There was, indeed, a bug
In looking at other battery run-down scenarios, I ran smack into a problem that’s likely at fault for many of the confusing battery life issues with the laptop, at least in macOS Sierra 10.12.2. On occasion, the laptop’s discrete GPU would just get stuck on and consume power even when it wasn’t used. Others had reported this too, but you’d really have to stumble onto it.
I was able to reproduce the issue in Safari by opening Google Maps, which would cause the laptop to switch over to the GPU for the WebGL workload. Opening additional browser tabs and then closing the Google Maps tab would, on occasion, leave the GPU consuming up to 10 watts of power while doing absolutely nothing.
Even though the MacBook Pro 15 had switched back to Intel graphics, the Radeon Pro would just suck up power. How much? I found after triggering this bug that I could kill the laptop in just 206 minutes, or a scant three and a half hours of basically idling with Safari open, Wi-Fi on, and eight websites loaded.
This isn’t just a blame-discrete-graphics moment, either. I performed an idle test with Safari open to Google Maps. This switched on the Radeon Pro, which would consume about 5 watts just idling. With Wi-Fi on and the screen set to 257 nits—but essentially nothing happening—the battery stretched out to 1,135 minutes, or just shy of 19 hours.
This particular power drain isn’t completely the fault of the Radeon Pro. I suspect this to be the cause of many of the problems people are reporting.
What Apple fixed
The Sierra 10.12.3 update seems to have fixed the problem. Apple officially says the update “improves automatic graphics switching on the MacBook Pro 15,” and it does.
I could no longer get the Radeon Pro to consume phantom power. It’s as though Apple basically increased the graphics load requirement before firing up the power-hungry GPU. Google Maps and even more intensive WebGL tasks would no longer fire up the Radeon Pro at all. Most of the time you really don’t need it, as the Intel IGP is generally more than enough for web graphics.
I tested the updated OS running the same task that earlier gave me the horrible 206 minutes of battery life under Sierra 10.12.2. The battery life stretched out to 631 minutes under Sierra 10.12.3.
Interestingly, I never initiated the developer mode that Consumer Reports did when it initially held back on recommending the MacBook Pros to consumers. Consumer Reports’ tests found the battery life to be all over the map and attributed it to the developer mode it used in its browser test.
Once Apple patched the OS, Consumer Reports reversed its decision. Those fixes eventually went into the final release of Sierra 10.12.3, but it’s not clear to me what exactly happened in the situation that Consumer Reports ran into.
The smaller battery is clearly a limitation
If you think the battery life situation (at least on the MacBook Pro 15) is solved, there’s something else you should consider that can’t be fixed in software: the smaller battery.
Prior to the 2016 MacBook Pro 15, Apple shoved giant 99.5-watt-hour batteries into the MacBook Pro 15 series. With the thinner, lighter late-2016 model, Apple reduced the size of the battery by almost 25 percent to 76 watt-hours. This downsizing has a real cost. While the new MacBook Pro 15 is impressively efficient, sometimes efficiency isn’t enough.
If you think of cars, many factors contribute to the driving range between refueling, including the size of the gas tank, how aerodynamic the car is, and how efficient the engine and transmission are. On a laptop, the battery capacity, measured in watt-hours, is akin to the fuel tank. The power efficiency of the CPU, GPU, and components could be the engine and transmission efficiency. The amount of power the display uses could be the aerodynamics.
As you saw from the previous numbers, the general efficiency of the display, CPU, GPU, and other system components on the 2016 MacBook Pro 15 are actually quite good.
It also compares quite favorably with older systems. To illustrate this, I grabbed a MacBook Pro 15 circa 2013. This laptop was built with a quad-core Ivy Bridge Core i7-3635QM and GeForce GT 650M graphics. It has 8GB of RAM and a 256GB SSD. This older MacBook Pro 15 also features just about the largest permanent laptop battery allowed by law onto a plane—99.5 watt hours. Despite its age, the battery itself was in good shape, with fewer than 10 discharge cycles on it and a health report of 99 percent.
A clean install of macOS Sierra 10.12.1 was used before testing. First came our 4K video run test at a brightness of 255 nits, Wi-Fi off, and our USB data logger in place. I should note that this model doesn’t have a Thunderbolt 3 chip, so there’s no power cost from having a USB device plugged in.
Even with its larger 99.5-watt-hour battery, the 2013 MacBook Pro 15 tapped out at only 330 minutes, which is well south of 6 hours. Compare that to the 2016 MacBook Pro, which, with its 25-percent-smaller battery, made it all the way to 474 minutes, or almost two hours more run time.
During video playback, both used the integrated Intel graphics and were mostly idling. The 2013 MacBook Pro 15 consumed about 18 watts, while the 2016 MacBook Pro 15 used about 13 watts.
The other factor in fuel mileage is how hard you drive it. No matter how efficient the engine is or how aerodynamic the car, sometimes having a larger fuel tank matters more.
To prove this point, I tasked the 2013 MacBook Pro with the same Prime95 torture test as the 2016 model. This test, if you recall, basically pushes all cores of the CPU as hard as possible. In car terms, it would be like flooring it fully loaded with passengers and luggage while going up a mountain. You’re burning a lot of fuel.
The older 2013 MacBook Pro has that in spades with its 99.5-watt-hour battery. It was able to run flat-out for 97 minutes before dying. The newer 2016 MacBook Pro, as efficient as it is, can’t push the CPU that hard for that long before emptying its 74-watt-hour battery at 83 minutes.
Increasing the battery capacity by 25 percent may not yield exactly 25 percent more battery life, but it would certainly last longer than it does now under heavy-duty loads. Those who actually use their MacBook Pro 15 in heavier tasks on battery may indeed experience shorter run times than with older models, especially if their point of reference is a more recent and more efficient Haswell- or Broadwell-based model.
And the tests say: Your mileage will vary
After all that battery testing, there are several takeaways:
- The 2016 MacBook Pro 15 did indeed have a bug that would sometimes cut short battery life. Apple appears to have corrected the problem.
- I found that plugging the MacBook Pro 15 into an external monitor engaged the discrete GPU full-time at a cost of 9 watts. While it’s unlikely you’d be at a desk using an external monitor without being plugged into AC, if you do, expect poor battery life.
- The newer MacBook Pro 15’s 500-nit screen is brighter. If you’re comparing it to an older MacBook Pro 15, use the same brightness settings to make the comparison fair.
- As reviewers originally found, the MacBook Pro 15 still has pretty decent life for a notebook in its class. Eight to nine hours of video playback at 257 nits on a quad-core equipped laptop with discrete graphics is very respectable. Comparable Windows laptops I’ve seen with quad-core CPUs, 4K screens, and discrete graphics generally offer worse battery life.
- Apple may have compromised too much by putting a smaller battery in the MacBook Pro 15. It’s clear from the testing on the older Ivy Bridge MacBook Pro 15 and the newest MacBook Pro 15 that the 76-watt-hour battery can yield shorter battery life on heavier loads.
- How much battery life should you expect? Well, as the tests here show, depending on what you’re doing, the answer is anywhere from an hour and a half to 18 hours. Not satisfied with that answer? Unfortunately, it’s the only correct one, as battery life on any laptop will vary depending on what you do.
This story, "Tested: The truth behind the MacBook Pro's 'terrible' battery life" was originally published by PCWorld.