Data retransmission method, electronic device and storage medium

Abstract

The present invention relates to the field of communication technology, and in particular to a data retransmission method, an electronic device and a storage medium, which are applied to a forward codec module in a physical layer of a first node, by obtaining an i-th codeword CW sent by a second node. i ; for CW i Perform verification, and if verification fails, extract CW i CWID i ; Set CWID i Add the next codeword CW that the first node is going to send to the second node j The second subcode cwB j Pad Bj , so that the pad Bj Carry CWID i and the transmission type for requesting the second node to retransmit; j Send it to the second node, so that the second node can use CWID i Find the target codeword and resend it according to the transmission type. Since the forward codec module needs to go through less logic for retransmission, the retransmission time can be greatly reduced.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a data retransmission method, electronic equipment and storage medium. Background Art

[0002] In the process of data transmission from the source to the destination, packet loss may occur due to bit errors. At present, the network congestion management mechanism is generally used to address the problem of packet loss, that is, combining PFC (Priority Flow Control) and ECN (Explicit Congestion Notification). Among them, PFC suspends low-priority traffic when the network is congested to ensure the transmission of high-priority traffic. ECN adds a mark to the data packet when the network is congested to notify the source and destination of the network congestion. PFC is mainly used to ensure the priority of data packet transmission, while ECN is used to control network congestion. These two mechanisms can be used in combination to minimize network congestion and data packet loss and improve network performance and reliability.

[0003] When the scale of the network is small, such as a network structure composed of dozens to hundreds of processors, the amount of data that needs to be retransmitted is small, and according to the above network congestion management mechanism, the data delay caused by retransmission is not prominent. When the scale of the network is large, such as a network structure composed of thousands of processors, the amount of data that needs to be retransmitted is large, and retransmission using the above network congestion management mechanism will cause a large retransmission delay. Summary of the invention

[0004] In view of the above technical problems, the technical solution adopted by the present invention is: a data retransmission method, applied to a forward encoding and decoding module in the physical layer of a first node, the method comprising the following steps:

[0005] S100, obtaining the i-th codeword CW sent by the second node i .

[0006] S200, for CW i Perform verification, and if verification fails, extract CW i CWID i .

[0007] S300, CWID i Add the next codeword CW that the first node is going to send to the second node j Among them, CW j Includes the first subcode cwA of the binding j and the second subcode cwB j , the cwA j Pad AjCarry CW j CWID j and The CW j The transmission type, the cwB j Pad Bj Carry CWID i and The CW i Transmission type; CW i The transmission type is a verification failure waiting for retransmission type requesting the second node to retransmit.

[0008] S400, CW j Send it to the second node, so that the second node can use CWID i Find the target codeword and use CW i The target codeword is retransmitted according to the transmission type.

[0009] In addition, the present invention also provides a non-transitory computer-readable storage medium, in which at least one instruction or at least one program is stored, and the at least one instruction or the at least one program is loaded and executed by a processor to implement the above method.

[0010] In addition, the present invention also provides an electronic device, including a processor and the above-mentioned non-transitory computer-readable storage medium.

[0011] The present invention has at least the following beneficial effects:

[0012] The present invention provides a data retransmission method, an electronic device and a storage medium, which are applied to a forward codec module in a physical layer of a first node, and transmit the identity information of the codeword that needs to be retransmitted by the opposite end by means of the padding bits in the second subcode cwB in the codeword CW transmitted by the forward codec module, so that the second node retransmits according to the identity information of the padding bits of cwB parsed, so as to achieve the purpose of retransmission through the forward codec module. The physical layer retransmission requires less logic, so the retransmission delay can be reduced to 125ns, greatly reducing the retransmission time, and the data transmitted within 125ns is less, compared with the retransmission delay of 100 microseconds in the prior art of the network congestion management mechanism, the transmission delay is greatly shortened, so the cache occupied by the retransmitted codeword is also smaller. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0014] Figure 1A flow chart of a data retransmission method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0016] In order to solve the technical problem of large retransmission delay, the present invention provides a data retransmission method, which is applied to the forward codec module in the physical layer of the first node. The codeword CW transmitted by the forward codec module includes two bound first subcodes cwA and second subcodes cwB, and the padding bits in cwA are used to transmit the identity information and transmission type of the current codeword; when retransmission is required, the identity information of the codeword that needs to be retransmitted by the other end is transmitted with the help of the padding bits in cwB. During the transmission process, the first node receives the codeword sent by the second node. When the verification fails, the identity information of the codeword that failed the verification is extracted, and the identity information is added to the padding bits of cwB in the next codeword to be sent; when the second node parses the identity information of the padding bits of cwB, the cached codeword with the same identity information is retransmitted to achieve the purpose of retransmission through the physical layer.

