CN105488011B - The memory access processing method and network-on-chip of network-on-chip - Google Patents
The memory access processing method and network-on-chip of network-on-chip Download PDFInfo
- Publication number
- CN105488011B CN105488011B CN201410483305.2A CN201410483305A CN105488011B CN 105488011 B CN105488011 B CN 105488011B CN 201410483305 A CN201410483305 A CN 201410483305A CN 105488011 B CN105488011 B CN 105488011B
- Authority
- CN
- China
- Prior art keywords
- core
- memory
- network
- data
- node
- 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.)
- Expired - Fee Related
Links
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The embodiment of the invention discloses the memory access processing methods and network-on-chip of a kind of network-on-chip.Present invention method includes:It determines the access operation request of the verification memory on network-on-chip, wherein including cable network and wireless network on the network-on-chip, at least one core is separated between two neighboring radio node in the wireless network;When access operation request is read request, the read request is sent to the memory by the core by the cable network, and the data fed back to the read request are sent to the core by the memory by the wireless network;When access operation request is write request, the data of the write request and write-in are sent to the memory by the core by the cable network respectively.The embodiment of the present invention can make full use of the advantages of Wi-Fi and cable network on network-on-chip to transmit data.
Description
Technical field
The present invention relates to the memory access processing methods and network-on-chip of moving communicating field more particularly to a kind of network-on-chip.
Background technology
Network-on-chip is a kind of communication means of system on chip, is the chief component of multi-core technology.Network-on-chip is normal
What is seen is using cable network, the i.e. network based on metal wire interconnection.As shown in Figure 1, Fig. 1 is wired online in the prior art
The common on piece Mesh structures in road.Multiple cores 10 in network-on-chip are arranged in a matrix, metal wire 11 respectively along horizontal and
Vertical direction is netted to form one by the connection of each core, wherein each core is a node.When appointing in the on piece network
When communicating between two cores of meaning, the data of communication are transmitted to another core from a core in two cores along metal wire.So
And as the quantity of the node on network-on-chip increases, the quantity of node skipped when communicating between core is bigger, and communication is caused to be prolonged
Late and power consumption increases.
The common another kind of network-on-chip is using wireless network.There are one its corresponding for each core on piece network
Wireless transmitter is communicated between two cores by the corresponding wireless transmitter of two cores, to make full use of channel radio
One benefit of letter, high bandwidth.However, wireless transmitter can occupy many chip areas, in the more situation of wireless transmitter
Under, the frequency between each wireless transmitter can interfere with each other.
Invention content
An embodiment of the present invention provides the memory access processing methods and network-on-chip of a kind of network-on-chip, can make full use of piece
The advantages of Wi-Fi and cable network on upper network, transmits data.
First aspect of the embodiment of the present invention provides a kind of memory access processing method of network-on-chip, including:
The access operation request of the verification memory on network-on-chip is determined, wherein including cable network on the network-on-chip
And wireless network, at least one core is separated between two neighboring radio node in the wireless network;
When access operation request is read request, the read request is sent to by the core by the cable network
The data fed back to the read request are sent to the core by the memory, the memory by the wireless network;
When the access operation request for write request when, the core respectively by the cable network by the write request with
And the data of write-in are sent to the memory.
With reference to the embodiment of the present invention in a first aspect, in the first realization method of the first aspect of the embodiment of the present invention,
The data fed back to the read request are sent to the core by the wireless network and specifically included by the memory:
The data fed back to the read request are sent to initial wireless node by the memory, wherein the initial wireless section
Point is the radio node nearest from the memory;
The initial wireless node determines the most short wireless path between the initial wireless node and purpose radio node
For transmitting path, wherein the purpose radio node is the radio node from the kernel nearest;
The data are sent to the purpose radio node by the initial wireless node by the transmitting path;
The data are sent to the core by the purpose radio node.
With reference to the first realization method of the first aspect of the embodiment of the present invention, the of the first aspect of the embodiment of the present invention
In two kinds of realization methods, the initial wireless node determines the shortest path between the initial wireless node and purpose radio node
Diameter is specifically included for transmitting path:
The initial wireless node determines all most short wireless between the initial wireless node and purpose radio node
Path;
The initial wireless node is by congestion aware by the minimum road of Congestion Level SPCC in all most short wireless paths
Diameter is as transmitting path.
