Skip to content
Snippets Groups Projects
Commit d2514e28 authored by Jakub Klinkovský's avatar Jakub Klinkovský
Browse files

Fixed order of fields printed to stdout in the CUDA benchmark

parent ee1b52d3
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
tests/benchmarks/tnl-cuda-benchmarks.h
+ 16
− 12
View file @ d2514e28
...@@ -100,25 +100,28 @@ int main( int argc, char* argv[] ) ...@@ -100,25 +100,28 @@ int main( int argc, char* argv[] )
cout << bandwidth << " GB/sec." << endl; cout << bandwidth << " GB/sec." << endl;
*/ */
Real resultHost, resultDevice; Real resultHost, resultDevice, timeHost, timeDevice;
cout << "Benchmarking scalar product on CPU: "; cout << "Benchmarking scalar product on CPU: ";
timer.reset(); timer.reset();
timer.start(); timer.start();
for( int i = 0; i < loops; i++ ) for( int i = 0; i < loops; i++ )
resultHost = hostVector.scalarProduct( hostVector2 ); resultHost = hostVector.scalarProduct( hostVector2 );
timer.stop(); timer.stop();
timeHost = timer.getTime();
bandwidth = 2 * datasetSize / timer.getTime(); bandwidth = 2 * datasetSize / timer.getTime();
cout << bandwidth << " GB/sec." << endl; cout << "bandwidth: " << bandwidth << " GB/sec, time: " << timer.getTime() << " sec." << endl;
cout << "Benchmarking scalar product on GPU: " << endl; cout << "Benchmarking scalar product on GPU: ";
timer.reset(); timer.reset();
timer.start(); timer.start();
for( int i = 0; i < loops; i++ ) for( int i = 0; i < loops; i++ )
resultDevice = deviceVector.scalarProduct( deviceVector ); resultDevice = deviceVector.scalarProduct( deviceVector2 );
cout << "Time: " << timer.getTime() << endl;
timer.stop(); timer.stop();
timeDevice = timer.getTime();
bandwidth = 2 * datasetSize / timer.getTime(); bandwidth = 2 * datasetSize / timer.getTime();
cout << "Time: " << timer.getTime() << " bandwidth: " << bandwidth << " GB/sec." << endl; cout << "bandwidth: " << bandwidth << " GB/sec, time: " << timer.getTime() << " sec." << endl;
cout << "CPU/GPU speedup: " << timeHost / timeDevice << endl;
if( resultHost != resultDevice ) if( resultHost != resultDevice )
{ {
cerr << "Error. " << resultHost << " != " << resultDevice << endl; cerr << "Error. " << resultHost << " != " << resultDevice << endl;
...@@ -144,23 +147,24 @@ int main( int argc, char* argv[] ) ...@@ -144,23 +147,24 @@ int main( int argc, char* argv[] )
#endif #endif
#endif #endif
cout << "Benchmarking prefix-sum on CPU ..." << endl; cout << "Benchmarking prefix-sum on CPU: ";
timer.reset(); timer.reset();
timer.start(); timer.start();
hostVector.computePrefixSum(); hostVector.computePrefixSum();
cout << "Time: " << timer.getTime() << endl;
timer.stop(); timer.stop();
timeHost = timer.getTime();
bandwidth = 2 * datasetSize / loops / timer.getTime(); bandwidth = 2 * datasetSize / loops / timer.getTime();
cout << "Time: " << timer.getTime() << " bandwidth: " << bandwidth << " GB/sec." << endl; cout << "bandwidth: " << bandwidth << " GB/sec, time: " << timer.getTime() << " sec." << endl;
cout << "Benchmarking prefix-sum on GPU ..." << endl; cout << "Benchmarking prefix-sum on GPU: ";
timer.reset(); timer.reset();
timer.start(); timer.start();
deviceVector.computePrefixSum(); deviceVector.computePrefixSum();
cout << "Time: " << timer.getTime() << endl;
timer.stop(); timer.stop();
timeDevice = timer.getTime();
bandwidth = 2 * datasetSize / loops / timer.getTime(); bandwidth = 2 * datasetSize / loops / timer.getTime();
cout << "Time: " << timer.getTime() << " bandwidth: " << bandwidth << " GB/sec." << endl; cout << "bandwidth: " << bandwidth << " GB/sec, time: " << timer.getTime() << " sec." << endl;
cout << "CPU/GPU speedup: " << timeHost / timeDevice << endl;
for( int i = 0; i < size; i++ ) for( int i = 0; i < size; i++ )
if( hostVector.getElement( i ) != deviceVector.getElement( i ) ) if( hostVector.getElement( i ) != deviceVector.getElement( i ) )
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment