EP0817042A2
Expired / Lapsed
EP
⏱️ 0.0 years remaining
A multiprocessing system including an apparatus for optimizing spin-lock operations
Assignee: Sun Microsystems Inc
•
Filing: 1997-06-27
•
Granted: —
•
Expires: 2017-06-27
Abstract
Abstract A multiprocessing system having a plurality of processing nodes interconnected by an interconnect network. To optimize performance during spin-lock operations, a home agent prioritizes the servicing of read-to-own (RTO) transaction requests over the servicing of certain read-to-share (RTS) transaction requests, even if the RTO transaction requests are received by the processing node after receipt of the RTS transaction requests. In one implementation, this is accomplished by providing a first queue within the home agent for receiving RTO transaction requests conveyed via the interconnect network which is separate from a second queue for receiving RTS transaction requests. The queues may each be implemented with FIFO buffers. A home agent control unit is configured to service a pending RTO transaction request within the RTO queue before servicing an RTS transaction request in the second queue, even though the RTS transaction request was received by the processing node from the interconnect network prior to receiving the RTO transaction request.
Inventors
- 👤 Erik E. Hagersten
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-10-30 20:54
CPC Classifications
G; G06; G06F; G06F12/00; G06F12/02; G06F12/08; G06F12/0802; G06F12/0806; G06F12/0815; G06F12/0817; G06F12/0828; G06F12/0813; G06F9/00; G06F9/06; G06F9/46; G06F9/466; G06F2212/00; G06F2212/25; G06F2212/254; G06F2212/2542
Claims Analysis
Total Claims: 21
Claim 1
Independent Claim
107 words
An apparatus for use within a home node of a multiprocessing
computer system, said multiprocessing computer system including a
plurality of processing nodes interconnected by a network forming
a distributed shared memory architecture, said apparatus
comprising: a first queue coupled to receive read-to-own transaction
requests from said plurality of processing nodes; a second queue coupled to receive read-to-share transaction
requests from said plurality of processing nodes; and a home agent control unit coupled to receive said read-to-own
and said read-to-share transaction requests and
configured to service said read-to-own transaction
requests stored by said first queue and to service said
read-to-share transaction requests stored by said
second queue.
Claim 2
Dependent Claim
49 words
The apparatus for use within a home node of a
multiprocessing computer system as recited in claim 1 wherein
said home agent control unit is configured to service a given RTO
transaction request from said first queue prior to servicing a
given RTS transaction request from said second queue.
Claim 3
Dependent Claim
61 words
The apparatus for use within a multiprocessing computer
system as recited in claim 2 wherein said home agent control unit
is configured to service said given RTO transaction request prior
to servicing said given RTS transaction request, even if said
given RTS request is transmitted through said network prior to a
transmission of said given RTO transaction request through said
network.
Claim 4
Dependent Claim
67 words
The apparatus for use within a home node of a
multiprocessing computer system as recited in claim 2 wherein
said home agent control unit is configured to service said given
RTO transaction request prior to servicing said given RTS
transaction request, even if said given RTS request is stored
within said second queue prior to a storage of said given RTO
transaction request within said first queue.
Claim 5
Dependent Claim
66 words
The apparatus for use within a home node of a
multiprocessing computer system as recited in claim 1 further
comprising a transaction blocking unit coupled to said first and
second queue, wherein said transaction blocking unit is
configured to block transaction requests to a particular
coherency unit if another transaction request to said particular
coherency unit is currently being serviced by said home agent
control unit.
Claim 6
Dependent Claim
43 words
The apparatus for use within a home node of a
multiprocessing computer system as recited in claim 1 wherein
said first queue is implemented using a first FIFO buffer unit,
and wherein said second queue is implemented using a second FIFO
buffer unit.
Claim 7
Dependent Claim
48 words
The apparatus for use within a home node of a
multiprocessing computer system as recited in claim 1 further
comprising a directory unit coupled to said home agent control
unit, wherein said directory unit is configured to store
coherency information corresponding to coherency units of said
home node.
