Prefetcher based speculative dynamic random-access memory read request technique

A memory system comprising:

• a penultimate-level cache;
• a last-level cache;
• a main memory;
• a first counter configured to store a request rate of speculative memory read requests from the penultimate-level cache to the main memory, each of the speculative memory read requests being in response to a miss in the penultimate-level cache of a corresponding data read request;
• a second counter configured to store an indicator of a hit rate of searches of the last-level cache for data of data read requests missed in the penultimate-level cache, and
• a controller associated with the penultimate-level cache, the controller being configured to:
• selectively enable core demand speculative memory read requests to the main memory in parallel with the searches based on the request rate, the hit rate, a predetermined request rate threshold value, and a predetermined hit rate threshold value;
• and
• selectively enable prefetch speculative memory read requests to the main memory in parallel with the searches based on the request rate, the hit rate, the predetermined request rate threshold value, and the predetermined hit rate threshold value,
• wherein each of the speculative memory read requests is a core demand speculative memory read request or a prefetch speculative memory read request.

A method for operating a memory system comprising:

• enabling core demand speculative memory read requests to a main memory in parallel with searching of a last-level cache for data requested by core demand data read requests that missed in a penultimate-level cache;
• disabling prefetch speculative memory read requests to the main memory for data requested by prefetch data read requests that missed in the penultimate-level cache;
• monitoring a request rate of speculative memory read requests issued from the penultimate-level cache to the main memory, the request rate being based on core demand speculative memory read requests and unrealized prefetch speculative memory read requests for prefetch data read requests that missed in the penultimate-level cache;
• monitoring a hit rate of the searching of the last-level cache;
• and
• in response to the request rate being less than a predetermined request rate threshold value and the hit rate being less than a predetermined hit rate threshold value:
• enabling the core demand speculative memory read requests;
• and
• disabling the prefetch speculative memory read requests.