Today NVIDIA began shipping a new product, the Jetson TX2 Development Kit. The Jetson TX2 is available on Amazon. Looky here:
Jetson TX2 Overview
The Jetson TX2 is a new iteration of the Jetson Development Kit which doubles the computing power and power efficiency of the earlier Jetson TX1.
The Jetson TX1 Dev Kit introduced a new module format, where a standardized Tegra Module is plugged into a carrier board. While the Jetson TX2 uses the same carrier board as the Jetson TX1, the actual Tegra TX2 Module itself is all new.
Hardware
The Jetson TX2 features a NVIDIA Pascal GPU with 256 CUDA capable cores. The CPU complex consists of two ARM v8 64 bit CPU clusters which are connected by a high-performance coherent interconnect fabric. The Denver 2 (Dual-Core) CPU cluster is optimized for higher single-thread performance; the second CPU cluster is an ARM Cortex-A57 QuadCore which is better suited for multi-threaded applications.
The memory subsystem incorporates a 128-bit memory controller, which provides high bandwidth LPDDR4 support. 8 GB LPDDR4 Main Memory and 32 GB eMMC Flash memory are integrated on the Module. Going to a 128-bit design from the TX1 64-bit is a major performance enhancement.
The Module also supports hardware video encoders and decoders which support 4K ultra-high-definition video at 60 fps in several different formats. This is slightly different than the hybrid Jetson TX1 module, which used both dedicated hardware and software which was running on the Tegra SoC for those tasks. Also included is an Audio Processing Engine with full hardware support for multi-channel audio.
The Jetson TX2 supports Wi-Fi and Bluetooth wireless connectivity. Wi-fi is much improved over the earlier Jetson TX1. Gigabit Ethernet BASE-T is included. Here’s a comparison between the TX1 and the TX2.
The carrier board, which is common between both the Jetson TX2 and the Jetson TX1, has the following I/O connectors:
- USB 3.0 Type A
- USB 2.0 Micro AB (supports recovery and host mode)
- HDMI
- M.2 Key E
- PCI-E x4
- Gigabit Ethernet
- Full size SD card reader
- SATA data+power
- Display expansion header
- Camera expansion header
There are two expansion headers, a 40 pin, 2.54mm spaced header with signals laid out similarly to the Raspberry Pi, and a 30 pin, 2.54mm spaced header for extra GPIO.
The Jetson also includes a 5MP camera in the camera expansion header, and a display expansion header for adding extra display panels.
The Jetson TX2 has added a CAN bus controller to the module. CAN is a network format that is frequently used in automobiles and other vehicles. The CAN bus signals are available directly on the GPIO Expansion Header.
Sippy or Speedy
This new generation brings a configurable amount of performance increase depending on power consumption requirements. NVIDIA has engineered two modes. Max-Q is the name of the energy efficiency mode which clocks the Parker SoC for efficiency over performance and draws about 7.5W, right before the bend in the power/performance curve. The result of this mode is that the TX2 has similar performance to a TX1 in max performance mode, while drawing about half the power!
In Max-P mode, the TX2 just flat out goes for it in the power budget of 15W. This provides about twice the performance of the Jetson TX1 at its maximum clock rate.
Software
There are several changes to the Jetson TX2 software stack. The Jetson TX2 runs a Developer Preview of an Ubuntu 16.04 variant named L4T 27.1. The Linux Kernel is 4.4, a newer version than the earlier Jetson TX1 version 3.10. There have been changes to the boot flow, with additional firmware managers added to the mix. The Jetson TX2 comes with a long list of software libraries, and a good selection of samples with source code.
The new JetPack 3.0 installer is available to flash and copy system software to the Jetson TX2.
Initial Impressions
NVIDIA claims that the Jetson TX2 is twice as fast as the Jetson TX1. After booting the machine, this surely seems the case. The entire experience feels very much like a desktop/laptop level machine. Doubling the memory (and the memory bus speed) surely helps with that feeling. Previous Jetsons experience quite a bit of memory pressure when running memory intensive, desktop applications like web browsers. The TX2 doesn’t even notice.
