GPU Performance & Power

We also finished up our GPU testing for the Galaxy S20 Ultra, and are able to showcase the Snapdragon 865’s true power efficiency as well as long-term performance for the first time.

The new Adreno 650 iGPU is a continuation of the architecture that has powered Qualcomm’s last two generations of SoCs. The company is advertising a 25% boost in performance, which we were able to verify in our performance preview of the SoC. What wasn’t very clear back then was the power efficiency of the new chip – again, our figures on the QRD865 weren’t very accurate – so testing out the new chip on an actual commercial device will be a much more representative measurement.

This is also the first time we’re able to showcase the sustained performance figures of the new generation SoC. Qualcomm had put a lot of emphasis on this aspect of the new chip, claiming it’s able to throttle a lot less than some other competing devices in the market. We hadn’t had the time to test this on the QRD865, so the S20 Ultra here will be the first device of 2020 going through this test. As a reminder, sustained performance not only is linked to the power consumption of the SoC, but also the thermal dissipation capabilities of the phone that it’s employed in, as well as the thermal limits imposed by the vendor.

3DMark Sling Shot 3.1 Extreme Unlimited - Physics

Starting off with the 3DMark Physics test, the new chip behaves fantastically, and even manages to perform better than the QRD865. The new Cortex-A77 cores are doing extremely well, and most interestingly, the sustained performance showcased by the unit is also excellent.

3DMark Sling Shot 3.1 Extreme Unlimited - Graphics

The graphics test of 3DMark is more GPU bound – and here again it showcased no throttling. This was a bit odd in the context that I did see more significant throttling in GFXBench…

GFXBench Aztec Ruins - High - Vulkan/Metal - Off-screen

In the Aztec High test, peak performance is identical to what we’ve seen on the QRD865, and sustained performance after a while throttles down by 25%. Although it’s a good showcase, the new Galaxy S20 Ultra here doesn’t fare any better than some of the best Snapdragon 855 devices out there, as well as lagging behind Apple’s iPhones.

GFXBench Aztec High Offscreen Power Efficiency
(System Active Power)
  Mfc. Process FPS Avg. Power
(W)
Perf/W
Efficiency
iPhone 11 Pro (A13) Warm N7P 26.14 3.83 6.82 fps/W
iPhone 11 Pro (A13) Cold / Peak N7P 34.00 6.21 5.47 fps/W
Galaxy S20 Ultra (Snapdragon 865) N7P 20.35 3.91 5.19 fps/W
iPhone XS (A12) Warm N7 19.32 3.81 5.07 fps/W
iPhone XS (A12) Cold / Peak N7 26.59 5.56 4.78 fps/W
Mate 30 Pro (Kirin 990 4G) N7 16.50 3.96 4.16 fps/W
Galaxy 10+ (Snapdragon 855) N7 16.17 4.69 3.44 fps/W
Galaxy 10+ (Exynos 9820) 8LPP 15.59 4.80 3.24 fps/W

Doing power measurements on the S20U, we see quite better figures than experienced on the QRD865. Thus, the Snapdragon 865 moves up in the efficiency scoreboard, inching closer to Apple’s A-series chipsets.

What’s interesting to see here is the absolute power figure itself coming in a little under 4W. This is an excellent improvement and reduction compared to the Snapdragon 855 last year, and it means heat generation should be a lot more in check compared to previous generation flagships.

GFXBench Aztec Ruins - Normal - Vulkan/Metal - Off-screen

GFXBench Aztec Normal Offscreen Power Efficiency
(System Active Power)
  Mfc. Process FPS Avg. Power
(W)
Perf/W
Efficiency
iPhone 11 Pro (A13) Warm N7P 73.27 4.07 18.00 fps/W
iPhone 11 Pro (A13) Cold / Peak N7P 91.62 6.08 15.06 fps/W
iPhone XS (A12) Warm N7 55.70 3.88 14.35 fps/W
Galaxy S20 Ultra (Snapdragon 865) N7P 54.09 3.91 13.75 fps/W
iPhone XS (A12) Cold / Peak N7 76.00 5.59 13.59 fps/W
Mate 30 Pro (Kirin 990 4G) N7 41.68 4.01 10.39 fps/W
Galaxy 10+ (Snapdragon 855) N7 40.63 4.14 9.81 fps/W
Galaxy 10+ (Exynos 9820) 8LPP 40.18 4.62 8.69 fps/W

The Aztec Normal test also shows excellent absolute peak power figures, and peak performance is where it’s expected to be at. Meanwhile, throttling on the S20 Ultra is again at around -25% of peak performance for sustained scenarios.

