• Why GNU grep is fast (2010)

  • 用 Boyer-Moore (而且做點 loop unroll)

  • 避免在搜尋前複製 input bytes

  • 在 match 之前不找 newlines

  • 看 “Fast String Searching” (by Andrew Hume and Daniel Sunday, in the November 1991 issue of Software Practice & Experience) 裡的 Boyer-Moore implementation tricks

  • Try to set things up (page-aligned buffers, page-sized read chunks, optionally use mmap) so the kernel can ALSO avoid copying the byte

• Wikipedia - Boyer–Moore string search algorithm