With reference to the embodiment of the present invention in a first aspect, in the third realization method of the first aspect of the embodiment of the present invention,
The data of the read request, the write request and said write are sent to described by the core by the cable network respectively
Memory specifically includes:
The core determines that the most short wireline pathway between the core and the memory is transmitting path;
The core is respectively passed the data of the read request, the write request and said write by the transmitting path
It send to the memory.
With reference to the third realization method of the first aspect of the embodiment of the present invention, the of the first aspect of the embodiment of the present invention
In four kinds of realization methods, the core determines that the most short wireline pathway between the core and the memory is specifically wrapped for transmitting path
It includes:
The core determines all most short wireline pathways between the core and the memory;
The core regard the minimum path of Congestion Level SPCC in all most short wireline pathways as transmission by congestion aware
Path.
Second aspect of the embodiment of the present invention provides a kind of network-on-chip, including:
The cable network and wireless network of multiple cores, memory and the multiple core of connection, wherein in the wireless network
At least one core is separated between two neighboring radio node;
The core is used for when sending read request, write request and the data of write-in to the memory, by described wired
The read request, write request and the data of write-in are sent to the memory by network;
The memory is used to that the data for feeding back the read request to be sent to the core by the wireless network.
With reference to the second aspect of the embodiment of the present invention, in the first realization method of the second aspect of the embodiment of the present invention,
The memory is specifically used for the data for feeding back the read request being sent to initial wireless node, wherein the initial wireless section
Point is the radio node nearest from the memory;
The initial wireless node is specifically used for determining most short between the initial wireless node and purpose radio node
Wireless path is transmitting path, the data is sent to the purpose radio node by the transmitting path, wherein described
Purpose radio node is the radio node from the kernel nearest;
The purpose radio node is specifically used for the data being sent to the core.
With reference to the first realization method of the second aspect of the embodiment of the present invention, the of the second aspect of the embodiment of the present invention
In two kinds of realization methods, the initial wireless node is specifically used for determining between the initial wireless node and purpose radio node
All most short wireless paths, the minimum path of Congestion Level SPCC in all most short wireless paths is used as by congestion aware
Transmitting path.
With reference to the second aspect of the embodiment of the present invention, in the third realization method of the second aspect of the embodiment of the present invention,
The core is specifically used for determining that the most short wireline pathway between the core and the memory is transmitting path, passes through the biography respectively
Send path that the data of the read request, the write request and said write are sent to the memory.
With reference to the third realization method of the second aspect of the embodiment of the present invention, the of the second aspect of the embodiment of the present invention
In four kinds of realization methods, the core is specifically used for determining all most short wireline pathways between the core and the memory, passes through
Congestion aware is using the minimum path of Congestion Level SPCC in all most short wireline pathways as transmitting path.
As can be seen from the above technical solutions, the embodiment of the present invention has the following advantages:
In the present invention, cable network and wireless network are provided with simultaneously on network-on-chip, it is wherein adjacent in wireless network
It is separated at least one core between two radio nodes, the arrangement density of radio node is reduced wireless compared to small in background technology
The degree that node interferes with each other;When check memory access operation request for read request when, the read request by cable network from
Core is sent to memory, and the data fed back to the read request are sent to core by memory by wireless network from memory;Work as access operation
When asking as write request, the data of the write request and write-in are sent to memory by cable network from core;Due to practice
The data volume of data that memory feeds back read request in middle big data processing is often much larger than the data volume of the data of write-in, therefore
The high bandwidth of wireless network can be made full use of to transmit the data that memory feeds back read request using wireless network, is avoided
Using power consumption is larger caused by need to skip multiple nodes during cable network transmission;And read request and write request are in itself
Data volume and the data volume of write-in are relatively small, which is transmitted by cable network, can ensure that data are front and rear and visit
Ask the consistency of sequence.
Description of the drawings
Fig. 1 is the structure diagram of the common on piece Mesh structures in wired on piece network in the prior art;
Fig. 2 is the flow chart of the memory access processing method of network-on-chip in one embodiment of the present of invention;
Fig. 3 is the structure diagram of network-on-chip in one embodiment of the present of invention.
