GPU Performance

GPU performance of the Pixel 3 should be relatively non-surprising for the most part – again we see the Snapdragon 845 at play and its Adreno 630 GPU should be performing excellently. This year the only real differences between devices was how vendors decided to set up their thermal throttling mechanisms and how the hardware itself is able to dissipate sufficient heat – as the SoC’s peak performance lies above the sustainable thermal envelope of a given device.

3DMark Sling Shot 3.1 Extreme Unlimited - Physics

In the 3DMark Physics test, the Pixel 3 performs as expected in peak performance, however we see a more than usual decline is sustained performance compared to other Snapdragon 845 phones. Here it is possible Google has more strict thermal limits on the CPU.

3DMark Sling Shot 3.1 Extreme Unlimited - Graphics

The graphics results on 3DMark are more in line with other S845 devices, still the small Pixel 3 does end up slightly lower in performance. It’s notable that the Pixel 3 here ends up with a lower sustained performance score as the Pixel 2 family – showcasing the worst-case scenario for the SoC.

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

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

In the new Aztec benchmark, the new GPU architecture does help quite a bit in terms of differentiating itself from last year’s Pixels, however still the Pixel 3 ends up at the lower end of S845 phones in terms of the sustained performance.

GFXBench Manhattan 3.1 Off-screen GFXBench T-Rex 2.7 Off-screen

Finally in Manhattan and T-Rex, the Pixel 3 ends up in the same ballpark range as last year’s Pixels – again a worst-case scenario for the new SoC.

Among Android devices the Pixel 3 doesn’t stand out too much from the competition, however is still going to be able to perform very well. One has to keep in mind GPU and gaming performance is very much tied to the hardware capabilities, and in this regard we’ll see major jumps with the next generation GPUs.

System Performance Display Measurement
Comments Locked


View All Comments

  • buxe2quec - Sunday, November 4, 2018 - link

    When I buy a phone I go to DxO Mobile and from the top ranking I go down until I find a phone within my budget.
    DxO however is sometimes controversial, so I would like to see a synthetic number from Anandtech, so that I can quickly do the same here as well.
  • misaki - Sunday, November 4, 2018 - link

    The Pixel3 4k EIS video is linked to the Pixel 2 video and the Pixel2 1080p30 is labeled incorrectly.
    I don't know if it shows up in the original video or if it's a Youtube encoding issue but the unstabilized Pixel 3 4k video retains a lot more detail than both of the the Pixel 2 4k videos, stabilized and unstabilized. I see it plainly without even needing to pixel peep. Pixel 2 video capture had good stabilization but was always disappointing in all other ways and behind other flagship phones. If they at least improved their encoder for the Pixel 3 like what I'm seeing then it's a good start.
  • cwolf78 - Monday, November 5, 2018 - link

    Thanks for the thorough review. One question I have is that there are many other reviewers out there that have praised the included ear buds and even had separate reviews for them they were so impressed. I don't think anyone has claimed the sound quality is top-tier by any means, but they were supposed to sound decent and have excellent Google Assistant support. Just curious if maybe you have a defective set or your standards are unrealistically high?
  • Andrei Frumusanu - Monday, November 5, 2018 - link

    I wouldn't say my standards are that high - and *maybe* my units are defective. Compared to any other bundled units they sound as if they're covered by tape and very muffled. A quick frequency response comparison ( and yes I know this isn't an absolute measurement, but a relative one to the Apple ones ) represented what I heard, with insanely weaker mid-ranges.
  • Fluffywings - Monday, November 5, 2018 - link

    Hi Andrei,

    Great review and clearly a lot of effort. As you mention in the camera review, excerpt below, the shadows are darker than should be, which is something I noticed with the Pixel 2 back in 2017. I have found a simple solution that changed how I use my Pixel 2 camera. When setting up the picture, click on a darker area for focus. It will bring up the shadows and the picture will look better than before but with much detail in the shadows. I will take usually 2 pictures, one clicking on the darkest portion of a scene, and the second, clicking on a slighter less darkest portion of the scene.
    Can you give it a try and let me know what you think?

    "A characteristic of Google’s phones we’ll see throughout the pictures is that the processing likes to darken the shadows more than what the sensor actually sees, and this most visible in the trees in these pictures, as the pines in the middle picture lose a lot of detail compared to any other phone, also something that happens throughout darker objects of the whole scene."
  • Andrei Frumusanu - Monday, November 5, 2018 - link

    I mean, sure, that's a way to do it, but then I'm no longer testing the default capabilities of the phone. I also said this year Samsung overexposes too much in many scenes, and I'm also not going around to adjust that for every shot.
  • Fluffywings - Sunday, November 11, 2018 - link

    Makes sense. Most people will use the phone in the same way.
  • s.yu - Thursday, November 15, 2018 - link

    I also notice that in my Gcam port to Note8, I set everything to max for maximum IQ but still sometimes shadows drop to pitch black too fast. Switching to the default camera app with the exact same framing I could get a much finer gradient in the shadows.
    The reason I still use GCam is higher DR in most instances and less noise, less smearing. Also the slight HDR effect applied works better with snapseed's set of adjustment sliders, especially ambience. The default camera works better with LR, snapseed's often a little quirky and counterintuitive.
  • stacey94 - Monday, November 5, 2018 - link

    Great review! I really wanted to see the display results.

    I don't know that you guys ever test this stuff, but it would be nice to see audio output comparisons between phones for the 3.5mm jack. (As it seems like a lot of digital dongles shipping with phones have lower quality than the Qualcomm DAC built into the phones).

    And some level of Bluetooth performance analysis would be helpful, whether that's through LDAC bitrate or signal strength. I know the original Pixel had an "antennagate" problem where touching the band on the top right of the phone would immediately cause BT audio to garble and cut. My Pixel 2 was also hit or miss with several BT devices compared to an iPhone 8.

    On another note, is testing the Surface Laptop 2 on the agenda at all? I'm mainly looking to see the color accuracy of the display and find out if they're still using that awful SSD the first gen shipped with.
  • dgtangman - Monday, November 5, 2018 - link

    I'm seeing a very strange behavior with the sample images and would like to know if anyone else has encountered it. If I right-click and select "Open Link in New Window" I get a new window that shows the full-size version of the currently displayed sample the first time I click on the image. If I then pick a different sample the in-page sample image changes and the link name shown when I hover on the sample image changes, but if I right click and open in a new window again I get the same image I saw the first time; the URL displayed in the new window is the same as for the first sample image and is not the URL that was displayed when I right clicked on the sample image. To see a different full-size image from a given set of samples I have to reload that page of the review.

    Has anyone else seen anything like this? I've been trying to figure out anything that could be wrong on my end to get these results, and I haven't had any luck. I'd like to blame it on my Comcast connection, but I can't figure out how they could accomplish this either.

    I've tried this on Firefox 63 with all add-ons disabled and on Chromium 70 with identical results.

Log in

Don't have an account? Sign up now