To show just how significant parallelized algorithms can be, today I discovered pxz, a parallelized version of the
xz compression utility, which I use constantly. The proof is in the numbers:
|xz pxz -9e||2937M 2937M||305M 281M||0.104 0.096||32m 4m(!)|
I put this alias in my
alias tar='tar --use-compress-program=pxz'
Note that to build
pxz on the Mac, I had to comment out a reference to
MAP_POPULATE, which the OS X’s
mmap function doesn’t support.