Nvidia comes with Jetson-TX1 development kit based on X1-soc




Nvidia has released a module for a self-learning system, the Jetson TX1. The module of 50x87mm was developed for GPUs based on the Maxwell-architecture. The company also introduces two new high-performance Tesla cards in the series.

The Jetson TX1 is designed to run artificial neural networks include computer vision, machine learning and navigation. Therefore the jewel offers one teraflops of computing power thanks to the Maxwell-architecture-based GPU with 256 CUDA cores. The module of the CPU is a soc with cores based on ARM A57 architecture.

With the GPU can encode the TX1 and 4K video decoding and a camera readout 1400 megapixels per second. The internal memory is of 4GB lpddr4 along with 16GB EMMC storage. A remote connection may be obtained via WiFi, Bluetooth or Gigabit Ethernet interface. It runs on Linux for Tegra.

nvidia jetson TX1

The purpose of the modules is especially the integration in robotic systems, such as partially autonomous drones. The integration into other types of robots is obvious, writes Nvidia in a notice on its website. The Jetson TX1-sdk for visual computing includes several Cuda libraries and frameworks. There is also support for OpenGL 4.5, OpenGL ES 3.1 and Vulkan and Cuda 7.0, which the GPU can be used as a general purpose processor.

The Jetson TX1 Developer Kit is available for $ 599 Thursday, with the version for schools will cost $ 299. The module will be available in early 2016 at a cost of $ 299 per unit for a minimum order of one thousand pieces.

Nvidia Jetson TX1
Nvidia Jetson TX1
Nvidia Jetson TX1
Nvidia Jetson TX1
Nvidia Jetson TX1
Nvidia Jetson TX1

Besides the TX1 announced Nvidia Tesla also two new accelerators to, among other calculations intended for artificial intelligence, the Tesla M4 and M40. The Tesla M4 card uses little energy, between 50 and 75watt, and is designed with data centers in mind. The property has 1024 CUDA cores, 4GB GDDR5 memory bandwidth of 88GB per second and can handle peaks of up to 2.2 teraflops.

The Tesla M40 however, is a lot less energy efficient and uses up to 250watt with a peak power of 7 teraflops. That reaches the accelerator to have the use of 3072 Cuda cores, 12 GB of GDDR5 memory with a bandwidth of 288GB per second. Both cards are based on the Maxwell architecture. The Tesla M40 is available end of 2015, the M4 early 2016. Both accelerators is no price announced.



In: Technology & Gadgets Asked By: [15572 Red Star Level]

Answer this Question

You must be Logged In to post an Answer.

Not a member yet? Sign Up Now »