Interpreter memory access optimizing method and device

Embodiments of the invention provide an interpreter memory access optimizing method and device, wherein the method comprises: generating interpreters respectively on at least two nodes of a non-uniform memory access NUMA system; according to numbers of nodes where interpreter threads are located, selecting one of the interpreters on the at least two nodes as a current interpreter; using the current interpreter to interpret and execute virtual machine commands called by the interpreter threads. According to the interpreter memory access optimizing method and device provided by the embodiments of the invention, the interpreters are generated respectively on at least two nodes, and one of the interpreters on the at least two nodes is selected as the current interpreter according to the numbers of the nodes where the interpreter threads are located during running, reducing the inter-node memory access quantity of the interpreters and therefore improving the performance of virtual machine interpreters.