WO1999022296A2 - Method and system for synchronizing block-organized data transfer - Google Patents
Method and system for synchronizing block-organized data transfer Download PDFInfo
- Publication number
- WO1999022296A2 WO1999022296A2 PCT/IB1998/001588 IB9801588W WO9922296A2 WO 1999022296 A2 WO1999022296 A2 WO 1999022296A2 IB 9801588 W IB9801588 W IB 9801588W WO 9922296 A2 WO9922296 A2 WO 9922296A2
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- station
- semaphore
- state
- block
- stations
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/52—Program synchronisation; Mutual exclusion, e.g. by means of semaphores
Definitions
- the invention relates to a method as recited in the preamble of Claim 1.
- United States Patent 4,594,657 describes a semaphore organization between first and second stations through a first binary semaphore that indicates ownership or non-ownership, and a second binary ownership semaphore that indicates which of the two stations previously owned the semaphore.
- First gating circuitry decodes control signals from the first station and generates a read or write signal.
- the read signal from the first gating circuit loads the semaphore and ownership signals into first and second flipfiops respectively, whose outputs are applied to the data bus of the first station.
- Second gating circuitry decodes control signals from the second station and generates a read or write signal.
- the read pulse signal from the second gating circuitry loads the semaphore and ownership signals into third and fourth flipfiops, respectively, whose outputs are applied to the data bus of the second station. If the first and second stations access the semaphore simultaneously, access is granted to the second station since the output of the third flipflop resets the first flipflop to deny access to the first station.
- the invention is characterized according to the characterizing part of Claim 1.
- the invention also relates to a data processing apparatus arranged for practising the method as recited in Claim 1. Further advantageous aspects of the invention are recited in dependent Claims.
- Prior art has a hardware semaphore that signals the previous mode of the system, which is not straightforwardly extendable to more than two stations, and which is in particular suitable for sharing a hardware resource.
- the present invention in particular shares data, and for block-organized transfer must ensure that reading is not started before writing has ended.
- the first station should not be allowed to execute a write-access with new data before all second stations will have read-accessed the old data.
- the second stations should all have filled in their respective and privileged parts of a collective data item, before the first station should be allowed to read.
- the fact that all semaphores have a single fixed owner station renders the scheme extendable.
- the invention may also be used in cases where two or more stations should synchronize their read accesses to a particular block of information that has been provided by a single write station. Note that the size of the block is inconsequential, and may go down to only a single sample such as a digital autio sample.
- the invention does not need specific and complex hardware for avoiding all conflicts between the various accessing operations.
- Figure 1 a hardware embodiment of the reference
- Figure 2 the mapping of the Kahn model
- FIGS. 3A-C a schematic of the twin-semaphore protocol
- Figure 4 a state diagram of two communicating stations
- FIG. 5 an exemplary code for a consumer station
- FIG. 6 an elementary semaphore arrangement.
- Figure 1 is a hardware embodiment of the reference with semaphore 100.
- Stations 102, 104 share memory facility 106.
- the DATA, ADdRess, Read/Write and CLocK signals from superviser station 102 are applied to memory 106 through transmission gates 108, 110, 112 and 120, in response to control signal CS from station 102.
- Symmetrical signals are applied to memory 106 via switches 114, 116, 118 and 122 in response to control signal CS from peripheral station 104. If both stations would simultaneously access memory 106, semaphore 100 is used to determine which of the stations get preference.
- any station Before access, any station must read the state of the semaphore and ownership signals.
- the semaphore has a zero state to signal that the memory is presently not owned.
- the ownership signal has a binary one state to signal that station 102 previously owned the memory. If any of the stations reads semaphore 100 to find that the memory is not owned, it may access the memory. If the station is ready to give up ownership, it flips the semaphore signal to indicate that the memory is no longer owned. If both stations would access the memory simultaneously, semaphore 100 grants access to peripheral station 104.
- Figure 2 shows the mapping of the Kahn model for use in two interconnected digital signal processors, for realizing the intercommunication at a low overhead tax.
