Somebody working for a Chinese manufacturer informed me Rockchip RK3288 specifications have been released on rk3288.com. We’ve already known the application processor was based on a quad core Cortex A17 processor with a Mali-T76X GPU, but this website provides some more details which appears to have been taken from RK3288 technical reference manual. I can’t confirm the source nor the accuracy of the data, but I doubt somebody would have just put up something with detailed fake information.
They mention RK3288 is an application processor destined to be used in high-end tablets, notebooks, all-in-one devices, smart monitors and TV Boxes, and that it features 4Kx2K support.. Let’s have a look at the details.
- CPU
- Quad-Core Cortex-A17
- Separately Integrated Neon and FPU per CPU
- 32KB/32KB L1 ICache/DCache per CPU
- Unified 1MB L2 Cache
- LPAE (Large Physical Address Extensions), support up to 8GB address space Virtualization Extensions Support
- DVFS support
- 3D GPU
- Quad-Core Mali-T7 series
- Support OpenGL ES 1.1/2.0/3.0, OpenVG 1.1, OpenCL 1.1 and Renderscript, Directx11
- DVFS support
- 2D GPU
- Multi-Core architecture
- Up to 8Kx8K input and 4Kx4K
- Output High-quality image scale
- Up/down Dither operation
- Image rotation with 90/180/270 degree or x/y-mirror BitBLT, Alpha Blending,Raster Operation
- VPU
- Video Decoder
- Support MPEG-2, MPEG-4, AVS, VC-1, VP8, MVC with up to 1080p@60fps (cnxsoft: It does not look correct or complete, as Rockchip showed slides with 4K2K H.265 decoder at CES 2014).
- Support multi-format video decoder with up to 4Kx2K
- High-quality deinterleave
- Video Encoder – Support muti-format video encoder with up to 1080p@30fps
- Video Decoder
- Video Input
- Dual-channel input for front and rear camera
- Dual-channel MIPI-CSI2 interface with 4-lane per channel
- 8/10/12 bits standard DVP interface
- Maximum 5MP for front camera
- Maximum 13MP for rear camera with high-performance ISP
- Video Display
- Dual-panel display with 2 separately interface
- Maximum resolution is 4Kx2K
- CABC support to decrease interface power
- Dual channel 8/10bits LVDS
- Dual channel MIPI-DSI
- HDMI 2.0 to support maximum 4Kx2K display
- Optional eDP1.1 interface
- Memory Interface
- Nand Flash Interface
- Dual-channel , 8bits per channel
- Compatible with all of SLC/MLC/TLC Nand Flash, including DDR Nand
- Embedded 60-bit hardware ECC
- 8 chip selects to support more device
- eMMC Interface
- compatible with eMMC4.5 standard 8-bit data width
- Support DDR-50, SDR-100
- DDR interface
- Dual channel 64bits interface
- Support DDR3/DDR3L/LPDDR2/LPDDR3
- Nand Flash Interface
- Peripherals and I/Os
- 3 SD/MMC/SDIO interfaces, compatible with SD3.0, SDIO3.0 and MMC4.5
- One 8-channels I2S/PCM interface, One 8-channels SPDIF interface
- One USB 2.0 OTG , Two USB 2.0 Host
- 100M/1000M RMII/RGMII Ethernet interface
- Dual channel TS stream interface, descramble and demux support
- Smart Card interface
- GPS baseband interface
- PS/2 master interface
- 5 UART, 3 SPI (master or slave), 6 I2C(up to 4Mbps), 5 PWM
- Misc
- Standalone crypto and decrypto, compatible with AES 128bits/DES/3DES/SHA-1/SHA-256/MD5/160bits PRNG
- Full security solution to support HDCP2.x,secure boot, secure debug and DRM
- Temperature Sensor to support better temperature controll inside chip
The CPU frequency has not been mentioned, but earlier reports indicate the four Cortex A17 core should be clocked at up to 1.8 GHz. The GPU is allegedly Mali-T760 or T-764, the latter does not exist yet, or has not announced yet. However, numbers previously leaked showing 300 MTri/s triangle rate and 2.4Gpix/s fill rate conflict with the 1390Mtri/s, and 11.2Gpix/s listed on ARM Mali-T760 page. My guess is that Mali-T764 does exist, but it has not been announced, and provides a lower cost and performance alternative to Mali-T760. There’s new “article” about Mali-T764 by Enybox, one manufacturer of TV box and HDMI TV sticks.