[0017] Optionally, the node is a processor or a switch. The processor may be a CPU, a GPU, or a GPGPU. The first node is a GPU, and the second node is a switch; or the first node and the second node are both GPUs; or the first node is a switch, and the second node is a GPU, etc. All other nodes capable of realizing data interaction in the prior art fall within the protection scope of the present invention.

[0018] Optionally, the first node and the second node communicate via the Ethernet protocol, and other local area network protocols in the prior art also fall within the protection scope of the present invention. The seven-layer framework of the OSI reference model is encapsulated into a four-layer framework architecture by the Ethernet protocol, and the four-layer network architecture of the Ethernet protocol includes: application layer, transport layer, network layer and network interface layer. The application layer in the Ethernet protocol encapsulates the application layer, presentation layer and session layer in the OSI reference model. The network interface layer in the Ethernet protocol encapsulates the data link layer and physical layer in the OSI reference model. The physical layer also includes a physical coding sublayer (Physical Coding Sublayer, PCS) and a forward error correction module (Forward Error Correction, FEC).

[0019] Optionally, the Ethernet protocol is the TCP / IP protocol, and other Ethernet protocols in the prior art fall within the protection scope of the present invention.

[0020] Optionally, the data retransmission method provided by the present invention is applied to a forward error correction (FEC) module in the physical layer. In the prior art, the data retransmission method is applied to other sublayers in the physical layer and falls within the protection scope of the present invention.

[0021] Optionally, the encoding of the forward encoding and decoding module adopts RS encoding. All other forward encoding and decoding technologies in the prior art fall within the protection scope of the present invention.

[0022] Optionally, the RS encoding is RS (272, 257) encoding, wherein 272 refers to the total number of data blocks in the codeword, and 257 refers to the total number of data blocks occupied by valid data. All other RS ​​encoding techniques in the prior art fall within the protection scope of the present invention.

[0023] See also Figure 1 , which shows a flow chart of a data retransmission method, the method is applied to a forward codec module in the physical layer of a first node, and the data retransmission method comprises the following steps:

[0024] S100, obtaining the i-th codeword CW sent by the second node i . Where i is greater than 0.

[0025] Among them, CW i Also includes: the first subcode cwA bound i and the second subcode cwB i ,cwA i Including cwA i Valid data bits data Ai , pad Ai and check digit P Ai ,pad Ai Including CW i CWID i and CW i Transmission type; cwB i Including cwB i Valid data bits data Bi , pad Bi and check digit P Bi Among them, CW iThe transmission type is any one of the normal transmission type, the verification failure and retransmission type, and the retransmission type. Among them, the encoding format of the first subcode and the second subcode is the same, and the encoding format of the codeword includes a total of 272 data blocks, each data block is 10 bits. Among them, the valid data bits of the codeword include M0-M256, a total of 257 codeword data blocks, and the codeword is the original data to be encrypted, also known as the valid data bit. The padding bit is a data block to ensure that the valid data bit and the data flag bit are even after verification, which meets the verification rules of RS encoding. The check bits are P0-P13, a total of 14 data blocks.

[0026] Among them, after encoding, the first subcode cwA of the binding is generated after forward error correction lane interleave and distribution (FEC laneinterleave&distribution) i and the second subcode cwB i The CW i Correspondingly, the first node needs to perform the de-interleaving step to obtain CW i .

[0027] Wherein, before S100, the following further includes: when the second node sends a CW to the first node i When CW i Add to the retransmission cache for backup. The retransmission cache is used to back up the codewords that the second node has sent out. If the codeword CW is i Data in Bi When an uncorrectable error occurs, you need to pass CWID i Look for the backup codeword in the cache for retransmission.

[0028] The codeword sent by the second node and the codeword sent by the first node have the same encoding format. The codewords sent by the second node and the first node both use the same encoding method.

[0029] The second node refers to the next node that communicates data with the first node. The communication between the first node and the second node is point-to-point, and the two cannot communicate across intermediate nodes. For example, the first GPU accesses the second GPU through a switch. At this time, it is divided into two sections of communication. The first section of communication is the communication between the first GPU and the switch, and the second section of communication is the communication between the switch and the second switch. In the first section of communication, the first GPU is the first node, and the switch is the second node of the first GPU. In the second section of communication, the switch is the first node, and the second GPU is the second node of the switch.

