CN104462009A - Multiprocessor system and full interconnection method thereof - Google Patents
Multiprocessor system and full interconnection method thereof Download PDFInfo
- Publication number
- CN104462009A CN104462009A CN201410801282.5A CN201410801282A CN104462009A CN 104462009 A CN104462009 A CN 104462009A CN 201410801282 A CN201410801282 A CN 201410801282A CN 104462009 A CN104462009 A CN 104462009A
- Authority
- CN
- China
- Prior art keywords
- minimum logical
- coprocessor
- coprocessors
- computing module
- logical block
- 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
Links
Landscapes
- Multi Processors (AREA)
Abstract
本发明提供一种多处理器系统及多处理器系统的全互连方法,多处理器系统包括:两个以上的最小逻辑单元,每一个最小逻辑单元的两侧分别连接有对应于该最小逻辑单元的协处理器;所述最小逻辑单元包括至少一路计算模块,每一路计算模块包括两个相互连接的中央处理器CPU;每一路计算模块分别与自身所在最小逻辑单元两侧对应的协处理器相连,且位于各个最小逻辑单元同一侧的各个协处理器相连,实现任意两个所述最小逻辑单元中CPU的连接。根据上述方案,通过将位于最小逻辑单元同一侧的各个协处理器实现了任意两个最小逻辑单元中CPU的连接,从而减少了任意两个最小逻辑单元中CPU之间的跳步,提高了CPU的通信效率。
The present invention provides a multiprocessor system and a full interconnection method of the multiprocessor system. The multiprocessor system includes: more than two minimum logic units, and the two sides of each minimum logic unit are respectively connected with The coprocessor of the unit; the minimum logic unit includes at least one computing module, and each computing module includes two interconnected central processing units CPU; each computing module is respectively associated with the corresponding coprocessors on both sides of the minimum logic unit where it is located Each coprocessor located on the same side of each minimum logic unit is connected to realize the connection of CPUs in any two minimum logic units. According to the above scheme, the connection of the CPUs in any two minimum logic units is realized by each coprocessor located on the same side of the minimum logic unit, thereby reducing the jump between the CPUs in any two minimum logic units and improving the CPU performance. communication efficiency.
Description
技术领域 technical field
本发明涉及计算机技术领域,特别涉及一种多处理器系统及多处理器系统的全互连方法。 The invention relates to the technical field of computers, in particular to a multiprocessor system and a full interconnection method of the multiprocessor system.
背景技术 Background technique
随着计算机技术的飞速发展,为了满足经济社会发展的需要,高性能的计算机系统成为制约社会发展的关键因素。金融、电信等关键领域对计算机系统的性能要求极高,因此需要构建多处理器系统以实现各个领域的性能要求。 With the rapid development of computer technology, in order to meet the needs of economic and social development, high-performance computer systems have become a key factor restricting social development. Key fields such as finance and telecommunications have extremely high performance requirements for computer systems, so it is necessary to build multi-processor systems to meet the performance requirements of various fields.
目前,多处理器系统中各个中央处理器(CPU)之间相互连接,当多处理器系统中的任意两个CPU之间相互通信时,需要根据该相互通信的两个CPU的连接关系,进行多个跳步才能实现,导致CPU之间相互通信的效率较低。 At present, each central processing unit (CPU) in a multiprocessor system is connected to each other. When any two CPUs in a multiprocessor system communicate with each other, it is necessary to perform Multiple hops can be realized, resulting in low efficiency of communication between CPUs.
发明内容 Contents of the invention
本发明提供一种多处理器系统及多处理器系统的全互连方法,以解决现有技术中CPU之间相互通信时效率较低的问题。 The invention provides a multi-processor system and a full interconnection method of the multi-processor system to solve the problem of low efficiency in communication between CPUs in the prior art.
