Yeah, high power mode since M1 has been a case of "I want every ounce of performance and I'm willing to have it make noise.
Low power mode is not much slower and much quieter.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M4 Max EATS battery
- Thread starter johnalan
- Start date
- Sort by reaction score
Yeah, high power mode since M1 has been a case of "I want every ounce of performance and I'm willing to have it make noise.
Low power mode is not much slower and much quieter.
There has been a follow-up video:
Last edited:
Along these lines:
Confirmed last night: I can run Witcher 3 on the M4 Max 14" (16/40) on low power mode at silky smooth frame rate at 1080p pretty much max settings with ZERO fan noise on low power mode.
Because, as I mentioned before in thread:
Heat goes up exponentially with clock speed. This isn't Mac specific, it is across the board for silicon based CPU/GPUs.
Adding more GPU cores is not an exponential increase in heat (its linear), ergo, running 40 GPU cores at 80-90% of the speed (low power mode) gets you 2x 80-90% of the GPU performance of a 20 GPU core Pro running flat out, with much less heat.
Or in layman's terms: doubling the heat output via 2x the cores (on a 40 core GPU) generates less heat in low power mode than pushing 20 cores 1.2x harder than 40.
Last edited:
More questions than answers.
- What was the defect, bad silicon lottery, inadequate thermal paste, uneven heat sink mounting, etc.?
- It's easy to test for so why didn't QC catch it before leaving the factory?
- How to determine if your unit is affected by this defect?
Per the comment in the updated video, is it normal or "defect"?
"More power more heat and noise. My 16" M4 Max gets plenty loud under full load or gaming. Plus a little coil whine is unfortunately present but not super annoying."
That's Apple's problem to determine. The machine was defective, it was RMA'd.
If you get a bad one, or are unsatisfied, RMA it.
As to why it wasn't caught by QC - I think you overestimate the level of QC on mass market products.
The machine would have been inspected, verified that it booted into an OS, charged, etc. Basic tests.
No manufacturer is going to be running benchmarks/stress tests on every single machine coming off the line. It's just not feasible. You select a percentage of machines for detailed testing, the majority get basic testing.
You're not going to catch EVERYTHING with a mass produced product during QC, this is why warranty exists.
Agreed, still it’s fine to run those tests. No different to actual workload in Blender render etc.
Just confirming this thing is hungry. M4 Max 16” 16/40. Export 1080p video it flies. I thought something had broken when it exported 1hr 20m of video 5 minutes. I was floored.
But that’s the hardware encoders. It sips battery. I also ran it through some VR workflows. Eos VR Utility and Topaz AI. It will hit 100w of power draw and ramp the fans upto 50% on auto and things throttle a little.
On high power mode fans go to 100% and it does keep it from throttling so the thermal system works. The fans are noisy though. This thing in a Mac Studio would fly.
I’m wondering if 100w is the maximum the system can draw or if it’s plugged in on the 140 MagSafe it could draw upto 140w??
I should also stress that unless you’re pushing it with a very demanding task over a prolonged period (I’m talking an hour) the fans barely came on. I only know the spin because I can’t hear them. If it’s not under load this is quieter than my M1 Max Studio which always made a very low 12db noise whether it was on the desktop or maxed out
So as noisy as the M4 Max can get at least it’s totally silent most of the time. 😅
But that’s the hardware encoders. It sips battery. I also ran it through some VR workflows. Eos VR Utility and Topaz AI. It will hit 100w of power draw and ramp the fans upto 50% on auto and things throttle a little.
On high power mode fans go to 100% and it does keep it from throttling so the thermal system works. The fans are noisy though. This thing in a Mac Studio would fly.
I’m wondering if 100w is the maximum the system can draw or if it’s plugged in on the 140 MagSafe it could draw upto 140w??
I should also stress that unless you’re pushing it with a very demanding task over a prolonged period (I’m talking an hour) the fans barely came on. I only know the spin because I can’t hear them. If it’s not under load this is quieter than my M1 Max Studio which always made a very low 12db noise whether it was on the desktop or maxed out
So as noisy as the M4 Max can get at least it’s totally silent most of the time. 😅
Last year, someone posted the following: https://forums.macrumors.com/threads/m3-max-low-power-mode-is-insane.2412621/
I'm curious how much of a difference the M4 Max in low power mode makes.
