The battery life of the phone gets worse or improves considerably

If you pay close attention to the phone launches, you may have noticed an unusual trend with the latest bumper harvest from Apple, Samsung and other OEMs. According to many critics, the life of the battery of new devices has globally declined. The exact numbers vary depending on the type of tests you run and the devices you are comparing, but in many cases the 2018 devices do not match the performance of previous models.

the Washington Post, who recently covered the subject, released his own Wi-Fi navigation tests:

During their tests, devices such as the iPhone 8 Plus beat the iPhone XS for an hour and the Pixel 2 outperformed that of the Pixel 3 by almost 90 minutes. This goes against the marketing discourse of these different companies and contradicts the general perception that each new generation of products is more energy efficient than the last. But the To post also notes that the trend is not universal and that Consumer reports, which uses a very different battery test methodology, has obtained exactly opposite results. According to CR, the XS and the Max are two significant improvements over the iPhone X. Why this huge disparity? the To post do not really respond to that, but we will do it.

Different methodologies, different results

Most critics test the battery life by performing a loop test (usually a series of web page loadings or video playback) until the battery dies. Different review sites use a different combination of web pages and video codecs. As a result, you will always see variations from one site to another. The majority of sites also calibrate the brightness to a specific value, usually between 150 and 200 nits.

Consumer reports does something very different. He writes:

To find out how long the battery of a phone can last Consumer reports uses a robotic finger programmed to run the phone in a series of tasks designed to simulate the average day of a consumer.

The robot navigates the Internet, takes photos, uses GPS navigation and, of course, makes phone calls. For the sake of consistency, we perform all smartphone battery tests with the display set to a brightness of 100%. However, if you want to extend the battery life of your phone, it is useful to turn on the automatic or adaptive brightness feature, which adjusts the display independently for the lighting environment. (Underlining added.)

I have emphasized this point because I want to illustrate that there are two important points of difference in the game rather than just one. Just for fun, I've loaded my 100% iPhone SE and connected it to the same Kill-A-Watt meter that we use to measure the system's power consumption. With the iPhone SE set to the minimum brightness, the power absorbed at the wall was 1.3 to 1.4 W. When the brightness is set to maximum, the power consumption reaches 1.8-1 9W. It's about an increase of 1.35. I do not claim to know how the OLED panels of new iPhones respond to changes in brightness, but we can safely assume that this configuration difference also affects the overall results. Nevertheless, the results of CR are quite different, with the iPhone XS lasting 24.5 hours and the XS Max 26 years, compared to its autonomy of 19.5 hours.

Why do phone battery tests look like they are

There are specific reasons why review sites tend to test batteries in this way. First, the use of a single test, such as playing a video or loading the same sequence of web pages, establishes a standard metric that can be used on multiple phones over a period of several years. It is true that very few people literally run a device from 100% to 0% doing a single activity, but the batteries do not discharge at a constant rate either. Testing from 100% to 0% allows the reviewer to avoid any errors by extrapolating the total discharge time from a partial measurement.

Secondly, the phones are in sandbox mode in the same way as personal computers and laptops. On the PC, you can use a reference harness to run a suite of applications to model a real-world use, with breaks and periods of inactivity to simulate the waiting time of the ################################################################################# 39 end user. This type of inter-application script is not available on phones, at least under normal operating conditions. Consumer reports"The robotic finger solves this problem, but the company probably needs to calibrate it for each test device.

Third, there is an astonishing variation in the way people use phones and under what conditions they use them. Do you use Wi-Fi or do you mainly use LTE? Do you use an advertising blocker? Do you use apps or do you prefer mobile browsers to access sites like Facebook? What level of brightness do you define? Which sites do you visit? In our own conversation on this topic, my colleague Ryan Whitwam pointed out that no one was sitting or surfing the web from 0 to 100%. I agree. For its part, iOS reports that Safari represents 51% of my personal battery consumption over the last 7 days, followed by Netflix (17%) and Amazon Prime Video (7%).

Fourth, companies with the best data on how customers use their products – Google, Qualcomm, Samsung, etc. – do not share this information with the press beyond the most general conditions. When a company launches a new SoC, it sometimes displays improvements in two or three specific areas, such as video playback, standby time, and web browsing, but it never shares them entirely. Companies that make general statements about improving the battery life throughout the day will tell you that their measurements take into account a large number of usage patterns, but do not detail exactly what are these patterns of use, what are their assumptions or where do the improvements come from. of.

Fifth, keep in mind that emptying the battery of a device can take hours, and no proofreader can sit for 8 to 10 hours performing exactly the same set of motion on 5 to 6 devices. # 39; row. The modern revision does not lend itself to such a playful evaluation schedule, even if you could find someone willing to follow it and stay on the job. The battery test must be automated. If you do not have access to a great robotic arm, you'll have to rely on scripted or loop tests.

Finally, there are a number of subtle factors that could influence the end results without a simple "correct" configuration option. Do you need to set up the phone for push notifications and frequent downloads by email, or for minimal notifications and only manual data uploads? Should Bluetooth and Wi-Fi be enabled or disabled? Should you leave low battery mode active when the manufacturer offers to turn it on or operate the device in full power mode until it dies? It's just a handful of relevant potential options and there is no "wrong" choice, but a different usage pattern.

Why improvements in battery life are difficult to predict

Once upon a time (the late '90s), processors were running at a constant clock speed and new process nodes still provided lower power processors and higher clocks. The world has evolved. A modern smartphone executes more complex orders of magnitude in power management than a notebook computer around 2001. As one can no longer rely on node reductions to make regular improvements, ARM and Intel have invested huge sums in the development of sophisticated silicon capabilities. disable portions of a chip when not in use, lower the clocks to minimum values ​​for responsive operation, and even deploy low-power, high-efficiency dedicated processor cores for optimal power performance and efficiency .

Instead of performing most of the software functions, the SoCs have built-in multiple accelerators to handle the same tasks in a fraction of the power envelope. With respect to software, Google and Apple have put in place best practices to minimize energy consumption and have improved their operating system planners so that they know which workloads need to be addressed. be executed on which hearts for maximum efficiency.

Even though the companies have made these improvements, they have increased the number of their cores, their clock speeds, the performance of the graphics processor, the screen resolutions, the size of the screen, the Cellular performance and Wi-Fi speeds. The back and forth between higher energy consumption needed to improve performance and improved efficiency based on technology is manifested differently according to domains. One of the reasons I love the SE iPhone so much is that he's married to a still relatively upscale SoC with a much more energy-efficient screen. The navigation tests gave a longer battery life than any other Apple device in its class.

This new mix of hardware acceleration, software best practices, scheduler enhancements, low-power processor cores, and improved power gating has undeniably brought benefits, but it has not been equal at all levels. That's why Maria Rerecich, director of CR's electronic testing, said: "You can not follow a clear trend," when asked if the battery life increases or decreases. It all depends largely on where you look. It depends on how you test. It depends on what matters to you and your use case. And unfortunately, this means that the trend lines will probably continue to point in both directions.

Now read: Apple will let users disable performance limitation on iPhones, replaceable batteries in Consumer Electronics from New Bill Mandates and why do we still use lame lithium-ion batteries after so many promising solutions?