Snapdragon 810 Performance Preview

Conclusion

Snapdragon 810 has been a very hot topic in the mobile tech world recently, with multiple rumors insinuating a faulty design. Fortunately, based on our limited initial testing, those rumors appear to be mostly false.

Regarding the supposed issue with the memory controller, our benchmark data shows the 810 performing mostly on par with the 805. The STREAM memory tests show a slight drop-off in multi-core memory performance, but it does not seem to have a noticeable impact on other benchmarks; the 810 performs very well in multi-core system tests. The challenging 3DMark Physics test suggests the 810's memory controller may be tuned for sequential rather than random requests, similar to Apple's Cyclone based SoCs. The one minor area of concern remaining is memory latency. This is something that definitely warrants further testing once shipping products arrive; however, based on all of our other benchmark results, even if there is more latency than normal, it's not the catastrophic issue the rumor suggests.


During the benchmarking workshop, Qualcomm discussed the importance of a device's skin temperature on user experience. The image above shows a skin temperature test Qualcomm performed with two devices playing the game "Asphalt 8," claiming that the device running a Snapdragon 810 SoC ran cooler than a device running Snapdragon 800. If frame rates were capped at 30fps, then the results seem plausible. With a much more powerful GPU, the 810 could underclock and reduce voltage significantly and still hit a 30fps target, while the older 800 probably needs to run at full speed to maintain 30fps. Unless more OEMs implement performance limits (like the Shield Power Control feature on Nvidia's Shield Tablet) rather than just thermal limits, this test isn't particularly realistic though.

We can't make a definitive conclusion about overheating issues until we can thoroughly test a shipping handset with intensive, longer running workloads. However, after running benchmarks back-to-back (rather than allowing a cool down period between runs like we normally do) I didn't see any performance degradation or noticeable skin temperature increase. Qualcomm's MDP tablet is a bit thicker than production tablets, but this anecdotal evidence is at least encouraging.

The performance of ARM's 64-bit A57 is a nice improvement over the previous Krait architecture, which has certainly reached the end of its lifespan. Snapdragon 810's multi-core performance is particularly impressive, its efficient AArch64 architecture and new cryptography instructions provide a significant boost for certain workloads. A57 is still IPC limited however, and trails both Apple's A8X and Nvidia's Denver architectures in single-core tasks.

GPU performance is also improved, with the Adreno 430 delivering on its promised 30% performance increase. The regression in Basemark X, along with several crashes while running gaming benchmarks, suggests there's more work to do on the software side, but the GPU hardware seems solid.

Based on this initial assessment, the rumored demise of the 810 seems to be extinguished. We still need to examine the memory controller performance and power consumption/thermals more thoroughly before we can give it our stamp of approval, but it seems like it's ready to power the next generation of flagship phones. Depending on power draw, the X1 may challenge it in the tablet space and if Samsung can get its Exynos SoC moved onto 14nm FinFET, Qualcomm could finally see some serious competition in handsets too. It should be an exciting year for mobile.

  • realjjj
    In Geekbench your result is by far the highest in the database , something is off there, it's overclocked or you are testing in a fridge or you got some new revision.What's certain is that something is way off.
    Reply
  • MobileEditor
    In Geekbench your result is by far the highest in the database , something is off there, it's overclocked or you are testing in a fridge or you got some new revision.What's certain is that something is way off.

    The 810 scores lower than both the Tegra K1 (Denver) and A8X in Geekbench single-core and 8% better than the A8X in multi-core. Looking at the table for the individual Integer test results shows that most of the 810's advantage is in the AES and SHA1 encryption tests, which AArch64 targets with new instructions, as I noted in the article.

    I used Geekbench 3 Pro v3.3.1 (as noted on the "Testing" page) and I definitely didn't test it in a fridge :) It was standing on a conference room table like shown in the picture on the "Testing" page.

    - Matt
    Reply
  • MrCommunistGen
    "With only half as many cores, both A8X and Tegra K1 (Denver) see their IPC advantage diminish in the multi-core tests"

    A8X is a tri-core CPU.
    Reply
  • airborn824
    This does not seem very promising at all. At this rate i am stuck with my S4 this year. I wont upgrade with such a small performance increase.
    Reply
  • MobileEditor
    "With only half as many cores, both A8X and Tegra K1 (Denver) see their IPC advantage diminish in the multi-core tests"

    A8X is a tri-core CPU.

    Doh! I was still thinking A8. That's what happens when writing at 4am with no sleep. I'll fix that.

    - Matt
    Reply
  • nebun
    tegra is a monster of a processor....nice numbers
    Reply
  • nebun
    battery life?????
    Reply
  • nebun
    how efficient is it?
    Reply
  • MobileEditor
    how efficient is it?

    If you're referring to Tegra X1, Nvidia claims between 5W-10W depending on application (tablet or car). We can't verify these claims since there aren't any shipping products yet and nobody outside of Nvidia has even been able to touch it.

    For the 810, we weren't given enough time to test battery life. We need to wait until products ship. With the 810 moving to 20nm and the fact that the Krait CPUs were pushed to their max frequency, I wouldn't be surprised to see the 810 use less power than the 805 for average tablet workloads. The Adreno 430 might use a little more power than 420 though.

    - Matt
    Reply
  • JeanLuc
    Are you checking to see if any of the devices are deliberately 'turboing/boosting' clockrates when certain benchmarks are run?
    Reply