Specific embodiment
An embodiment of the present invention provides a kind of memory access processing method of network-on-chip, for making full use of on network-on-chip
The advantages of Wi-Fi and cable network, transmits data.
In order to which those skilled in the art is made to more fully understand the present invention program, below in conjunction in the embodiment of the present invention
The technical solution in the embodiment of the present invention is clearly and completely described in attached drawing, it is clear that described embodiment is only
The embodiment of a part of the invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
Member's all other embodiments obtained without making creative work should all belong to the model that the present invention protects
It encloses.
Term " comprising " and " having " in description and claims of this specification and above-mentioned attached drawing and they
Any deformation, it is intended that cover it is non-exclusive include, for example, containing the process of series of steps or unit, method, being
System, product or equipment are not necessarily limited to those steps clearly listed or unit, but may include not listing clearly or
For the intrinsic other steps of these processes, method, system, product or equipment or unit.
The memory access processing method of the network-on-chip of the present invention is described in detail below.
Referring to Fig. 2, the memory access processing method of network-on-chip includes in one embodiment of the present of invention:
201st, the access operation request of the verification memory on network-on-chip is determined, wherein including on the network-on-chip wired
Network and wireless network are separated at least one core in the wireless network between two neighboring radio node;
In the present embodiment, the topological structure of network-on-chip can be any regular or irregular structure, herein not
It is restricted.Network-on-chip includes multiple cores, wherein being connected between each core by metal wire, to form a cable network.
Data are transmitted by the metal wire that is connected between two cores and core when communicating between any two core.
Multiple wireless transmitters, wherein one on each wireless transmitter and network-on-chip are additionally provided on network-on-chip
Nuclear phase corresponds to, and a wireless generator is a radio node.It is right compared to each core on the network-on-chip in background technology
Answer a radio node, in the present embodiment, network-on-chip only has part core to have corresponding radio node, and two neighboring
At least every there are one cores between radio node.
When transmitting data, each radio node can only be transferred data to positioned at apart from the wireless section two radio nodes
Radio node in point presetting range.If mutually passing the distance between two radio nodes of data is less than or equal to preset model
It encloses, then the radio node of transmission data side directly transfers data to purpose radio node.Mutually pass two radio nodes of data
The distance between if more than presetting range, then the radio node of sender first send data to positioned at the radio node it is preset away from
From an interior radio node, referred to as first radio node.First radio node send data to again positioned at this
A radio node in the preset distance of one radio node, referred to as second radio node.Gone down with this, until the data
It is sent to purpose radio node.
Between mutually two radio nodes of data are passed, each radio node that data are passed through successively is linked to be one without circuit
Diameter.At least two nodes are equipped in the preset distance of at least part radio node if can be seen that, then mutually pass two of data
A plurality of wireless path can be formed between radio node.Mutually any one nothing can be selected between two radio nodes of biography data
Thread path transmits data.
202nd, when access operation request is read request, the core is passed the read request by the cable network
It send to the memory, the memory is sent the data fed back to the read request to the core by the wireless network;
In the present embodiment, the access operation for checking memory includes two kinds of reading and write-in.When access operation is reads, core
Read request message is sent to memory by cable network.In practice, core, can be first slow when needing to read some data
Deposit middle lookup.If being stored with the data that need to be read in caching, core is directly read from caching, if not being stored in caching needs to read
The data taken, core just need read request being sent to memory by cable network.Read request message data volume itself is smaller, therefore
It is to be transmitted by cable network.
When interior presence receives read request, the data fed back to the read request are sent to core.In practice, especially
When big data is handled, the data volume for the data that memory is fed back often is much larger than the data volume of read request message in itself.Therefore, it is interior
It deposits and core to is sent the data of the feedback by wireless network, the advantages of to make full use of wireless network high bandwidth.
203rd, when access operation request is write request, the core respectively please by described write by the cable network
The data asked and be written are sent to the memory;
When access operation is write-in, write request message is passed and is sent to memory by cable network by core.Memory, which receives, to be write
It is ready during request for data write-in, so that core data to be written by cable network sends memory to store.It needs to note
Meaning, the specific circuit that core write request when write request is sent to memory by cable network transmits in cable network,
The specific circuit that the data being written with core when transmitting memory by cable network data to be written transmit in cable network
Can be same or different, this is not restricted.