GFXBench Manhattan 3.1 Off-screen

GFXBench Manhattan 3.1 Offscreen Power Efficiency
(System Active Power)
  Mfc. Process FPS Avg. Power
(W)
Perf/W
Efficiency
iPhone 11 Pro (A13) Warm N7P 100.58 4.21 23.89 fps/W
Galaxy S20 Ultra (Snapdragon 865) N7P 88.93 4.20 21.15 fps/W
iPhone 11 Pro (A13) Cold / Peak N7P 123.54 6.04 20.45 fps/W
iPhone XS (A12) Warm N7 76.51 3.79 20.18 fps/W
iPhone XS (A12) Cold / Peak N7 103.83 5.98 17.36 fps/W
Mate 30 Pro (Kirin 990 4G) N7 75.69 5.04 15.01 fps/W
Galaxy 10+ (Snapdragon 855) N7 70.67 4.88 14.46 fps/W
Galaxy 10+ (Exynos 9820) 8LPP 68.87 5.10 13.48 fps/W
Galaxy S9+ (Snapdragon 845) 10LPP 61.16 5.01 11.99 fps/W
Mate 20 Pro (Kirin 980) N7 54.54 4.57 11.93 fps/W
Galaxy S9 (Exynos 9810) 10LPP 46.04 4.08 11.28 fps/W
Galaxy S8 (Snapdragon 835) 10LPE 38.90 3.79 10.26 fps/W
Galaxy S8 (Exynos 8895) 10LPE 42.49 7.35 5.78 fps/W

In Manhattan 3.1, things are also looking better for the Snapdragon 865. The improved power figures do have it now able to leap the A12 in power efficiency. The sustained performance comes in at 73% of peak performance in our testing conditions.

GFXBench T-Rex 2.7 Off-screen

GFXBench T-Rex Offscreen Power Efficiency
(System Active Power)
  Mfc. Process FPS Avg. Power
(W)
Perf/W
Efficiency
iPhone 11 Pro (A13) Warm N7P 289.03 4.78 60.46 fps/W
iPhone 11 Pro (A13) Cold / Peak N7P 328.90 5.93 55.46 fps/W
Galaxy S20 Ultra (Snapdragon 865) N7P 205.37 3.83 53.30 fps/W
iPhone XS (A12) Warm N7 197.80 3.95 50.07 fps/W
iPhone XS (A12) Cold / Peak N7 271.86 6.10 44.56 fps/W
Galaxy 10+ (Snapdragon 855) N7 167.16 4.10 40.70 fps/W
Mate 30 Pro  (Kirin 990 4G) N7 152.27 4.34 35.08 fps/W
Galaxy S9+ (Snapdragon 845) 10LPP 150.40 4.42 34.00 fps/W
Galaxy 10+ (Exynos 9820) 8LPP 166.00 4.96 33.40fps/W
Galaxy S9 (Exynos 9810) 10LPP 141.91 4.34 32.67 fps/W
Galaxy S8 (Snapdragon 835) 10LPE 108.20 3.45 31.31 fps/W
Mate 20 Pro (Kirin 980) N7 135.75 4.64 29.25 fps/W
Galaxy S8 (Exynos 8895) 10LPE 121.00 5.86 20.65 fps/W

Finally, in T-Rex, we’re again seeing a healthy boost in efficiency compared to previous figures, but in terms of sustained performance, the S865 is still largely behind the A13 chips.

Besides the improved power efficiency numbers, we’ve now measured on the Galaxy S20 Ultra, the one thing that I really liked about the new device’s behavior is its thermal behavior. The phone wouldn’t exceed 41-42°C peak skin temperature, remaining only relatively lukewarm no matter the kind of workload you’d throw at it. This is some excellent thermal management and I appreciate this a lot more than having higher performance that leads to the phone getting uncomfortably hot. Samsung deserves some praise here.

