A method and apparatus for pre-fetching data into a cache using a hardware element that includes registers for receiving a reference for an initial pre-fetch and a stride-indicator. The initial pre-fetch reference allows for direct pre-fetch of a first portion of memory. A stride-indicator is also received and is used along with the initial pre-fetch reference in order to generate a new pre-fetch reference. The new pre-fetch reference is used to fetch a second portion of memory.