In the present embodiment, cable network and wireless network, wherein phase in wireless network are provided with simultaneously on network-on-chip
At least one core is separated between adjacent two radio nodes, the arrangement density of radio node is compared to small in background technology, reduction nothing
The degree that line node interferes with each other;When checking the access operation request of memory as read request, which passes through cable network
Memory is sent to from core, the data fed back to the read request are sent to core by memory by wireless network from memory;It is grasped when accessing
When making request as write request, the data of the write request and write-in are sent to memory by cable network from core;Due to practical fortune
The data volume of data that memory feeds back read request in being handled with middle big data is often much larger than the data volume of the data of write-in, because
This can make full use of the high bandwidth of wireless network using the data fed back to transmit memory to read request of wireless network,
It avoids using power consumption is larger caused by it need to skip multiple nodes during cable network transmission;And read request and write request sheet
The data volume of body and the data volume of write-in are relatively small, which is transmitted by cable network, before can ensureing data
The consistency of access order afterwards.
In the present embodiment, the data fed back to read request by wireless network are sent to and send read request by memory
Core.There are many ways to transmission by wireless network.In the present embodiment, it is preferred that memory passes through the wireless network in step 102
The data fed back to the read request are sent to the core and specifically included by network:
Memory will send the data that read request is fed back to initial wireless node, and the wherein initial wireless node is from this
The closest radio node of memory.Specifically, memory is transferred data to by cable network at radio node.
Initial wireless node checks distance sends the nearest radio node of the core of read request namely purpose wirelessly saves
Point, it is transmitting path to determine the most short wireless path between the initial wireless node and purpose radio node.In the present embodiment, by
In not being that each core is corresponding there are one radio node on network-on-chip, if therefore sending the core of read request with corresponding nothing
Line node, then the corresponding radio node of the core is the radio node apart from the kernel nearest;If the core of the transmission read request does not have
Corresponding radio node, then the radio node corresponding to the core of the kernel nearest apart from transmission request is what is asked apart from the transmission
The radio node of kernel nearest.Have between initial wireless node and purpose radio node without several wireless paths, wherein without circuit
Radio node is fewer included in diameter, then wireless path is shorter.Comprising the minimum wireless path of radio node be most short nothing
Thread path.In practice, there may be at least two most short wireless path between initial wireless node and purpose radio node.
Preferably, initial wireless node can determine that the path that Congestion Level SPCC is minimum in all most short wireless paths is made by congestion aware
For transmitting path.
After determining transmitting path, initial wireless node transfers data to purpose radio node by the transmitting path.
Then the data are sent to the core for sending read request by purpose radio node.In practice, if the purpose radio node is is somebody's turn to do
Radio node corresponding to core, then the data can be conveyed directly to core.If the purpose radio node is not corresponding to the core
Radio node, then purpose radio node the data are sent to by the core by cable network again.
Described above is memory described in step 102 how the number that the read request will be fed back by the wireless network
According to being sent to the core.In a step 102, read request is sent to memory by core by cable network, in step 103, core point
Write request and the data of write-in are not sent to by memory by cable network.In practice, in this method, core can have more
The data of the read request, the write request and said write are sent to institute by kind of method by the cable network respectively
State memory.Preferably, when read request is sent to memory by core by cable network, alternatively, core by cable network by write request
When being sent to memory, alternatively, when core is sent to memory by cable network data to be written, specifically include:
Core determines that the most short wireline pathway between the core and memory is transmitting path.In the present embodiment, on network-on-chip
Between each core by metal wire be linked to be one it is netted.Wireline pathway between core and memory is the metal that the core is connected with memory
Line and other cores.As can be seen that in network-on-chip, there is a plurality of wireline pathway, gold wherein included between core and memory
The wireline pathway for belonging to the minimum number of line is most short wireline pathway.In practice, may have at least two between core and memory
The most short wireline pathway of item.Preferably, core can determine that Congestion Level SPCC is minimum in all most short wireline pathways by congestion aware
Path is as transmitting path.
