TouchMarks II: Touchscreen Latencies in Flagship Tablets

In our last TouchMarks report, we looked at the touchscreen latencies of the flagship smartphones from different manufacturers. In this report, we’ll benchmark the touchscreen latencies of the leading tablets, including iPads, Microsoft’s Surface and leading Android tablets including Amazon’s newest Kindle Fire HD.

Methodology

Before we get into the results, a little bit about our methodology (for a full recap please check out our first post. In TouchMarks, we use our Touchscope to benchmark the App Response Time of touchscreen devices by measuring the time between when the user touches the screen and the device updates the display. We place the light sensor and touch at the center of the screen to account for the device’s 60 Hz refresh rate, the devices are in airplane mode with full brightness and they have had all background apps closed to make the test as fair as possible.

For this report, we will specifically be measuring the Minimum App Response Time by using OpenGL/DirectX-based optimized apps with minimal logic to quickly flash the screen white in response to a touch. We’ve open sourced the apps here for review. If you think there is a material discrepancy between our test apps on the different operating systems that gives one device an unfair advantage over the other, please let us know in the comments or submit a pull request.

With that, let’s take a look at the results!

Results

For this report, we benchmarked the latest versions of popular tablets from various manufacturers, including Amazon’s brand new Kindle Fire HD. We also threw Nvidia’s new Tegra 4-based SHIELD in there even though it’s not a traditional tablet. Here’s a graph of the results:

touchmark_graph_ftablets_rev2

In a result that’s perhaps now unsurprising, the iOS devices are more responsive than its competitors. Interestingly, the iPad mini – with its smaller screen and 1024×768 resolution – performs similarly to the larger fourth generation iPad and its 2048×1536 resolution, suggesting that responsiveness is not reduced by the larger screen size or resolution.

Perhaps the more interesting results are in the non-iOS devices. Nvidia’s SHIELD performed much better (though strangely with much higher variance) than the other Android devices, suggesting that Nvidia went through special effort to optimize the responsiveness of the device. It’ll be interesting to see whether the improvement carries over to other, non-Nvidia Tegra 4-based devices.

Amazon’s Kindle Fire HD was a surprising leader in the Android section, especially given that it’s $90+ cheaper than the other two Android tablets. This could be a result of Amazon’s willingness to sell expensive hardware at a loss. Finally, Microsoft’s Surface RT also performed much better than we expected, especially given that Nokia’s Lumia also running Windows didn’t fare significantly better than Android devices in our smartphone comparison. Given these results, I’m especially excited to run Amazon’s higher-end Kindle HDX and Microsoft’s new Surface 2 through our Touchscope. I’ll be updating this post with the results as soon as I can get my hands on them.

Our previous speculations continue to apply –  more responsive devices may process touches earlier in the stack, poll for touches more frequently or have touchscreens optimized or calibrated to be more responsive. In our latest review of our test apps, we discovered an optimization that suggests that the GPU or GPU drivers in the devices might also add significant latency. We will explore the ramifications of this and possible explanations in a future TouchMarks, but note our current test apps do not include the optimization in order to better represent the performance of normal OpenGL/DirectX apps.

Conclusion: if you primarily use your tablet for reading, watching videos or browsing the web, then shop around and pick the best tablet – iOS, Android, Windows 8 – that suits your needs. With their lower price points and high PPI screens you may find an Android tablet works perfectly for you. If, however, you’re into latency-sensitive applications like games or interactive music apps, then your best bet might be an iPad.

46 thoughts on “TouchMarks II: Touchscreen Latencies in Flagship Tablets

  1. Hi!

    It is possible to increase the sample time on the Surface touchscreen. Would be cool if you could test that and see if there is a difference.

    Found a key for touch prediction that when edited showed a marked improvement in keyboard responsiveness and small item manipulation ie classic desktop, file explorer, etc.
    The key is: HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\TouchPredict ion

    Edit key for latency from 8 to 2.
    Edit sample time from 8 to 2
    Restart

    1. David – That change does NOT affect the latency of the touch screen and just breaks the touch experience of your Surface (or other Windows touch device).

      That setting has no affect on the latency or sample time of the touch sensor. What it does is change the value given to the prediction engine, so it *thinks* the sensor is operating at a lower latency and faster sample time. This will cause the prediction engine to be wrong more of the time, giving you a worse experience, particularly for operations like dragging/panning.

      You should not make that change, and testing it will yield no change in the results of this test (since the prediction engine isn’t involved in the reporting of a single touch contact).

