WO2011051074A1 - Data processing method, device, and system for processing requests in a multi-core system - Google Patents
Data processing method, device, and system for processing requests in a multi-core system Download PDFInfo
- Publication number
- WO2011051074A1 WO2011051074A1 PCT/EP2010/064536 EP2010064536W WO2011051074A1 WO 2011051074 A1 WO2011051074 A1 WO 2011051074A1 EP 2010064536 W EP2010064536 W EP 2010064536W WO 2011051074 A1 WO2011051074 A1 WO 2011051074A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- logical partition
- request
- filter
- record
- unit
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F12/00—Accessing, addressing or allocating within memory systems or architectures
- G06F12/02—Addressing or allocation; Relocation
- G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
- G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
- G06F12/0806—Multiuser, multiprocessor or multiprocessing cache systems
- G06F12/0815—Cache consistency protocols
- G06F12/0831—Cache consistency protocols using a bus scheme, e.g. with bus monitoring or watching means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
- G06F2212/10—Providing a specific technical effect
- G06F2212/1028—Power efficiency
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
- G06F2212/10—Providing a specific technical effect
- G06F2212/1041—Resource optimization
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Definitions
- the present invention relates to the field of data processing in a computer system, and in particular, relates to a data processing method, device, and system for processing requests in a multi-core system.
- each core on a processor has a private LI cache (not shown in Fig. 1) for holding some data.
- Multiple cores may share one L2 cache, and then a plurality of L2 caches 130A, 130B, 130C, and 130D may form a cache coherence system.
- the current computer system has a cache coherence mechanism.
- the cache coherence mechanism is used to help the system to maintain data coherence, which means that if more than one backup of a data block is located in more than one LI and L2 cache at the same time, the coherence mechanism may guarantee that the values of all backups at each time are completely identical.
- the process of the cache coherence mechanism may be described as follows:
- a core 1 on the L2 cache 13 OB intends to load a data block and searches the data block in its private LI cache A (not shown). Then, if backup of the data block is found in the private LI or L2 cache 130B, then the data loading process ends, and the coherence process will not be called. However, if backup of the data block is not found in the private LI or L2 cache 13 OB, then the coherence process will be called.
- the request for the data block will be transmitted by the L2 cache 130B to all other L2 caches 130A, 130C, and 130D. Next, the L2 caches 130A, 130C, and 130D will search the requested data block in their own arrays.
- the cache which finds the backup will issue a confirmation message to the L2 cache 13 OB. Meanwhile, the cache which finds the backup will remove its own data backup so as to guarantee that the values of the data block in all caches are identical. However, if none of other L2 caches 130A, 130C, and 130D finds the backup of the data block, then a memory will issue the data to the L2 cache 13 OB.
- the cache coherence mechanism aims at enabling data backups among all L2 caches to be completely identical, thereby achieving coherency of data.
- One solution thereof is based on a hash table.
- This solution uses a hash table to encode and record logical partition IDs (LPID) of all logical partitions running on L2 caches.
- LPID logical partition IDs
- the LPID attached to the data request is extracted through a hash function and mapped to a value.
- the value then is compared with all entries recorded in the hash table.
- Another solution is based on historical records of data access. This solution uses a historical record table for maintaining records of data access.
- the filtration ratio obtained through this solution depends on the size of the historical record table to too great an extent. The more the requests for filtration are, the larger should be the size of the table. Thus, it is easily seen that a large-sized historical record table per se will cost hardware module size and power consumption. Thus, in order to reduce power consumption, it is desirable to provide a system and method for reducing and even eliminating the above- mentioned unnecessary process of searching L2 caches.
- An object of the present invention is to provide a method, device and system to implement the above objects with a filter mechanism in caches.
- a feature of the present invention is to provide a pipe stage-based data processing method, device and system.
- a method for processing a request in a multi-core system comprising steps of: receiving a request for data by a filter from a requesting unit; comparing an indicator indicative of a logical partition in the request with an indicator indicative of a logical partition in a record of the filter; searching in a unit where the filter is located based on the request and returning a search result to the requesting unit if the comparison result matches; and returning a NONE response to the requesting unit from the filter.