The read request is sent to the memory or by the transmitting path by the core by the transmitting path
The write request is sent to the memory or the data of said write are sent in described by the transmitting path
It deposits.
For ease of understanding, below with a practical application scene to the memory access processing side of the network-on-chip of the embodiment of the present invention
Method is described.
Network-on-chip includes memory and multiple cores, wherein being connected between each core by metal wire, to form one
Cable network.Data are transmitted by the metal wire that is connected between two cores and core when communicating between any two core.
Multiple wireless transmitters, a nuclear phase pair wherein on each wireless transmitter and network-on-chip are additionally provided on network-on-chip
Should, a wireless generator is a radio node.Network-on-chip only has part core to have corresponding radio node, and adjacent
At least every there are one cores between two radio nodes.
One of core (hereinafter referred to as the first core) in network-on-chip need to from the memory read data in network-on-chip,
In first core not be equipped with corresponding radio node.First core first searches the data to be read in the buffer.It is if slow
The data are stored in depositing, then the first core is directly read from caching, if not being stored with the data in caching, the first core determines
All most short wireline pathways between first core and memory, and pass through congestion aware and determine in all most short wireline pathways
Then read request is sent to memory by the minimum path of Congestion Level SPCC as transmitting path by the transmitting path.
After memory receives read request, the data which feeds back will be transmitted to by cable network apart from the memory
Nearest radio node, referred to as initial wireless node.Initial wireless node checks distance sends the nearest of the core of read request
Radio node namely purpose radio node determine all most short wireless between the initial wireless node and purpose radio node
Path namely comprising the minimum wireless path of radio node, and pass through congestion aware and determine all most short wireless paths
The minimum path of middle Congestion Level SPCC is as transmitting path.After determining transmitting path, initial wireless node passes through the transmitting path
Purpose radio node is transferred data to, which is sent to the first core by purpose radio node by cable network again.
When the first core needs to write data to memory, what the first core determined between first core and memory all most short has
Thread path, and determine that transmission path is used as in the path that Congestion Level SPCC is minimum in all most short wireline pathways by congestion aware
Then write request is sent to memory by diameter by the transmitting path, be ready when memory receives write request for data write-in.
First core determines that Congestion Level SPCC is minimum in current all most short wireline pathways again by congestion aware
Then path is sent to memory to store as transmitting path by the transmitting path data to be written.
The memory access processing method of the network-on-chip in the embodiment of the present invention is described above, below to of the invention real
The network-on-chip applied in example is described, referring to Fig. 3, network-on-chip 300 includes in the embodiment of the present invention:
The cable network 303 and wireless network 304 of multiple cores 301, memory 302 and the multiple core of connection, wherein institute
It states and is separated at least one core in wireless network between two neighboring radio node;
The core 301 is used to, when sending the memory 302 read request, write request and the data of write-in, pass through institute
It states cable network 303 and the read request, write request and the data of write-in is sent to the memory 302;
The memory 302 is used to be sent to the data for feeding back the read request by the wireless network 304 described
Core 301.
In the present embodiment, cable network and wireless network are provided with simultaneously on network-on-chip, it is wherein adjacent in wireless network
It is separated at least one core between two radio nodes, the arrangement density of radio node is reduced wireless compared to small in background technology
The degree that node interferes with each other;When check memory access operation request for read request when, the read request by cable network from
Core is sent to memory, and the data fed back to the read request are sent to core by memory by wireless network from memory;Work as access operation
When asking as write request, the data of the write request and write-in are sent to memory by cable network from core;Due to practice
The data volume of data that memory feeds back read request in middle big data processing is often much larger than the data volume of the data of write-in, therefore
The high bandwidth of wireless network can be made full use of using the data fed back to transmit memory to read request of wireless network, is kept away
Exempt from using power consumption is larger caused by it need to skip multiple nodes during cable network transmission;And read request and write request are in itself
Data volume and write-in data volume it is relatively small, which is transmitted by cable network, before and after can ensureing data
The consistency of access order.
