You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
See Doornik and the wikipedia page; it's apparently faster to use a integer table of x[i] / x[i - 1] then compare the rng integer value against that. Requires that we keep two tables, one for uint32 and uint64 of course, and some benchmarking and explanation would be nice
The text was updated successfully, but these errors were encountered:
richfitz
changed the title
Use integers for cutoff comparison
Use integers for ziggurat cutoff comparison
Nov 8, 2021
This is going to be really hard as we can't easily get two versions of the coefficients for different integer types given the big fight in #327/#328; I still think it could be worth doing if faster but don't believe it will be that much faster...
See Doornik and the wikipedia page; it's apparently faster to use a integer table of
x[i] / x[i - 1]
then compare the rng integer value against that. Requires that we keep two tables, one for uint32 and uint64 of course, and some benchmarking and explanation would be niceThe text was updated successfully, but these errors were encountered: