I am glad to see this participation on this issue :)
The hints about trying another compiler and std.mmfile turned out
to be very effective.
Even this simple code is faster then my systems "wc -l" now:
void main() {
import std.stdio;
writeln(lcs("benchmark.dat"));
}
size_t lcs(string filename) {
import std.mmfile;
auto f = new MmFile(filename);
auto data = cast(ubyte[]) f[];
size_t c = 0;
foreach(ref i ; data) {
if (i == '\n') c++;
}
return c;
}
time wc -l ./benchmark.dat
10500000 ./benchmark.dat
wc -l ./benchmark.dat 0,06s user 0,03s system 99% cpu 0,089
total
time ./lcs
10500000
./lcs 0,05s user 0,01s system 99% cpu 0,061 total
