Nvidia will award 50 of its 2.3GHz Tegra K1-based “Jetson TK1” SBCs to winners of a “CUDA Vision Challenge,” but all entries must be received by April 30.
Nvidia unveiled the $192 Linux-based Jetson TK1 single-board computer, touted as the “world’s first embedded supercomputer,” in January, and demoed its use in a self-driving Audi. The board’s Tegra K1 SoC integrates four ARM Cortex-A15 cores, a 192-core Mobile Kepler GPU, and an ARM7 power management core.
Nvidia Jetson TK1 development board
(click image to enlarge)
Nvidia claims the 5.0 x 5.0-inch Jetson TK1 board can pump out more than 300 gigaflops of processing power — “nearly three times” that of “any similar embedded platform,” says the company. In addition to its potent Tegra K1 SoC, the SBC comes with 2GB RAM (expandable to 4GB) and 16GB eMMC flash storage, plus a generous helping of I/O ports including HDMI, USB 3.0, SATA, and GPIO.
Jetson TK1 SBC details, block diagram
(click images to enlarge)
The Jetson TK1 Developer Kit includes a developer tools supporting the Tegra K1’s 192-core Mobile Kepler GPU, including a version of the C/C++ “VisionWorks” toolkit based on the CUDA 6.0 architecture. The tools include debuggers and profilers to develop massively parallel applications. Nvidia claims the included accelerated libraries for FFTs, linear algebra, sparse matrix, and image and video processing make the Mobile Kepler GPU “much easier to program” than typical FPGAs, custom ASICs, and DSPs. The also Developer Kit also includes a ready-to-boot Ubuntu desktop OS.
Entering the Challenge
Entrants in the Tegra K1 CUDA Vision Challenge are asked to submit a brief description (500 words or less) of an embedded application idea based on the TK1 SBC by April 30. Winners be selected based their idea’s “innovation, impact on research or industry, public availability, and quality of work,” says Nvidia.
Nvidia will announce 50 winners by the end of May. Each winner will receive “one of the first Jetson TK1 DevKits to roll off the production line,” along with access to various technical support “documents and assets,” says the company. Additionally, the five “most noteworthy Jetson TK1 breakthroughs” may be offered the opportunity to present their projects at next year’s Nvidia GPU Technology Conference.
Further details and a form for submitting entries are available at Nvidia’s Tegra K1 Vision Challenge page.