- the model maps communication between processes 20, 30, with infinite FIFO buffering 42. If physical buffer space is limited, it is only necessary that the outputting process will block if the buffer is full and the inputting process will block if the buffer is empty. As shown, the FIFO buffer is distributed into blocks and mapped on the local memory 26, 28, 36, of one or both of the two stations. Coupling between the stations gets tighter if the FIFO is smaller, and/or the block size is larger.
- Figure 3 shows a schematic of the twin-semaphore protocol.
- the two processors are synchronized by a simple flow control mechanism that has two semaphores for each buffer block.
- the use of two semaphores avoids shared-resource conflicts and the necessity for complex hardware solutions.
- the protocol is as follows. Initially, the block is empty and both semaphores are set to the same value, cf. Figure 3A.
- the data producer must read the remote consumer semaphore and compare it with its own semaphore. If the two semaphores are equal, the producer is allowed to write data into the block that is being guarded by these particular semaphores.
- the producer toggles its own local semaphore. If the semaphores would have been unequal, the block contained unread data, and the producer has to wait in a blocked-write state as in Figure 3B. Likewise, the data consumer compares its own semaphore with the remote producer semaphore whenever it desires to read data. The data guarded by the semaphores may be read whenever the semaphores are unequal. Otherwise the consumer has to wait in blocked-read state until new data becomes available. After the data has been read and the consumer station has toggled its semaphore, the data in the block may be overwritten again, cf. Figure 3C.
- twin-semaphore protocol uses the fact that buffer blocks are accessed in alternating fashion (producer-consumer- producer ..). This avoids the necessity for hardware semaphores and/or spin-locks. Note that the set-up in Figures 3A-C applies to a series of three shared blocks, of which in Figure 3B the third one is yet free.
- a station may poll the remote semaphore to see whether the status of the buffer has changed.
- an interrupt is provided for the consuming (producing) station by the producing (consuming) station after it has changed its own semaphore.
- a similar procedure may apply if either the number of producers is plural or the number of consumers is plural, as has been hinted supra.
- a station may go into low-power mode whenever it has to wait for completion of the accessing by the other station. Receiving an interrupt will wake the idle station and will cause it to re-evaluate the semaphore that had caused the idle state, cf. the diagram in Figure 4.
- Figure 4 shows a state diagram of two communicating stations.
- the read-and-check of the remote semaphores, inclusive of the transition to the idle state should resemble an atomic action.
- This relates to the embodiment with interrupts. That is, if a communication interrupt arrives after the remote semaphore has been read, it should remain pending until the idle state has been reached. It is important that the interrupt should arrive after the state of the semaphore had changed, but not earlier. The transmitting may therefore be effected, provided that it can be guaranteed that the interrupt will not arrive before the not yet changed semaphore may have been read.
- These communication interrupts carry no information and their only purpose is to wake up the communicating station in question from the idle state.
- Figure 5 is a self-explanatory code example for a reading station. Again, the other station's code is nearly identical.
- Figure 6 is an elementary semaphore arrangement for two stations. At left is the first station that may toggle flipflop 70. At right is the second station that may toggle flipflop 72. If they are unequal as checked in EXOR 74, the right hand side station gets a logic "1". If they are equal, the left hand station gets a logic "1" through inverter 78. For simplicity, the generating of the interrupt has been ignored.