Fantastic Performance; Stay Tuned For The Full Review

When it comes to performance, the Galaxy S20 Ultra has so far been an immensely impressive device, easily being the single fastest, most responsive, and most fluid phone I’ve had the pleasure to use, quite far ahead of any second contender. The combination of a 120Hz screen with the new Snapdragon 865 is a match made in heaven, and people prioritizing this aspect of a phone will seemingly not be disappointed in the S20 line-up.

There’s still a lot of work to do until we get the full review finished, notably still waiting on the Exynos variants of the devices. One aspect that I want to make note of right now is that there’s been some concerns in regards to the battery life aspect of the new 120Hz mode of the phones, and I can confirm that in this setting the S20 Ultra showcases worse battery life than the S10+ (-10% in our web test), even though the new unit has a more efficient SoC as well as a 21% bigger battery. Samsung’s implementation of the 120Hz mode comes at quite a large power hit that’s very disappointing in how it behaves, and the device in general seems not quite as efficient even at 60Hz.

I’ll be posting a separate battery preview once I get more battery testing numbers together, and of course we’ll have a full feature review out once all the thorough testing is completed.

Intro & System Performance, CPU Power
Comments Locked

117 Comments

View All Comments

  • flyingpants265 - Sunday, March 15, 2020 - link

    AHHHHHHHHH!!!!!!!
  • N Zaljov - Friday, March 13, 2020 - link

    The casually included MT6885Z benchmark results are really sticking out to me. Are you going to review the Redmi K30 or a similar device? If so, I'm really looking forward to it.

    Following the shambles of S.LSI's previous "adventures" I'm kind of curious as to how SARC's M5 cores end up performing compared to QC's products, but considering the fact that they've scrapped it entirely along with the fact that they're not shipping the Exynos version in their own market (South Korea) along with the struggles of Samsung's Foundry business, I highly doubt that Samsung will be able to be competitive for a very long time...

    Anyway: I appreciate your work and hope that you'll continue to deliver excellent content like this in the future.
  • MrCommunistGen - Friday, March 13, 2020 - link

    I also came here to comment on the inclusion of the Dimensity 1000L in the SPEC numbers. Are we getting a full review and/or analysis of a device with this SoC?
  • anonomouse - Friday, March 13, 2020 - link

    Also interested, it appears that the supposed 2.6Ghz A77 is performing worse than the 2.4Ghz A77 in SD865, the 2.86Ghz A76 in Kirin 990, and then 2.84Ghz A76 in SD855... which would point to some deficiencies in how Mediatek implemented their memory hierarchy beyond the L1 in their SoC?
  • N Zaljov - Friday, March 13, 2020 - link

    Keep in mind that the A77 cores in the Dimensity 1000L are only running with up to 2.2 GHz (it's a slightly downrated version of the Dimensity 1000). If you adjust the SPECSpeed for clocks, the A77 cores should fall in line with the other A77-based SoCs, assuming that there are no bottlenecks that would affect the performance scaling negatively.
  • anonomouse - Monday, March 16, 2020 - link

    Ah, forgot about the 1000L vs 1000.
  • Andrei Frumusanu - Friday, March 13, 2020 - link

    As mentioned, the D1000L is running at 2.2GHz, not 2.6.
  • Andrei Frumusanu - Friday, March 13, 2020 - link

    I'll be doing a piece on that when I'm done with the S20's, yes.
  • Psyside - Saturday, March 14, 2020 - link

    Please, please include Flnd X2 Pro in the camera tests, the IMX689 looks amazing so far
  • Dragonstongue - Friday, March 13, 2020 - link

    IMO I believe for my $ anything considered "flagship" class, should dang well include 3.5mm jack, BT / wireless is "cool" and all that, latency wise however (not to mention connectivity options) 3.5mm jack still has so much more going for it.

    BT/Wireless has a place, no doubt, but gosh dang, over $1100 phone (above and beyond most full out laptops, which tend to carry audio jacks LOL) then you have to go and buy "respectable" quality earbuds on top of this usually in the $50+ as well.

    the world is made of money and we like trashing things every year or 2 "cause that is how we roll"

    Oppio (or whatever they called, I typo I know) and some others have very near same spec (not Exynos chip of course) less pricey and many still carry the venerable audio/aux jack not make stupid excuses about not having it either

Log in

Don't have an account? Sign up now