METHOD FOR CONTROLLING MEMORY IN GPU ACCELERATED IMAGE COMPOSITING SOFTWARE

Disclosed is a memory management method for graphics processing unit (GPU) accelerated image compositing software. The memory management method includes a step in which a processor analyzes a plurality of nodes to be processed using a current node and first data generated at the current node; a step in which the processor selects one or more of the nodes directly connected to the nodes to be processed based on the current node and first data generated at the current node; and a step in which a GPU memory is managed to initiate the processor to save second data generated at the selected one or more nodes. [Reference numerals] (AA) Start; (BB) End; (S10) Analyze nodes; (S20) Determine a node directly connected to a render pass; (S30) Store data generated at the node in a GPU cash; (S40) Change the current node?; (S50) Backup?; (S60) Compression?; (S70) Compress; (S80) Store in the GPU cash; (S90) Remove from the GPU cash