padhz.com compared three of the fastest Quad Core Cortex A9 processors: Rockchip RK3188, Samsung Exynos 4412, and Nvidia Tegra 3. In their detailed report (in Chinese) they compare the performance, power consumption, video play back and more. As expected, it turns out RK3188 is the best, followed by Exynos 4412, and then Tegra 3, which correspond to the inverse of their release dates. Keep reading to find the details.
The Contestants
Nvidia | Samsung | Rockchip | |
Model | Tegra 3 | Exynos 4412 | RK3188 |
CPU | Cortex A9 Quad Core | Cortex A9 Quad Core | Cortex A9 Quad Core |
Process | 40nm | 32nm | 28nm |
Frequency | 1.3 GHz | 1.4 GHz | 1.6 / 1.8 GHz |
GPU | Geforce ULP | ARM Mali-400MP4 | ARM Mali-400MP4 |
GPU Frequency | 416MHz | 440MHz | 533 / 600 MHz |
GPU GFLOPS | 10 | 17.6 | 24 |
Level 2 Cache | 1MB | 1MB | 512KB |
Memory Bus | 32-bit | 64-bit | 32-bit |
HW Video Decode | 1080p | 1080p | 1080p |
Release date | November 2011 | April 2012 | January 2013 |
They use 3 tablets for the comparison: Nexus 7 (Tegra 3), Ramos W42 (Exynos 4412), and Cube U30GT2 (RK3188).
Nexus 7 | Ramos W42 | Cube U30GT2 | |
Display | 7.0” | 9.4” | 10.1” |
Resolution | 1280×800 | 1280×800 | 1920×1200 |
CPU Frequency | 51~1300 MHz | 200~1400MHz | 312~1608MHz |
RAM | 1GB DDR3 | 1GB DDR3 | 2GB DDR3 |
Free RAM | 974.5MB | 628.1MB | 1387MB |
OS | Android 4.2.2 | Android 4.0.4 | Android 4.1.1 |
Battery | 3.7V/4326mAh | 3.7V/6000mAh | 7.4V/3600mAh |
Firmware Version | Q39 (13/2/2013) | V1.2 (25/02/2013) | V1.01 (11/03/2013) |
That’s quite a combination of different resolutions, available memory, Android versions and battery capacities, but let’s carry on.
Benchmark Results (Antutu, Nbench, GLBenchmark…)
RK3188 is the clear winner in all categories (except flash write), even with 2D/3D tests despite the higher resolution. For some reasons Ramos W42 seems to have a dismal IO performance, but this is not the point of this test.
RK3188 is also the clear winner with NBench which test the integer and floating point performance of the CPU, as well as memory bandwidth.
GL Benchmark shows similar results for GPU performance. The benchmark has been run in 1080p offscreen mode, so the tablet screen resolution does not affect the results.
padhz also ran Nenamark which puts RK3188 at the top, but this time, closely followed by Tegra 3, with Exynos 4412 a bit further behind.
Application Tests
Instead of just running benchmark to compare the best, they also performed some usual tasks, such as installing and uninstalling large applications, loading a PDF, loading games…
The results are quite surprising as installing large application on the Cube tablet can be about twice as slow as on the other 2 tablets. I guess this has less to do with RK3188 performance than with the poor flash performance we’ve detected in Antutu. PDF loading times are about the same in all tablet, yet slightly slower in RK3188. The guys at padhz think it’s because of the larger screen resolution, which could be a valid explanation. There’s very little difference between games loading times in either of the 3 tablets. For testing web performance, they used pages with Flash which is not supported in Nexus 7, but they still found that it’s faster to load in the Exynos based tablet than the Rockchip based tablet.
Video Playback
They’ve also testing video playback of videos with different codecs and container formats using MXPlayer. To understand the results we need to learn a bit a Chinese first. The top fields are self-explanatory, but the Chinese text in the result fields translate as follows (Correct me If I’m wrong):
- – “No”, in the sense “N/A”.
- – “Not supported”
- – “Play smoothly”
- – “No sound”
- – “Lots of lag”
- – “Obvious lag”
- – “Minor lag”
- – “Audio cuts”
That means RK3188 does not support 2160p videos, nor EVO container aka Enhanced VOB (this should be fixable), but would still seem to be very capable in a media player, as it supports all other files perfectly.
Game Testing
Need for Speed achieved an average framerate above 30fps in Cube U30GT2, beating both the Nexus 7 (25 fps) and Ramos W42 (27fps). However in other games, such as Gangstar Rio, the RK3188 based tablet was the slowest of the three tablets, presumably because of the larger screen resolution.
Power Consumption
In the first test, they put the tablets into standby at 18:00, turn the devices on at 10:00 the following day, and let it on for a further 30 minutes, and see the battery charge levels. They did not start off with the same battery levels on all devices, so I’m not sure the comparison is 100% valid, as battery discharge may not be linear, and each device has a difference battery capacity.
Yet, it seems when it comes to power consumption, the Tegra 3 easily beats the other two processors in standby mode, most probably because of the 5th companion core, and Ramos W42 appears to consume more energy when the screen is on than the other two.
Finally they ran the battery test from GLBenchmark after charging all tablets to 100% with the following results:
The 3 line is the time it took the test to fully discharge the battery, and the last line correspond to the power consumed per hour, and has been calculated by dividing the battery capacity by the time it took complete the benchmark. The Nexus 7 is clearly the best performer, followed by Cube U30GT2, and Ramos W32.
However, the battery test a above are more a tablet battery test than a processor power consumption test, as there are so many components affecting the power consumption, one of the most important being the display itself.
Rockchip RK318 will bring us high performance devices, with great multimedia capabilities, at a much lower cost than either Tegra 3 and Exynos 4412 based designs. So I’m looking forward to putting my hands on one of the RK3188 mini PCs to do my own testing…
Via Liliputing
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress
2013 28nm SoC against 40nm and 32nm 2011 / 2012 SoCs.
Sure the 28nm at significantly higher clock speed (CPU up to 500MHz difference) is better – it would be a shame if not.
I also don’t want to know how hot it will get at 1.8GHz 😉
For Battery life: Rockchip Tablet uses a 26Wh Battery,
the Tegra Tablet uses 16Wh.
10Wh difference!
(and as you said there are a lot of different components in there and you can mess around with a lot
of settings)
–> Pointless test.
Anyway to slot in the iMX6 &/ A31 somewhere?
@onebir
That would be quite a lot of work, and I don’t have the tablets… If padhz.com did the same test on i.mx6 and a31 tablets, then that would be a possibility.
@o_O
I understand the RK3188 tablet is running @1.6GHz for now. For the battery test, they normalized it, so it’s still kind of useful of evaluate the power consumption “efficiency” of the tablet, although I’m not sure we can draw conclusions about the processors’ power consumption.
In real life all Rockchip have a very bad wifi and not standard Android Rom. These roms are full of useless software that ruins the Android experience.
Now i buy only exynos, in this mode i can install cyanogenmod and all the exynos board have a good wifi module integration.
@strauzo
CyanogenMod for one is highly non-standard Android “Rom”, and lots of software there ruins Android experience (simply because if you get a device w/o CyanogenMod support, there’s no choice to install separate CyanogenMod tools on it, so if you grew to depend on them, you won’t be able to comfortably use normal Android device, and your experience will be ruined). Now CyanogenMod will even unconditionally spy on you (https://blogs.fsfe.org/torsten.grote/2013/04/03/cyanogenmod-removes-tracking-opt-out-reverses-decision-still-uses-google-analytics/), something which even Google wasn’t accused of. So, definitely enjoy CM.