Arteris, Inc.

Applied Micro Circuits Corporation

David Alan Kruckemyer

Craig Stephen Forrest

John Gregory Favor

Matthew Ashcraft

In accordance with various aspects of the invention, a recall transaction is issued if a tag filter entry needs to be freed up for an incoming transaction. Directory entries chosen for a recall transaction are pushed into a fully associative structure called victim buffer. If this structure gets full, then an entry is selected from entries inside the victim buffer for the recall.

Cache coherent system implementing victim buffers

Method for using victim buffer in cache coherent systems

A cache coherent system includes a directory with more than one snoop filter, each of which stores information in a different set of snoop filter entries. Each snoop filter is associated with a subset of all caching agents within the system. Each snoop filter uses an algorithm chosen for best performance on the caching agents associated with the snoop filter. The number of snoop filter entries in each snoop filter is primarily chosen based on the caching capacity of just the caching agents associated with the snoop filter. The type of information stored in each snoop filter entry of each snoop filter is chosen to meet the desired filtering function of the specific snoop filter.

Configurable snoop filters for cache coherent systems

Victim buffer for cache coherent systems

A system and method are disclosed that include a bridge that translates non-coherent transactions, which are received from a non-coherent subsystem, into one or more coherent transactions to be issued to a coherent subsystem. The bridge also buffers data coherently in an internal cache, also known as a proxy cache, based on certain attributes of the non-coherent transaction. The invention may be applied to any cache, which receives read and write transactions that become coherent transactions.

Proxy cache conditional allocation

Compression of address bits within a cache coherent subsystem of a chip is performed, enabling a cache coherent subsystem to avoid transmitting, storing, and operating upon unnecessary address information. Compression is performed according to any appropriate lossless algorithm, such as discarding of bits or code book lookup. The algorithm may be chosen according to constraints on logic delay and silicon area. An algorithm for minimum area would use a number of bits equal to the rounded up binary logarithm of the sum of all addresses of all memory regions. A configuration tool generates a logic description of the compression algorithm. The algorithm may be chosen automatically by the configuration tool. Decompression may be performed on addresses exiting the coherent subsystem.

Compression of hardware cache coherent addresses

A system and method are disclosed for multiple coherent caches supporting agents that use different, incompatible coherence models. Compatibility is implemented by translators that accept coherency requests and snoop responses from an agent and accept snoop requests and coherency responses from a coherence controller. The translators issue corresponding coherency requests and snoop responses to the coherence controller and issue corresponding coherency responses and snoop requests to the agent. Interaction between translators and the coherence controller accord with a generic coherence model, which may be a subset, superset, or partially inclusive of features of any native coherence model. A generic coherence protocol may include binary values for each of characteristics: valid or invalid, owned or non-owned, unique or shared, and clean or dirty.

System and method for adaptation of coherence models between agents

A memory system having increased throughput is disclosed. Specifically, the memory system includes a first level write combining queue that reduces the number of data transfers between a level one cache and a level two cache. In addition, a second level write merging buffer can further reduce the number of data transfers within the memory system. The first level write combining queue receives data from the level one cache. The second level write merging buffer receives data from the first level write combining queue. The level two cache receives data from both the first level write combining queue and the second level write merging buffer. Specifically, the first level write combining queue combines multiple store transactions from the load store units to associated addresses. In addition, the second level write merging buffer merges data from the first level write combining queue.

Multi-level store merging in a cache and memory hierarchy

Various aspects provide for implementing a cache coherence protocol. A system comprises at least one processing component and a centralized controller. The at least one processing component comprises a cache controller. The cache controller is configured to manage a cache memory associated with a processor. The centralized controller is configured to communicate with the cache controller based on a power state of the processor.

Broadcast messaging and acknowledgment messaging for power management in a multiprocessor system

A load/store unit with an outstanding load miss buffer and a load miss result buffer is configured to read data from a memory system having a level one cache. Missed load instructions are stored in the outstanding load miss buffer. The load/store unit retrieves data for multiple dependent missed load instructions using a single cache access and stores the data in the load miss result buffer. The outstanding load miss buffer stores a first missed load instruction in a first primary entry. Additional missed load instructions that are dependent on the first missed load instructions are stored in dependent entries of the first primary entry or in shared entries. If a shared entry is used for a missed load instruction the shared entry is associated with the primary entry.

Outstanding load miss buffer with shared entries

A load/store unit with an outstanding load miss buffer and a load miss result buffer is configured to read data from a memory system having a level one cache. Missed load instructions are stored in the outstanding load miss buffer. The load/store unit retrieves data for multiple dependent missed load instructions using a single cache access and stores the data in the load miss result buffer. When missed load instructions are reissued from the outstanding load miss buffer, data for the missed load instructions are read from the load miss result buffer rather than the level one cache. Because the data is stored in the load miss result buffer, other instructions that may change the data in level one cache do not cause data hazards with the missed load instructions.

Load store unit with load miss result buffer

A load/store unit with an outstanding load miss buffer and a load miss result buffer is configured to read data from a memory system having a level one cache. Missed load instructions are stored in the outstanding load miss buffer. The load/store unit retrieves data for multiple dependent missed load instructions using a single memory access and stores the data in the load miss result buffer. The load miss result buffer includes dependent data lines, dependent data selection circuits, shared data lines and shared data selection circuits. The dependent data selection circuits are configured to select a subset of data from the memory system for storing in an associated dependent data line. Similarly, the shared data selection circuits are configured to select a subset of data from the memory system for storing in an associated shared data line.

Growing community of inventors

San Jose, CA, United States of America

David Alan Kruckemyer