- a device for processing a request in a multi-core system comprising: a filter configured to receive a request for data transmitted from a requesting unit and to compare an indicator indicative of a logical partition in the request with an indicator indicative of a logical partition in a record of the filter; wherein the filter is further configured to search in a unit where the filter is located based on the request and return a search result to the requesting unit if the comparison result matches; and to return a NONE response to the requesting unit if the search result does not match.
- a multi-core system comprising a first cache, and one or more second caches, wherein the first cache is configured to issue a request for data to the one or more second caches; each of the one or more second caches comprises a filter configured to receive the request and compare an indicator indicative of a logical partition in the request with an indicator indicative of the logical partition in a record of the filter; wherein the filter is further configured to search in the first cache based on the request and return a search result to the one or more second caches if the comparison result matches, and to return a NONE response to the one or more second caches from the filter if the search result does not match.
- Fig. 1 shows a multi-core system having a cache coherence mechanism
- Fig. 2 shows a processing device for a request for data according to the present invention
- Fig. 3 shows a flow chart of a processing method for a request for data according to the present invention
- Fig. 4 shows a specific configuration of a processing device for a request for data according to an exemplary embodiment of the present invention
- FIG. 5 shows a flow chart of a processing method for a request for data according to an exemplary embodiment of the present invention
- Fig. 6 shows a specific configuration of a processing device for a request for data according to another exemplary embodiment of the present invention
- Fig. 7 shows a flow chart of a processing method for a request for data according to another exemplary embodiment of the present invention.
- Fig. 8 shows a structural schematic diagram of a multi-core system in which the present invention may be implemented.
- Fig. 2 shows a data processing system 200 in a multi-core system according to a preferred embodiment of the present invention.
- the system 200 comprises a requesting unit 202 configured to issue a request for data; and a filter 201-1 for receiving the request for data from the requesting unit 202 and comparing an indicator indicative of a logical partition in the request with an indicator indicative of a logical partition in records of the filter 201-1.
- the requesting unit 202 performs data search in its own logical partition before issuing a request for data, and in case of not finding the required data, the requesting unit 202 issues a data request to other units, for example unit 203, 204, and 206, wherein a coherence mechanism may exist between the requesting unit and other units.
- the filter 201-1 is further configured to search in the unit 206 where the filter 201-1 is located based on the request and return a search result to the requesting unit 202 which issues the request if the comparison result matches, and to directly return a NONE response to the unit 202 which issues the request from the filter 201-1 if the search result does not match. In this way, in the case that the comparison result does not match, it is not necessary to access the unit 206, thereby reducing data access in the multi-core system so as to reduce power consumption.
- the structure of the requesting unit 202 may be identical to the structure of the unit 206. Additionally, there may be many units 203, 204, etc., which have similar structures.
- the requesting unit 202 and the unit 206 are caches. Preferably, the caches may be L2 caches.
- the structure of the requesting unit 202 may be identical to the structure of the unit 206.
- the requesting unit 202 may also comprise a filter 201-4. More preferably, the units 203 and 204 may comprise a filter 201-2 and a filter 201-3, respectively.
- the indicator indicative of a logical partition in a record of the filter 201-1 is dynamically updated.
- the dynamic update process may be implemented under the control of a manager (not shown) similar to the manager 140 as shown in Fig. 1.
- the system 200 of the present invention as shown in Fig. 2 further comprises other components, such as a bus and other hierarchy, etc. In order not to confuse the present invention, description of such components is omitted.
- Fig. 3 shows a flow chart of a data processing method in a cache according to an embodiment of the present invention.
- a request for data is received by a filter from the requesting unit.
- the indicator indicative of a logical partition in the request is compared with the indicator indicative of a logical partition in the record of the filter. If the comparison result matches, the process continues to step 303, where search is performed in the unit where the filter is located based on the request, and the search result is returned to the requesting unit. If the comparison result does not match, then the process continues to step 304, where a NONE response is returned to the requesting unit from the filter.
- Fig. 4 shows a specific configuration of a processing device for a request for data according to an exemplary embodiment of the present invention.
