From a723e1c62a27aeb39aaf7fcdeb3beb4e89fba32d Mon Sep 17 00:00:00 2001
From: Alexey <AlexeyAB@users.noreply.github.com>
Date: Wed, 15 Aug 2018 20:52:09 +0000
Subject: [PATCH] Merge pull request #766 from HotChick91/AlexeyAB-mask

---
 src/network_kernels.cu |   32 +++++++++++++++++++++++++++++++-
 1 files changed, 31 insertions(+), 1 deletions(-)

diff --git a/src/network_kernels.cu b/src/network_kernels.cu
index 7a261c5..681542f 100644
--- a/src/network_kernels.cu
+++ b/src/network_kernels.cu
@@ -36,6 +36,10 @@
 #include "blas.h"
 }
 
+#ifdef OPENCV
+#include "opencv2/highgui/highgui_c.h"
+#endif
+
 float * get_network_output_gpu_layer(network net, int i);
 float * get_network_delta_gpu_layer(network net, int i);
 float * get_network_output_gpu(network net);
@@ -51,7 +55,24 @@
             fill_ongpu(l.outputs * l.batch, 0, l.delta_gpu, 1);
         }
         l.forward_gpu(l, state);
+        if(net.wait_stream)
+            cudaStreamSynchronize(get_cuda_stream());
         state.input = l.output_gpu;
+/*
+        cuda_pull_array(l.output_gpu, l.output, l.batch*l.outputs);
+        if (l.out_w >= 0 && l.out_h >= 1 && l.c >= 3) {
+            int j;
+            for (j = 0; j < l.out_c; ++j) {
+                image img = make_image(l.out_w, l.out_h, 3);
+                memcpy(img.data, l.output+ l.out_w*l.out_h*j, l.out_w*l.out_h * 1 * sizeof(float));
+                char buff[256];
+                sprintf(buff, "layer-%d slice-%d", i, j);
+                show_image(img, buff);
+            }
+            cvWaitKey(0); // wait press-key in console
+            cvDestroyAllWindows();
+        }
+*/
     }
 }
 
@@ -111,7 +132,15 @@
     state.delta = 0;
     state.truth = *net.truth_gpu;
     state.train = 1;
+#ifdef CUDNN_HALF
+    int i;
+    for (i = 0; i < net.n; ++i) {
+        layer l = net.layers[i];
+        cuda_convert_f32_to_f16(l.weights_gpu, l.c*l.n*l.size*l.size, l.weights_gpu16);
+    }
+#endif
     forward_network_gpu(net, state);
+    //cudaStreamSynchronize(get_cuda_stream());
     backward_network_gpu(net, state);
 }
 
@@ -392,7 +421,8 @@
 
 float *network_predict_gpu(network net, float *input)
 {
-    cuda_set_device(net.gpu_index);
+    if (net.gpu_index != cuda_get_device())
+        cuda_set_device(net.gpu_index);
     int size = get_network_input_size(net) * net.batch;
     network_state state;
     state.index = 0;

--
Gitblit v1.10.0