Home > Rockchip RK29XX, Rockchip RK30xx, Rockchip RK31xx, Rockchip RK32xx > Rockchip SoCs Comparison Table, RK32xx Quad Core Cortex A12 Coming Up in 2014

Rockchip SoCs Comparison Table, RK32xx Quad Core Cortex A12 Coming Up in 2014

They guys at Linux Rockchip have setup a comparison table between Rockchip processors, but it’s not quite complete yet, and Rockchip noticed it, so they’ve just sent them the most recent Company Presentation that provides more details, including a roadmap.

In this documentation, we can see the company targets different types of products with their SoCs:

  • Android OTT dongle/box (aka mini PC, HDMI TV dongles…) – RK2928, RK3066, RK3068, and RK3188
  • Tablets – RK2926, RK2928, RK3028, RK3066, RK3168, and RK3188
  • EPD E-Book – RK2818, RK2906 and RK2918
  • MP3 player – RKnanoC

Rockchip_Product_Structure

On page 12 of the presentation, they have a convenient comparison table for the SoC used in mini PCs and Tablets.

Series RK2926 /RK2928 RK3066 /RK3068 RK3168 RK3188
Process 55nm 40nm 28nm HKGM 28nm HKGM
CPU Cortex A9 Dual Cortex A9 Dual Cortex A9 Quad Cortex A9
MAX CPU Frequency 1.0 GHz 1.6 GHz 1.2 GHz 1.6 GHz
L1 32KB 32KB 32KB 32KB
L2 128KB 512KB 256KB 512KB
GPU Mali 400 Mali-400 MP4 PowerVR 540 Mali-400 MP4
MAX GPU Frequency 330 MHz 400 MHz 600 MHz 600 MHz
Memory DDR3/L LPDDR2 DDR3/L LPDDR2 DDR3/L LPDDR2 DDR3/L
FLASH support eMMC, MLC, NAND eMMC, MLC, NAND eMMC, MLC, NAND eMMC, MLC, NAND
Video Decoder Multi-Format, VP8, H.264 Multi-Format, VP8, H.264 MVC Multi-Format, VP8, H.264 Multi-Format, VP8, H.264, MVC
Video Encoder H.264 H.264, VP8, MVC H.264 H.264, VP8, MVC

We also learn Rockchip is working on a new RK32xx SoC (RK3288?) that will become available in Q2 2014 (Engineering Samples) with four of the latest ARM Cortex 12 cores, and the following features:

  • Rockchip_RK32xx_Block_Diagram28nm HKMG Process
  • 1.8GHz Quad ARM A12 core
  • ARM Mali-T624 GPU, support OPENGL ES1.1/2.0/Halti, OPEN VG1.1, OPENCL, Directx11
  • High performance dedicated 2D processor
  • Dual channels DDR controller, support, DDR3, DDR3L, LPDDR2, LPDDR3
  • Display resolution up to 3840X2160, MIPI, LVDS, HDMI transmitter
  • Embed 1300M ISP and MIPI-CSI interface
  • 4K video multi format video decoding
  • 1080P video encoding for H.264 and VP8, MVC
  • Support MLC NAND, eMMC
  • Image enhancement HW
  • Low power display improvement HW
  • Rich peripheral and connectivity

Interestingly, there’s no hardware support for H.265/HEVC and VP9 even in this 2014 processor, so they’ll have to decode these codecs by software, just like they plan to do with RK3188 up to 1080p.

Other interesting bits (to me) include:

  • RK3168 and RK3188 are pin-to-pin compatible
  • RK3068 has more pins than RK3066, yet it’s much smaller thanks to a smaller pitch (BGA548 14X14 0.5mm pitch vs BGA453 19X19, 0.8mm pitch)
  • RK292X allows for low cost TV-BOX solution (“wireless HDMI”) with a PCBA cost of about $10 US.
Digg This
Reddit This
Stumble Now!
Buzz This
Vote on DZone
Share on Facebook
Bookmark this on Delicious
Kick It on DotNetKicks.com
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter

  1. onebir
    July 27th, 2013 at 13:28 | #1

    I wondered:
    - why the RK3188 can be clocked faster than the RK3168, when they both use the same process. (& presumably 4 cores means more heat…)
    - what kind of performance improvements over the RK3188 might come from shifting to ARM Cortex 12?

  2. mj
    July 27th, 2013 at 15:53 | #2

    @onebir
    >why the RK3188 can be clocked faster than the RK3168
    quality control/binning ? RK3168 are bad RK3188 ? (Intel does that)
    power consumption ?

    Cortex 12 has interesting stuff, like GPGPU and virtualization. and more RAM. So nice little servers. Remember that AllWinner Eva said one time to charbax that AW was working on a “server” CPU. I am sure Rockchip is preparing themselves too. Plus it looks “better” that A7 :), since it A12. So better than AW and MTK !

  3. July 27th, 2013 at 16:12 | #3

    @onebir
    Maybe lower frequency means better yields, and better prices?

    Cortex A12 processor will provide 40 percent more performance than Cortex-A9 at the same frequency, and there are also the new features mentioned by mj.

  4. Galllus
    July 27th, 2013 at 16:13 | #4

    Atleast no Mali-400 anymore :)

  5. anon
    July 27th, 2013 at 16:53 | #5

    I wonder will they still use the CedarX video codec with their otherwise ARM’s techs (Cortex-A12 + Mali-T624)?

    Sure the ARM Mali-V500 also lacks h.265, but at least that way it would become “all in-house ARM” based SoC.

    Not really sure why people are calling the V500 a DRM nightmare, as ALL hardcoded video codec blocks are equally DRM-ridden. People are just bummed out that they still do not get a DRM-free video acceleration? Unfortunately that will never happen by using MPEG-LA standards.

  6. vtomin
    July 27th, 2013 at 18:14 | #6

    Just a small correction: They say Q2 2014 for RK32xx not Q4. :)

  7. July 27th, 2013 at 18:19 | #7

    @vtomin
    Thanks. I’ve also updated the post to say Q2 2014 is for engineering samples.

  8. Andrew
    July 27th, 2013 at 22:09 | #8

    So Rockchip will continue to be a year (or more) behind. Snapdragon 800 is about to be released. I expect Rockchip won’t be matching it’s performance for at least two more years.

  9. July 28th, 2013 at 09:53 | #9

    @Andrew
    It seems Rockchip is going its own way.. No cortex A7, no Cortex A15, contrary to most other companies.

  10. Rocker
    July 29th, 2013 at 22:28 | #10

    anon :
    I wonder will they still use the CedarX video codec with their otherwise ARM’s techs (Cortex-A12 + Mali-T624)?

    CedarX is only used by Allwinner SoCs, not by Rockchip.

    Rockchip uses Stagefright API (libstagefright) on Android.

    And I do not think Rockchip officially supports Linux.

  11. anon
    July 31st, 2013 at 07:48 | #11

    @Rocker
    Whoops, my Allwinner past creeps up on me. ;-)

    Anywho, I have always thought that the stagefright is nothing but an API interface to the hardware itself, so the actual hardware block is something else, have not yet looked up libstagefright code does it actually do “all” work with direct access to the VPU, or is it just an common API to some (interchangeable) binary blob that interfaces to the VPU.

  1. July 27th, 2013 at 15:29 | #1
  2. July 28th, 2013 at 16:13 | #2
  3. January 4th, 2014 at 14:45 | #3
  4. January 10th, 2014 at 19:55 | #4
  5. January 11th, 2014 at 23:30 | #5
  6. March 14th, 2014 at 04:47 | #6
  7. August 17th, 2014 at 10:27 | #7