Battery Life: Good (60Hz) to Average (120Hz)

Last week we had published our initial battery life report on the S20 series, with some interesting findings. First of all, what needs to be mentioned again is that the new 120Hz display modes on the phones come with a quite large battery life impact. The behavior is exhibited on all our S20 models at hand and I think it’s likely due to the panel itself or the DDIC. Samsung had included various display refresh modes varying from 48, 60, 96 and 120Hz, however we have yet to find evidence of any mechanism that actively switches between the various modes. 

Device-On Black Screen Power Consumption (Airplane Mode)

As such, even on a black static screen, running at 120Hz comes with a quite steep power penalty that’s always present whenever the display is on, costing around 160mW of power.

I had noted that I found our variant of the Snapdragon 865 Galaxy S20 Ultra to have worse idle power than our Exynos phone version. Initially I had attributed this to possibly the SoC or even the nature of the external X55 modem, but since then I’ve also received an LG V60 and that device’s idle power is perfectly normal. The only other thing that differentiates our S20 Ultra here is the fact that it has the extra mmWave antennas and RF systems. It would be interesting to see if non-mmWave variants of the Snapdragon S20 Ultra behave any differently (Tip at our Chinese or Korean readers).

I also had made mention that the “Performance” mode of the Exynos S20 phones seemingly behaved quite overzealously in terms of its scheduling settings, and there was a quite drastic increase in power draw for what was not nearly an as drastic increase in performance. I’ve rerun the battery tests in the “Optimized” settings which doesn’t have the “Increased system speed” option enabled, and I’ve confirmed my suspicion as the battery life figures did improve by some notable amounts. I’ve also tested the Snapdragon in the “Optimized” setting and the runtimes only differed by 2% - for users having the Snapdragon versions it’s thus safe to simply leave that enabled.

Web Browsing Battery Life 2016 (WiFi)

In our web test, the new S20 series end up right about where you’d expect them to. The Snapdragon 865 Galaxy S20 Ultra at 60Hz fares the best amongst the tested models, and now represents Samsung’s longest lasting flagship device. Slightly behind it we find the Exynos S20 Ultra at 60Hz. The difference between the two phones here isn’t very big in this test, and I attribute this to the higher constant idle power draw of the Snapdragon phone which counteracts the much higher compute efficiency of the SoC. The Eyxnos S20+ ends up slightly behind the S10+ phones, but still lasts a good 12.65h in this test.

Once we turn on the 120Hz display modes, the battery life results on all the phones drops quite notably. The Snapdragon S20 Ultra goes from 14h to 11.3h, a 20% drop. The same applies to the Exynos S20 Ultra, with a 20% drop, but for some reason the S20+ sees a larger drop of 25%. In the systems performance section I did mention that there’s some software configuration differences between the Exynos S20 Ultra and S20+, maybe some of that plays part here in the results.

PCMark Work 2.0 - Battery Life

Overall, the conclusion on battery life isn’t quite as black & white as we thought it would be. The key point is to stay away from the seemingly broken Performance mode on the Exynos chipset and you’ll have roughly similar battery life results between the two SoC variants of the S20. Naturally, that’s only being achieved by the fact that the Exynos does showcase worse performance, saving energy by using the more efficient lower performance states more.

What’s valid for all variants of the phones is that the 120Hz display mode is quite the power hog. Samsung probably has the opportunity to improve this by introducing a better managed variable refresh rate mode that actually changes between the different refresh rates based on content, something that seemingly isn’t happening right now. Also switching to lower refresh rates when showcasing static content would be a huge power saver, but I’m not sure if Samsung would be able to actually deploy such a mechanism.

Display Measurement Camera Architecture: Huge New Sensors
Comments Locked

137 Comments

View All Comments

  • airdrifting - Monday, April 6, 2020 - link

    I am actually a very forgiving person, I gave them 4 chances but they botched all. And then there was the Samsung Note catching fire left and right while Samsung PR was lying to cover up, that was the last straw for me. I was very skeptical when I switched to Oneplus, but I was very impressed with the build quality and speedy updates, the only weak point was the camera.
  • tuxRoller - Sunday, April 5, 2020 - link

    Any thoughts as to why Huawei was having issues with the A77?
  • BedfordTim - Sunday, April 5, 2020 - link

    They say the battery life penalty means the extra performance isn't worth it. This fits with their energy use figures in the benchmarks which are very good. It also makes sense from a user perspective as the extra performance will not be noticeable in everyday use.
  • iphonebestgamephone - Monday, April 6, 2020 - link

    How come qualcomm improved efficiency then? Its more like hisilicon had no time.
  • Andrei Frumusanu - Monday, April 6, 2020 - link

    I don't think they had issue, they probably didn't have time.
  • tuxRoller - Tuesday, April 7, 2020 - link

    You're probably right.
    Any idea as to why they were running late this time? The arm update cadence has been reliably in May since at least the A72.
  • ** A - R ** - Sunday, April 5, 2020 - link

    Andrei,
    For storage benchmarks, how about this (cpdt) app ?

    https://play.google.com/store/apps/details?id=com....
  • Andrei Frumusanu - Monday, April 6, 2020 - link

    It's not useful - there's very different aspects of storage performance on Android that's not covered by these traditional tests.
  • Arian_amg - Sunday, April 5, 2020 - link

    Thanks for great reviews:)
    I waited for this one like hell... With these Corona situations it's hard to find entertaining content and as always you nailed it best chipset reviews I've ever seen most will getaway with some benchmark and thats it
    Im really looking forward for the dimensity 1000 and kirin 820 to get a proper understanding of new Valhalla cores with a proper node
    (btw do you agree with huawei that they said they didn't use a77 because it didn't reach a certain clock(I know it had much higher ipc than a76)?)
  • Andrei Frumusanu - Monday, April 6, 2020 - link

    HiSilicon probably had less time to integrate the A77 and thus didn't reach good clocks.

Log in

Don't have an account? Sign up now