JPH0354660A - Shared memory management system for multiprocessor system - Google Patents

Shared memory management system for multiprocessor system

Info

Publication number
JPH0354660A
JPH0354660A JP18954189A JP18954189A JPH0354660A JP H0354660 A JPH0354660 A JP H0354660A JP 18954189 A JP18954189 A JP 18954189A JP 18954189 A JP18954189 A JP 18954189A JP H0354660 A JPH0354660 A JP H0354660A
Authority
JP
Japan
Prior art keywords
memory
processor
bus
shared memory
access
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
JP18954189A
Other languages
Japanese (ja)
Inventor
Fujio Yoshida
富士夫 吉田
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NEC Corp
Original Assignee
NEC Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by NEC Corp filed Critical NEC Corp
Priority to JP18954189A priority Critical patent/JPH0354660A/en
Publication of JPH0354660A publication Critical patent/JPH0354660A/en
Pending legal-status Critical Current

Links

Abstract

PURPOSE:To exclusively control the access to a shared memory with a simple circuit by connecting a memory bus to a bus gate circuit corresponding to the processor to which a memory acquisition response is returned. CONSTITUTION:When a CPU 1 turns on a memory request signal a0 because requiring the access to a shared memory 4, a priority control circuit 3 performs the exclusive control based on access requests from plural processors and returns the memory acquisition response to one processor. When the indication to connect a bus 7 to a bus gate circuit 5 or 6 corresponding to this processor to which the memory acquisition response is returned is issued, this bus gate circuit 5 or 6 connects the processor and the memory. Thus, the access request to the shared memory is exclusively controlled without degrading the performance of the whole of the multiprocessor system.