本发明实施例提供了一种多处理器系统,包括:两个以上的最小逻辑单元,每一个最小逻辑单元的两侧分别连接有对应于该最小逻辑单元的协处理器;所述最小逻辑单元包括至少一路计算模块,每一路计算模块包括两个相互连接的中央处理器CPU; The embodiment of the present invention provides a multiprocessor system, comprising: more than two minimum logic units, each of the two sides of the minimum logic unit is respectively connected with a coprocessor corresponding to the minimum logic unit; the minimum logic unit Including at least one computing module, each computing module includes two interconnected central processing units CPU;
其中,每一路计算模块分别与自身所在最小逻辑单元两侧对应的协处理器相连,且位于各个最小逻辑单元同一侧的各个协处理器相连,实现任意两个所述最小逻辑单元中CPU的连接。 Wherein, each calculation module is connected to the corresponding coprocessors on both sides of the minimum logic unit where it is located, and each coprocessor located on the same side of each minimum logic unit is connected to realize the connection of CPUs in any two minimum logic units .
优选地,包括8个最小逻辑单元; Preferably, it includes 8 minimum logic units;
在位于所述8个最小逻辑单元的同一侧,对应于所述8个最小逻辑单元的8个协处理器两两连接。 On the same side of the 8 minimum logic units, 8 coprocessors corresponding to the 8 minimum logic units are connected in pairs.
优选地,包括16个最小逻辑单元; Preferably, it includes 16 minimum logic units;
在位于所述16个最小逻辑单元的同一侧,对应于该16个最小逻辑单元的16个协处理器分别位于两个分组内,每个分组包括8个协处理器,位于同一分组的8个协处理器两两连接,且一个分组中的协处理器与另一个分组中的协处理器一一对应连接。 On the same side of the 16 smallest logical units, the 16 coprocessors corresponding to the 16 smallest logical units are respectively located in two groups, each group includes 8 coprocessors, and the 8 coprocessors located in the same group The coprocessors are connected in pairs, and the coprocessors in one group are connected to the coprocessors in the other group in one-to-one correspondence.
优选地,所述最小逻辑单元包括两路计算模块,其中一路计算模块中的CPU与另一路计算模块中的CPU一一对应连接。 Preferably, the minimum logic unit includes two computing modules, wherein the CPU in one computing module is connected to the CPU in the other computing module in a one-to-one correspondence.
本发明实施例还提供了一种多处理器系统的全互连方法,应用于上述的多处理器系统,所述多处理器系统包括:两个以上的最小逻辑单元,每一个最小逻辑单元的两侧分别连接有对应于该最小逻辑单元的协处理器;所述最小逻辑单元包括至少一路计算模块,每一路计算模块包括两个相互连接的中央处理器CPU;所述全互连方法包括: The embodiment of the present invention also provides a method for full interconnection of a multiprocessor system, which is applied to the above-mentioned multiprocessor system. The multiprocessor system includes: more than two minimum logic units, each of the minimum logic units Both sides are respectively connected with coprocessors corresponding to the minimum logic unit; the minimum logic unit includes at least one computing module, and each computing module includes two interconnected central processing units CPU; the full interconnection method includes:
将每一路计算模块分别与该路计算模块所在最小逻辑单元两侧对应的协处理器相连; Connect each computing module to the corresponding coprocessors on both sides of the smallest logic unit where the computing module is located;
将位于各个最小逻辑单元同一侧的各个协处理器相连,以实现任意两个所述最小逻辑单元中的CPU相连。 The coprocessors located on the same side of each minimum logic unit are connected to realize the connection of CPUs in any two minimum logic units.
优选地,所述将位于各个最小逻辑单元同一侧的各个协处理器相连,包括: Preferably, the connecting the coprocessors located on the same side of each minimum logic unit includes:
在所述多处理器系统包括8个最小逻辑单元时,将位于所述8个最小逻辑单元同一侧,对应于所述8个最小逻辑单元的8个协处理器进行两两相连。 When the multiprocessor system includes 8 minimum logic units, the 8 coprocessors located on the same side of the 8 minimum logic units and corresponding to the 8 minimum logic units are connected in pairs.
