Open Models Almanac
Methodsoptimizationtraining runtime

single source · optimization

Double-buffered CUDA-event evaluation

Evaluates CUDA events from the preceding training step while current-step work proceeds, using double buffering.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See training runtime for the methods filed beside it.

Also called use double-buffering, double-buffering.

Filed under optimization › training runtime