相同代码在不同的gcc编译器中具有巨大的性能差异

Question

Trying to understand the cause of the performance difference. I'm reading a ~70M file with the function below. Running the code with: gcc 4.4.6 takes less than a second gcc 3.2.3 takes more than 6 minutes

Most of the time is spend in the assign part. What was changed to make this speed difference between the 2 compilers ?

bool ReadFile(const string& path, string& file_data)
{
    ifstream ifs(path.c_str(), ifstream::in | ifstream::ate);
    if (!ifs) return false;
    int size = ifs.tellg();
    if (size==0) return false;
    ifs.seekg(0, ios::beg);
    file_data.assign((istreambuf_iterator<char>(ifs)),
       istreambuf_iterator<char>());

    return true;
}

Answer 1

Could you try tweak a bit this code (one extra line):

bool ReadFile(const string& path, string& file_data)
{
    ifstream ifs(path.c_str(), ifstream::in | ifstream::ate);
    if (!ifs) return false;
    int size = ifs.tellg();
    if (size==0) return false;
    ifs.seekg(0, ios::beg);

    file_data.reserve(size);
    file_data.assign((istreambuf_iterator<char>(ifs)),
       istreambuf_iterator<char>());

    return true;
}

and do measurement again
In second attempt you can try this:

bool ReadFile(const string& path, string& file_data)
{
    ifstream ifs(path.c_str(), ifstream::in | ifstream::ate);
    if (!ifs) return false;
    int size = ifs.tellg();
    if (size==0) return false;
    ifs.seekg(0, ios::beg);

    file_data.resize(size);
    return ifs.read(&file_data[0], size);
}