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multi Processors (AREA)
- Small-Scale Networks (AREA)
- Time-Division Multiplex Systems (AREA)
Abstract
Description
Claims
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP98945480A EP0960372A2 (en) | 1997-10-29 | 1998-10-12 | Method and system for synchronizing block-organized data transfer |
JP52356699A JP2001508214A (en) | 1997-10-29 | 1998-10-12 | Block structured data transfer synchronization method and system |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP97202568.8 | 1997-10-29 | ||
EP97202568 | 1997-10-29 |
Publications (2)
Publication Number | Publication Date |
---|---|
WO1999022296A2 true WO1999022296A2 (en) | 1999-05-06 |
WO1999022296A3 WO1999022296A3 (en) | 1999-07-22 |
Family
ID=8228659
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/IB1998/001588 WO1999022296A2 (en) | 1997-10-29 | 1998-10-12 | Method and system for synchronizing block-organized data transfer |
Country Status (4)
Country | Link |
---|---|
US (1) | US6493805B2 (en) |
EP (1) | EP0960372A2 (en) |
JP (1) | JP2001508214A (en) |
WO (1) | WO1999022296A2 (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2003005197A1 (en) * | 2001-07-06 | 2003-01-16 | Koninklijke Philips Electronics N.V. | Method and system of processing a media file |
EP1653346A2 (en) * | 2004-10-28 | 2006-05-03 | Microsoft Corporation | Message-passing processor based on the Pi-calculus |
US9098270B1 (en) | 2011-11-01 | 2015-08-04 | Cypress Semiconductor Corporation | Device and method of establishing sleep mode architecture for NVSRAMs |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6892258B1 (en) * | 2001-10-26 | 2005-05-10 | Lsi Logic Corporation | Hardware semaphores for a multi-processor system within a shared memory architecture |
US20030115476A1 (en) * | 2001-10-31 | 2003-06-19 | Mckee Bret | Hardware-enforced control of access to memory within a computer using hardware-enforced semaphores and other similar, hardware-enforced serialization and sequencing mechanisms |
US7174552B2 (en) * | 2002-01-12 | 2007-02-06 | Intel Corporation | Method of accessing a resource by a process based on a semaphore of another process |
US7184468B2 (en) * | 2002-01-24 | 2007-02-27 | Broadcom Corporation | Method and system for implementing a conditional one's complement of partial address |
US20050120185A1 (en) * | 2003-12-01 | 2005-06-02 | Sony Computer Entertainment Inc. | Methods and apparatus for efficient multi-tasking |
US7890736B2 (en) * | 2005-11-08 | 2011-02-15 | St-Ericsson Sa | Control device with flag registers for synchronization of communications between cores |
JP2012508424A (en) * | 2008-11-07 | 2012-04-05 | オパンガ ネットワークス インコーポレイテッド | Portable data storage device using host device to start data transfer |
US11144368B2 (en) * | 2019-06-18 | 2021-10-12 | Qualcomm Incorproated | Providing self-resetting multi-producer multi-consumer semaphores in distributed processor-based systems |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4594657A (en) * | 1983-04-22 | 1986-06-10 | Motorola, Inc. | Semaphore for memory shared by two asynchronous microcomputers |
FR2731533A1 (en) * | 1995-03-06 | 1996-09-13 | Andre Francis | Synchronisation, conjoint and concomitant communication controlled by semaphore |
US5592673A (en) * | 1991-02-27 | 1997-01-07 | Kabushiki Kaisha Toshiba | Loosely coupled compound computer system using lock and semaphore mechanism for performing exclusive control of shared resource which is accessed through a distinct bus |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5263161A (en) * | 1989-07-26 | 1993-11-16 | Massachusetts Institute Of Technology | Non-busy waiting resource control |
US5432911A (en) * | 1991-07-15 | 1995-07-11 | Matsushita Electric Works, Ltd. | Controllers request access within one bus cycle causing hardware-wait to stall second controller when first controller is accessing and second controller is still requesting access |
US5313638A (en) * | 1992-03-24 | 1994-05-17 | International Business Machines Corp. | Method using semaphores for synchronizing communication between programs or processes resident in a computer system |
WO1996023317A1 (en) * | 1995-01-23 | 1996-08-01 | Tandem Computers Incorporated | A method for accessing a file in a multi-processor computer system using pipes and fifos |
US5946387A (en) * | 1997-02-10 | 1999-08-31 | Genesys Telecommunications Laboratories, Inc, | Agent-level network call routing |
-
1998
- 1998-10-12 WO PCT/IB1998/001588 patent/WO1999022296A2/en not_active Application Discontinuation
