Research Article

Multi-Softcore Architecture on FPGA

Table 8

Performance results of Mali-T604 GPU and 16-core architecture. Numbers in brackets denote the difference between GPU’s and our architecture’s values.

Application GPU 16-core configuration
Exec. time % peak perf. Exec. time % peak perf.

FIR filter
−32 taps, 131 072 samples
655 s 18.8% 461 s
(29% faster)
84.2%
(4.5x better)

Matrix-matrix multiplication
−256 × 256 matrix size
2 634 s 18.7% 2 011 s
(23% faster)
77.37%
(4x better)