优选地,所述将位于各个最小逻辑单元同一侧的各个协处理器相连,包括: Preferably, the connecting the coprocessors located on the same side of each minimum logic unit includes:
在所述多处理器系统包括16个最小逻辑单元时,将位于所述16个最小逻辑单元同一侧,对应于所述16个最小逻辑单元的16个协处理器分成两个分组,其中,每个分组包括8个协处理器; When the multiprocessor system includes 16 minimum logic units, the 16 coprocessors corresponding to the 16 minimum logic units on the same side of the 16 minimum logic units are divided into two groups, wherein each A group includes 8 coprocessors;
将位于同一分组的8个协处理器进行两两相连,以及将一个分组中的协处理器与另一个分组中的协处理器一一对应相连。 The 8 coprocessors in the same group are connected two by two, and the coprocessors in one group are connected with the coprocessors in another group in one-to-one correspondence.
优选地,进一步包括: Preferably, further comprising:
在所述最小逻辑单元包括两路计算模块时,将一路计算模块中的CPU与另一路计算模块中的CPU一一对应连接。 When the minimum logic unit includes two calculation modules, the CPU in one calculation module is connected to the CPU in the other calculation module in one-to-one correspondence.
本发明实施例提供了一种多处理器系统及多处理器系统的全互连方法,该多处理器系统包括两个以上的最小逻辑单元,每一个最小逻辑单元的两侧分别连接有对应于该最小逻辑单元的协处理器,通过将位于最小逻辑单元同一侧的各个协处理器实现了任意两个最小逻辑单元中CPU的连接,从而减少了任意两个最小逻辑单元中CPU之间的跳步,提高了CPU的通信效率。 The embodiment of the present invention provides a multiprocessor system and a full interconnection method of the multiprocessor system. The multiprocessor system includes more than two minimum logic units, and the two sides of each minimum logic unit are respectively connected with corresponding The coprocessor of the minimum logic unit realizes the connection of CPUs in any two minimum logic units by connecting each coprocessor on the same side of the minimum logic unit, thereby reducing the jump between CPUs in any two minimum logic units Step, improve the communication efficiency of CPU.
附图说明 Description of drawings
图1是本发明实施例提供的多处理器系统结构示意图; FIG. 1 is a schematic structural diagram of a multiprocessor system provided by an embodiment of the present invention;
图2是本发明实施例提供的单侧8个协议处理器的连接关系示意图; Fig. 2 is a schematic diagram of the connection relationship of 8 protocol processors on one side provided by the embodiment of the present invention;
图3是本发明实施例提供的单侧16个协议处理器的连接关系示意图; Fig. 3 is a schematic diagram of the connection relationship of 16 protocol processors on one side provided by the embodiment of the present invention;
图4是本发明实施例提供的多处理器系统的全互连方法流程图。 FIG. 4 is a flowchart of a full interconnection method for a multiprocessor system provided by an embodiment of the present invention.
具体实施方式 Detailed ways
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述。显然,所描述的实施例仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。 The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Apparently, the described embodiments are only some of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.
