NVIDIA has expanded its family of compact compute modules with the announcement of the Jetson TX2 NX SO-DIMM solution. The product is designed to handle a variety of artificial intelligence (AI) applications.

The central processor of the novelty combines two NVIDIA Denver 2 cores and four ARM Cortex-A57 MPCore cores with a clock speed of up to 2,0 GHz. A graphics accelerator with NVIDIA Pascal architecture, with 256 CUDA cores, is used.

The equipment includes 4 GB of LPDDR4 RAM and a 5.1 GB eMMC 16 flash drive. There are three PCIe 2.0 lanes through which two devices (one and two lanes) can be connected.

Performance for tasks related to artificial intelligence reaches 1,33 teraflops (trillion floating point operations per second).

We are talking about support for Gigabit Ethernet, SDIO, SPI, UART, I2S, I2C, GPIO interfaces. The module allows you to use up to five cameras. Dimensions are 69,6 × 45 mm.

The novelty is able to cope with the encoding and decoding of video materials in H.265 / H.264 format with 4K resolution and 60 frames per second. The supply voltage is 5 V; operation in 7,5 and 15 W modes is possible.

Samples of the Jetson TX2 NX are now shipping, priced at approximately $ 200.