Running a handful of compiles and tests on applications like Caffe proved that the Jetson TX2 is indeed quite a bit faster than the earlier Jetson TX1 (see the video for one of the tests).
One of the fun samples that comes with the Jetson TX2 is an object recognition example which is demonstrated in the video. The deep learning sample uses Caffe along with ImageNet and uses the onboard camera to grab imagery.
Note that we haven’t performed any performance tuning for the demos, this is how it runs fresh out the box!
If you want some hardcore numbers, go over to Phoronix and check out NVIDIA Jetson TX2 Linux Benchmarks
Conclusion
Stay tuned as we begin working with the TX2 to better understand how to take advantage of the extra performance. Find out more on the NVIDIA Developers site.
Pictures, Natch!
37 Responses
Hi,
Do you have recommendation for drone to use it with Jetson TX2 development kit with less hardware work and easily drone fly control from TX2?
Hi Ram,
That’s a good question to ask in the Jetson TX2 forum: https://devtalk.nvidia.com/default/board/188/jetson-tx2/
There are people who already manufacture drones with Jetson TX modules, perhaps that will suit your needs. For example:
https://tealdrones.com
http://enroute1.com/
Thanks for the response to earlier queries. Because you have access to both TX1 and TX2 hardware, I think you are in a better position to answer following question:
Does the software written for TX2 is backward compatible with TX1 except hardware specific features?
The answer is a qualified yes. The TX2 is running L4T 27.1 with a 4.4 kernel, the TX1 is currently on L4T 24.2.1 and a 3.10 kernel. I believe that they plan to update the TX1 to 27.1 and 4.4 in coming releases. Other than that, they are both the same.
i want to stream video from the hololens, and other sources. can i do that with the tx1? i read something about gstreamer but that seems to run only on the device.
what do i have to think about when trying to stream from a hololens for example.
also is all the software for tx2 available on the tx1?
I don’t have any experience with Hololens. Please ask this question on the official NVIDIA Jetson Development forums where a large group of developers and NVIDIA engineers share their knowledge. The forum is located at: https://devtalk.nvidia.com/default/board/139/
Hi,
Today I received TX2 dev board and setup was straight forward and also your video helped me. Thank you for that. Problem is with Wi-Fi connection. it shows my broadband Wi-Fi network but would not connect with correct password. Do you have similar experience with on board Wi-Fi? with Cable it was straight forward method to connect to internet. I would like to switch to Wi-Fi for flexibility. any suggestions to use on board Wi-Fi?
Hi Ram,
Please ask support questions in the official NVIDIA Jetson TX2 support forum: https://devtalk.nvidia.com/default/board/188/jetson-tx2/
That way you’ll get a wider community of people, including NVIDIA engineers, to help.
Hi,
Is it possible to attach the TX2 module to an existing TX1 carrier board?
Thanks.
To be honest, I don’t know. You should be able to get a definitive answer in the Jetson TX2 forum where there are some NVIDIA engineers monitoring the questions: https://devtalk.nvidia.com/default/board/188/jetson-tx2/
Thanks for reading!
Hi,
Do you know estimated price of Jetson TX2 SoC, not complete board?
For example in the US: https://www.arrow.com/en/products/900-83310-0001-000/nvidia
That is for Single board computer module. Don’t they sell just SOC?
From what I understand, they do not. I’m sure if you’re a large enough customer or have high enough quantity needs that they might make exceptions. You can contact NVIDIA directly for more information.
Sorry formas any inconvenience, but I would kike to answer you if a jetson tx2 is a recommended board for an student who are going to do her final project of an electronical engineering. She loves everything which is related to artificial vision and similar. Maybe a jetson is too complicated.