为了解决现有技术的缺点,本发明考虑到多个处理器能够共享内存的特点,以及考虑到多处理器系统随着系统规模增大,扩展系统性能的线性增长的需求,如图1所示,本发明实施例提供了一种多处理器系统,该多处理器系统可以包括: In order to solve the shortcomings of the prior art, the present invention takes into account the characteristics that multiple processors can share memory, and considers the requirement of linear growth of multi-processor system as the system scale increases, expanding the performance of the system, as shown in Figure 1 , the embodiment of the present invention provides a multiprocessor system, the multiprocessor system may include:
两个以上的最小逻辑单元10,每一个最小逻辑单元10的两侧分别连接有对应于该最小逻辑单元10的协处理器(CC,coprocessor)11,其中,多处理器系统是以各个最小逻辑单元10为中心的对称结构。最小逻辑单元10包括至少一路计算模块12,每一路计算模块12包括两个相互连接的中央处理器;每一路计算模块12分别与自身所在最小逻辑单元10两侧对应的协处理器11相连,且位于各个最小逻辑单元10同一侧的各个协处理器11相连,实现任意两个最小逻辑单元10中CPU的连接。其中,各个最小逻辑单元10两侧的各个协处理器11的连接关系相同。 More than two minimum logic units 10, each of the two sides of the minimum logic unit 10 is respectively connected with a coprocessor (CC, coprocessor) 11 corresponding to the minimum logic unit 10, wherein the multiprocessor system uses each minimum logic unit Unit 10 is a centrally symmetrical structure. The minimum logic unit 10 includes at least one path computing module 12, and each path computing module 12 includes two interconnected central processing units; each path computing module 12 is connected to the corresponding coprocessor 11 on both sides of the minimum logic unit 10 where it is located, and The coprocessors 11 located on the same side of each minimum logic unit 10 are connected to realize the connection of CPUs in any two minimum logic units 10 . Wherein, the connection relationship of each coprocessor 11 on both sides of each minimum logic unit 10 is the same.
在图1中,仅示出了两个最小逻辑单元10,且每个最小逻辑单元10中包括两路计算模块12,每一路计算模块12包括两个相互连接的中央处理器(CPU,Central Processing Unit)。 In Fig. 1, only two minimum logic units 10 are shown, and each minimum logic unit 10 includes two-way calculation modules 12, and each way calculation module 12 includes two interconnected central processing units (CPU, Central Processing Unit).
其中,每个最小逻辑单元10中包括的两路计算模块12之间的连接关系是:一路计算模块12与另一路计算模块12中的CPU一一对应连接。从而保证了每个最小逻辑单元10中的所有CPU能够直接或间接连接。另外,在每个最小逻辑单元10中还包括:独立的计算资源、内存资源与I/O 资源。最小逻辑单元10中的所有CPU共享这些资源,从而实现每个最小逻辑单元10独立运行的操作系统。 Wherein, the connection relationship between the two calculation modules 12 included in each minimum logic unit 10 is: one calculation module 12 is connected to the CPU in the other calculation module 12 in a one-to-one correspondence. This ensures that all CPUs in each minimum logic unit 10 can be directly or indirectly connected. In addition, each minimum logic unit 10 also includes: independent computing resources, memory resources and I/O resources. All CPUs in the minimum logic unit 10 share these resources, so as to implement an operating system that each minimum logic unit 10 runs independently.
在本实施例中,为了保证各个最小逻辑单元10之间的CPU能够通信,将各个最小逻辑单元10同一侧的各个协处理器11通过内部高速互连网络相连,用于维护多处理器系统的cache一致性, In this embodiment, in order to ensure that the CPUs between each minimum logic unit 10 can communicate, each coprocessor 11 on the same side of each minimum logic unit 10 is connected through an internal high-speed interconnection network, which is used to maintain the multiprocessor system. cache consistency,
在本实施例中,多处理器系统为一个紧耦合形式,即当每次扩展时,需要对各个最小逻辑单元10两侧的协处理器的连接关系进行改变。 In this embodiment, the multiprocessor system is in a tightly coupled form, that is, the connection relationship between the coprocessors on both sides of each minimum logic unit 10 needs to be changed every time it is expanded.
