Convolutional working on GPU

Joseph Redmon

2014-10-13 787d5345609459f21fd65d2d8b4fcd55201e21a1

 src/network.c

@@ -8,7 +8,9 @@
#include "connected_layer.h"
#include "convolutional_layer.h"
#include "maxpool_layer.h"
#include "cost_layer.h"
#include "normalization_layer.h"
#include "freeweight_layer.h"
#include "softmax_layer.h"
#include "dropout_layer.h"

@@ -28,14 +30,18 @@
}

#ifdef GPU
void forward_network_gpu(network net, cl_mem input_cl, int train)
void forward_network_gpu(network net, cl_mem input, cl_mem truth, int train)
{
    int i;
    for(i = 0; i < net.n; ++i){
        if(net.types[i] == CONVOLUTIONAL){
            convolutional_layer layer = *(convolutional_layer *)net.layers[i];
            forward_convolutional_layer_gpu(layer, input_cl);
            input_cl = layer.output_cl;
            forward_convolutional_layer_gpu(layer, input);
            input = layer.output_cl;
        }
        else if(net.types[i] == COST){
            cost_layer layer = *(cost_layer *)net.layers[i];
            forward_cost_layer_gpu(layer, input, truth);
        }
        /*
        else if(net.types[i] == CONNECTED){
@@ -67,9 +73,75 @@
    }
}

void backward_network_gpu(network net, cl_mem input)
{
    int i;
    cl_mem prev_input;
    cl_mem prev_delta;
    for(i = net.n-1; i >= 0; --i){
        if(i == 0){
            prev_input = input;
            prev_delta = 0;
        }else{
            prev_input = get_network_output_cl_layer(net, i-1);
            prev_delta = get_network_delta_cl_layer(net, i-1);
        }
        if(net.types[i] == CONVOLUTIONAL){
            convolutional_layer layer = *(convolutional_layer *)net.layers[i];
            backward_convolutional_layer_gpu(layer, prev_delta);
        }
        else if(net.types[i] == COST){
            cost_layer layer = *(cost_layer *)net.layers[i];
            backward_cost_layer_gpu(layer, prev_input, prev_delta);
        }
    }
}

void update_network_gpu(network net)
{
    int i;
    for(i = 0; i < net.n; ++i){
        if(net.types[i] == CONVOLUTIONAL){
            convolutional_layer layer = *(convolutional_layer *)net.layers[i];
            update_convolutional_layer_gpu(layer);
        }
        else if(net.types[i] == MAXPOOL){
            //maxpool_layer layer = *(maxpool_layer *)net.layers[i];
        }
        else if(net.types[i] == SOFTMAX){
            //maxpool_layer layer = *(maxpool_layer *)net.layers[i];
        }
        else if(net.types[i] == NORMALIZATION){
            //maxpool_layer layer = *(maxpool_layer *)net.layers[i];
        }
        else if(net.types[i] == CONNECTED){
            connected_layer layer = *(connected_layer *)net.layers[i];
            update_connected_layer(layer);
        }
    }
}

cl_mem get_network_output_cl_layer(network net, int i)
{
    if(net.types[i] == CONVOLUTIONAL){
        convolutional_layer layer = *(convolutional_layer *)net.layers[i];
        return layer.output_cl;
    }
    return 0;
}

cl_mem get_network_delta_cl_layer(network net, int i)
{
    if(net.types[i] == CONVOLUTIONAL){
        convolutional_layer layer = *(convolutional_layer *)net.layers[i];
        return layer.delta_cl;
    }
    return 0;
}

#endif

void forward_network(network net, float *input, int train)
void forward_network(network net, float *input, float *truth, int train)
{
    int i;
    for(i = 0; i < net.n; ++i){
@@ -88,6 +160,10 @@
            forward_crop_layer(layer, input);
            input = layer.output;
        }
        else if(net.types[i] == COST){
            cost_layer layer = *(cost_layer *)net.layers[i];
            forward_cost_layer(layer, input, truth);
        }
        else if(net.types[i] == SOFTMAX){
            softmax_layer layer = *(softmax_layer *)net.layers[i];
            forward_softmax_layer(layer, input);
@@ -108,6 +184,11 @@
            dropout_layer layer = *(dropout_layer *)net.layers[i];
            forward_dropout_layer(layer, input);
        }
        else if(net.types[i] == FREEWEIGHT){
            if(!train) continue;
            freeweight_layer layer = *(freeweight_layer *)net.layers[i];
            forward_freeweight_layer(layer, input);
        }
    }
}

@@ -159,7 +240,9 @@
}
float *get_network_output(network net)
{
    return get_network_output_layer(net, net.n-1);
    int i;
    for(i = net.n-1; i > 0; --i) if(net.types[i] != COST) break;
    return get_network_output_layer(net, i);
}

float *get_network_delta_layer(network net, int i)
@@ -182,6 +265,14 @@
    return 0;
}

float get_network_cost(network net)
{
    if(net.types[net.n-1] == COST){
        return ((cost_layer *)net.layers[net.n-1])->output[0];
    }
    return 0;
}

float *get_network_delta(network net)
{
    return get_network_delta_layer(net, net.n-1);
@@ -212,9 +303,8 @@
    return max_index(out, k);
}

float backward_network(network net, float *input, float *truth)
void backward_network(network net, float *input)
{
    float error = calculate_error_network(net, truth);
    int i;
    float *prev_input;
    float *prev_delta;
@@ -246,15 +336,19 @@
            connected_layer layer = *(connected_layer *)net.layers[i];
            backward_connected_layer(layer, prev_input, prev_delta);
        }
        else if(net.types[i] == COST){
            cost_layer layer = *(cost_layer *)net.layers[i];
            backward_cost_layer(layer, prev_input, prev_delta);
        }
    }
    return error;
}

float train_network_datum(network net, float *x, float *y)
{
    forward_network(net, x, 1);
    forward_network(net, x, y, 1);
    //int class = get_predicted_class_network(net);
    float error = backward_network(net, x, y);
    backward_network(net, x);
    float error = get_network_cost(net);
    update_network(net);
    //return (y[class]?1:0);
    return error;
@@ -287,8 +381,9 @@
            int index = rand()%d.X.rows;
            float *x = d.X.vals[index];
            float *y = d.y.vals[index];
            forward_network(net, x, 1);
            sum += backward_network(net, x, y);
            forward_network(net, x, y, 1);
            backward_network(net, x);
            sum += get_network_cost(net);
        }
        update_network(net);
    }
@@ -351,7 +446,8 @@
    else if(net.types[i] == CONNECTED){
        connected_layer layer = *(connected_layer *)net.layers[i];
        return layer.outputs;
    } else if(net.types[i] == DROPOUT){
    }
    else if(net.types[i] == DROPOUT){
        dropout_layer layer = *(dropout_layer *) net.layers[i];
        return layer.inputs;
    }
@@ -396,7 +492,8 @@

int get_network_output_size(network net)
{
    int i = net.n-1;
    int i;
    for(i = net.n-1; i > 0; --i) if(net.types[i] != COST) break;
    return get_network_output_size_layer(net, i);
}

@@ -457,7 +554,7 @@

float *network_predict(network net, float *input)
{
    forward_network(net, input, 0);
    forward_network(net, input, 0, 0);
    float *out = get_network_output(net);
    return out;
}