DConf 2014 talks slides

[Era Scarecrow via Digitalmars-d]

  While we're on the topic of slides and it's slightly related, 
Walter Bright brought up an example of code that showed poor 
optimization that the compiler should be doing. Like using bit 
shifts to do multiply/divide by a power of 2.

  i/8  vs  i>>3
  i*8  vs  i<<3

  Is there resource of similar optimizations both as informational 
and things to avoid since it unintentionally obfuscates and makes 
the compiler's job harder? I know there was a page on low level 
optimizations and how to write code so the compiler made simpler 
better code on it's preferred types, but off hand i can't find 
it...