- the system 400 preferably comprises: a requesting unit 402 configured to issue a request for data, wherein the request for data is attached to a logical partition ID; and a filter 401-1 for receiving the request for data from the requesting unit 402, the filter 401-1 comprising a register 405 in which a record having an indicator indicative of a logical partition is held.
- the record in the register 405 is associated with a plurality of logical partitions, wherein some logical partitions of the plurality of logical partitions are located in the unit 406 where the filter 401-1 is located, and the plurality of logical partitions have their own logical partition ID, respectively.
- the number of logical partitions may be random or determined as required.
- the record in the filter 401-1 is associated with respective logical partition IDs of the plurality of logical partitions in the multi-core system, and one bit in the indicator indicative of a logical partition in the record is used to represent a logical partition or a logical partition ID.
- the bit location in the indicator indicative of a logical partition in the record of the filter 401-1 and associated with this logical partition is set as 1; if a logical partition is not located in a unit 406 where the filter 401-1 is located, the bit location in the indicator indicative of a logical partition in the record of the filter 401-1 and associated with this logical partition (or the logical partition ID) is set as 0.
- the filter 401-1 is further configured to split the logical partition ID attached to the request and shift it so as to form a mask, wherein the shift may be left shift or right shift.
- the bit width of the mask is identical to the bit width as recorded in the filter 401-1.
- the bit indicative of the logical partition ID split from the request is set as "1", and all remaining bits are set as zero; if a logical partition is located in the unit 406 where the filter 401-1 is located, then the corresponding bit of the indicator indicative of the logical partition in the record is set as "1".
- the mask comprises a plurality of bits, wherein if, for example, it is a second bit that indicates the split logical partition ID, then the second bit is set as "1".
- the second bit in the record is also set as "1" so as to indicate that the logical partition is located in the unit 406.
- the filter 401-1 is further configured to perform an AND operation to the mask and the indicator indicative of a logical partition in the record which have the same bit width. If the result of the AND operation is none-zero, it indicates matching, and then based on the request, search is performed in unit 406 where the filter 401-1 is located, and the search result is returned to the requesting unit 402 which issues the request. If the result of the AND operation is zero, it indicates no matching, and then a NONE response is directly returned from the filter 401-1 to the unit 402 which issues the request, without searching the unit 406, thereby reducing data access thereby achieving the object of reducing power consumption.
- the bit positions corresponding to the two logical partitions in the indicators indicative of the logical partitions in the record of the filter 401-1 are set as 1, while the bit positions corresponding to other logical partitions are set as zero.
- the requesting unit 402 and the unit 406 may be caches.
- the caches may be L2 caches.
- the structure of the requesting unit 402 may be identical to the structure of the unit 406.
- the indicator indicative of the logical partition in the record of the filter 401 is dynamically updated.
- the dynamic update process may be implemented under the control of a manager (not shown) similar to the manager 140 as shown in Fig. 1.
- the requesting units 402, 403, and 404 may have filters 401-4, 401-2, and 401-3, respectively.
- the system 400 of the present invention as shown in Fig. 4 further comprises other units, such as a bus and other hierarchy, etc. In order not to confuse the present invention, description of such components is omitted.
- FIG. 5 shows a flow chart of a processing method for a request for data according to an exemplary embodiment of the present invention.
- the requesting unit issues a request for data.
- the request comprises a logical partition ID.
- the request for data is received by a filter in which a record associated with a plurality of logical partitions, the record having indicators indicative of logical partitions is held.
- the plurality of logical partitions has their own logical ID, respectively.
- the number of logical partitions may be random or determined as required.
- the record in the filter is associated with respective logical partition ID of the plurality of logical partitions in the multi-core system, and one bit in the indicator indicative of a logical partition in the record is used to represent a logical partition or a logical partition ID. If a logical partition is located in the unit where the filter is located, the bit location in the indicator indicative of a logical partition in the record of the filter and associated with this logical partition (or the logical partition ID) is set as 1; if a logical partition is not located in a unit where the filter is located, the bit location in the indicator indicative of a logical partition in the record of the filter and associated with this logical partition (or the logical partition ID) is set as 0.
- the logical partition ID attached to the request is split from the request and shifted so as to form a mask, wherein shift may be left shift or right shift.
