I posted this as part of a 'bug' earlier, but since that was not actually a compiler bug (but one in my code ;), the performance part of it may have been lost in the shuffle.

Attached are the .d and .cod files for both a 'for' and 'foreach' version of a benchmark using dynamic int[][]'s.

Although the 'foreach' version is faster, both versions are significantly slower than the equivalent for DMC.

Relative times are in the 'results' file.

BTW - Should I be posting these here, or in digitalmars.D?

Thanks,

Dave