- 1998-10-12 EP EP98945480A patent/EP0960372A2/en not_active Withdrawn
- 1998-10-12 JP JP52356699A patent/JP2001508214A/en not_active Withdrawn
- 1998-10-29 US US09/182,699 patent/US6493805B2/en not_active Expired - Lifetime
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4594657A (en) * | 1983-04-22 | 1986-06-10 | Motorola, Inc. | Semaphore for memory shared by two asynchronous microcomputers |
US5592673A (en) * | 1991-02-27 | 1997-01-07 | Kabushiki Kaisha Toshiba | Loosely coupled compound computer system using lock and semaphore mechanism for performing exclusive control of shared resource which is accessed through a distinct bus |
FR2731533A1 (en) * | 1995-03-06 | 1996-09-13 | Andre Francis | Synchronisation, conjoint and concomitant communication controlled by semaphore |
Non-Patent Citations (1)
Title |
---|
PATENT ABSTRACTS OF JAPAN; & JP 4225433 A (FUJITSU LTD) 14 August 1992. * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2003005197A1 (en) * | 2001-07-06 | 2003-01-16 | Koninklijke Philips Electronics N.V. | Method and system of processing a media file |
EP1653346A2 (en) * | 2004-10-28 | 2006-05-03 | Microsoft Corporation | Message-passing processor based on the Pi-calculus |
EP1653346A3 (en) * | 2004-10-28 | 2006-05-10 | Microsoft Corporation | Message-passing processor based on the Pi-calculus |
US9098270B1 (en) | 2011-11-01 | 2015-08-04 | Cypress Semiconductor Corporation | Device and method of establishing sleep mode architecture for NVSRAMs |
Also Published As
Publication number | Publication date |
---|---|
JP2001508214A (en) | 2001-06-19 |
WO1999022296A3 (en) | 1999-07-22 |
US6493805B2 (en) | 2002-12-10 |
US20020083276A1 (en) | 2002-06-27 |
EP0960372A2 (en) | 1999-12-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US4754398A (en) | System for multiprocessor communication using local and common semaphore and information registers | |
US4204251A (en) | Interconnection unit for multiple data processing systems | |
US5434970A (en) | System for distributed multiprocessor communication | |
CA2036688C (en) | Multiple cluster signal processor | |
US4591977A (en) | Plurality of processors where access to the common memory requires only a single clock interval | |
US4481572A (en) | Multiconfigural computers utilizing a time-shared bus | |
EP0477364B1 (en) | Distributed computer system | |
CA1129110A (en) | Apparatus and method for providing byte and word compatible information transfers | |
KR0133236B1 (en) | Shared memory system and arbitration method and system | |
JPH01147647A (en) | Data processor | |
JPS6270964A (en) | Data hub | |
US6493805B2 (en) | Method and system for synchronizing block-organized data transfer amongst a plurality of producer and consumer stations | |
US5210828A (en) | Multiprocessing system with interprocessor communications facility | |
GB2148563A (en) | Multiprocessor system | |
US5255369A (en) | Multiprocessor system with reflective memory data transfer device | |
JP3431941B2 (en) | Method and apparatus for determining instruction execution order in a data processing system | |
WO1981002798A1 (en) | Computer system and interface therefor | |
EP0291907B1 (en) | Interprocessor communication system in information processing system enabling communication between execution processor units during communication between other processor units | |
JPH0731666B2 (en) | Inter-processor communication method | |
KR100362061B1 (en) | Device for controlling local bus | |
Del Corso et al. | An integrated controller for modified inter-integrated circuit protocol | |
JPH01291343A (en) | Memory managing device | |
KR950010947B1 (en) | Bus protocol mapping circuit | |
JP3588966B2 (en) | Information processing system | |
JPH10171770A (en) | Multiprocessor system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A2 Designated state(s): JP |
|
AL | Designated countries for regional patents |
Kind code of ref document: A2 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1998945480 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref country code: JP Ref document number: 1999 523566 Kind code of ref document: A Format of ref document f/p: F |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
AK | Designated states |
Kind code of ref document: A3 Designated state(s): JP |
|
AL | Designated countries for regional patents |
Kind code of ref document: A3 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
WWP | Wipo information: published in national office |
Ref document number: 1998945480 Country of ref document: EP |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: 1998945480 Country of ref document: EP |