- the bit width of the mask is identical to the bit width of the indicator indicative of the logical partition in the record of the filter.
- the mask comprises a plurality of bits, wherein if, for example, it is a second bit that indicates the split logical partition ID, then the second bit is set as "1".
- the second bit in the record is also set as "1" so as to indicate that the logical partition is located in the unit.
- the filter performs AND operation to the mask and the indicator indicative of a logical partition in the record which have the same bit width to check whether the result of AND operation is non-zero. If the result of AND operation is non-zero, it indicates matching, and then at step 505, based on the request, search is performed in the unit where the filter is located, and the search result is returned to the requesting unit which issues the request. If the result of AND operation is zero, it indicates no matching, and then at step 506, a NONE response is directly returned from the filter to the unit which issues the request, without searching, thereby reducing data access to achieve the object of reducing power consumption.
- Fig. 6 shows a specific configuration of a processing device for a request for data according to another exemplary embodiment of the present invention.
- the system 600 preferably comprises: a requesting unit 602 configured to issue a request for data, wherein the request for data is attached with a logical partition ID; and a filter 601-1 for receiving the request for data from the requesting unit 602, comprising a group of registers 605, each register in the group of registers 605 of the filter 601-1 holding a record, respectively, the record being associated with a plurality of logical partitions, wherein the plurality of logical partitions are located in the unit 606 where the filter 601-1 is located, and the plurality of logical partitions have their own logical partition ID, respectively.
- the record has an indicator indicative of logical partitions.
- the number of logical partitions may be random or determined as required.
- the record of each register (register 1, register 2, ..., register n) in the group of registers 605 is associated with the respective logical partition ID of the plurality of logical partitions in the unit 606, and the indicator indicative of logical partitions in the record is the logical partition ID of each logical partition.
- the filter 601 is further configured to split the logical ID attached to the request, to compare the split logical partition ID with each logical partition ID in the record, and to perform OR operation to each comparison result. If the result of OR operation is non-zero, it indicates matching, and then based on the request, search is performed in unit 606 where the filter 601-1 is located, and the search result is returned to the requesting unit 602 which issues the request. If the result of OR operation is zero, it indicates no matching, and then a NONE response is directly returned from the filter 601-1 to the unit 602 which issues the request, without searching the unit 606, thereby reducing data access to achieve the object of reducing power consumption.
- the requesting unit 602 and the unit 606 are caches.
- the caches may be L2 caches.
- the structure of the requesting unit 602 may be identical to the structure of the unit 606.
- the requesting unit 602 may comprise a filter.
- the indicator indicative of the logical partition in the record of the filter 601 is dynamically updated.
- the dynamic update process may be implemented under the control of a manager (not shown) similar to the manager 140 as shown in Fig. 1.
- the requesting units 602, 603, and 604 may have filters 601-4, 601-2, and 601-3,
- the system 600 of the present invention as shown in Fig. 6 further comprises other units, such as a bus 601 and other hierarchy, etc.
- FIG. 7 shows a flow chart of a processing method for a request for data according to another exemplary embodiment of the present invention.
- a requesting unit issues a request for data.
- the request comprises a logical partition ID.
- the request for data is received by a filter holding therein a record associated with a plurality of logical partitions.
- the record has indicators indicative of logical partitions, and the plurality of logical partitions has their own logical partition ID, respectively.
- the number of logical partitions may be random or determined as required.
- the record in the filter is associated with the respective logical partition ID of the plurality of logical partitions, and the indicators indicative of logical partitions in the record are the logical partition ID of each logical partition.
- the logical partition ID attached to the request is split, and the split logical partition ID is compared with each logical partition ID in the record.
- OR operation is performed to all comparison result so as to check whether the result of OR operation is none-zero. If the result of OR operation is non-zero, it indicates matching, and then at step 705, based on the request, search is performed in the unit where the filter is located, and the search result is returned to the requesting unit which issues the request. If the result of OR operation is zero, it indicates no matching, and then at step 706, a NONE response is directly returned from the filter to the unit which issues the request, without searching the unit, thereby reducing data access to achieve the object of reducing power consumption.