Description

【発明の詳細な説明】 (産業上の利用分野) 本発明は一つの記憶装置を複数のプロセッサが共有する
マルチプロセッサシステムにおける共有メモリ管理方式
に関する. (従来の技術) 処理能力、計算機資源の使用効率等を高めるため、複数
のプロセッサで一つの記憶装置(共有メモリ)を共有す
るマルチプロセッサシステムが構築されている.このマ
ルチブロセ・ソサシステムにおいては、記憶装置に対す
るアクセス権をどのプロセッサに与えるかを制御する排
他制御を行なっている,そして、この排m M m方式
には次のような方式がある. ■共有メモリにアクセス要求をしたプロセッサが、この
共有メモリをアクセスする可能性のある全てのプロセッ
サを停止させ、他のプロセッサからアクセス要求が発生
しないようにしてから共有メモリへアクセスする. ■共有メモリのアクセス要求を管理する専用のプロセッ
サ(管理プロセッサ)を設け、共有メモリへのアクセス
しようとするプロセッサはこの管理プロセッサにアクセ
ス要求を出す.そうすると、この管理プロセッサは、排
他制御を行なっていずれか一つのプロセッサに対しての
みアクセスを許可する. (発明が解決しようとする課題) 上述したように、従来の共有メモリ管理方式には次のよ
うな解決すべき課題があった.従来方式■では、いずれ
か一つのプロセッサが共有メモリにアクセスする毎に他
の全てのプロセッサが停止させられるので、マルチプロ
セッサシステム全体の性能が低下してしまう. 従来方式■も同様に管理プロセッサとの通信のためのオ
ーバヘッドが大きくなりマルチプロセッサシステム全体
の性能が低下してしまう.本一発明は、このような事情
に鑑みてなされたものであり、その目的は、マノレチプ
ロセッサシステム全体の性能を低下させることなく共有
メモリへのアクセス要求の排他制御を行なうマルチプロ
セッサシステムにおける共有メモリ管理方式を提供する
ことにある. 〈課題を解決するための手Pi> 本発明のマルチプロセッサシステムにおける共有メモリ
管理方式は、上記目的を達戒するために、複数のプロセ
ッサと該複数のプロセッサが共有する一つの記憶装置と
がメモリバスに接続され、前記複数のプロセッサから前
記記憶装置にアクセス要求があるといずれか一つのプロ
セ・yサに対してアクセスを許可するマルチプロセッサ
システムにおける共有メモリ管理方式において、 プロセッサと前記メモリバスとを接続状態または切断状
態にするバスゲート回路と、 前記複数のプロセッサからのアクセス要求に基づいて排
他制御を行なっていずれか一つのプロセッサに対してメ
モリ獲得応答を返送すると共に、該メモリ獲得応答を返
した前記プロセッサに対応する前記バスゲート回路にメ
モリバスを接続状態にする旨の指示を出す優先順位制御
回路とを有する. (作用) 本発明のマルチプロセッサシステムにおける共有メモリ
管理方式においては、優先1項位制御回路が、複数のプ
ロセッサからのアクセス要求に基づいて排他制御を行な
っていずれか一つのプロセッサに対してメモリ獲得応答
を返送すると共にこの該メモリ獲得応答を返したプロセ
ッサに対応するバスゲート回路にメモリバスを接続状態
にする旨の指示を出す.そうすると、バスゲート回路が
、プロセッサとメモリバスとを接続状態にする.(実施
例) 次に、本発明の実施例について図面を参照して詳細に説
明する. 第1図は本発明の一実雄例の梢或図である.同図におい
て、1.2はプロセッサ(CPUと称する)、3は排他
制御を行う優先順位制御回路、4は共有メモリ、5.6
はバスゲート、7はメモリバスである. 次に第1図の実施例においてCPUIから共有メモリ4
に対してアクセス要求が発生した際の動作について説明
する,cpuiは共有メモリ4にアクセスする必要があ
ると、メモリリクエスト信号aOをON状態にする.そ
うすると、優先順位制御回路3は、メモリリクエスト信
号aOがON状態になったことを認識して次のような排
他制御を行なう. ■CPU2からのメモリリクエスト信号a1がON状態
でないとき このようなとき優先順位制御回路3は、メモリ獲得応答
信号bOt.oN状態してCPUIに共有メモリ4への
アクセスを許可した旨を通知すると共に、バスゲート制
御信号COをON状態にする.このON状態のバスゲー
ト制m信号を受け取ったバスゲート5は、内部のゲート
を導通状態にする.この結果、CPUIと共有メモリ4
とがメモリバス7を介して接続される. ON状態のメモリ獲得応答信号boを受け取ったCPU
1は、共有メモリ4ヘアクセスする.そして、アクセス
が終了するとメモリリクエスト信号aOをOFF状態に
する.従って、次の新たなアクセス要求が受けつけられ
るようになる.■CPU2からのメモリリクエスト信号
a1がON状態のとき このようなとき優先順位制御回路3は、メモリリクエス
ト信号a1がON状態からOFF状態になるのを待つ.
そしてメモリリクエスト信号a1がOFF状態になると
、上述したように、メモリ獲得応答信号bOおよびバス
ゲート制御信号COをON状態にする.そしてCPU1
は、メモリ獲得応答信号がON状態になると共有メモリ
4ヘアクセスし、アクセスが終了するとメモリリクエス
ト信号aOをOFF状態にする.このようにして共有メ
モリに対する排他制御が行なわれる.CPU2が共有メ
モリ4にアクセスするときも同様の処理が行われる. なお、本実施例では、マルチプロセッサシステムが2つ
のCPUで構成されるとしたが、CPUの数が3以上で
もよいことは勿論のことである.(発明の効果) 以上に説明したように、本発明のマルチプロセッサシス
テムにおける共有メモリ管理方式によれば、共有メモリ
に対するアクセスの排他制御を簡単な回路により実現で
き、且つ処理も高速に行える.従って、マルチプロセッ
サの各プロセッサが独立に動作出来る時間が大きくなり
、システム全体の性能の向上が図れる。
Detailed Description of the Invention (Field of Industrial Application) The present invention relates to a shared memory management method in a multiprocessor system in which a single storage device is shared by multiple processors. (Prior Art) In order to increase processing power and efficiency in using computer resources, multiprocessor systems have been constructed in which multiple processors share a single storage device (shared memory). In this multi-processor system, exclusive control is performed to control which processor is given access rights to the storage device, and this exclusion method includes the following methods. ■A processor that requests access to shared memory stops all processors that may access this shared memory and prevents access requests from other processors before accessing shared memory. ■A dedicated processor (management processor) is provided to manage shared memory access requests, and a processor attempting to access shared memory issues an access request to this management processor. Then, this management processor performs exclusive control and grants access to only one processor. (Problems to be Solved by the Invention) As mentioned above, the conventional shared memory management method has the following problems to be solved. In conventional method (■), every time one processor accesses shared memory, all other processors are stopped, resulting in a decline in the performance of the entire multiprocessor system. Conventional method ■ also requires a large amount of overhead for communication with the management processor, which reduces the performance of the entire multiprocessor system. The present invention has been made in view of these circumstances, and its purpose is to provide shared memory in a multiprocessor system that performs exclusive control