Okay, here we go...
1) Don't use upper case letters in module names
(http://dlang.org/module.html#ModuleDeclaration)
2) As has already been suggested, if you're targeting raw
performance, don't use GC. You can always malloc and free your
storage. Using C heap has certain implications, but they're not
important right now.
3) Ditch extra constructors, they're completely unnecessary. For
matrix you only need your non-trivial ctor, postblit and lvalue
assignment operator.
4) This:
ref Matrix transpose() const {
return Matrix(this).transposeAssign();
}
just doesn't make any sense. You're returning a reference to a
temporary.
5) Use scoped imports, they're good :)
6) Use writefln for formatted output, it's good :)
With the above suggestions your code transfrorms into this:
http://dpaste.dzfl.pl/9d7feeab59f6
And here are the timings on my machine:
$ rdmd -O -release -inline -noboundscheck main.d
allocationTest ...
Time required: 1 sec, 112 ms, 827 μs, and 3 hnsecs
multiplicationTest ...
Time required: 1 sec, 234 ms, 417 μs, and 8 hnsecs
toStringTest ...
Time required: 998 ms, 16 μs, and 2 hnsecs
transposeTest ...
Time required: 813 ms, 947 μs, and 3 hnsecs
scalarMultiplicationTest ...
Time required: 105 ms, 828 μs, and 5 hnsecs
matrixAddSubTest ...
Time required: 240 ms and 384 μs
matrixEqualsTest ...
Time required: 244 ms, 249 μs, and 8 hnsecs
identityMatrixTest ...
Time required: 249 ms, 897 μs, and 4 hnsecs
LDC yields roughly the same times.