Preferably, in the present embodiment, the memory is specifically used for the data for feeding back the read request being sent to initially
Radio node, wherein the initial wireless node is the radio node nearest from the memory.The initial wireless node is specific
For determining that the most short wireless path between the initial wireless node and purpose radio node for transmitting path, passes through the biography
Send path that the data are sent to the purpose radio node, wherein the purpose radio node is the nothing from the kernel nearest
Line node.The purpose radio node is specifically used for the data being sent to the core.Further, it is preferred that described initial
Radio node is specifically used for determining all most short wireless paths between the initial wireless node and purpose radio node, passes through
Congestion aware is using the minimum path of Congestion Level SPCC in all most short wireless paths as transmitting path.
Preferably, in the present embodiment, the core is specifically used for determining that most short between the core and the memory has circuit
Diameter is transmitting path, is respectively passed the data of the read request, the write request and said write by the transmitting path
It send to the memory.Further, it is preferred that the core was specifically used for determining between the core and the memory all most short has
Thread path regard the minimum path of Congestion Level SPCC in all most short wireline pathways as transmitting path by congestion aware.
For ease of understanding, the network-on-chip of the embodiment of the present invention is described with a practical application scene below.
Network-on-chip includes memory and multiple cores, wherein being connected between each core by metal wire, to form one
Cable network.Data are transmitted by the metal wire that is connected between two cores and core when communicating between any two core.
Multiple wireless transmitters, a nuclear phase pair wherein on each wireless transmitter and network-on-chip are additionally provided on network-on-chip
Should, a wireless generator is a radio node.Network-on-chip only has part core to have corresponding radio node, and adjacent
At least every there are one cores between two radio nodes.
One of core (hereinafter referred to as the first core) in network-on-chip need to from the memory read data in network-on-chip,
In first core not be equipped with corresponding radio node.First core first searches the data to be read in the buffer.It is if slow
The data are stored in depositing, then the first core is directly read from caching, if not being stored with the data in caching, the first core determines
All most short wireline pathways between first core and memory, and pass through congestion aware and determine in all most short wireline pathways
Then read request is sent to memory by the minimum path of Congestion Level SPCC as transmitting path by the transmitting path.
After memory receives read request, the data which feeds back will be transmitted to by cable network apart from the memory
Nearest radio node, referred to as initial wireless node.Initial wireless node checks distance sends the nearest of the core of read request
Radio node namely purpose radio node determine all most short wireless between the initial wireless node and purpose radio node
Path namely comprising the minimum wireless path of radio node, and pass through congestion aware and determine all most short wireless paths
The minimum path of middle Congestion Level SPCC is as transmitting path.After determining transmitting path, initial wireless node passes through the transmitting path
Purpose radio node is transferred data to, which is sent to the first core by purpose radio node by cable network again.
When the first core needs to write data to memory, what the first core determined between first core and memory all most short has
Thread path, and determine that transmission path is used as in the path that Congestion Level SPCC is minimum in all most short wireline pathways by congestion aware
Then write request is sent to memory by diameter by the transmitting path, be ready when memory receives write request for data write-in.
First core determines that Congestion Level SPCC is minimum in current all most short wireline pathways again by congestion aware
Then path is sent to memory to store as transmitting path by the transmitting path data to be written.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit can refer to the corresponding process in preceding method embodiment, and details are not described herein.
In several embodiments provided herein, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of division of logic function can have other dividing mode, such as multiple units or component in actual implementation
It may be combined or can be integrated into another system or some features can be ignored or does not perform.Another point, it is shown or
The mutual coupling, direct-coupling or communication connection discussed can be the indirect coupling by some interfaces, device or unit
It closes or communicates to connect, can be electrical, machinery or other forms.
The unit illustrated as separating component may or may not be physically separate, be shown as unit
The component shown may or may not be physical unit, you can be located at a place or can also be distributed to multiple
In network element.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it can also
That each unit is individually physically present, can also two or more units integrate in a unit.Above-mentioned integrated list
The form that hardware had both may be used in member is realized, can also be realized in the form of SFU software functional unit.
If the integrated unit is realized in the form of SFU software functional unit and is independent product sale or uses
When, it can be stored in a computer read/write memory medium.Based on such understanding, technical scheme of the present invention is substantially
The part to contribute in other words to the prior art or all or part of the technical solution can be in the form of software products
It embodies, which is stored in a storage medium, is used including some instructions so that a computer
Equipment (can be personal computer, server or the network equipment etc.) performs the complete of each embodiment the method for the present invention
Portion or part steps.And aforementioned storage medium includes:USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only
Memory), random access memory (RAM, Random Access Memory), magnetic disc or CD etc. are various can store journey
The medium of sequence code.