of access requests to shared memory without degrading the performance of the entire processor system. Its purpose is to provide a memory management method. <Measures Pi for Solving the Problems> In order to achieve the above object, the shared memory management method in the multiprocessor system of the present invention allows multiple processors and one storage device shared by the multiple processors to In a shared memory management method in a multiprocessor system that is connected to a bus and grants access to any one processor when there is an access request from the plurality of processors to the storage device, the processor and the memory bus a bus gate circuit that connects or disconnects the processor; and a bus gate circuit that performs exclusive control based on access requests from the plurality of processors, returns a memory acquisition response to any one of the processors, and sends the memory acquisition response to the processor. and a priority control circuit that issues an instruction to connect the memory bus to the bus gate circuit corresponding to the returned processor. (Operation) In the shared memory management method in the multiprocessor system of the present invention, the priority 1 control circuit performs exclusive control based on access requests from multiple processors to acquire memory for any one processor. A response is sent back and an instruction is issued to the bus gate circuit corresponding to the processor that returned this memory acquisition response to connect the memory bus. Then, the bus gate circuit connects the processor and the memory bus. (Example) Next, an example of the present invention will be described in detail with reference to the drawings. Figure 1 is a treetop view of one example of the present invention. In the figure, 1.2 is a processor (referred to as CPU), 3 is a priority control circuit that performs exclusive control, 4 is a shared memory, and 5.6
is a bus gate, and 7 is a memory bus. Next, in the embodiment of FIG. 1, from the CPU to the shared memory 4
The following describes the operation when an access request is made to the shared memory 4.When the CPU needs to access the shared memory 4, it turns on the memory request signal aO. Then, the priority control circuit 3 recognizes that the memory request signal aO is in the ON state and performs the following exclusive control. (2) When the memory request signal a1 from the CPU 2 is not in the ON state In such a case, the priority control circuit 3 outputs the memory acquisition response signal bOt. It turns on and notifies the CPU that access to the shared memory 4 is permitted, and turns the bus gate control signal CO on. The bus gate 5, which receives this ON-state bus gate control m signal, turns on the internal gate. As a result, CPUI and shared memory 4
are connected via a memory bus 7. The CPU that received the memory acquisition response signal bo in the ON state
1 accesses shared memory 4. Then, when the access is completed, the memory request signal aO is turned OFF. Therefore, the next new access request will be accepted. (2) When the memory request signal a1 from the CPU 2 is in the ON state In such a case, the priority control circuit 3 waits for the memory request signal a1 to change from the ON state to the OFF state.
When the memory request signal a1 becomes OFF, the memory acquisition response signal bO and bus gate control signal CO are turned ON, as described above. and CPU1
accesses the shared memory 4 when the memory acquisition response signal turns ON, and turns the memory request signal aO OFF when the access is completed. In this way, exclusive control over shared memory is performed. Similar processing is performed when the CPU 2 accesses the shared memory 4. In this embodiment, the multiprocessor system is configured with two CPUs, but it goes without saying that the number of CPUs may be three or more. (Effects of the Invention) As explained above, according to the shared memory management method in the multiprocessor system of the present invention, exclusive control of access to the shared memory can be realized with a simple circuit, and processing can be performed at high speed. Therefore, the time during which each processor in the multiprocessor can operate independently becomes longer, and the performance of the entire system can be improved.

