Disabling of Caching on faster systems possible?

I’m using a very fast system with internal NVME arrays for extremely fast playback of 4K DPX sequences.

Dragging the view slider gives me nearly realtime playback from the drives. However, hitting the “play” button plays back significantly slower. I suspect that this is because caching is moving frames into memory, taking time. This is also an issue when the cache is full; then a clip is played back and trying to put itself into the cache as cache is freed up.

Is there a method of 100% disabling any caching mechanisms to allow the storage device(s) to operate at full speed without anything getting in the way (other than rendering, of course)? Theoretically I should be able to hit the ‘play’ button and get realtime playback unless there is something else that I am unaware of taking place…?

Welcome @sam.marrocco Are there other nodes connected to your source clip? If so and you are viewing the last node in the chain, each node has to be processed before it is cached. This is in addition to reading the file from disk. And no, there is not currently a way to disable the caching.

Thank you.

No other nodes are involved. I literally dragged in the sequence from a windows file explorer into sSilhouette and doubled-clicked it in Silhouette to cause it to view……then dragged the slider or hit the play button.

Scrubbing definitely displays more frames per second than ‘playing’.


Assuming that the only thing preventing complete disabling of the image cache system is the setting in Preferences/Cache/%Total Physical RAM being limited to a minimum of 10%, perhaps a future feature enhancement would be for that minimum to be removed so it may be set to “0%”. If caching has more ‘underpinnings’ that that, a “Global Cache Enable/Disable” would be useful for systems (like mine) that have fast enough storage/infrastructure that do not require the cache. This would also apply to systems using large RAM drives for media as well.

I am sure the caching can be disabled. However, disabling it would never allow you to cache any of the attached nodes.

It would be a welcome feature addition for us; thank you for the consideration.