Very clean and high-level functional optimization techniques and hints. Erlang doesn't seem to have a way to circumvent the compiler for people who REALLY know what they're doing.
We Lispers have high-level optimization guidelines too, sure, but we also have the sort of "feel the bits between your toes stuff", like:
[Edit: HN doesn't do code snippets, shell output or anything even remotely hackish. I tried the three-spaces-before-a-line trick and it doesn't work. oh well]
Much of it seems dated - before R7B. Just program as usual and measure in order to optimize. That works as a general formula for all languages. For more fun, look at the Haskell optimization guide! http://haskell.org/haskellwiki/Performance
We Lispers have high-level optimization guidelines too, sure, but we also have the sort of "feel the bits between your toes stuff", like:
http://common-lisp.net/project/cmucl/doc/cmu-user/
The canonical example is in psuedo C: f(x,y) = (x|y)&(x^y)
See the x86 disassembly output of that in
http://www.cons.org/cmucl/doc/reading-disassembly.html
[Edit: HN doesn't do code snippets, shell output or anything even remotely hackish. I tried the three-spaces-before-a-line trick and it doesn't work. oh well]