mmr/tkDNN
Archived
1
0
Fork 0
Code Issues 46 Pull Requests 12 Releases 4 Activity

all test ok

Browse Source
This commit is contained in:
Francesco Gatti
2020-06-01 16:15:29 +02:00
parent d4e0d07e09
commit c8ed6d782a
16 changed files with 63 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files
include/tkDNN/DarknetParser.h
+6 -6
View File
@@ -124,7 +124,7 @@ namespace tk { namespace dnn {
}
tk::dnn::Network *darknetAddNet(darknetFields_t &fields) {
std::cout<<"Add Net: "<<fields.type<<"\n";
//std::cout<<"Add Net: "<<fields.type<<"\n";
dataDim_t dim(1, fields.channels, fields.height, fields.width);
return new tk::dnn::Network(dim);
}
@@ -138,10 +138,10 @@ namespace tk { namespace dnn {
if(f.pad == 1) {
f.padding_x = f.padding_y = f.size_x /2;
}
std::cout<<"Add layer: "<<f.type<<"\n";
//std::cout<<"Add layer: "<<f.type<<"\n";
if(f.type == "convolutional") {
std::string wgs = wgs_path + "/c" + std::to_string(netLayers.size()) + ".bin";
printf("%d (%d,%d) (%d,%d) (%d,%d) %s %d %d\n", f.filters, f.size_x, f.size_y, f.stride_x, f.stride_y, f.padding_x, f.padding_y, wgs.c_str(), f.batch_normalize, f.groups);
//printf("%d (%d,%d) (%d,%d) (%d,%d) %s %d %d\n", f.filters, f.size_x, f.size_y, f.stride_x, f.stride_y, f.padding_x, f.padding_y, wgs.c_str(), f.batch_normalize, f.groups);
tk::dnn::Conv2d *l= new tk::dnn::Conv2d(net, f.filters, f.size_x, f.size_y, f.stride_x,
f.stride_y, f.padding_x, f.padding_y, wgs, f.batch_normalize, false, f.groups);
netLayers.push_back(l);
@@ -163,7 +163,7 @@ namespace tk { namespace dnn {
if(layerIdx < 0)
layerIdx = netLayers.size() + layerIdx;
if(layerIdx < 0 || layerIdx >= netLayers.size()) FatalError("impossible to shortcut\n");
std::cout<<"shortcut to "<<layerIdx<<" "<<netLayers[layerIdx]->getLayerName()<<"\n";
//std::cout<<"shortcut to "<<layerIdx<<" "<<netLayers[layerIdx]->getLayerName()<<"\n";
netLayers.push_back(new tk::dnn::Shortcut(net, netLayers[layerIdx]));
} else if(f.type == "upsample") {
@@ -177,7 +177,7 @@ namespace tk { namespace dnn {
if(layerIdx < 0)
layerIdx = netLayers.size() + layerIdx;
if(layerIdx < 0 || layerIdx >= netLayers.size()) FatalError("impossible to route\n");
std::cout<<"Route to "<<layerIdx<<" "<<netLayers[layerIdx]->getLayerName()<<"\n";
//std::cout<<"Route to "<<layerIdx<<" "<<netLayers[layerIdx]->getLayerName()<<"\n";
layers.push_back(netLayers[layerIdx]);
}
netLayers.push_back(new tk::dnn::Route(net, layers.data(), layers.size()));
@@ -190,7 +190,7 @@ namespace tk { namespace dnn {
} else if(f.type == "yolo") {
std::string wgs = wgs_path + "/g" + std::to_string(netLayers.size()) + ".bin";
printf("%d %d %s %d %f\n", f.classes, f.num/f.n_mask, wgs.c_str(), f.n_mask, f.scale_xy);
//printf("%d %d %s %d %f\n", f.classes, f.num/f.n_mask, wgs.c_str(), f.n_mask, f.scale_xy);
tk::dnn::Yolo *l = new tk::dnn::Yolo(net, f.classes, f.num/f.n_mask, wgs, f.n_mask, f.scale_xy);
if(names.size() != f.classes)
FatalError("Mismatch between number of classes and names");
include/tkDNN/DetectionNN.h
+6 -6
View File
@@ -106,14 +106,14 @@ class DetectionNN {
originalSize.clear();
if(TKDNN_VERBOSE) printCenteredTitle(" TENSORRT detection ", '=', 30);
{
TIMER_START
TKDNN_TSTART
for(int bi=0; bi<cur_batches;++bi){
if(!frames[bi].data)
FatalError("No image data feed to detection");
originalSize.push_back(frames[bi].size());
preprocess(frames[bi], bi);
}
TIMER_STOP
TKDNN_TSTOP
if(save_times) *times<<t_ns<<";";
}
@@ -122,9 +122,9 @@ class DetectionNN {
dim.n = cur_batches;
{
if(TKDNN_VERBOSE) dim.print();
TIMER_START
TKDNN_TSTART
netRT->infer(dim, input_d);
TIMER_STOP
TKDNN_TSTOP
if(TKDNN_VERBOSE) dim.print();