其中,由于每个最小逻辑单元10中包括4个CPU,在多处理器系统中包括8个最小逻辑单元时,为多处理器系统扩展至32路系统。如图2所示,为8个最小逻辑单元10同一侧的各个协处理器11的连接关系示意图。该连接关系为:在位于8个最小逻辑单元10的同一侧,对应于所述8个最小逻辑单元10的8个协处理器11两两连接,即任一个协处理器11分别与另外7个协处理器11相连,从而保障了系统性能。由于最小处理器单元10两侧的协处理器11没有访问的需要,因此8个最小逻辑单元10两侧的协处理器11不相连。 Wherein, since each minimum logic unit 10 includes 4 CPUs, when the multiprocessor system includes 8 minimum logic units, the multiprocessor system is extended to a 32-way system. As shown in FIG. 2 , it is a schematic diagram of the connection relationship of each coprocessor 11 on the same side of the eight minimum logic units 10 . The connection relationship is: on the same side of the 8 smallest logic units 10, the 8 coprocessors 11 corresponding to the 8 smallest logic units 10 are connected in pairs, that is, any coprocessor 11 is connected to the other 7 Coprocessors 11 are connected to ensure system performance. Since the coprocessors 11 on both sides of the smallest processor unit 10 do not need to access, the coprocessors 11 on both sides of the eight smallest logic units 10 are not connected.
其中,在多处理器系统中包括16个最小逻辑单元时,多处理器系统扩展至64路系统。如图3所示,为16个最小逻辑单元10同一侧的各个协处理器11的连接关系示意图。该连接关系为:在位于所述16个最小逻辑单元的同一侧,对应于该16个最小逻辑单元的16个协处理器分别位于两个分组内,如分组一和分组二,分组一和分组二分部包括8个协处理器,分组一中的8个协处理器两两连接,该两两连接的示意图如图2所示,以及分组二中的8个协处理器两两连接,该两两连接的示意图如图2所示。且一个分组中的协处理器与另一个分组中的协处理器一一对应连接。例如,分组一中的8个协处理器的编号为100、110、120、……170,分组二的8个协处理器的编号为200、210、220……270,那么分组一和分组二中的连接关系为如图3所示,协处理器100与协处理器200相连,协处理器110与协处理器210相连,协处理器120与协处理器220相连,……,协处理170与协处理器270相连。由于最小处理器单元10两侧的协处理器11没有访问的需要,因此16个最小逻辑单元10两侧的协处理器11不相连。 Among them, when the multiprocessor system includes 16 minimum logic units, the multiprocessor system is extended to a 64-way system. As shown in FIG. 3 , it is a schematic diagram of the connection relationship of each coprocessor 11 on the same side of the 16 minimum logic units 10 . The connection relationship is: on the same side of the 16 smallest logic units, the 16 coprocessors corresponding to the 16 smallest logic units are respectively located in two groups, such as group one and group two, group one and group two The second division includes 8 coprocessors, the 8 coprocessors in the group one are connected in pairs, the schematic diagram of the pairwise connection is shown in Figure 2, and the 8 coprocessors in the group two are connected in pairs, the two A schematic diagram of the two connections is shown in Figure 2. And the coprocessors in one group are connected with the coprocessors in another group in one-to-one correspondence. For example, the numbers of the 8 coprocessors in group 1 are 100, 110, 120, ... 170, and the numbers of the 8 coprocessors in group 2 are 200, 210, 220 ... 270, then group 1 and group 2 The connection relationship among is as shown in Figure 3, coprocessor 100 is connected with coprocessor 200, coprocessor 110 is connected with coprocessor 210, coprocessor 120 is connected with coprocessor 220, ..., coprocessor 170 Connected to coprocessor 270 . Since the coprocessors 11 on both sides of the smallest processor unit 10 do not need to be accessed, the coprocessors 11 on both sides of the 16 smallest logic units 10 are not connected.
根据上述方案,该多处理器系统包括两个以上的最小逻辑单元,每一个最小逻辑单元的两侧分别连接有对应于该最小逻辑单元的协处理器,通过将位于最小逻辑单元同一侧的各个协处理器实现了任意两个最小逻辑单元中CPU的连接,从而减少了任意两个最小逻辑单元中CPU之间的跳步,提高了CPU的通信效率。 According to the above scheme, the multiprocessor system includes more than two minimum logic units, and the two sides of each minimum logic unit are respectively connected with coprocessors corresponding to the minimum logic unit. The coprocessor realizes the connection of the CPUs in any two smallest logic units, thereby reducing the jump between the CPUs in any two smallest logic units, and improving the communication efficiency of the CPUs.
