Accelarated FFT Implementation using Tensilica TIE Extension
This project contains:
- reference C implementation of DIT and DIF algorithm (fft and inverse fft)
- accelerated implementation using TIE extensions (fused operations, SIMD, FLIX)
Performance Results for DIT algorithm
The following results apply for compilation with optimization level -O2 and feedback optimization.
- N=8: 243
- N=256: 9533
- N=1024: 41573
The speedup in comparison to the c implementation: 10-20
Note: This task was done for the HW/SW Codesign Lab at TU Dresden.