Low power mode on my max is a lot more severe than it was on my M1 Pro.
It becomes silent under load. Haven’t tested performance impact or battery impact though yet.
I did specifically not worry about heat and noise and decided to go Max though for this reason. Low power mode drops clocks a bit but a lower clocked max is still going to be faster than a pro most likely clocked high at normal or high power mode.
Heat and energy consumption goes up exponentially vs Clock speed but only linearly vs core count, and idle cores are basically switched off.
Last edited:
Exactly like my most 'intense' would be rendering 1:1 or standard previews in Lightroom for multiple images (500-1000). Fans would come on for a minute or two, but then they would shut very quickly once done.
I've been having intermittent problems with the battery draining fast on my new M4 Pro. I finally bothered to look deeper and see what apps were chewing up energy and it turns out I had a terminal process watching filesystem changes that were spiking energy usage to the levels that I'd normally see if I was doing furiously fast editing of RAW images.
This could have been a problem all along and I just never noticed it, but wow. When you actually audit your usage sometimes the thing that you find is the last thing you expected.
On the plus side, it's been cold lately and I have cold hands. Now I know how to activate my built in hand warmers. I had been missing this feature in the winter ever since I stopped using Intel Macs.
My M4 Pro chews through battery when I'm pushing it.
But it also chews through battery when I'm just developing apps. It likes to max out at its top speed which is nice to save 1 or 2 seconds of compile time, but it also eats battery life. It's a compromise I'm not happy about.
But for light usage, the battery life is absolutely fantastic.
But it also chews through battery when I'm just developing apps. It likes to max out at its top speed which is nice to save 1 or 2 seconds of compile time, but it also eats battery life. It's a compromise I'm not happy about.
But for light usage, the battery life is absolutely fantastic.
Yeah, but when you compare that benchmark to the biggest Intel tower, your laptop still beats it. You have an extremely fast computer, laptop or otherwise.
Processing power per volume is perhaps the most interesting metric I've heard of yet.
It's not a lemon, it's what a benchmark is supposed to do, hence why they're not indicative at all of any real world performance.
I do heavy stuff on my M3 and it lasts all day easily (and longer). Most of my work is 3D gamdev at the moment. If I were to sit there and try and render a complex animation or do simulations in Houdini my battery would drop like crazy because it's a very intensive task.
I find it very helpful to run coconutBattery in the menu bar and set it so that it shows how many watts I'm using, along with the battery life estimate. It helps as a bit of a red flag if battery drain starts creeping up without an obvious reason.
The most my little MacBook Air M1 will pull is like 10-15 watts (on a Zoom call with brightness at 100%), but when doing normal stuff with the brightness around 50%, it settles down to like 2-4 watts, which gives me an estimated battery life well above 10 hours.
Obviously, that would all get adjusted WAY upward for the kind of stuff we're talking about in this thread with an M4 Max. I'd be curious, though, to hear what the wattage is for that machine doing basic productivity stuff, as a baseline.
The most my little MacBook Air M1 will pull is like 10-15 watts (on a Zoom call with brightness at 100%), but when doing normal stuff with the brightness around 50%, it settles down to like 2-4 watts, which gives me an estimated battery life well above 10 hours.
Obviously, that would all get adjusted WAY upward for the kind of stuff we're talking about in this thread with an M4 Max. I'd be curious, though, to hear what the wattage is for that machine doing basic productivity stuff, as a baseline.
I don’t quite follow.Processing power per volume is perhaps the most interesting metric I've heard of yet.
Maybe you don’t understand that large tower PC’s are traditionally faster than laptops, because they can use massive CPU’s which can run at full speed due to massive cooling fans and power supplies. Yet here we have a laptop which beats just about any tower PC you can build. And when it’s doing that… yes, it takes up a lot of battery, although it’s still far lower power consumption than the Intel/AMD towers.