如图4所示,本发明实施例还提供了一种多处理器系统的全互连方法,应用于上述多处理器系统,所述多处理器系统包括:两个以上的最小逻辑单元,每一个最小逻辑单元的两侧分别连接有对应于该最小逻辑单元的协处理器;所述最小逻辑单元包括至少一路计算模块,每一路计算模块包括两个相互连接的中央处理器CPU;所述全互连方法包括: As shown in FIG. 4 , the embodiment of the present invention also provides a method for full interconnection of a multiprocessor system, which is applied to the above multiprocessor system. The multiprocessor system includes: more than two minimum logic units, each Coprocessors corresponding to the minimum logic unit are respectively connected to both sides of a minimum logic unit; the minimum logic unit includes at least one computing module, and each computing module includes two interconnected central processing units CPU; Interconnection methods include:
步骤401:将每一路计算模块分别与该路计算模块所在最小逻辑单元两侧对应的协处理器相连; Step 401: Connect each computing module to the corresponding coprocessors on both sides of the smallest logic unit where the computing module is located;
步骤402:将位于各个最小逻辑单元同一侧的各个协处理器相连,以实现任意两个所述最小逻辑单元中的CPU相连。 Step 402: Connect the coprocessors located on the same side of each minimum logic unit, so as to realize the connection of CPUs in any two minimum logic units.
所述将位于各个最小逻辑单元同一侧的各个协处理器相连,包括:在所述多处理器系统包括8个最小逻辑单元时,将位于所述8个最小逻辑单元同一侧,对应于所述8个最小逻辑单元的8个协处理器进行两两相连。 Said connecting the coprocessors located on the same side of each minimum logic unit includes: when the multiprocessor system includes 8 minimum logic units, connecting the coprocessors located on the same side of the 8 minimum logic units, corresponding to the The 8 coprocessors of the 8 smallest logic units are connected in pairs.
所述将位于各个最小逻辑单元同一侧的各个协处理器相连,包括:在所述多处理器系统包括16个最小逻辑单元时,将位于所述16个最小逻辑单元同一侧,对应于所述16个最小逻辑单元的16个协处理器分成两个分组,其中,每个分组包括8个协处理器; The connecting the coprocessors located on the same side of each minimum logic unit includes: when the multiprocessor system includes 16 minimum logic units, connecting the coprocessors located on the same side of the 16 minimum logic units, corresponding to the The 16 coprocessors of the 16 smallest logical units are divided into two groups, wherein each group includes 8 coprocessors;
将位于同一分组的8个协处理器进行两两相连,以及将一个分组中的协处理器与另一个分组中的协处理器一一对应相连。 The 8 coprocessors in the same group are connected two by two, and the coprocessors in one group are connected with the coprocessors in another group in one-to-one correspondence.
进一步包括: Further includes:
在所述最小逻辑单元包括两路计算模块时,将一路计算模块中的CPU与另一路计算模块中的CPU一一对应连接。 When the minimum logic unit includes two calculation modules, the CPU in one calculation module is connected to the CPU in the other calculation module in one-to-one correspondence.
以上所述仅为本发明的较佳实施例而已,并不用以限制本发明,凡在本发明的精神和原则之内,所做的任何修改、等同替换、改进等,均应包含在本发明保护的范围之内。 The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the present invention. within the scope of protection.
