From f39160f6e8f4465e9325e235e18f0d9413d1f672 Mon Sep 17 00:00:00 2001
From: AlexeyAB <alexeyab84@gmail.com>
Date: Sun, 04 Mar 2018 00:09:40 +0000
Subject: [PATCH] Added: calc_anchors
---
src/network.c | 36 +++++++++++++++++++++++++++---------
1 files changed, 27 insertions(+), 9 deletions(-)
diff --git a/src/network.c b/src/network.c
index 56a316c..d23468d 100644
--- a/src/network.c
+++ b/src/network.c
@@ -50,6 +50,7 @@
int batch_num = get_current_batch(net);
int i;
float rate;
+ if (batch_num < net.burn_in) return net.learning_rate * pow((float)batch_num / net.burn_in, net.power);
switch (net.policy) {
case CONSTANT:
return net.learning_rate;
@@ -66,8 +67,9 @@
case EXP:
return net.learning_rate * pow(net.gamma, batch_num);
case POLY:
- if (batch_num < net.burn_in) return net.learning_rate * pow((float)batch_num / net.burn_in, net.power);
- return net.learning_rate * pow(1 - (float)batch_num / net.max_batches, net.power);
+ return net.learning_rate * pow(1 - (float)batch_num / net.max_batches, net.power);
+ //if (batch_num < net.burn_in) return net.learning_rate * pow((float)batch_num / net.burn_in, net.power);
+ //return net.learning_rate * pow(1 - (float)batch_num / net.max_batches, net.power);
case RANDOM:
return net.learning_rate * pow(rand_uniform(0,1), net.power);
case SIG:
@@ -314,7 +316,20 @@
net->layers[i].batch = b;
#ifdef CUDNN
if(net->layers[i].type == CONVOLUTIONAL){
- cudnn_convolutional_setup(net->layers + i);
+ cudnn_convolutional_setup(net->layers + i, cudnn_fastest);
+ /*
+ layer *l = net->layers + i;
+ cudnn_convolutional_setup(l, cudnn_fastest);
+ // check for excessive memory consumption
+ size_t free_byte;
+ size_t total_byte;
+ check_error(cudaMemGetInfo(&free_byte, &total_byte));
+ if (l->workspace_size > free_byte || l->workspace_size >= total_byte / 2) {
+ printf(" used slow CUDNN algo without Workspace! \n");
+ cudnn_convolutional_setup(l, cudnn_smallest);
+ l->workspace_size = get_workspace_size(*l);
+ }
+ */
}
#endif
}
@@ -326,6 +341,12 @@
cuda_set_device(net->gpu_index);
if(gpu_index >= 0){
cuda_free(net->workspace);
+ if (net->input_gpu) {
+ cuda_free(*net->input_gpu);
+ *net->input_gpu = 0;
+ cuda_free(*net->truth_gpu);
+ *net->truth_gpu = 0;
+ }
}
#endif
int i;
@@ -338,6 +359,7 @@
//fflush(stderr);
for (i = 0; i < net->n; ++i){
layer l = net->layers[i];
+ //printf(" %d: layer = %d,", i, l.type);
if(l.type == CONVOLUTIONAL){
resize_convolutional_layer(&l, w, h);
}else if(l.type == CROP){
@@ -369,13 +391,9 @@
}
#ifdef GPU
if(gpu_index >= 0){
- if(net->input_gpu) {
- cuda_free(*net->input_gpu);
- *net->input_gpu = 0;
- cuda_free(*net->truth_gpu);
- *net->truth_gpu = 0;
- }
+ printf(" try to allocate workspace = %zu * sizeof(float), ", (workspace_size - 1) / sizeof(float) + 1);
net->workspace = cuda_make_array(0, (workspace_size-1)/sizeof(float)+1);
+ printf(" CUDA allocate done! \n");
}else {
free(net->workspace);
net->workspace = calloc(1, workspace_size);
--
Gitblit v1.10.0