【図面の簡単な説明】[Brief explanation of drawings]

第1図は本発明の実施例の構成図である.1,2・・・
CPU、3・・・優先順位制御回路、4・・・共有メモ
リ、5.6・・・バスゲート、7・・・メモリバス.
Figure 1 is a configuration diagram of an embodiment of the present invention. 1, 2...
CPU, 3... Priority control circuit, 4... Shared memory, 5.6... Bus gate, 7... Memory bus.

Claims (1)

【特許請求の範囲】 複数のプロセッサと該複数のプロセッサが共有する一つ
の記憶装置とがメモリバスに接続され、前記複数のプロ
セッサから前記記憶装置にアクセス要求があるといずれ
か一つのプロセッサに対してアクセスを許可するマルチ
プロセッサシステムにおける共有メモリ管理方式におい
て、 プロセッサと前記メモリバスとを接続状態または切断状
態にするバスゲート回路と、 前記複数のプロセッサからのアクセス要求に基づいて排
他制御を行なっていずれか一つのプロセッサに対してメ
モリ獲得応答を返送すると共に、該メモリ獲得応答を返
した前記プロセッサに対応する前記バスゲート回路にメ
モリバスを接続状態にする旨の指示を出す優先順位制御
回路と を設けたことを特徴とするマルチプロセッサシステムに
おける共有メモリ管理方式。
[Claims] A plurality of processors and a storage device shared by the plurality of processors are connected to a memory bus, and when the plurality of processors makes an access request to the storage device, one of the processors receives an access request from the plurality of processors. A shared memory management method in a multiprocessor system in which access is permitted in a multiprocessor system includes: a bus gate circuit that connects or disconnects a processor and the memory bus; and a bus gate circuit that performs exclusive control based on access requests from the plurality of processors. a priority control circuit that returns a memory acquisition response to any one processor and instructs the bus gate circuit corresponding to the processor that has returned the memory acquisition response to connect a memory bus; A shared memory management method in a multiprocessor system characterized by the provision of.
JP18954189A 1989-07-21 1989-07-21 Shared memory management system for multiprocessor system Pending JPH0354660A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP18954189A JPH0354660A (en) 1989-07-21 1989-07-21 Shared memory management system for multiprocessor system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP18954189A JPH0354660A (en) 1989-07-21 1989-07-21 Shared memory management system for multiprocessor system

Publications (1)

Publication Number Publication Date
JPH0354660A true JPH0354660A (en) 1991-03-08

Family

ID=16243038

Family Applications (1)

Application Number Title Priority Date Filing Date
JP18954189A Pending JPH0354660A (en) 1989-07-21 1989-07-21 Shared memory management system for multiprocessor system

Country Status (1)

Country Link
JP (1) JPH0354660A (en)

Cited By (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100293594B1 (en) * 1992-01-31 2001-09-17 가나이 쓰도무 Multiprocessor system running exclusive access to shared memory
US7076583B2 (en) 2001-02-20 2006-07-11 Nec Corporation Multiprocessor system, shared-memory controlling method, recording medium and data signal embedded in a carrier wave
JP2009251871A (en) * 2008-04-04 2009-10-29 Nec Corp Contention analysis device, contention analysis method, and program
US8347065B1 (en) 2006-11-01 2013-01-01 Glasco David B System and method for concurrently managing memory access requests
US8347064B1 (en) 2006-09-19 2013-01-01 Nvidia Corporation Memory access techniques in an aperture mapped memory space
US8352709B1 (en) 2006-09-19 2013-01-08 Nvidia Corporation Direct memory access techniques that include caching segmentation data
US8359454B2 (en) 2005-12-05 2013-01-22 Nvidia Corporation Memory access techniques providing for override of page table attributes
US8504794B1 (en) 2006-11-01 2013-08-06 Nvidia Corporation Override system and method for memory access management
US8533425B1 (en) 2006-11-01 2013-09-10 Nvidia Corporation Age based miss replay system and method
US8601223B1 (en) 2006-09-19 2013-12-03 Nvidia Corporation Techniques for servicing fetch requests utilizing coalesing page table entries
US8607008B1 (en) 2006-11-01 2013-12-10 Nvidia Corporation System and method for independent invalidation on a per engine basis
US8700883B1 (en) 2006-10-24 2014-04-15 Nvidia Corporation Memory access techniques providing for override of a page table
US8700865B1 (en) 2006-11-02 2014-04-15 Nvidia Corporation Compressed data access system and method
US8706975B1 (en) 2006-11-01 2014-04-22 Nvidia Corporation Memory access management block bind system and method
US8707011B1 (en) 2006-10-24 2014-04-22 Nvidia Corporation Memory access techniques utilizing a set-associative translation lookaside buffer
US9880846B2 (en) 2012-04-11 2018-01-30 Nvidia Corporation Improving hit rate of code translation redirection table with replacement strategy based on usage history table of evicted entries
US10108424B2 (en) 2013-03-14 2018-10-23 Nvidia Corporation Profiling code portions to generate translations
US10146545B2 (en) 2012-03-13 2018-12-04 Nvidia Corporation Translation address cache for a microprocessor
US10241810B2 (en) 2012-05-18 2019-03-26 Nvidia Corporation Instruction-optimizing processor with branch-count table in hardware
US10324725B2 (en) 2012-12-27 2019-06-18 Nvidia Corporation Fault detection in instruction translations