Claims (8)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410801282.5A CN104462009A (en) | 2014-12-22 | 2014-12-22 | Multiprocessor system and full interconnection method thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410801282.5A CN104462009A (en) | 2014-12-22 | 2014-12-22 | Multiprocessor system and full interconnection method thereof |
Publications (1)
Publication Number | Publication Date |
---|---|
CN104462009A true CN104462009A (en) | 2015-03-25 |
Family
ID=52908084
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410801282.5A Pending CN104462009A (en) | 2014-12-22 | 2014-12-22 | Multiprocessor system and full interconnection method thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104462009A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112416850A (en) * | 2020-11-20 | 2021-02-26 | 新华三云计算技术有限公司 | A multi-processor interconnected system and its communication method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5630162A (en) * | 1990-11-13 | 1997-05-13 | International Business Machines Corporation | Array processor dotted communication network based on H-DOTs |
US5761455A (en) * | 1995-02-06 | 1998-06-02 | Cpu Technology, Inc. | Dynamic bus reconfiguration logic |
CN1937782A (en) * | 2006-08-04 | 2007-03-28 | 华为数字技术有限公司 | Frame device and its message transmitting method |
CN101216815A (en) * | 2008-01-07 | 2008-07-09 | 浪潮电子信息产业股份有限公司 | A Two-Wing Scalable Multiprocessor Tightly Coupled Shared Memory Architecture |
-
2014
- 2014-12-22 CN CN201410801282.5A patent/CN104462009A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5630162A (en) * | 1990-11-13 | 1997-05-13 | International Business Machines Corporation | Array processor dotted communication network based on H-DOTs |
US5761455A (en) * | 1995-02-06 | 1998-06-02 | Cpu Technology, Inc. | Dynamic bus reconfiguration logic |
CN1937782A (en) * | 2006-08-04 | 2007-03-28 | 华为数字技术有限公司 | Frame device and its message transmitting method |
CN101216815A (en) * | 2008-01-07 | 2008-07-09 | 浪潮电子信息产业股份有限公司 | A Two-Wing Scalable Multiprocessor Tightly Coupled Shared Memory Architecture |
Non-Patent Citations (1)
Title |
---|
王洪玉 等: "全互连立方体网络在并行处理系统中的应用", 《计算机研究与发展》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112416850A (en) * | 2020-11-20 | 2021-02-26 | 新华三云计算技术有限公司 | A multi-processor interconnected system and its communication method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10409766B2 (en) | Computer subsystem and computer system with composite nodes in an interconnection structure | |
EP3140748B1 (en) | Interconnect systems and methods using hybrid memory cube links | |
CN102129418B (en) | High-performance fault-tolerant computer system and implementation method | |
US8769458B2 (en) | Prototype verification system and verification method for high-end fault-tolerant computer | |
CN104360982B (en) | A kind of host computer system bibliographic structure method and system based on restructural chip technology | |
US20160378548A1 (en) | Hybrid heterogeneous host system, resource configuration method and task scheduling method | |
CN103336756B (en) | A kind of generating apparatus of data computational node | |
CN115994107B (en) | Access Acceleration System for Storage Devices | |
CN107247623A (en) | A kind of distributed cluster system and data connecting method based on multi-core CPU | |
CN104408014A (en) | System and method for interconnecting processing units of calculation systems | |
Akgun et al. | Scalable memory fabric for silicon interposer-based multi-core systems | |
US11461234B2 (en) | Coherent node controller | |
CN106844263B (en) | Configurable multiprocessor-based computer system and implementation method | |
CN106951390B (en) | A NUMA system construction method that can reduce memory access latency across nodes | |
CN104199521A (en) | Blade node and extension method thereof | |
CN104035525A (en) | Computational node | |
CN105045670A (en) | Method and system for balancing loads of central processing units and graphic processing units | |
WO2021027745A1 (en) | Graph reconstruction method and apparatus | |
CN118714020B (en) | Computer network system and computer system | |
CN104462009A (en) | Multiprocessor system and full interconnection method thereof | |
CN208766658U (en) | A kind of server system | |
CN103294639A (en) | CPU+MIC mixed heterogeneous cluster system for achieving large-scale computing | |
CN112486402B (en) | Storage node and system | |
CN107239432A (en) | A kind of server with novel topological structure | |
US20250112698A1 (en) | Multi-fiber cable connectivity |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150325 |
|
WD01 | Invention patent application deemed withdrawn after publication |