Caching
Last updated
Rendering a frame can become expensive. After a frame is rendered, this value is stored to reuse later. This helps improve realtime playback, and reduces the time required to preview work that you’ve already done.
Cache hashing
Section titled “Cache hashing”Cached items are unique and match to the hash key calculated when rendering the frame. If a frame’s hash matches a cache entry this can be sent to the GPU to be displayed, preventing the need to re-render a frame.
Cache tiers
Section titled “Cache tiers”There are three tiers to the cache. The cache works on a first in, first out approach. When a cache is filled, it removes the oldest frame, and the frame is demoted to the next cache. If the frame is demoted out of the disk cache, it is deleted.
Caches can be cleared by clicking on the cache in Lumit’s bottom strip, or within settings.
The VRAM cache stores cached frames directly on the GPU. These allow for the fastest loading of files to display, and the most recently cached frames are stored on it. When Lumit closes, these cached frames are cleared.
The RAM cache remains relatively fast to load a preview frame. When Lumit closes, these cached frames are cleared.
The disk cache is the slowest, but remains faster for intensive compositions. When Lumit closes, these cached frames do not clear, meaning they can fill up the RAM and VRAM caches upon loading. While using Lumit, the other caches will send cached frames to the disk cache to keep them across sessions.