Searched refs:shared_memory_size (Results 1 – 3 of 3) sorted by relevance
| /external/tensorflow/tensorflow/core/kernels/ |
| D | bias_op_gpu.cu.cc | 211 int32 shared_memory_size = 0; in compute() local
213 shared_memory_size = bias_size * sizeof(typename AccumulatorType<T>::type); in compute()
216 if (shared_memory_size <= max_shared_memory_size) { in compute()
219 <<<config.block_count, config.thread_per_block, shared_memory_size, in compute()
|
| D | topk_op_gpu.h | 383 constexpr auto shared_memory_size = 48 << 10; // 48 KB
386 num_shards = shared_memory_size / heap_size - 1;
402 auto shared_memory_size = (num_shards + 1) * k * sizeof(Entry<T>);
405 shared_memory_size, stream, input, length, k,
|
| D | depthwise_conv_op_gpu.h | 635 const int shared_memory_size = in LaunchDepthwiseConv2dGPUSmall() local
640 num_outputs, device, kernel, shared_memory_size, in LaunchDepthwiseConv2dGPUSmall()
642 kernel<<<config.block_count, block_dim, shared_memory_size, in LaunchDepthwiseConv2dGPUSmall()
1594 const int shared_memory_size =
1596 if (shared_memory_size > device.sharedMemPerBlock()) {
1626 num_out_backprop, device, kernel, shared_memory_size,
1628 kernel<<<config.block_count, block_dim, shared_memory_size,
|