Conversation

Notices

1. Embed this notice
   linear cannon (linear@nya.social)'s status on Tuesday, 21-Nov-2023 02:44:22 JST linear cannon

   in reply to

   @chjara@akko.wtf the overhead of branch prediction would be maybe 3 or 4 cycles, compared to hundreds or thousands for the execution of the function

   • Haelwenn /элвэн/ :triskell: likes this.
   • Embed this notice
     linear cannon (linear@nya.social)'s status on Tuesday, 21-Nov-2023 02:44:25 JST linear cannon

     @chjara@akko.wtf if i were running on a real cpu instead of dosbox i might expect that
     
     but also, 486-class cpus have very limited branch prediction (basically none) and it shouldn't have enough of an effect here to matter