RDK X5 first hands-on experience! It really is so cool!!!
This Blog was synchronized and published on:
- Groundhog Robotics Developer Forum:
- CSDN：
More than a year ago, I accidentally learned about the RDK X3, after which I was quickly attracted by the excellent ecology of the Horizon Robotics Developers Forum (now renamed as the Gourd Robotics Developers Forum), the beautiful official website interface, detailed development information, a wealth of development cases, as well as robotics development and exchange group Chao and other people in the rapid technical support so that I quickly became a loyal fan of the RDK, the RDK X3, also accompanied me through two competitions to witness my transformation from a novice to the present. RDK X3 also accompanied me through two competitions witnessed me from a small white to today's metamorphosis, heard that to come out of the X5 is very excited about wanting to get an early X5 hands-on experience, and finally also in today's get the RDK X5! Hahahahahahahaha!
RDK X5's price is really very beautiful, debut this time you can also enjoy 100 discount reduction, the price of 4GB as long as 449, 8GB is also as long as 599, the price of this configuration of this ecological comparison of other friends can only say Win~!
Here's a link to buy and a link to the official Groundhog Robot website:
- Groundhog Robotics Developer Kit RDK X5 - Higher Arithmetic Power, Superb Algorithms ()
- DIGGUARD DEVELOPER COMMUNITY Home ()
Then we get to the meat of today's review!!!!
Getting the RDK X5 I was really amazed with the artistic design of the RDK, whether it's the X3 or the X5, this box design is really artistic! It's beautiful! Kudos to the artistic design!
Then open the box to take out our X5, I can only say that the first look in the hand is really too handsome!!!! The CPU in the middle of the glass mirror design, a sense of technology came up, coupled with the layout of the X5 board alignment is simply a beautiful crafts! (Whispering bb: I don't dare to hold it in my hand for fear of sticking fingerprints on it and affecting its face value...)
Let's take a look at the configuration of the X5, compared to the X3 two years ago, the RDK X5's CPU has been upgraded from a 4-core A53 to an 8-core A55, and the BPU's arithmetic power has been upgraded to 10 TOPS along with a
32GFlops of GPU and most of all a CAN port!!!!
I've organized the specific parameters:
| CPU | Octa-Core CortesA55 |
|---|---|
| BPU | 10 TOPS |
| GPU | 32Gflops |
| random access memory (RAM) | 4GB/8GB LPDDR4 |
| save | NA, supports external Micro SD card Peripheral |
| Sensor | 2 x 4-lane MIPI CSI |
| USB Host | 4 x USB 3.0 Host interfaces（Type-A） |
| USB Device | 1 x USB 2.0 Device interface（Type-C） |
| Debug Serial | 1 x Debug serial port（Micro USB） |
| earphones | 1 x 3.5mm headphone jack audio input/output |
| wireless network | Wi-Fi 6 & Bluetooth 5.4 |
| wired network | 1 x Gigabit Ethernet RJ45 port with POE |
| CAN | 1x CAN FD |
| IO | 28 GPIOs (Reusable support 5 x UART, 8 x PWM, 3 x I2C, 2 x SPI, 1x I2S |
| demonstrate | 1 x MIPI DSI 4 Lane，1 x HDMI Type-A port supporting up to 1080p60 |
| System Support | Ubuntu22.04 |
I would have liked to use RDK X5 the latest support for the flash connection technology, but I do not know why my computer download RDK Studio will have an error and then say that I can not find the chip burning tool, but also understand the flash connection is still a just released tool, the future of RDK Studio this piece of groundhog's staff there is still a lot of optimization of space, optimized for a good time after it should be a very high productivity development tools.
Then we used the traditional method to download the image and then burned it with BalenaEtcher to make a TF card, plugged it in and turned on the screen!!!!
Next, let's connect our X5 to the network with a keyboard and mouse, then use MobaXterm to remotely SSH to our X5, and then use the following command to upgrade the apt after connecting:
sudo apt update
sudo apt upgrade
X5 desktop version of the system experience than the X3 a lot better than the CPU upgrade brought about by the change is very obvious, before using the X3 when the desktop version of the system is not overclocked, then the old card card, remote connection to the X3 will also be inexplicable reboot from time to time, but now the X5 can feel the speed of the very smooth after all, the 8-core A55 is still very good da, but really accustomed to overclocking, so it is still copying it by hand! (doge), we first according to the official manual in the frequency management section of the statement, use the following command on the X5 overclocking:
echo 1 >/sys/devices/system/cpu/cpufreq/boost
echo performance >/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
After overclocking, let's experience the X5's 10TOPS power! Since I have not yet matched the official toolchain of the groundnut so we today directly to experience the groundnut official website NodeHub inside the routine bla, first of all, we open the official website of the NodeHub:Groundhog Robotics - NodeHubIn the past, I found the Yolov8 target detection column, according to the official introduction of the super brother on the X5 Yolov8 can run up to a staggering 220FPS! (In the lab for two years, never fought the battle of so much powerwwwwww), we follow the deployment tutorials inside to start step by step configuration (whisper bb: a lot of text error ah)
Let's enter the command step by step:
# Configuration environment
source /opt/tros/humble/
# Configure the type of camera to use（usb or mipi）
export CAM_TYPE=usb
# activate (a plan)launchfile
ros2 launch dnn_node_example dnn_node_example. dnn_example_config_file:=config/ dnn_example_image_width:=480 dnn_example_image_height:=272
Enter after running the printout of the results as above, you can see that the preprocessing part of the 7ms, the reasoning part of the use of 27ms, post-processing took 3ms like, compared to the X3 reasoning part of the time is really a lot faster, and then we open the X5 web interface to view the results of real-time inference, we open the browser in the same network environment, any device, type in the following address can be viewed in X5's real-time reasoning results:
Board IP:8000 #Port is 8000
You can see that the results of our reasoning are still very accurate~
As I still need to attend class today, I can only experience the performance of Yolov8 on X5 superficially, and I will release more experience tips and some projects, you can follow my blog first, so stay tuned!