From 23cb35e6c8eae8b59fab161036ae3f417a55c8db Mon Sep 17 00:00:00 2001
From: AlexeyAB <alexeyab84@gmail.com>
Date: Fri, 30 Mar 2018 11:46:51 +0000
Subject: [PATCH] Changed small_object
---
src/network_kernels.cu | 13 ++++++++++++-
1 files changed, 12 insertions(+), 1 deletions(-)
diff --git a/src/network_kernels.cu b/src/network_kernels.cu
index 7a261c5..d6bb294 100644
--- a/src/network_kernels.cu
+++ b/src/network_kernels.cu
@@ -51,6 +51,8 @@
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;
}
}
@@ -111,7 +113,15 @@
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 +402,8 @@
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