- any manner that is capable of determining whether the indicator indicative of a logical partition in the request matches the indicator indicative of the logical partition as held in the cache may be adopted, which is not limited to the above two operations as specifically provided by the present invention.
- Fig. 8 shows a structural schematic diagram of a multi-core system 800 in which the present invention may be implemented.
- the multi-core system 800 comprises: a first cache such as L2 cache 830A, and one or more second caches such as L2 caches 830B, 830C, and 830D, wherein the first cache is configured to issue a request for data to the one or more second caches.
- Each of the one or more second caches comprises a filter (for example, filters 801B, 801C, and 80 ID), which is configured to receive the request and compares the indicator indicative of a logical partition in the request with the indicator indicative of the logical partition in a record of the filter.
- the filter is further configured to perform search in the first cache based on the request if the comparison result matches and to return the search result to the one or more second caches.
- a NONE response is returned from the filter to the one or more second caches if the comparison result does not match.
- the multi-core system 800 comprises: a memory 810, a bus 820, a plurality of L2 caches 830A, 830B, 830C, and 830D, a manager 840, and a plurality of applications 850, wherein each of a plurality of L2 caches 830A, 830B, 830C, and 830D comprises one or more logical partitions and filters 801A, 801B, 801C, and 801D, respectively.
- a request for data is issued from a L2 cache 830A thereof to other L2 caches 830B, 830C, and 830D
- the request for data is transmitted from the system bus 820 to the L2 caches 830B, 830C, and 830D.
- the request for data is firstly processed at the filter. Specifically, for example, when the request for data reaches the filter 80 IB of the L2 cache 830B, the filter 801B compares the indicator indicative of a logical partition in the request with the indicator indicative of the logical partition in the record as held in the filter 801B. If the comparison result matches, then based on the request, search is performed in the L2 cache 830B, and the search result is returned to the L2 cache 830A which issues the request. If the comparison does not match, a NONE response is directly returned from the filter 80 IB to the L2 cache 830A. Likewise, similar operations are also performed in the L2 caches 830C and 830D. Though this filter mechanism, vain data access operation may be avoided, thereby reducing power consumption in the L2 cache and improving the power efficiency of the processing system.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Memory System Of A Hierarchy Structure (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Hardware Redundancy (AREA)
- Debugging And Monitoring (AREA)
Abstract
Description
Claims
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
DE112010004198T DE112010004198T5 (en) | 2009-10-30 | 2010-09-30 | Data processing method, unit and system for processing requests in a multi-core system |
GB1202832.0A GB2487313B (en) | 2009-10-30 | 2010-09-30 | Data processing method, device, and system for processing requests in a multi-core system |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200910211381.7 | 2009-10-30 | ||
CN200910211381.7A CN102053946B (en) | 2009-10-30 | 2009-10-30 | Method, equipment and system for processing requests in multi-core system |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2011051074A1 true WO2011051074A1 (en) | 2011-05-05 |
Family
ID=43502513
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/EP2010/064536 WO2011051074A1 (en) | 2009-10-30 | 2010-09-30 | Data processing method, device, and system for processing requests in a multi-core system |
Country Status (4)
Country | Link |
---|---|
CN (1) | CN102053946B (en) |
DE (1) | DE112010004198T5 (en) |
GB (1) | GB2487313B (en) |
WO (1) | WO2011051074A1 (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2013012760A1 (en) | 2011-07-15 | 2013-01-24 | Numeta Sciences, Inc. | Compositions and methods for modulating metabolic pathways |
US9086980B2 (en) | 2012-08-01 | 2015-07-21 | International Business Machines Corporation | Data processing, method, device, and system for processing requests in a multi-core system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5307495A (en) * | 1987-10-23 | 1994-04-26 | Hitachi, Ltd. | Multiprocessor system statically dividing processors into groups allowing processor of selected group to send task requests only to processors of selected group |
US20040230976A1 (en) * | 2003-05-12 | 2004-11-18 | International Business Machines Corporation | Filtering processor requests based on identifiers |