Claim 8
Dependent Claim
42 words
The apparatus for use within a home node of a
multiprocessing computer system as recited in claim 1 wherein
said read-to-own transaction requests are routed through said
first queue independent of a routing of said read-to-own
transaction requests through said second queue.
Claim 9
Dependent Claim
71 words
The apparatus for use within a home node of a
multiprocessing computer system as recited in claim 1 wherein
said home agent control unit is configured to service requests in
a ping-pong fashion such that a pending RTS transaction request
within said second queue is serviced after servicing a first RTO
transaction request pending within said first queue and before
servicing a second RTO transaction request pending within said
first queue.
Claim 10
Dependent Claim
35 words
The apparatus for use within a home node of a
multiprocessing computer system as recited in claim 1 further
comprising a third queue coupled to receive I/O transaction
requests from said plurality of processing nodes.
Claim 11
Independent Claim
67 words
A method of processing transaction requests at a home node
within a multiprocessing system having a distributed shared
memory architecture, said method comprising: receiving a read-to-share transaction request at said home
node; receiving a read-to-own transaction request at said home
node subsequent to said home node receiving said read-to-share
transaction request; and said home node servicing said read-to-own transaction
request prior to servicing said read-to-share
transaction request.
Claim 12
Dependent Claim
22 words
The method of processing transaction requests as recited in
claim 11 further comprising storing said read-to-own transaction
request within a first queue.
Claim 13
Dependent Claim
22 words
The method of processing transaction requests as recited in
claim 12 further comprising storing said read-to-share
transaction request in a second queue.
Claim 14
Dependent Claim
21 words
the method of processing transaction requests as recited in
claim 13 wherein said first queue is separate from said second
queue.
Claim 15
Dependent Claim
44 words
The method of processing transaction requests as recited in
claim 11 wherein said home node services said read-to-own
transaction request prior to servicing said RTS transaction
request, even if said read-to-own transaction request corresponds
to a coherency unit associated with said read-to-share
transaction request.
Claim 16
Dependent Claim
39 words
The method of processing transaction requests as recited in
claim 11 further comprising blocking a particular transaction
request for a given coherency unit if another transaction request
to said coherency unit is already being serviced by said home
node.
Claim 17
Dependent Claim
96 words
A home agent for use within a node of a multiprocessing
computer system comprising: a plurality of storage elements configured to receive
transaction requests from other nodes of said
multiprocessing computer system; and a home agent control unit coupled to receive said
transaction requests stored by said plurality of
storage elements, wherein said home agent control unit
is configured to service a given read-to-own
transaction request prior to servicing a given read-to-share
transaction request, even if said given read-to-share
transaction request is received by said node
prior to said node receiving said given read-to-own
transaction request.
Claim 18
Dependent Claim
28 words
The home agent as recited in claim 17 wherein said given
read-to-own transaction request and said given read-to-share
transaction request are each directed to a common coherency unit.
Claim 19
Dependent Claim
23 words
The home agent as recited in claim 17 wherein said plurality
of storage elements from a first queue for receiving read-to-own
transaction requests.
Claim 20
Dependent Claim
24 words
The home agent as recited in claim 19 wherein said plurality
of storage elements further form a second queue for storing read-to-share
transaction requests.
Claim 21
Dependent Claim
99 words
A multiprocessing computer system comprising: a plurality of processing nodes interconnected by an
interconnect network, wherein one of said processing nodes
includes: a plurality of storage elements configured to receive
transaction requests from other nodes of said
multiprocessing computer system; and a home agent control unit coupled to receive said
transaction requests stored by said plurality of
storage elements, wherein said home agent control unit
is configured to service a given read-to-own
transaction request with a higher priority than a given
read-to-share transaction request received by said one
of said processing nodes prior to said given read-to-own
transaction request.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.