I'm interested in working with the D programming language for simulating physics. How suitable is it? Is DMD's floating point code generation mature enough to compete with compilers available for c++ and fortran? Have fast matrix libraries been ported to D (or are they accessible from D)? Is there anything in the language itself that would inhibit/improve the floating point performance of the language?
I love programming in the language, however I'm worried that the compilers/libraries might not be suitable yet for numerical work.