Invention Grant
US08217810B2 Memory management for prediction by partial matching context models 有权
通过部分匹配上下文模型进行预测的内存管理

Memory management for prediction by partial matching context models
Abstract:
Techniques for resource management of a PPM context model are described herein. According to one embodiment, in response to a sequence of symbols to be coded, contexts are allocated, each having multiple entries and each entry representing a symbol that the current context is able to encode, including a counter value representing a frequency of each entry being used. For each symbol coded by a context, a local counter value and a global counter value are maintained. The global counter value represents a total number of symbols that have been coded by the context model and the local counter value represents a number symbols that have been coded by the respective context. Thereafter, a resource management operation is performed for system resources associated with the plurality of contexts based on a global counter value and a local counter value associated with each of the plurality of contexts.
Information query
Patent Agency Ranking
0/0