stats.push_back(t_ns);
if(save_times) *times<<t_ns<<";";
@@ -132,10 +132,10 @@ class DetectionNN {
batchDetected.clear();
{
TIMER_START
TKDNN_TSTART
for(int bi=0; bi<cur_batches;++bi)
postprocess(bi, mAP);
TIMER_STOP
TKDNN_TSTOP
if(save_times) *times<<t_ns<<"\n";
}
}
include/tkDNN/test.h
+4 -4
View File
@@ -34,9 +34,9 @@ int testInference(std::vector<std::string> input_bins, std::vector<std::string>
tk::dnn::dataDim_t dim1 = net->input_dim; //input dim
printCenteredTitle(" CUDNN inference ", '=', 30); {
dim1.print();
TIMER_START
TKDNN_TSTART
net->infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
for(int i=0; i<outputs.size(); i++) cudnn_out[i] = outputs[i]->dstData;
@@ -45,9 +45,9 @@ int testInference(std::vector<std::string> input_bins, std::vector<std::string>
tk::dnn::dataDim_t dim2 = net->input_dim;
printCenteredTitle(" TENSORRT inference ", '=', 30); {
dim2.print();
TIMER_START
TKDNN_TSTART
netRT->infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
for(int i=0; i<outputs.size(); i++) rt_out[i] = (dnnType*)netRT->buffersRT[i+1];
include/tkDNN/utils.h
+3 -3
View File
@@ -39,15 +39,15 @@
#define TKDNN_VERBOSE 0
// Simple Timer
#define TIMER_START timespec start, end; \
#define TKDNN_TSTART timespec start, end; \
clock_gettime(CLOCK_MONOTONIC, &start);
#define TIMER_STOP_C(col, show) clock_gettime(CLOCK_MONOTONIC, &end); \
#define TKDNN_TSTOP_C(col, show) clock_gettime(CLOCK_MONOTONIC, &end); \
double t_ns = ((double)(end.tv_sec - start.tv_sec) * 1.0e9 + \
(double)(end.tv_nsec - start.tv_nsec))/1.0e6; \
if(show) std::cout<<col<<"Time:"<<std::setw(16)<<t_ns<<" ms\n"<<COL_END;
#define TIMER_STOP TIMER_STOP_C(COL_CYANB, TKDNN_VERBOSE)
#define TKDNN_TSTOP TKDNN_TSTOP_C(COL_CYANB, TKDNN_VERBOSE)
/********************************************************
* Prints the error message, and exits
tests/backbones/dla34/dla34.cpp
+4 -4
View File
@@ -312,9 +312,9 @@ int main()
printCenteredTitle(" CUDNN inference ", '=', 30);
{
dim1.print();
TIMER_START
TKDNN_TSTART
net.infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
cudnn_out = net.layers[net.num_layers-1]->dstData;
@@ -326,9 +326,9 @@ int main()
printCenteredTitle(" TENSORRT inference ", '=', 30);
{
dim2.print();
TIMER_START
TKDNN_TSTART
netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
rt_out = (dnnType *)netRT.buffersRT[1];
tests/backbones/resnet101/resnet101.cpp
+4 -4
View File
@@ -301,9 +301,9 @@ int main()
printCenteredTitle(" CUDNN inference ", '=', 30);
{
dim1.print();
TIMER_START
TKDNN_TSTART
net.infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
cudnn_out = net.layers[net.num_layers-1]->dstData;
@@ -314,9 +314,9 @@ int main()
printCenteredTitle(" TENSORRT inference ", '=', 30);
{
dim2.print();
TIMER_START
TKDNN_TSTART
netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
rt_out = (dnnType *)netRT.buffersRT[1];
tests/centernet/dla34_cnet/dla34_cnet.cpp
+4 -4
View File
@@ -486,9 +486,9 @@ int main()
printCenteredTitle(" CUDNN inference ", '=', 30);
{
dim1.print();
TIMER_START
TKDNN_TSTART
net.infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
@@ -496,9 +496,9 @@ int main()
printCenteredTitle(" TENSORRT inference ", '=', 30);
{
dim2.print();
TIMER_START
TKDNN_TSTART
netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
tests/centernet/resnet101_cnet/resnet101_cnet.cpp
+4 -4
View File
@@ -361,9 +361,9 @@ int main()
printCenteredTitle(" CUDNN inference ", '=', 30);
{
dim1.print();
TIMER_START
TKDNN_TSTART
net.infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
@@ -373,9 +373,9 @@ int main()
printCenteredTitle(" TENSORRT inference ", '=', 30);
{
dim2.print();
