I'm looking to do some physics computation and I'm kind of interested in exploring the fabled numerical powers of FORTRAN, but I'm not really wanting to dig through piles and piles of pages all about FORTRAN 70. Is there a good, clean, modern reference?
Thanks in advance!
I'd be doubtful that FORTRAN was the bees knees anymore in terms of numerical prowess. Add to that the fact that it's probably not used that wisely (this was meant to be "widely" but it's funnier as-is, so I'll leave it in).
I'd be looking at using a more modern language with a numerical computation package, such as Python with NumPy, C/C++ with GMP or even MatLab.
GNU has their own FORTRAN compiler, part of GCC, if you're interested. You can get directly to the documentation from this link, part of the previous linked page.
There are several good books, alas somewhat pricey; I haven't seen them all (far from it), but it seems to me that this one is quite good if "money's no object". I don't really know what to recommend in terms of online materials, but this page points to some tutorials as well as compilers (I haven't checked out the tutorials so I don't know how good they are).
I do most of my numerical computing with Python (and numpy / scipy -- or gmpy when I want GMP for combinatorics and other advanced number theory stuff) but I don't mind turning to Fortran now and then for simple fast stuff. However, I generally use Fortran '77 for the purpose; the advanced features of Fortran '90 and beyond don't seem worth the candle to me, compared with using simple and highly optimizable F77 for some underlying algorithms (if nor already available in the above frameworks) and doing the "orchestration" &c in Python.
It seems that the standard text has been updated again.
But I tend to agree with Pax - unless you have just been handed a 30 year old million line project (in which case it will be F77 if you are lucky) I would look at python+libs or something like matlab.