Cited By (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100293594B1 (en) * 1992-01-31 2001-09-17 가나이 쓰도무 Multiprocessor system running exclusive access to shared memory
US7076583B2 (en) 2001-02-20 2006-07-11 Nec Corporation Multiprocessor system, shared-memory controlling method, recording medium and data signal embedded in a carrier wave
US8359454B2 (en) 2005-12-05 2013-01-22 Nvidia Corporation Memory access techniques providing for override of page table attributes
US8601223B1 (en) 2006-09-19 2013-12-03 Nvidia Corporation Techniques for servicing fetch requests utilizing coalesing page table entries
US8347064B1 (en) 2006-09-19 2013-01-01 Nvidia Corporation Memory access techniques in an aperture mapped memory space
US8352709B1 (en) 2006-09-19 2013-01-08 Nvidia Corporation Direct memory access techniques that include caching segmentation data
US8707011B1 (en) 2006-10-24 2014-04-22 Nvidia Corporation Memory access techniques utilizing a set-associative translation lookaside buffer
US8700883B1 (en) 2006-10-24 2014-04-15 Nvidia Corporation Memory access techniques providing for override of a page table
US8706975B1 (en) 2006-11-01 2014-04-22 Nvidia Corporation Memory access management block bind system and method
US8533425B1 (en) 2006-11-01 2013-09-10 Nvidia Corporation Age based miss replay system and method
US8347065B1 (en) 2006-11-01 2013-01-01 Glasco David B System and method for concurrently managing memory access requests
US8601235B2 (en) 2006-11-01 2013-12-03 Nvidia Corporation System and method for concurrently managing memory access requests
US8607008B1 (en) 2006-11-01 2013-12-10 Nvidia Corporation System and method for independent invalidation on a per engine basis
US8504794B1 (en) 2006-11-01 2013-08-06 Nvidia Corporation Override system and method for memory access management
US8700865B1 (en) 2006-11-02 2014-04-15 Nvidia Corporation Compressed data access system and method
JP2009251871A (en) * 2008-04-04 2009-10-29 Nec Corp Contention analysis device, contention analysis method, and program
US10146545B2 (en) 2012-03-13 2018-12-04 Nvidia Corporation Translation address cache for a microprocessor
US9880846B2 (en) 2012-04-11 2018-01-30 Nvidia Corporation Improving hit rate of code translation redirection table with replacement strategy based on usage history table of evicted entries
US10241810B2 (en) 2012-05-18 2019-03-26 Nvidia Corporation Instruction-optimizing processor with branch-count table in hardware
US10324725B2 (en) 2012-12-27 2019-06-18 Nvidia Corporation Fault detection in instruction translations
US10108424B2 (en) 2013-03-14 2018-10-23 Nvidia Corporation Profiling code portions to generate translations

Similar Documents

Publication Publication Date Title
JPH0354660A (en) Shared memory management system for multiprocessor system
US6986005B2 (en) Low latency lock for multiprocessor computer system
JPH04246745A (en) Memory access system
JP2539352B2 (en) Hierarchical multi-computer system
US4084233A (en) Microcomputer apparatus
JPH0479026B2 (en)
JPH0525135B2 (en)
JPH0666060B2 (en) Bus priority control method
JPH0348544B2 (en)
US20230315636A1 (en) Multiprocessor system cache management with non-authority designation
JPH0234062B2 (en) MARUCHIPUROSETSUSASHISUTEMUNIOKERUMEMORIAKUSESUSEIGYOHOSHIKI
JPS6240565A (en) Memory control system
JPH02101560A (en) Bus interface device
JPS5839343B2 (en) Multiprocessor system instructions
JPH01263858A (en) Multi-processor system
JPS6160162A (en) Bus arbitration system
JPH0215152Y2 (en)
JPS62262159A (en) Electronic computer
JPH04236651A (en) Bus control system
JPS60129872A (en) Distributed processor of data
JPH0213339B2 (en)
JPH0374755A (en) Data access system
JPH04326453A (en) Multi-processor system
JPH02191057A (en) Multiplex processor system
JPH10247182A (en) Multiprocessor system