Experimental Features

GNOLL has the following experimental features:

Central Limit Therom + Box-Muller

Providing USE_CLT=1 to make will enable an optimization for huge dice rolls.

Regular Roll Distribution CLT Roll Distribution
Distribution of GNOLL before this feature is enabled Distribution of GNOLL before this feature is enabled. It is a similar graph

We can roll on a normalized random distribution once, rather than a scaling amount of times on a uniform random distribution. This makes GNOLL perform at a near constant time for large dice pools.

Performance Diagram - CLT is a constant speed, Previously performance scaled with number of dice

Warning: There are slight differences in data distribution. The CLT method tends slightly less to the center of the graph than the regular algorithm.

Warning: As CLT calculates the result of many dice in a single step, the information on a per-die basis is lost. Operations such as dropping/introspection are limited or nonfunctional when CLT is enabled.