US Patent:
20200293595, Sep 17, 2020
Inventors:
- New York NY, US
Hai Li - New York NY, US
Xiaoyang Liu - New York NY, US
International Classification:
G06F 17/14
G06T 1/20
G06F 12/06
G06F 17/16
Abstract:
The present disclosure provides a CPU-based third-order low-rank tensor calculation method, Operation steps of the method include: transmitting, by a CPU, s third-order real value tensor input data DATA to a CPU; performing, by the GPU, Fourier transforms on the DATA to obtain third-order complex value tensor data DATA performing, by the CPU, matrix operations on the DATA to obtain third-order complex value tensor data DATA performing, by the CPU, inverse Fourier transforms on the DATA to obtain third-order real value tensor output to data DATA and transmitting, by the GPU, the DATA to the CPU. In the present disclosure, in the third-order low-rank tensor calculation, a computational task with high concurrent processes is accelerated by using the CPU to improve computational efficiency. Compared with conventional CPU-based third-order low-rank tensor calculation, computational efficiency is significantly improved, is and same calculation can be completed by using less time.
