From e92f7d301c971b4d27aa3dcd1e4047e94f04b3fc Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Wed, 25 Mar 2015 01:27:12 +0000
Subject: [PATCH] smaller gridsize in bias
---
src/cost_layer.c | 87 +++++++++++++------------------------------
1 files changed, 26 insertions(+), 61 deletions(-)
diff --git a/src/cost_layer.c b/src/cost_layer.c
index 08d3bb5..d2c616f 100644
--- a/src/cost_layer.c
+++ b/src/cost_layer.c
@@ -1,6 +1,7 @@
#include "cost_layer.h"
#include "utils.h"
-#include "mini_blas.h"
+#include "cuda.h"
+#include "blas.h"
#include <math.h>
#include <string.h>
#include <stdlib.h>
@@ -9,7 +10,6 @@
COST_TYPE get_cost_type(char *s)
{
if (strcmp(s, "sse")==0) return SSE;
- if (strcmp(s, "detection")==0) return DETECTION;
fprintf(stderr, "Couldn't find activation function %s, going with SSE\n", s);
return SSE;
}
@@ -19,8 +19,6 @@
switch(a){
case SSE:
return "sse";
- case DETECTION:
- return "detection";
}
return "sse";
}
@@ -35,83 +33,50 @@
layer->delta = calloc(inputs*batch, sizeof(float));
layer->output = calloc(1, sizeof(float));
#ifdef GPU
- layer->delta_cl = cl_make_array(layer->delta, inputs*batch);
+ layer->delta_gpu = cuda_make_array(layer->delta, inputs*batch);
#endif
return layer;
}
-void forward_cost_layer(cost_layer layer, float *input, float *truth)
+void pull_cost_layer(cost_layer layer)
{
- if (!truth) return;
- copy_cpu(layer.batch*layer.inputs, truth, 1, layer.delta, 1);
- axpy_cpu(layer.batch*layer.inputs, -1, input, 1, layer.delta, 1);
- if(layer.type == DETECTION){
- int i;
- for(i = 0; i < layer.batch*layer.inputs; ++i){
- if((i%5) && !truth[(i/5)*5]) layer.delta[i] = 0;
- }
- }
+ cuda_pull_array(layer.delta_gpu, layer.delta, layer.batch*layer.inputs);
+}
+void push_cost_layer(cost_layer layer)
+{
+ cuda_push_array(layer.delta_gpu, layer.delta, layer.batch*layer.inputs);
+}
+
+void forward_cost_layer(cost_layer layer, network_state state)
+{
+ if (!state.truth) return;
+ copy_cpu(layer.batch*layer.inputs, state.truth, 1, layer.delta, 1);
+ axpy_cpu(layer.batch*layer.inputs, -1, state.input, 1, layer.delta, 1);
*(layer.output) = dot_cpu(layer.batch*layer.inputs, layer.delta, 1, layer.delta, 1);
//printf("cost: %f\n", *layer.output);
}
-void backward_cost_layer(const cost_layer layer, float *input, float *delta)
+void backward_cost_layer(const cost_layer layer, network_state state)
{
- copy_cpu(layer.batch*layer.inputs, layer.delta, 1, delta, 1);
+ copy_cpu(layer.batch*layer.inputs, layer.delta, 1, state.delta, 1);
}
#ifdef GPU
-cl_kernel get_mask_kernel()
+void forward_cost_layer_gpu(cost_layer layer, network_state state)
{
- static int init = 0;
- static cl_kernel kernel;
- if(!init){
- kernel = get_kernel("src/axpy.cl", "mask", 0);
- init = 1;
- }
- return kernel;
-}
+ if (!state.truth) return;
+
+ copy_ongpu(layer.batch*layer.inputs, state.truth, 1, layer.delta_gpu, 1);
+ axpy_ongpu(layer.batch*layer.inputs, -1, state.input, 1, layer.delta_gpu, 1);
-void mask_ongpu(int n, cl_mem x, cl_mem mask, int mod)
-{
- cl_setup();
- cl_kernel kernel = get_mask_kernel();
- cl_command_queue queue = cl.queue;
-
- cl_uint i = 0;
- cl.error = clSetKernelArg(kernel, i++, sizeof(n), (void*) &n);
- cl.error = clSetKernelArg(kernel, i++, sizeof(x), (void*) &x);
- cl.error = clSetKernelArg(kernel, i++, sizeof(mask), (void*) &mask);
- cl.error = clSetKernelArg(kernel, i++, sizeof(mod), (void*) &mod);
- check_error(cl);
-
- const size_t global_size[] = {n};
-
- cl.error = clEnqueueNDRangeKernel(queue, kernel, 1, 0, global_size, 0, 0, 0, 0);
- check_error(cl);
-
-}
-
-void forward_cost_layer_gpu(cost_layer layer, cl_mem input, cl_mem truth)
-{
- if (!truth) return;
-
- copy_ongpu(layer.batch*layer.inputs, truth, 1, layer.delta_cl, 1);
- axpy_ongpu(layer.batch*layer.inputs, -1, input, 1, layer.delta_cl, 1);
-
- if(layer.type==DETECTION){
- mask_ongpu(layer.inputs*layer.batch, layer.delta_cl, truth, 5);
- }
-
- cl_read_array(layer.delta_cl, layer.delta, layer.batch*layer.inputs);
+ cuda_pull_array(layer.delta_gpu, layer.delta, layer.batch*layer.inputs);
*(layer.output) = dot_cpu(layer.batch*layer.inputs, layer.delta, 1, layer.delta, 1);
- //printf("cost: %f\n", *layer.output);
}
-void backward_cost_layer_gpu(const cost_layer layer, cl_mem input, cl_mem delta)
+void backward_cost_layer_gpu(const cost_layer layer, network_state state)
{
- copy_ongpu(layer.batch*layer.inputs, layer.delta_cl, 1, delta, 1);
+ copy_ongpu(layer.batch*layer.inputs, layer.delta_gpu, 1, state.delta, 1);
}
#endif
--
Gitblit v1.10.0