FAQ/Caching

From VFXPedia

< FAQ
Revision as of 23:46, 18 September 2007 by Daniel (Talk | contribs)
(diff) ← Older revision | Current revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Contents

Basics

As of version 4, Fusion uses an intelligent caching system that attempts to retain the results of all processed tools, at different stages and times. This system is described in brief in the manual.

The philosophy behind Fusion's caching system is, "If the memory is available, why not take full advantage of it to avoid re-processing tools unnecessarily?". To this end, even the simplest of comps will try to retain all relevant images and results, and RAM usage will climb rapidly up to the limits defined in Preferences/Memory (by default, this is up to 60% of your RAM).

Size Limits

It is important to remember that Fusion's cache will not grow beyond the boundaries set by these limits, and that you can set them to reduce any impact on other applications you may be running, or even turn off the cache altogether. Fusion will take only as much RAM as it requires for rendering, and the balance between this and the Prefs/Memory limits will be used for caching. If the comp is particularly demanding of RAM, the cache will be correspondingly reduced, and overall memory usage will remain relatively constant. Note that in extreme cases, Fusion's total memory usage may sometimes be greater that the cache boundaries set in Prefs, but if this happens, no memory will be used by the cache at all. In some cases, Fusion may automatically decrease the cache limits in Prefs to reduce the overall system impact of the cache.

Controls

The two main controls for limiting the cache usage are in Preferences/Memory.

  • Limit Caching determines the maximum percentage of the total RAM to use for the cache (note that 32 bit versions of Fusion are further limited to 2-4 GB depending on your system). A setting of 60% on a 2 GB system will limit the cache to 1.2 GB. In most cases, a maximum of 80% is recommended, as this still leaves some spare space for other apps and operating system needs.
  • Leave at least n MBytes will further limit caching when the system's available free RAM drops below n megabytes. Setting this to 100MB will try to leave at least 100 MB of free memory at all times, allowing RAM for the system or other apps to use. Setting it to 0 will allow Fusion to use the full amount of RAM specified by the Limit Caching control, ignoring other apps. This may cause Windows to swap out other applications as cache levels increase.

Priorities

Since not all tool results can usually fit into the cache, Fusion assigns priorities to the objects it caches. The algorithm for determining the best objects to cache is complex, and takes into account how slow the tool is to render, how often its results are needed, whether playback or rendering is occurring, and many other factors. The lowest priority objects are dropped from the cache.