The above, the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations;Although with reference to before
Embodiment is stated the present invention is described in detail, it will be understood by those of ordinary skill in the art that:It still can be to preceding
The technical solution recorded in each embodiment is stated to modify or carry out equivalent replacement to which part technical characteristic;And these
Modification is replaced, the spirit and scope for various embodiments of the present invention technical solution that it does not separate the essence of the corresponding technical solution.
Claims (8)
1. a kind of memory access processing method of network-on-chip, which is characterized in that including:
The access operation request of the verification memory on network-on-chip is determined, wherein including cable network and nothing on the network-on-chip
Gauze network is separated at least one core in the wireless network between two neighboring radio node;
When access operation request is read request, the read request is sent to described by the core by the cable network
The data fed back to the read request are sent to initial wireless node by memory, the memory, wherein the initial wireless node
For the radio node nearest from the memory;The initial wireless node determines the initial wireless node and purpose radio node
Between most short wireless path for transmitting path, wherein the purpose radio node is the radio node from the kernel nearest;Institute
It states initial wireless node and the data is sent to by the purpose radio node by the transmitting path;The purpose wirelessly saves
The data are sent to the core by point;
When access operation request is write request, the core by the write request and is write respectively by the cable network
The data entered are sent to the memory.
2. the memory access processing method of network-on-chip according to claim 1, which is characterized in that the initial wireless node is true
Fixed shortest path between the initial wireless node and purpose radio node is specifically included for transmitting path:
The initial wireless node determines all most short wireless paths between the initial wireless node and purpose radio node;
The initial wireless node is made the minimum path of Congestion Level SPCC in all most short wireless paths by congestion aware
For transmitting path.
3. the memory access processing method of network-on-chip according to claim 1, which is characterized in that the core is respectively by described
The data of the read request, the write request and said write are sent to the memory and specifically included by cable network:
The core determines that the most short wireline pathway between the core and the memory is transmitting path;
The data of the read request, the write request and said write are sent to by the core by the transmitting path respectively
The memory.
4. the memory access processing method of network-on-chip according to claim 3, which is characterized in that the core determine the core with
Most short wireline pathway between the memory is specifically included for transmitting path:
The core determines all most short wireline pathways between the core and the memory;
The core regard the minimum path of Congestion Level SPCC in all most short wireline pathways as transmitting path by congestion aware.
5. a kind of network-on-chip, which is characterized in that including:
The cable network and wireless network of multiple cores, memory and the multiple core of connection, wherein adjacent in the wireless network
At least one core is separated between two radio nodes;
The core is used to, when sending the memory read request, write request and the data of write-in, pass through the cable network
The read request, write request and the data of write-in are sent to the memory;
The memory is specifically used for the data for feeding back the read request being sent to initial wireless node, wherein the initial nothing
Line node is the radio node nearest from the memory;
The initial wireless node is specifically used for determining most short wireless between the initial wireless node and purpose radio node
Path is transmitting path, the data is sent to the purpose radio node by the transmitting path, wherein the purpose
Radio node is the radio node from the kernel nearest;
The purpose radio node is specifically used for the data being sent to the core.
6. network-on-chip according to claim 5, which is characterized in that
The initial wireless node is specifically used for determining all most short between the initial wireless node and purpose radio node
Wireless path regard the minimum path of Congestion Level SPCC in all most short wireless paths as transmitting path by congestion aware.
7. network-on-chip according to claim 5, which is characterized in that
The core is specifically used for determining that the most short wireline pathway between the core and the memory is transmitting path, passes through institute respectively
It states transmitting path and the data of the read request, the write request and said write is sent to the memory.
