| 71 | // cpu function of batchedQR |
| 72 | template <typename T> |
| 73 | void batchedQR( |
| 74 | DeviceDnTen<T>& d_Q, DeviceDnTen<T>& d_A, size_s cols, size_s batchSize, |
| 75 | const cudaStream_t& stream = (cudaStream_t) 0, int block_size = 1024 |
| 76 | ) { |
| 77 | int num_block = (batchSize + block_size - 1) / block_size; |
| 78 | batchedQRKernel<<<num_block, block_size, 0, stream>>>( |
| 79 | d_Q.vals, d_A.vals, d_A.dimensions[0], cols, batchSize |
| 80 | ); |
| 81 | return; |
| 82 | } |
| 83 | |
| 84 | #endif // BATCHEDQR_H |