US20040268044A1 (en) * | 2003-06-25 | 2004-12-30 | International Business Machines Corporation | Multiprocessor system with dynamic cache coherency regions |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8527709B2 (en) * | 2007-07-20 | 2013-09-03 | Intel Corporation | Technique for preserving cached information during a low power mode |
-
2009
- 2009-10-30 CN CN200910211381.7A patent/CN102053946B/en active Active
-
2010
- 2010-09-30 DE DE112010004198T patent/DE112010004198T5/en not_active Ceased
- 2010-09-30 WO PCT/EP2010/064536 patent/WO2011051074A1/en active Application Filing
- 2010-09-30 GB GB1202832.0A patent/GB2487313B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5307495A (en) * | 1987-10-23 | 1994-04-26 | Hitachi, Ltd. | Multiprocessor system statically dividing processors into groups allowing processor of selected group to send task requests only to processors of selected group |
US20040230976A1 (en) * | 2003-05-12 | 2004-11-18 | International Business Machines Corporation | Filtering processor requests based on identifiers |
US20040268044A1 (en) * | 2003-06-25 | 2004-12-30 | International Business Machines Corporation | Multiprocessor system with dynamic cache coherency regions |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2013012760A1 (en) | 2011-07-15 | 2013-01-24 | Numeta Sciences, Inc. | Compositions and methods for modulating metabolic pathways |
US9086980B2 (en) | 2012-08-01 | 2015-07-21 | International Business Machines Corporation | Data processing, method, device, and system for processing requests in a multi-core system |
Also Published As
Publication number | Publication date |
---|---|
GB2487313A (en) | 2012-07-18 |
GB2487313B (en) | 2016-06-29 |
GB201202832D0 (en) | 2012-04-04 |
CN102053946A (en) | 2011-05-11 |
CN102053946B (en) | 2014-05-28 |
DE112010004198T5 (en) | 2012-08-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9164923B2 (en) | Dynamic pinning of virtual pages shared between different type processors of a heterogeneous computing platform | |
CN100422940C (en) | System and method of arbitrating access of threads to shared resources within a data processing system | |
CN1991793B (en) | Method and system for proximity caching in a multiple-core system | |
EP2798502B1 (en) | Data migration for composite non-volatile storage device | |
US20090287901A1 (en) | System and method for content replication detection and elimination in main memory | |
US20130073811A1 (en) | Region privatization in directory-based cache coherence | |
US8051250B2 (en) | Systems and methods for pushing data | |
US7962615B1 (en) | Multi-system deadlock reduction | |
KR20170131274A (en) | Method and apparatus for enabling larger memory capacity than physical memory size | |
JPH08235061A (en) | Multiprocessor data-processing system | |
EP3404537A1 (en) | Processing node, computer system and transaction conflict detection method | |
US8904102B2 (en) | Process identifier-based cache information transfer | |
US10902014B1 (en) | Reducing network traffic when replicating memory data across hosts | |
WO2011051074A1 (en) | Data processing method, device, and system for processing requests in a multi-core system | |
US11381400B2 (en) | Using double hashing schema to reduce short hash handle collisions and improve memory allocation in content-addressable storage systems | |
US8990507B2 (en) | Storing data in a system memory for a subsequent cache flush | |
US11113194B2 (en) | Producer-to-consumer active direct cache transfers | |
CN116719813A (en) | Hash table processing method and device and electronic equipment | |
US9274955B2 (en) | Reduced scalable cache directory | |
US9086980B2 (en) | Data processing, method, device, and system for processing requests in a multi-core system | |
US10579519B2 (en) | Interleaved access of memory | |
US8838909B2 (en) | Dynamic initial cache line coherency state assignment in multi-processor systems | |
CN115237954A (en) | Cuckoo Hash query method based on PIM device, PIM device and system | |
US20140297960A1 (en) | Multi-core system and method of data consistency | |
US10432727B1 (en) | Reducing network traffic when replicating memory data across hosts |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 10768199 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 1202832 Country of ref document: GB Kind code of ref document: A Free format text: PCT FILING DATE = 20100930 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1202832.0 Country of ref document: GB |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1120100041989 Country of ref document: DE Ref document number: 112010004198 Country of ref document: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 10768199 Country of ref document: EP Kind code of ref document: A1 |