| KEYWORD |
Software development techniques for resilient embedded GPUs
keywords GPU, NVIDIA, ACCELERATOR, GPU, PEDESTRIAN RECOGNITION, DEPENDABILITY, SAFETY
Reference persons LUCA STERPONE
Research Groups GR-05 - ELECTRONIC CAD & RELIABILITY GROUP - CAD
Description Autonomous driving is today largely supported by embedded high performance General Purpose Graphical Processing Unit (GPGPU) that executes image elaboration and control algorithms enabling self-driving cars. The activity of this thesis, done in collaboration with NVIDIA, is based on the development of applications and analysis of errors by the development of new fault injection methods on one of the state-of-the-art NVIDIA device (Jetson nano) which has a Maxwell NVIDIA architecture with 128 cores.
Required skills C, C++, basic Knowledge of the GPU architecture
Deadline 20/09/2019
PROPONI LA TUA CANDIDATURA