[0030] It should be noted that the execution subject of S100-S400 is the forward codec module in the physical layer of the first node.

[0031] S200, for CW i Perform verification, and if verification fails, extract CW i CWID i .

[0032] It should be noted that codewords may have errors during transmission, which may cause verification failure. Causes of errors include voltage signals that decay during signal transmission, noise, transmission equipment failure, or other factors. When verification fails, the codeword that caused the error needs to be retransmitted.

[0033] All methods for checking codewords fall within the protection scope of the present invention. On this basis, the present invention also provides a new checking step, which includes:

[0034] S210, for data Ai and data Bi Check and correct errors respectively. Ai and / or data Bi If the verification fails and cannot be corrected, the verification fails; otherwise, the verification passes. Ai and / or data Bi A checksum error that cannot be corrected refers to data Ai The checksum is wrong and cannot be corrected, or data Bi The checksum is wrong and cannot be corrected, or data Ai and data Bi All verifications are wrong and cannot be corrected. It should be noted that when a verification error occurs and the codeword that generates the error can be corrected, no retransmission is required. When the number of data blocks with verification errors exceeds the range that can be corrected, it is determined that the error is uncorrectable, and the verification fails at this time. The present invention provides two verification methods, one is verification without error correction function, and the verification fails when a verification error occurs. The other is a verification with a correction function, which corrects the verification error when it occurs. When the erroneous codeword can be corrected, no retransmission is required; when the verification error occurs and cannot be corrected, the verification fails and retransmission is required. All verification methods used for verification or with a correction function in the prior art fall within the protection scope of the present invention.

[0035] Optionally, the check algorithm is parity check, checksum, cyclic redundancy check (CRC), longitudinal redundancy check (LRC) or block check character (BCC). All other check algorithms for generating check codes fall within the protection scope of the present invention.

[0036] S300, CWID i Add the next codeword CW sent by the first node to the second node j Among them, CW j Includes the first subcode cwA of the binding j and the second subcode cwB j , the cwA j Pad Aj Carry CW j CWID j and The CW j The transmission type, the cwB j Pad Bj Carry CWID i and The CW i Transmission type; CW i The transmission type of CW is a verification failure waiting for retransmission type requesting the second node to retransmit. j It is the j-th codeword sent by the first node, where j is greater than 0.

[0037] It should be noted that the next codeword CW sent by the first node to the second node j Not specifically used to send the identity information CWID of the codeword to be retransmitted i The codeword is a codeword that the first node needs to send to the second node. The transmission type of the codeword can be any one of the normal transmission type, the check error waiting for retransmission type, and the retransmission type. j Pad Aj The type of transmission that carries this codeword. j Pad Bj The CWID that needs to be retransmitted i Passed to the second node.

[0038] S400, CW j Send it to the second node, so that the second node can use CWID i Find the target codeword and use CW i The target codeword is retransmitted according to the transmission type.

[0039] All according to CWID i Methods for searching for a codeword to be retransmitted and retransmitting it all fall within the protection scope of the present invention. The embodiment of the present invention further provides a retransmission step of a second node, including S420-S440.

[0040] S420: The second node receives the received CWID i , query the cache to get the target codeword;

[0041] S440, taking out the target codeword for retransmission;

[0042] S460, sequentially retrieve all codewords after the target codeword in the cache, wherein the kth codeword CW retransmitted after the target codeword i+k Includes the first subcode cwA of the binding i+k and the second subcode cwB i+k ,cwA i+k Pad i+k Includes CWID i+k and The CW i+k Transmission type, among which CW i+k The transmission type is the retransmission type, where the value of k ranges from 1 to K, and K is the number of times the second node sends a CW to the first node. i The number of codewords that have been sent to the first node before retransmitting the target codeword.

[0043] It should be noted that the cache stores the codewords that have been sent within a preset time or a preset number. When a code error occurs, the target codeword can be found through the identity information of the codeword, and the target codeword in the cache and all the codewords that have been sent after the target codeword are retransmitted. When retransmitting, the transmission type of the retransmitted codeword needs to be set to the retransmission type.

[0044] As a preferred embodiment, S400 further includes:

[0045] S410, the second node receives the CWID i Afterwards, a flag signal for suspending transmission is sent to the system cache to control the codewords in the system cache to suspend transmission.

[0046] S450, when retry in retransmission buffer i When the codeword retransmission is completed, the second node continues to send the codeword in the system cache.

