Use half_float16 instead of float32 if defined both CUDNN and CUDNN_HALF. Use Tensor Cores.

src/convolutional_kernels.cu

@@ -74,6 +74,38 @@ void binarize_weights_gpu(float *weights, int n, int size, float *binary)
     check_error(cudaPeekAtLastError());
 }
 
+__global__ void cuda_f32_to_f16(float* input_f32, size_t size, half *output_f16)
+{
+	int idx = blockIdx.x * blockDim.x + threadIdx.x;
+	if (idx < size) output_f16[idx] = input_f32[idx];
+}
+
+void cuda_convert_f32_to_f16(float* input_f32, size_t size, half *output_f16) {
+	cuda_f32_to_f16 <<< size / BLOCK + 1, BLOCK, 0, get_cuda_stream() >>> (input_f32, size, output_f16);
+}
+
+__global__ void cuda_f16_to_f32(half* input_f16, size_t size, float *output_f32)
+{
+	int idx = blockIdx.x * blockDim.x + threadIdx.x;
+	if (idx < size) output_f32[idx] = input_f16[idx];
+}
+
+void cuda_convert_f16_to_f32(half* input_f16, size_t size, float *output_f32) {
+	cuda_f16_to_f32 <<< size / BLOCK + 1, BLOCK, 0, get_cuda_stream() >>> (input_f16, size, output_f32);
+}
+
+half *cuda_make_f16_from_f32_array(float *src, size_t n)
+{
+	half *dst16;
+	size_t size = sizeof(half)*n;
+	check_error(cudaMalloc((void **)&dst16, size));
+	if (src) {
+		cuda_convert_f32_to_f16(src, n, dst16);
+	}
+	if (!dst16) error("Cuda malloc failed\n");
+	return dst16;
+}
+
 void forward_convolutional_layer_gpu(convolutional_layer l, network_state state)
 {
     fill_ongpu(l.outputs*l.batch, 0, l.output_gpu, 1);
@@ -90,9 +122,57 @@ void forward_convolutional_layer_gpu(convolutional_layer l, network_state state)
     }
 
 #ifdef CUDNN
-    float one = 1;
+	//float one = 1;	// alpha[0], beta[0] is float for HALF and FLOAT
+	float alpha = 1, beta = 0;
+
+#ifdef CUDNN_HALF
+	// Note: For improved performance it is advised to use beta[0] = 0.0. 
+	// For Tensor Core: cudnnSetConvolutionMathType() where cudnnMathType_t mathType = CUDNN_TENSOR_OP_MATH;
+	// 1. or CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_PRECOMP_GEMM and use CUDNN_DATA_HALF
+	// 2. or CUDNN_CONVOLUTION_FWD_ALGO_WINOGRAD_NONFUSED
+	// More: http://docs.nvidia.com/deeplearning/sdk/cudnn-developer-guide/index.html#tensor_ops
+
+	const size_t input16_size = l.batch*l.c*l.w*l.h;
+	static size_t max_input16_size = input16_size;
+	static half* input16 = cuda_make_f16_from_f32_array(NULL, max_input16_size);
+
+	const size_t output16_size = l.batch*l.out_c*l.out_h*l.out_w;
+	static size_t max_output16_size = output16_size;
+	static half* output16 = cuda_make_f16_from_f32_array(NULL, max_output16_size);
+
+	if (max_input16_size < input16_size) {
+		max_input16_size = input16_size;
+		cuda_free((float *)input16);
+		input16 = cuda_make_f16_from_f32_array(state.input, max_input16_size);
+	}
+
+	if (max_output16_size < output16_size) {
+		max_output16_size = output16_size;
+		cuda_free((float *)output16);
+		output16 = cuda_make_f16_from_f32_array(NULL, max_output16_size);
+	}
+
+	cuda_convert_f32_to_f16(state.input, input16_size, input16);
+
+	cudnnConvolutionForward(cudnn_handle(),
+		&alpha,
+		l.srcTensorDesc,
+		input16,
+		l.weightDesc,
+		l.weights_gpu16,
+		l.convDesc,
+		l.fw_algo,
+		state.workspace,
+		l.workspace_size,
+		&beta,
+		l.dstTensorDesc,
+		output16);
+	
+	cuda_convert_f16_to_f32(output16, output16_size, l.output_gpu);
+#else
+
     cudnnConvolutionForward(cudnn_handle(),
-                &one,
+                &alpha,
                 l.srcTensorDesc,
                 state.input,
                 l.weightDesc,
@@ -101,9 +181,11 @@ void forward_convolutional_layer_gpu(convolutional_layer l, network_state state)
                 l.fw_algo,
                 state.workspace,
                 l.workspace_size,
-                &one,
+                &beta,
                 l.dstTensorDesc,
                 l.output_gpu);
+#endif
+
 
 #else
     int i;
@@ -232,6 +314,9 @@ void pull_convolutional_layer(convolutional_layer layer)
 void push_convolutional_layer(convolutional_layer layer)
 {
     cuda_push_array(layer.weights_gpu, layer.weights, layer.c*layer.n*layer.size*layer.size);
+#ifdef CUDNN_HALF
+	cuda_convert_f32_to_f16(layer.weights_gpu, layer.c*layer.n*layer.size*layer.size, (half *)layer.weights_gpu16);
+#endif
     cuda_push_array(layer.biases_gpu, layer.biases, layer.n);
     cuda_push_array(layer.weight_updates_gpu, layer.weight_updates, layer.c*layer.n*layer.size*layer.size);
     cuda_push_array(layer.bias_updates_gpu, layer.bias_updates, layer.n);