TIMER_START
TKDNN_TSTART
netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
tests/imuodom/imuodom.cpp
+2 -2
View File
@@ -46,10 +46,10 @@ int main() {
int ret_cudnn = 0;
for(int i=0; i<N; i++) {
std::cout<<"i: "<<i<<"\n";
//TIMER_START
//TKDNN_TSTART
// Inference
ImuNet.update(i0_h, i1_h, i2_h);
//TIMER_STOP
//TKDNN_TSTOP
// log path
path<<ImuNet.odomPOS(0)<<" "<<ImuNet.odomPOS(1)<<" "<< ImuNet.odomPOS(2)<<" ";
tests/mnist/test_mnist.cpp
+4 -4
View File
@@ -35,9 +35,9 @@ int main() {
std::cout<<"CUDNN inference:\n"; {
dim.print(); //print initial dimension
TIMER_START
TKDNN_TSTART
out_data = net.infer(dim, data);
TIMER_STOP
TKDNN_TSTOP
dim.print();
}
@@ -49,9 +49,9 @@ int main() {
std::cout<<"TENSORRT inference:\n"; {
dim2.print();
TIMER_START
TKDNN_TSTART
out_data2 = netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
tests/mnist/test_mnistRT.cpp
+4 -4
View File
@@ -49,9 +49,9 @@ int main() {
// Inference
{
TIMER_START
TKDNN_TSTART
data = net.infer(dim, data);
TIMER_STOP
TKDNN_TSTOP
dim.print();
}
@@ -157,9 +157,9 @@ int main() {
{
checkCuda(cudaMemcpyAsync(buffers[inputIndex], input_h, 1 * 28*28* sizeof(float), cudaMemcpyHostToDevice, stream));
cudaStreamSynchronize(stream); //want to test only the inference time
TIMER_START
TKDNN_TSTART
context->enqueue(1, buffers, stream, nullptr);
TIMER_STOP
TKDNN_TSTOP
checkCuda(cudaMemcpyAsync(output, buffers[outputIndex],10*sizeof(float), cudaMemcpyDeviceToHost, stream));
cudaStreamSynchronize(stream);
}
tests/mobilenet/bdd-mobilenetv2ssd/bdd-mobilenetv2ssd.cpp
+4 -4
View File
@@ -477,9 +477,9 @@ int main()
printCenteredTitle(" CUDNN inference ", '=', 30);
{
dim1.print();
TIMER_START
TKDNN_TSTART
net.infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
@@ -492,9 +492,9 @@ int main()
printCenteredTitle(" TENSORRT inference ", '=', 30);
{
dim2.print();
TIMER_START
TKDNN_TSTART
netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
tests/mobilenet/mobilenetv2ssd/mobilenetv2ssd.cpp
+4 -4
View File
@@ -477,9 +477,9 @@ int main()
printCenteredTitle(" CUDNN inference ", '=', 30);
{
dim1.print();
TIMER_START
TKDNN_TSTART
net.infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
@@ -492,9 +492,9 @@ int main()
printCenteredTitle(" TENSORRT inference ", '=', 30);
{
dim2.print();
TIMER_START
TKDNN_TSTART
netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
tests/mobilenet/mobilenetv2ssd512/mobilenetv2ssd512.cpp
+4 -4
View File
@@ -476,9 +476,9 @@ int main()
printCenteredTitle(" CUDNN inference ", '=', 30);
{
dim1.print();
TIMER_START
TKDNN_TSTART
net.infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
@@ -491,9 +491,9 @@ int main()
printCenteredTitle(" TENSORRT inference ", '=', 30);
{
dim2.print();
TIMER_START
TKDNN_TSTART
netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
tests/simple/test_simple.cpp
+4 -4
View File
@@ -38,18 +38,18 @@ int main() {
tk::dnn::dataDim_t dim1 = dim; //input dim
printCenteredTitle(" CUDNN inference ", '=', 30); {
dim1.print();
TIMER_START
TKDNN_TSTART
out_data = net.infer(dim1, data);
TIMER_STOP
TKDNN_TSTOP
dim1.print();
}
tk::dnn::dataDim_t dim2 = dim;
printCenteredTitle(" TENSORRT inference ", '=', 30); {
dim2.print();
TIMER_START
TKDNN_TSTART
out_data2 = netRT.infer(dim2, data);
TIMER_STOP
TKDNN_TSTOP
dim2.print();
}
tests/test_rtinference/rtinference.cpp
+2 -2
View File
@@ -42,9 +42,9 @@ int main(int argc, char *argv[]) {
checkCuda(cudaMemcpy(input_d, input, idim.tot()*sizeof(dnnType), cudaMemcpyHostToDevice));
tk::dnn::dataDim_t dim = idim;
TIMER_START
TKDNN_TSTART
netRT.infer(dim, input_d);
TIMER_STOP
TKDNN_TSTOP
total_time+= t_ns;
// control output