Let v1 be the target vector, v2 needs to be appended to the back of it.
I\'m now doing:
v1.reserve(v1.size() + v2.size());
copy(v2.begin(), v2.end()
I simply did a quick performance measurement with the following code and
v1.insert( v1.end(), v2.begin(), v2.end() );
seems to be the right choice (as already stated above). Nevertheless, you find the reported performance below.
Test code:
#include
#include
#include
//==============================================================================
//
//==============================================================================
/// Returns a vector containing the sequence [ 0, ... , n-1 ].
inline std::vector _range(const int n)
{
std::vector tmp(n);
for(int i=0; i testdata1 = _range(100000000);
const vector testdata2 = _range(100000000);
vector testdata;
printf("--------------------------------------------------------------\n");
printf(" METHOD: push_back()\n");
printf("--------------------------------------------------------------\n");
testdata.clear();
{ vector().swap(testdata); }
testdata = testdata1;
{
boost::timer::auto_cpu_timer t;
for(size_t i=0; i().swap(testdata); }
testdata = testdata1;
{
boost::timer::auto_cpu_timer t;
testdata.reserve(testdata.size() + testdata2.size());
for(size_t i=0; i().swap(testdata); }
testdata = testdata1;
{
boost::timer::auto_cpu_timer t;
testdata.insert( testdata.end(), testdata2.begin(), testdata2.end() );
}
printf("--------------------------------------------------------------\n");
printf(" METHOD: reserve() + insert()\n");
printf("--------------------------------------------------------------\n");
testdata.clear();
{ vector().swap(testdata); }
testdata = testdata1;
{
boost::timer::auto_cpu_timer t;
testdata.reserve( testdata.size() + testdata.size() );
testdata.insert( testdata.end(), testdata2.begin(), testdata2.end() );
}
printf("--------------------------------------------------------------\n");
printf(" METHOD: copy() + back_inserter()\n");
printf("--------------------------------------------------------------\n");
testdata.clear();
{ vector().swap(testdata); }
testdata = testdata1;
{
boost::timer::auto_cpu_timer t;
testdata.reserve(testdata.size() + testdata2.size());
copy(testdata2.begin(), testdata2.end(), back_inserter(testdata));
}
printf("--------------------------------------------------------------\n");
printf(" METHOD: reserve() + copy() + back_inserter()\n");
printf("--------------------------------------------------------------\n");
testdata.clear();
{ vector().swap(testdata); }
testdata = testdata1;
{
boost::timer::auto_cpu_timer t;
testdata.reserve(testdata.size() + testdata2.size());
copy(testdata2.begin(), testdata2.end(), back_inserter(testdata));
}
}
With Visual Studio 2008 SP1, x64, Release mode, /O2 /LTCG the output is as follows:
--------------------------------------------------------------
METHOD: push_back()
--------------------------------------------------------------
0.933077s wall, 0.577204s user + 0.343202s system = 0.920406s CPU (98.6%)
--------------------------------------------------------------
METHOD: reserve() + push_back()
--------------------------------------------------------------
0.612753s wall, 0.452403s user + 0.171601s system = 0.624004s CPU (101.8%)
--------------------------------------------------------------
METHOD: insert()
--------------------------------------------------------------
0.424065s wall, 0.280802s user + 0.140401s system = 0.421203s CPU (99.3%)
--------------------------------------------------------------
METHOD: reserve() + insert()
--------------------------------------------------------------
0.637081s wall, 0.421203s user + 0.218401s system = 0.639604s CPU (100.4%)
--------------------------------------------------------------
METHOD: copy() + back_inserter()
--------------------------------------------------------------
0.743658s wall, 0.639604s user + 0.109201s system = 0.748805s CPU (100.7%)
--------------------------------------------------------------
METHOD: reserve() + copy() + back_inserter()
--------------------------------------------------------------
0.748560s wall, 0.624004s user + 0.124801s system = 0.748805s CPU (100.0%)