UE E9TS-D - Tools and algorithms

Level of knowledge :

N1: beginner
N2: intermediate
N3: advanced
N4: expert

The knowledge expected at the end of the course

Acquire notions concerning pointers in C language in signal processing algorithms in the broadest sense: (C1, N3)
Know the syntaxes allowing the implementation of data and function pointers in C language: (C1, N4)
Know the different methods of accelerating computing algorithms, GPU architecture and the performance achievable by GPUs compared with CPUs: (C1, N2)
Know the principles of implementing parallel processing accelerated on GPUs using CUDA: (C1, N3)
Know the different block-matching algorithms (search strategies and similarity criteria): (C1, N4)

Learning outcomes in terms of abilities, skills and attitudes expected at the end of the course

Master the implementation of pointers in C language in signal processing algorithms in the broadest sense: (C2, N4)
Analyze, correct and develop an implementation using pointers in C language: (C2, N3)
Implement classical signal processing algorithms in the broadest sense in CUDA: (C2, N3)
Measure the execution times of the various blocks of a video compression chain: (C2, N3)
Evaluate the performance obtained (execution time and quality) by replacing a motion vector calculation block of high computational complexity with an adapted alternative algorithm: (C2, N4)
Present a time/precision compromise: (C5, N3)