Simple program to benchmark different compiler performance.
Current algorithm is the Pi(n) function.
which is finding the number of prime numbers less than n.
For following compilers are tested:
| Language | Compiler | Flags | time in secs (N=5000000) |
timer |
|---|---|---|---|---|
| Fortran | gfortran |
-O3 |
0.923 | system_clock |
| C++ | g++ |
-O3 |
0.929 | chrono |
| Haskell | ghc |
-O2 |
3.860 | Data.Time |
| Rust | rustc |
-C opt-level=3 |
0.940 | Instant |
| Python | cython3 + gcc |
-O2 |
17.73 | time |
| Julia |
unit: Microsoft Surface Pro 9
system: Ubuntu 24.04 x86-64, Linux 5.15-microsoft-standard-WSL2, Windows 11
hardware: Intel i7-1255U, 2600Mhz, 2P+8E Cores, 8GB RAM
Note: The .py python script file is first converted to .c using cython3, then compiled as a standalone executable using gcc compiler.
Haskell time resolution might be wrong for WSL2, see this article.
I tried to compile all the codes, then run them simultaneously by opening a new shell for each executable.
Not very elegant, you might have to uncomment this option inMakefile.