LC> This scheme has proven itself to be quite efficient on at least the Intel
LC> CPU. With Linux/Glibc and Solaris/Sun-libc, I can save up to 15% memory
LC> with block allocation. The allocation is a sligtly faster than plain
LC> malloc. I have a 3% speedup with a CPU intensive script and 10% speedup
LC> with a memory intensive application that allocates a lot of small memory
LC> memory segments (E.g., creating a hash with hundreds of thousands of
LC> keys).

Very nice.  I wonder how it would interact with the hash presizing
patch I proposed:

http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/~poffice/mail/ruby-talk/17470

thanks,
-joe