The present U.S. patent application is related to the following U.S. patent applications filed concurrently:
(1) Application Ser. No. 11/479,179 filed Jun. 30, 2006, entitled “REQUESTER-GENERATED FORWARD OPERATION IN A CACHE COHERENCY PROTOCOL” and
(2) Application Ser. No. 11/479,576 filed Jun. 30, 2006, entitled “REPEATED CONFLICT ACKNOWLEDGEMENTS IN A CACHE COHERENCY PROTOCOL.”
Embodiments of the invention relate to a cache coherency protocol. More particularly, embodiments of the invention relate to resolution of conflicts in a system having multiple caching agents.
When an electronic system includes multiple cache memories, the validity of the data available for use must be maintained. This is typically accomplished by manipulating data according to a cache coherency protocol. As the number of caches and/or processors increases, the complexity of maintaining cache coherency also increases.
When multiple components (e.g., a cache memory, a processor) request the same block of data the conflict between the multiple components must be resolved in a manner that maintains the validity of the data. Current cache coherency protocols typically have a single component that is responsible for conflict resolution. However, as the complexity of the system increases, reliance on a single component for conflict resolution can decrease overall system performance.
Embodiments of the invention are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings in which like reference numerals refer to similar elements.
Techniques for distributed cache coherency conflict resolution in a multi-node system are described. In the following description, numerous specific details are set forth. However, embodiments of the invention may be practiced without these specific details. In other instances, well-known circuits, structures and techniques have not been shown in detail in order not to obscure the understanding of this description.
The cache coherency protocol described herein provides a snooping protocol without the limitations of a single, serializing bus. The protocol relies on nodes with cached copies of data to maintain coherency. In one embodiment, nodes are interconnected through point-to-point links between caching agents. In one embodiment, the cache coherency protocol may provide distributed conflict resolution by having nodes identify potential conflicts and involve the Home node in the resolution process. The Home node may participate in every transaction without being on the critical path in order to resolve conflicts.
The cache coherency protocol involves broadcasting an initial request to all peer nodes as well as the Home node. A conflict will be detected if each node monitors for conflicts after making a request. Because all nodes must respond to a broadcast request either by supplying the data or indicating they do not have a copy, the response can include an indication of a conflict, so conflicting nodes will detect the conflict.
Under certain conditions a requesting node may have received the requested data and acknowledged receipt to the Home node. However, before a completion message is received from the Home node, a conflicting request may be received from a peer node. This may be referred to as a “late conflict” condition. Techniques for resolution of the late conflict condition are described herein.
Specifically, the physical layer may provide communication between two ports over a physical interconnect comprising two uni-directional links. Specifically, one uni-directional link 104 from a first transmit port 150 of a first integrated device to a first receiver port 150 of a second integrated device. Likewise, a second uni-directional link 106 from a first transmit port 150 of the second integrated device to a first receiver port 150 of the first integrated device. However, the claimed subject matter is not limited to two uni-directional links.
The four nodes (220, 240, 260 and 280) may be any type of system component having a cache memory, for example, a processor. In one embodiment, the nodes and home agent may be interconnected via multiple point-to-point links (290, 291, 292, 293, 294, 295, 296, 297, 298, and 299) as described with respect to
In one embodiment, node controller 110 may include processing/control agent 114. Processing/control agent 114 may provide all or a portion of the functionality provided by node controller 110. Processing/control agent 114 may be implemented as hardware, software, firmware, or any combination thereof. Node controller 110 may also include additional circuits and functionality. In one embodiment, node controller 110 may be a gateway for communication beyond the cluster. Node controller 110 may also operate as a proxy home or caching agent for the home agents of remote clusters, if any. Node controller 110 may also serve as a proxy for the caching agents in the local cluster.
To request a block of data, Node A may transmit a Data Request (Data Req(A)) message to the Home Agent and a Snoop Request (SnpReq(A)) message to all peer nodes. The requested block of data may be any size block of data, for example, a cache line or a byte of data.
After sending the Response message, but before Node A has received ownership of the requested data, Node B may request the block of data by sending a Data Request (Data Req(B)) message to the Home Agent and a Snoop Request (SnpReq(B)) message to all peer nodes.
In response to the Snoop Request message, Node B may send a Response Forward (RspFwd) message to the Home Agent to indicate that the requested data is (or will be) forwarded to Node A. Node B also sends the data to Node A. In response to the Response Forward message, the Home Agent may send a Complete (Cmp) message to Node B to complete the request phases for Node A and Node B with respect to the requested block of data.
In one embodiment, each cluster (410, 420, 430, 440) is configured similarly to the cluster of
Reference in the specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.
While the invention has been described in terms of several embodiments, those skilled in the art will recognize that the invention is not limited to the embodiments described, but can be practiced with modification and alteration within the spirit and scope of the appended claims. The description is thus to be regarded as illustrative instead of limiting.
Number | Name | Date | Kind |
---|---|---|---|
5966729 | Phelps | Oct 1999 | A |
6009488 | Kavipurapu | Dec 1999 | A |
6631401 | Keller et al. | Oct 2003 | B1 |
6631448 | Weber | Oct 2003 | B2 |
6704842 | Janakiraman et al. | Mar 2004 | B1 |
6922756 | Hum et al. | Jul 2005 | B2 |
6954829 | Beers et al. | Oct 2005 | B2 |
7107409 | Glasco | Sep 2006 | B2 |
7536515 | Spink et al. | May 2009 | B2 |
20040122966 | Hum et al. | Jun 2004 | A1 |
20040123045 | Hum et al. | Jun 2004 | A1 |
20040123047 | Hum et al. | Jun 2004 | A1 |
20050144400 | Hum et al. | Jun 2005 | A1 |
20050160230 | Doren et al. | Jul 2005 | A1 |
20050237941 | Hum et al. | Oct 2005 | A1 |
20050240734 | Batson et al. | Oct 2005 | A1 |
20060053257 | Sistla et al. | Mar 2006 | A1 |
20080005483 | Spink et al. | Jan 2008 | A1 |
Number | Date | Country | |
---|---|---|---|
20080005487 A1 | Jan 2008 | US |