[0047] In summary, the present invention provides a data retransmission method, which is applied to the forward codec module in the physical layer of the first node, and the codeword CW transmitted by the forward codec module includes two bound first subcodes cwA and second subcodes cwB. When retransmission is required, the identity information of the codeword that needs to be retransmitted by the other end is transmitted with the help of the padding bit in cwB. When the second node parses the identity information of the padding bit of cwB, the codeword with the same identity information stored in the retransmission cache is retransmitted to achieve the purpose of retransmission through the forward codec module. Since the forward codec module needs to go through less logic for retransmission, the retransmission time can be greatly reduced, and the retransmission delay can be reduced to 125ns. Since the data transmitted within 125ns is less, the cache occupied by the retransmitted codeword is also smaller. And the retransmission method provided by the present invention does not need to send other additional codewords or data packets, and only needs to cleverly use the data blocks in the normally transmitted codewords to transmit the identity information of the codewords that need to be retransmitted back, which can reduce communication pressure and congestion.

[0048] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to implementing a method in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiment.

[0049] An embodiment of the present invention further provides an electronic device, comprising a processor and the aforementioned non-transitory computer-readable storage medium.

[0050] An embodiment of the present invention further provides a computer program product, which includes program code. When the program product is run on an electronic device, the program code is used to enable the electronic device to execute the steps of the method according to various exemplary embodiments of the present invention described above in this specification.

[0051] Although some specific embodiments of the present invention have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A data retransmission method, characterized in that: A forward codec module applied to a physical layer of a first node, the method comprising the following steps: S100, obtaining the i-th codeword CW sent by the second node i ; S200, for CW i Perform verification, and if verification fails, extract CW i CWID i ; S300, CWID i Add the next codeword CW sent by the first node to the second node j Among them, CW j Includes the first subcode cwA of the binding j and the second subcode cwB j , the cwA j Pad Aj Carry CW j CWID j and The CW j The transmission type, the cwB j Pad Bj Carry CWID i and The CW i The type of transmission; CW i The transmission type is a verification failure waiting for retransmission type requesting the second node to retransmit; S400, CW j Send it to the second node, so that the second node can use CWID i Find the target codeword and use CW i The target codeword is retransmitted according to the transmission type; Wherein, S400 also includes a retransmission step of the second node: S410, the second node receives the CWID i Afterwards, a flag signal for suspending transmission is sent to the system cache to control the codewords in the system cache to suspend transmission; S420: The second node receives the received CWID i , query the cache to get the target codeword; S440, taking out the target codeword for retransmission; S450, when retry in retransmission buffer i When the codeword retransmission is completed, the second node continues to send the codeword in the system cache; S460, sequentially retrieve all codewords after the target codeword in the cache, wherein the kth codeword CW retransmitted after the target codeword i+k Includes the first subcode cwA of the binding i+k and the second subcode cwB i+k ,cwA i+k Pad i+k Includes CWID i+k and The CW i+k Transmission type, among which CW i+k The transmission type is the retransmission type, where the value of k ranges from 1 to K, and K is the number of times the second node sends a CW to the first node. i The number of codewords that have been sent to the first node before retransmitting the target codeword.

2. The method according to claim 1, characterized in that S100 CW i Includes: The first subcode of the binding cwA i and the second subcode cwB i ,cwA i Including cwA i Valid data bits data Ai , pad Ai and check digit P Ai ,pad Ai Including CW i CWID i and CW i Transmission type; cwB i Including cwB i Valid data bits data Bi , pad Bi and check digit P Bi .

3. The method according to claim 2, characterized in that The CW i The transmission type is any one of the normal transmission type, the check failure to be retransmitted type and the retransmission type.

4. The method according to claim 2, characterized in that: S200 also includes a verification step: S210, for data Ai and data Bi Check and correct errors respectively. Ai and / or data Bi When the calibration error cannot be corrected, CW i Verification fails; otherwise verification passes.

5. The method according to claim 1, characterized in that Before S100, the process also includes: when the second node sends a CW to the first node i When CW i Add backup to retransmission cache.

6. The method according to claim 1, characterized in that The forward encoding and decoding module adopts RS encoding.

7. A non-transitory computer-readable storage medium, wherein at least one instruction or at least one program is stored in the storage medium, characterized in that: The at least one instruction or the at least one program is loaded and executed by the processor to implement the method according to any one of claims 1 to 6.

8. An electronic device, characterized in that: The invention comprises a processor and the non-transitory computer-readable storage medium as claimed in claim 7.

Images