| | |
| | | |
| | | #ifdef GPU |
| | | void axpy_ongpu(int N, float ALPHA, cl_mem X, int INCX, cl_mem Y, int INCY); |
| | | void axpy_ongpu_offset(int N, float ALPHA, cl_mem X, int OFFX, int INCX, cl_mem Y, int OFFY, int INCY); |
| | | void copy_ongpu(int N, cl_mem X, int INCX, cl_mem Y, int INCY); |
| | | void copy_ongpu_offset(int N, cl_mem X, int OFFX, int INCX, cl_mem Y, int OFFY, int INCY); |
| | | void scal_ongpu(int N, float ALPHA, cl_mem X, int INCX); |
| | | void im2col_ongpu(cl_mem data_im, int batch, |
| | | int channels, int height, int width, |
| | |
| | | int channels, int height, int width, |
| | | int ksize, int stride, int pad, float *data_col); |
| | | |
| | | void gemm_ongpu_offset(int TA, int TB, int M, int N, int K, float ALPHA, |
| | | cl_mem A_gpu, int a_off, int lda, |
| | | cl_mem B_gpu, int b_off, int ldb, |
| | | float BETA, |
| | | cl_mem C_gpu, int c_off, int ldc); |
| | | |
| | | void gemm_ongpu(int TA, int TB, int M, int N, int K, float ALPHA, |
| | | cl_mem A_gpu, int lda, |
| | | cl_mem B_gpu, int ldb, |