Are there any other better sbc for her.
It would be a present for her.
Thank you for your help.
Sorry there are two mistakes, formas is for and instead of answer is quiestion.
The website do not allow me to edit the comment
The Jetson TX2 is a good choice for vision algorithms on embedded systems. Thanks for reading!
Hi Marco,
Raspberry pi is a good board for students to do computer Vision, Robotics, etc.,
We are looking for TX1 or TX2 developers for a Blockchain/Bitcoin Mining Project. Do you know of any resources?
You can ask on the Jetson Developers forum: https://devtalk.nvidia.com/default/board/139/jetson-embedded-systems/
Thanks!
Why would someone chooses Jetson over TitanX GPU cards for mining.Jetson are targeted for deep Learning inference in which training the model with large data set happens on TitanX kind of system and after that model will be transferred to the Jetson. I think same rule applies to your task.
Good question, @ram. In our blockchain app, custom ASIC chip is ultimate goal. However, SW control and proprietary algos are used in conjunction to accelerate mining at lower power consumption. You are welcome to download our WP which describes project and router specs here: http://ammbr.com
If your ultimate goal is to build ASIC then I would recommend Xilinx Zynq SoC (ARM mulcicore CPUs + FPGA) for offloading algorithms to FPGA by writing code in C/C++ with their HLS tool.For my work, I started with Jetson TX2 and then switched to Xilinx Zynq SoC. The problem is with Jetson Tx2 is that Nvidia only supports Linux for their CUDA, not bare metal system. Note: I am not complaining about it, it does not meet my requirement. I would like to have flexibility to run either bare metal system or RTOS.
@ram thanks for the info. Yes, we are aware of vendor lockin here. We are in evaluation mode with a few vendors, including Xilinx. Your suggested approach will be examined further. Stay tuned and stay in touch.
Hi All,
I have Jetson TX2 module and Jetson TK1 development kit. Is there any problem if I substitute TK1 module from the development kit with TX2 module so that I can develop on TX2 instead of TK1? Or the carrier board of TK1 is not compatible with TX2 and I have to buy another development kit?
Thank you
The carrier board for the Jetson TK1 is not compatible with the Jetson TX2 module. Thanks for reading!
This is just an unbelievably bad*ss machine. I plugged a 256GB SATA SSD into it and now I use it as a GP machine. Great to develop on.
I think it’s an inflection point in computing. The processors/GPU like the Jetson and high end smartphones are equivalent to what a desktop was a few short years ago. Which means that you can develop on the machine itself and not have to go through all the emulation headache. Development becomes easier and more fun, which in turn can lead to better applications.
Hi All,
I was wondering how we can connect a smartphone lcd with MIPI DSI input to the Jetson, there are 2 DSIs on the Jetson TX1/TX2 but I don’t know how we can use this interface on the developer board(I know we can use HDMI but I want to use DSI)
Thanks
I do not have any experience with that type of project. You will probably reach a wider developer audience, along with some NVIDIA engineers, by posting this question in the official NVIDIA Jetson TX2 forum: https://devtalk.nvidia.com/default/board/188/jetson-tx2/
Thanks for reading!
Your description of the CPUs, GPUs, and the memory controller, etc. in the Hardware section was the first instance when I actually understood what is available in the Jetsons TX2. So clear and very simply explained.
Thank you.
I am glad you found the explanation useful. Thanks for reading!
Thank you for your videos,
Do you have any article or video on how to use Jetson TX2, I opened the box like you did, but how to proceed from there ?
Please ask this question on the official NVIDIA Jetson forum where a large number of developers and NVIDIA engineers share their experience. The forum is here: https://devtalk.nvidia.com/default/board/188/jetson-tx2/
Is there an alternative to NVIDIA Jetson Tx1 Development Kit?
The NVIDIA Jetson TX2 Development Kit and the newly announced NVIDIA Jetson Xavier Kit. Thanks for reading!