Searched refs:grid_dim (Results 1 – 3 of 3) sorted by relevance
/external/tensorflow/tensorflow/core/kernels/ |
D | reduction_gpu_kernels.cu.h | 622 dim3 grid_dim(1, 627 grid_dim.y = std::min((int)grid_dim.y, 32); 629 if (grid_dim.y > 2 && grid_dim.y < 32) { 630 int log2 = Log2Floor(grid_dim.y); 631 grid_dim.y = 1 << log2; 634 if (grid_dim.y == 1) { 636 ColumnReduceMax16ColumnsKernel<IN_T, OUT_T, Op>, grid_dim, block_dim, 0, 643 sizeof(T) * extent_y * grid_dim.y)}), 646 grid_dim, block_dim, 0, cu_stream, in, 650 dim3 new_grid_dim((grid_dim.y * extent_y + 31) / 32, 1, 1); [all …]
|
/external/tensorflow/tensorflow/core/util/ |
D | cuda_launch_config.h | 331 Status CudaLaunchKernel(void (*function)(Ts...), dim3 grid_dim, dim3 block_dim, in CudaLaunchKernel() argument 340 auto result = cudaLaunchKernel(func_ptr, grid_dim, block_dim, arg_ptrs.data(), in CudaLaunchKernel()
|
/external/tensorflow/tensorflow/contrib/rnn/kernels/ |
D | lstm_ops_gpu.cu.cc | 243 const int grid_dim = in LSTMBlockCellFpropWithCUDA() local 245 TF_CHECK_OK(CudaLaunchKernel(concat_xh<T>, grid_dim, block_dim, 0, cu_stream, in LSTMBlockCellFpropWithCUDA()
|