, researching, accessing shipping information, analyzing stocks, creating complex spreadsheets, etc. All of this can be done in my experience using less than 3 watts average on an M1 with very fast performance. This will get you the kind of insane battery life numbers that M1 owners know and love (10, 12, 15 hours).

This is a big deal, and it has to do with the big/little architecture of the M1 among other things. Let's not start saying "M1 MacBooks are the same specs and battery usage as the other hated manufactures" when this is clearly not true. They CAN use 15 watts when asked to, but they also CAN run quite fast optimized chores at < 3 watts; and that is the impressive part.

Apple kindly lets us turn the volume up to 11 when needed, but that doesn't mean we won't spend the vast majority of our time listening at 3.

 

[MK500]

An example of optimization/hardware acceleration:

I stream a lot of steam games from my gaming PC to my M1 Air. Steam had serious optimization issues for the first few months of M1 life. I was pulling 15 watts streaming my games. The Air got hot. I would get maybe 3 hours on battery.

In the last few weeks they pushed an update and made a small step toward optimization. I believe they implemented rudimentary M1 compatible GPU acceleration for streaming. My usage went from 15 watts to 6-8 watts.

The performance of the stream is identical. Quality is identical. But more than 2x the battery life.

Adobe can do this as well, and if they have enough customers on M1 they may. The handbrake team can do so as well, although their optimization may increase frames at the same wattage vs. decreasing wattage. At some point it may be nice to have some sort of slider control to be able to choose to save some battery. I mean; if you can encode at the same speed as a desktop i7 from a year ago maybe that is sometimes fast enough

 

[samotivad]

Yeah no one is taking away how great the M1 laptop's battery can be, it's not really a debate, the battery life is exceptional. This thread is to demonstrate what happens when the laptop is rendering video, 3D scenes and motion graphics. Application developers can improve their software but rendering is always a power hungry process, which is why these tests are useful to do.

There will be people like film, animation and graphics students who should know what will happen if they decide to render using only battery power on an M1. Professionals may know it's not realistic to expect the battery to last the same amount of time but others might not be aware.

 

[ed.]

There will be people like film, animation and graphics students who should know what will happen if they decide to render using only battery power on an M1

I'm one of those: I usually work on an iMac but when I'm working on site sometimes I need to render stuff and power outlet might be scarce (some adventurous settings on trade shows or movie sets), so it's good to know that I can render (marginally) faster than the iMac, because I usually have the reference of the desktop render times, so I can make decisions based on that.

 

[Parikshit Panwar]

I have a m1 air macbook and it lastes for around 3 hrs while doing basic 4 k video editing, am I getting a good battery life?

 

[ed.]

Bump for relevant news about After Effects users: I recently tested AE beta for apple silicon, and besides being super snappy in usage, the render times are cut almost in half compared to my 2017 i7 iMac 5k. This might also apply to other, newer machines, though: Barefeats had a test comparing the Rosetta version of AE against the beta rendering the same project on the same machine (an M1 max), and the test took 171 seconds on the rosetta version and 96 on the native one.
My own project took 8:43 on the intel mac and 4:55 on the M1 running the beta, which is proportionally exactly the same gain as the Barefeats test. Curious to see AE comparisons to pc-based workstations now.