8. network-on-chip according to claim 7, which is characterized in that
The core is specifically used for determining all most short wireline pathways between the core and the memory, by congestion aware by institute
Congestion Level SPCC is minimum in all most short wireline pathways path is stated as transmitting path.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410483305.2A CN105488011B (en) | 2014-09-19 | 2014-09-19 | The memory access processing method and network-on-chip of network-on-chip |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410483305.2A CN105488011B (en) | 2014-09-19 | 2014-09-19 | The memory access processing method and network-on-chip of network-on-chip |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105488011A CN105488011A (en) | 2016-04-13 |
CN105488011B true CN105488011B (en) | 2018-06-19 |
Family
ID=55674995
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410483305.2A Expired - Fee Related CN105488011B (en) | 2014-09-19 | 2014-09-19 | The memory access processing method and network-on-chip of network-on-chip |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105488011B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102035723A (en) * | 2009-09-28 | 2011-04-27 | 清华大学 | On-chip network router and realization method |
CN103124420A (en) * | 2013-01-21 | 2013-05-29 | 电子科技大学 | Wireless on-chip network structuring method |
CN103188158A (en) * | 2011-12-28 | 2013-07-03 | 清华大学 | On-chip network router and on-chip network routing method |
CN103984674A (en) * | 2014-05-15 | 2014-08-13 | 厦门大学 | Central controller used for wireless on-chip network communication and control method thereof |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090109996A1 (en) * | 2007-10-29 | 2009-04-30 | Hoover Russell D | Network on Chip |
US8526422B2 (en) * | 2007-11-27 | 2013-09-03 | International Business Machines Corporation | Network on chip with partitions |
-
2014
- 2014-09-19 CN CN201410483305.2A patent/CN105488011B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102035723A (en) * | 2009-09-28 | 2011-04-27 | 清华大学 | On-chip network router and realization method |
CN103188158A (en) * | 2011-12-28 | 2013-07-03 | 清华大学 | On-chip network router and on-chip network routing method |
CN103124420A (en) * | 2013-01-21 | 2013-05-29 | 电子科技大学 | Wireless on-chip network structuring method |
CN103984674A (en) * | 2014-05-15 | 2014-08-13 | 厦门大学 | Central controller used for wireless on-chip network communication and control method thereof |
Non-Patent Citations (1)
Title |
---|
混合无线片上网络结构及通信机制的研究;吕鹏;《中国优秀硕士论文全文数据库》;20140315;第I135-223页 * |
Also Published As
Publication number | Publication date |
---|---|
CN105488011A (en) | 2016-04-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103905447B (en) | Service link selects control method and equipment | |
CN105264930B (en) | Sending node and its reporting cached state method | |
CN109218352B (en) | Consensus confirmation method and device for transaction information in blockchain network | |
CN112468372B (en) | Method and device for detecting equipment state in power line communication network | |
CN104486407B (en) | The generation method and device of numbering | |
CN105260128B (en) | A kind of method and storage device for writing data into storage device | |
CN109743142A (en) | Information communication method and device | |
CN106101264B (en) | Content distributing network log method for pushing, device and system | |
CN104935641B (en) | A kind of method and apparatus for accelerating file transmission | |
CN106936662A (en) | A kind of method for realizing heartbeat mechanism, apparatus and system | |
CN104935612A (en) | Data processing method and server | |
CN111092925B (en) | Block chain capacity expansion processing method, device and equipment | |
CN105488011B (en) | The memory access processing method and network-on-chip of network-on-chip | |
CN105099753B (en) | The method of Network Management System and its processing business | |
CN105681230A (en) | Data processing method and equipment for use in service chain | |
CN104363269B (en) | It is a kind of to pass through FC link transmissions, the method and device of reception NAS data | |
CN103874230A (en) | Processing method and device for data transmission | |
CN104754760B (en) | A kind of Packet Service method for reconstructing and terminal | |
CN104123261B (en) | A kind of electronic equipment and information transferring method | |
CN107682423A (en) | A kind of data forwarding method and device | |
CN105519055A (en) | Dynamic equilibrium method and apparatus for QoS of I/O channel | |
CN107579804A (en) | A kind of method and system for filtering duplicate data | |
CN104679687B (en) | A kind of method and device for identifying interrupt source | |
CN103975572B (en) | Radio communication device, wireless communications method and radio communication control program | |
CN104679688B (en) | Data access method, apparatus and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180619 Termination date: 20190919 |
|
CF01 | Termination of patent right due to non-payment of annual fee |