Is it possible to run multipathnet on Nvidia Jetson TK1?

[nara007]

The compute capability of this board is 3.2，but requirements of multipathnet claims for compute capability 3.5+. If i install multipathnet on this board, what will happen? Thanks !

I have not jet bought this board.

some information about JetsonTK1:

NVIDIA Kepler GPU with 192 CUDA cores
• NVIDIA 4-Plus-1 quad-core ARM Cortex-A15 CPU
• 2 GB x 16 memory with 64-bit width and 16 GB 4.51 eMMC memory

Tegra K1 SoC

• NVIDIA Kepler GPU with 192 CUDA cores
• NVIDIA 4-Plus-1 quad-core ARM Cortex-A15 CPU
• 2 GB x 16 memory with 64-bit width
• 16 GB 4.51 eMMC memory
• Half mini-PCIE slot 1
• Full size SD/MMC connector
• 1 USB 2.0 port, micro AB 1
• 1 Full-size HDMI port
• RS232 serial port
• 1 ALC5639 Realtek Audio codec with Mic in and Line out
• 1 RTL8111GS Realtek GigE LAN
• 1 SATA data port
• SPI 4MByte boot flash

CUDA Developer Information

• CUDA Version: 6.0
• CUDA Cores:

Computational Capability: sm_32
Number of cores: 192

• CUDA libraries:
cudart, cufft, cublas, curand, cusparse, npp, opencv4tegra for registered developers
Visionworks: available on request

• CUDA tools:

for local development, all the command line tools (compiler, cuda-gdb, cuda-memcheck, command-line profiler
for remote development, all the command-line tools and the visual tools too (NSight Eclipse Edition, Visual Profiler)

[szagoruyko]

Hi, never used Jetson, if it fully supports Torch and CUDNN it's likely to run mulipathnet too.