      1. Hi!

        Thanx for the responce. I also thought it was too good to be true and wanted it tested to be busted more or less. I´m sure microsoft have maked sure to get the most out of the hardware in this aspect but was curious if it was possible they hadnt.
        I havent made the change to my surface and frankly I´m more than happy by the responce of the device.

        /David

  2. ” If, however, you’re into latency-sensitive applications like games or interactive music apps, then your best bet might be an iPad.”

    What defines a “latency-sensitive application” and what evidence is there that these measured differences actually matter to them, if they are correct at all?

    There sure is a lot of pretense here.

  3. Why did you test the Surface RT and not the Pro the first time around.. The RT is NOT the flagship, but the low end model?

  4. Latency sensitivity is one of the most pressing problems in the touch screen world at present. There are several ways of going forward with the problem but all have their pros and cons.

    My personal favourite method is to make sure that your touchscreen is slightly damp when you use it. The dampness doesn’t harm the glass yet it lubricates the screen sufficiently to make it less laggy. Don’t believe me? Try it. Get a damp rag and wipe it on the screen and then use the device. The device will respond 10-15% faster than a dry screen.

    1. No thanks, using a dry iPad is plenty fast enough, no need to use my electronics while they’re wet. By the slowest tablet on the list, you’re suggesting to perceive a 15 ms improvement in latency over a dry screen, I can’t say I buy it.

      This questionable “home remedy” could possibly help on an $80 tablet from Walgreens, but you especially don’t want to use a cheap tablet while wet, yikes!

      If you must create extra capacitance between your fingertips and touchscreen, it’s much safer and effective to moisturize regularly. :)

    1. Using a different app, I get 75 to 100 ms for the 2013 Nexus 7. The large variation makes me think the software i’m using to change the screen brightness on press events may not be ideal.

      1. I got 90-100ms on my nexus 7 (60 fps video so not too precise). I didnt use any app; i just enabled Show Touches in dev options and measured the drag latency

        latency in apps is a frame or so higher than Show Touches

  5. Thanks for the review and previous test.

    Do you have access to the Wacom Cintiq Companion (Windows one) and can test its Touchscreen Latency?

    Would be interested in a future test for the Surface 2 Pro for sure, too :-)

    One Question. I’m in research for over half a year. Are there ANY devices with less then 50ms Latency out there? From what I have seen in the video “Applied Sciences Group: High Performance Touch” on youtube and my experience with touchscreen devices, I hope there will be one with a good response soon. Why is there such a big lack of research/dev with Pen-Input and its lag?

    1. using a mice + CRT monitor on linux or windows gives effectively no lag (less than the refresh period which would be 16.7ms for 60hz) for the movement of the cursor.

  6. This test seems well suited for testing latency as it applies to games (since it focuses on time to change visuals on the screen), but I don’t think it reflects the latency exhibited in music creation apps. Even the fastest of the tested tablets in the chart would be well beyond the acceptable latency for a musical instrument. Have you considered a test that would measure touch-to-beep latency?

  7. I recently co-authored a blog post for Microsoft focused on how to reduce input latency in DirectX apps. We are averaging less than 60ms of latency on Surface 2. Details can be found in this blog post.

  8. Very great post. I simply stumbled upon your blog and wanted to say that I have truly enjoyed surfing around your weblog posts.
    After all I’ll be subscribing on your feed and I am hoping you write once more very soon!

  9. Another parent shared with her instructor and got angry.

    I couldn’t have asked for something more,” Sadie’s handler, Gabriel Rangel, told reporters afterwards. The video you decide to watch it as-is as much as you.

  10. An outstanding share! I’ve just forwarded this onto a colleague who was conducting a little research on this. And he in fact ordered me breakfast due to the fact that I found it for him… lol. So let me reword this…. Thanks for the meal!! But yeah, thanx for spending the time to discuss this topic here on your web page.

  11. What’s Going down i’m new to this, I stumbled upon this I have
    discovered It positively useful and it has helped me
    out loads. I hope to give a contribution & aid other
    customers like its aided me. Good job.

  12. The photos are apparently being utilized to established up pre-order
    sales. Appear carefully at the pictures for dents, cracks, scratches on the screen, etc.
    This Apple iphone situation is priced in between $35-$39.

  13. By now, you must surely have known how-to view free movies online.
    But certainly, being a dad brings about the mush in even the absolute most matter-of-factly a man!

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>