I’ve been coding with plain C for the last few months because of our university curriculum. The programs I write are quite small and very not-resource-demanding. As a result, compiling with gcc has always been an almost instantaneous process (40ms on my "biggest" program so far).
Recently though, I enrolled in a local competition, for which C++ is apparently the norm, and so I started practicing. Despite my programs still being extremely small (not even half a kB), I noticed that compiling with g++ has consistently been almost like a 0.2s pause, followed by the instantaneous compilation I was used to with gcc.
Here’s the code I’m currently testing it on:
//#include <bits/stdc++.h> // I commented this out thinking that maybe the problem was due to too many header files being included
#include <iostream>
#include <vector>
#include <algorithm>
#include <climits>
using namespace std; // I know this is kind of a bad practice but I've been told it's recommended in competitions for ease of writing
int mex(vector<int> arr, vector<int> subarr) {
int min = INT_MAX;
for (auto i : arr) {
// If i doesn't exist in subarr
if (find(subarr.begin(), subarr.end(), i) == subarr.end()) {
if (i < min) {
min = i;
}
}
}
return min;
}
int main () {
vector<int> a = {0,1,2,3,4,5};
vector<int> suba = {0, 1,3,4};
cout << mex(a, suba) << endl;
return 0;
}
This compiles in 240ms.
Here’s g++’s ftime-report
:
Time variable usr sys wall GGC
phase setup : 0.00 ( 0%) 0.00 ( 0%) 0.00 ( 0%) 1562k ( 2%)
phase parsing : 0.16 ( 84%) 0.13 ( 87%) 0.29 ( 85%) 60M ( 83%)
phase lang. deferred : 0.01 ( 5%) 0.01 ( 7%) 0.02 ( 6%) 5818k ( 8%)
phase opt and generate : 0.02 ( 11%) 0.01 ( 7%) 0.03 ( 9%) 5468k ( 7%)
|name lookup : 0.05 ( 26%) 0.02 ( 13%) 0.03 ( 9%) 2550k ( 3%)
|overload resolution : 0.01 ( 5%) 0.00 ( 0%) 0.02 ( 6%) 6891k ( 9%)
callgraph construction : 0.01 ( 5%) 0.00 ( 0%) 0.01 ( 3%) 1448k ( 2%)
df scan insns : 0.01 ( 5%) 0.00 ( 0%) 0.00 ( 0%) 4128 ( 0%)
preprocessing : 0.01 ( 5%) 0.05 ( 33%) 0.05 ( 15%) 1805k ( 2%)
parser (global) : 0.03 ( 16%) 0.02 ( 13%) 0.07 ( 21%) 16M ( 23%)
parser struct body : 0.00 ( 0%) 0.03 ( 20%) 0.03 ( 9%) 11M ( 15%)
parser function body : 0.03 ( 16%) 0.00 ( 0%) 0.03 ( 9%) 3231k ( 4%)
parser inl. func. body : 0.01 ( 5%) 0.02 ( 13%) 0.01 ( 3%) 1474k ( 2%)
parser inl. meth. body : 0.01 ( 5%) 0.00 ( 0%) 0.03 ( 9%) 5790k ( 8%)
template instantiation : 0.07 ( 37%) 0.02 ( 13%) 0.08 ( 24%) 19M ( 27%)
constant expression evaluation : 0.00 ( 0%) 0.00 ( 0%) 0.01 ( 3%) 85k ( 0%)
expand : 0.00 ( 0%) 0.01 ( 7%) 0.00 ( 0%) 367k ( 0%)
integrated RA : 0.00 ( 0%) 0.00 ( 0%) 0.00 ( 0%) 2092k ( 3%)
LRA non-specific : 0.00 ( 0%) 0.00 ( 0%) 0.01 ( 3%) 14k ( 0%)
symout : 0.01 ( 5%) 0.00 ( 0%) 0.00 ( 0%) 6907k ( 9%)
rest of compilation : 0.00 ( 0%) 0.00 ( 0%) 0.01 ( 3%) 212k ( 0%)
TOTAL : 0.19 0.15 0.34 73M
I’m using gcc and g++ versions 11.4.0 on Ubuntu 22.04.
Is this a normal occurrence? Is g++ normally this noticeably slower than gcc? Or is there an underlying issue on my computer?
2
Answers
Yes, this is normal . The C++ language is far more complicated than the C language, and there’s more going on in the standard headers.
Enterprise-level projects can take hours to build, so be very glad yours is only taking 0.2 seconds !
Yes it is pretty common from my experience as C++ has many features and more complicated syntax.
If your code doesn’t use any C++ features the compile time may be almost identical, but using C++ introduces many more checks and logic that adds to the complexity of the process.
You can find a more detailed explaination for the reason here