There are some nice features from the specifications like LPAE support for up to 8 GB RAM, Gigabit Ethernet support, HDMI 2.0 for True 4K2K up to 60 Hz, 4K2K decoding, but the lack of support for SATA and USB 3.0 is a little disappointing.
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
i hope tronsmart makes a tv box from this very soon
it looks a little fake to me, there is no mention to A17 cores and the rk3288.com website is no longer accessible, wait & see then
Will be a very interesting chip, once mainline Linux Kernel supports it.
@Slackstick
Mainline kernel ? Keep dreaming … Allwinner has a community around it and their SOCs are starting to work with mainline kernel buy Rockchip, they’ve got almost nothing as far as mainline goes. At least not that I know of.
I think information correctly, I am see same info in Ugoos Industrial blog, then asked Ugoos team. They confirmed what information is true and they started to research new android box based on this SoC
This SoC supports 4K2K, just I don’t know about frames. Begining to design Android TV Box soon 😉
HDMI 2.0 to support maximum 4Kx2K display sounds great!
I thought the initial claim of the RK3288 being based on Cortex-A17 was wrong and in fact, it’s the Cortex-A12 (www.cnx-software.com/2014/01/10/rockchip-rk3288-vs-rk3188-performance-comparison/).
The same about the GPU announcement, which was corrected in other articles, reporting about the new chip back in January, from a Mali-T76X to only be a Mali-T624.
Now I am confused.
@Tux
It’s highly confused, the short answer is that Chinese company tend to announce product a bit too early (before specs are frozen), then tend to indirectly announce new CPU and GPU cores from ARM before ARM does it officially…
I can explain the full timeline.
1. July 2013. Rockchip released a marketing document with RK3288 to be based on Cortex A12 and Mali-T624. http://www.cnx-software.com/2013/07/27/rockchip-socs-comparison-table-rk32xx-quad-core-cortex-a12-coming-up-in-2014/
2. 4 Jan 2014 – Some marketing materials surface with Cortex A17, which has not been announced at the time. http://www.cnx-software.com/2014/01/04/what-to-expect-from-allwinner-rockchip-mediatek-and-other-silicon-vendors-at-ces-2014/
3. 10 Jan 2014 – Rockchip has covered their marketing materials with “Cortex A17” with a “Cortex A12” sticker, and they also mention a T76X GPU, with some test showing Mali-T764, which does not exist, or has not been announced, and it also does not match the performance number for T760, so I assumed it must be another mistake, and be Mali-T624 as initially announced.
4. 11 Feb 2014 – ARM officially announces Cortex A17, which probably means the Cortex A17 on Rockchip marketing material was not a mistake after all, they were just not supposed to talk about it until then. This also leads credence to the theory Mali-T764 does exist, but has not been announced yet. http://www.cnx-software.com/2014/02/11/arm-unveils-cortex-a17-processor-first-used-in-mediatek-mt6595-and-rockchip-rk3288-socs/
5. 27 February 2014 (this post), a third party site rk3288.com, possibly made by Ugoos, shows the specs being Cortex A17 and mali-T7 series with Mali-T760 block diagram. It can’t be Mali-T760 with all cores, as it should be faster than the performance figure announced, so it must be Mali-T764, which has yet to be announced (So Rockchip can’t talk about it, which is why they use Mali-T76X).
cnxsoft, thanks for taking the time and clearing things up by providing that nice timeline. Much appreciated 🙂
Some products appeared on Alibaba(large quantity order only):
http://www.alibaba.com/product-detail/Android-MINI-PC-RK3288_1728219033.html
http://www.alibaba.com/product-detail/Android-MINI-PC-RK3288-support-android4_1728156894.html
Link to Rockchip RK3288 technical reference manual -> http://rockchip.fr/RK3288%20TRM/