Flow control methods and chips

By generating and sending flow control information based on the cache queue status, the problem of data overflow in multi-chip communication is solved, fine-grained flow control is achieved, and communication efficiency is improved.

CN122069228BActive Publication Date: 2026-11-10BEIJING XINGYUN INTEGRATED CIRCUIT CO LTD +1
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202610209648.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-02-12
Publication Date
2026-11-10
Estimated Expiration
2046-02-12

AI Technical Summary

Technical Problem

In multi-chip communication scenarios, existing technologies only send traffic status to the current data sender, resulting in other potential data senders not receiving the information, which may lead to data overflow.

Method used

Flow control information is generated by the cache queue status of the first chip and sent to multiple second chips that can communicate with it, ensuring that each chip performs flow control according to the congestion level, including generating flow control information on congestion level and capacity change trends, thus achieving fine-grained flow control.

Benefits of technology

This reduces data overflow in the first chip, ensures that the message sending rate and processing rate of multiple chips match, and improves message transmission efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122069228B_ABST
    Figure CN122069228B_ABST
Patent Text Reader

Abstract

This application provides a flow control method and chip, relating to the field of chip communication technology. The method includes: selectively generating first flow control information based on the buffer queue state of a first chip, the first flow control information including at least a first congestion level corresponding to the buffer queue state; and sending the first flow control information to multiple second chips capable of communicating with the first chip, so that each second chip performs flow control on messages sent to the first chip according to the first congestion level. This application can prevent data overflow.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of chip communication technology, and more specifically, to a flow control method and a chip. Background Technology

[0002] In multi-chip communication scenarios, each chip typically needs to connect with multiple other chips to achieve various information exchanges, including command transmission and data exchange. However, during data transmission, if the amount of data sent by the transmitting chip to the receiving chip exceeds its actual processing capacity, data will accumulate at the receiving end, causing communication congestion. To avoid the resulting delays or data loss, an effective flow control mechanism must be implemented to coordinate the transmission and reception rates, ensuring the reliability and efficiency of communication.

[0003] However, existing technology only sends the traffic status to the current data sender, which means that other potential data senders do not receive this information. Other chips may continue to send information to the receiving chip, resulting in data overflow. Summary of the Invention

[0004] The purpose of this application is to provide a flow control method and chip to address the shortcomings of the prior art, thereby reducing the occurrence of data overflow.

[0005] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a flow control method applied to a first chip, the method comprising: Based on the cache queue state of the first chip, first flow control information is selectively generated, and the first flow control information includes at least: the first congestion level corresponding to the cache queue state; The first flow control information is sent to a plurality of second chips that can communicate with the first chip, so that each second chip performs flow control on messages sent to the first chip according to the first congestion level.

[0006] Optionally, the step of selectively generating the first flow control information based on the cache queue state of the first chip includes: Based on the used capacity of the cache queue of the first chip and the first capacity change trend, the first flow control information is generated, which includes the first congestion level and the first capacity change trend.

[0007] Optionally, the method further includes: Receive the second flow control information sent by the third chip; Based on the second flow control information, determine the flow-limiting channel and back pressure rules; According to the current limiting channel and the back pressure rule, the message sent from the first chip to the third chip is subject to back pressure control.

[0008] Optionally, determining the flow-limiting channel and backpressure rule based on the second flow control information includes: The second flow control information is analyzed to determine the flow-limiting channel, the second congestion level, and the second capacity change trend; The backpressure rule is determined based on the second congestion level and the second capacity change trend, and the backpressure rule includes the backpressure cycle percentage.

[0009] Optionally, the congestion level includes at least three levels. Under the same congestion level, the proportion of backpressure cycles with an upward trend in capacity change is greater than the proportion of backpressure cycles with a downward trend.

[0010] Optionally, the method further includes: The system determines whether to forward the second flow control information to the fourth chip according to the preset forwarding rules. The fourth chip and the first chip are located in different computing cards, and the fourth chip communicates with the third chip through the first chip. When the forwarding rule is satisfied, remote flow control information is generated based on the second flow control information; The remote flow control information is forwarded to the fourth chip so that the fourth chip can perform backpressure control on the messages sent to the third chip according to the remote flow control information.

[0011] Optionally, the method further includes: If the second flow control information is remote flow control information and the forwarding enable state of the first chip is enabled, the remote flow control information is broadcast to the fifth chip located on the same computing card as the first chip, so that the fifth chip performs backpressure control on the messages sent to the third chip according to the remote flow control information.

[0012] Optionally, the method further includes: If the third flow control information sent by the third chip is received within the preset flow control period, the backpressure rule is updated according to the third flow control information after the preset flow control period ends. Based on the current limiting channel and the updated backpressure rules, backpressure control is applied to the messages sent from the first chip to the third chip.

[0013] Optionally, the method further includes: Receive write requests sent by the host processor; If the target chip of the write request is the first chip, the credit value of the first virtual channel between the host processor and the first chip is reduced so that the host processor can perform backpressure control on the write request sent to the first chip. If the target chip of the write request is another chip, the write request is forwarded to the other chip through the second virtual channel between the host processor and the first chip.

[0014] Secondly, embodiments of this application also provide a chip, the chip including a flow control module, the flow control module being used to perform the flow control method as described in any of the first aspects.

[0015] The beneficial effects of this application include at least the following: The flow control method and chip provided in this application generate first flow control information based on the buffer queue state of the first chip, and send the first flow control information to multiple second chips that can communicate with the first chip. This ensures that multiple second chips can perform backpressure control on messages to be sent in advance, reducing the occurrence of data overflow in the first chip. Based on the buffer queue state, a first congestion level is determined to generate flow control information that matches the buffer queue state, thereby realizing fine-grained flow control of messages sent to the first chip by multiple second chips. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic flowchart of a flow control method provided in an embodiment of this application; Figure 2 A schematic block diagram of a flow control information sending unit provided in an embodiment of this application; Figure 3 A system architecture diagram provided for one embodiment of this application; Figure 4 A schematic flowchart illustrating a flow control method provided in another embodiment of this application; Figure 5 A schematic block diagram of a flow control information receiving unit provided in an embodiment of this application; Figure 6 A schematic flowchart illustrating a flow control method provided in another embodiment of this application; Figure 7A schematic flowchart illustrating a flow control method provided in another embodiment of this application; Figure 8 A schematic diagram of the flow control method provided in another embodiment of this application; Figure 9 This is an architecture diagram of a computing system provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of this application, but not all embodiments.

[0019] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] Furthermore, the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Additionally, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] The modules and units mentioned in this application may include one or more hardware circuits, including but not limited to: application-specific integrated circuits, digital signal processors, field-programmable gate arrays, discrete logic circuits, state machines, and any combination of the foregoing circuits; these circuits are specifically designed, configured and interconnected to perform one or more specific functions disclosed in this application.

[0022] It should be noted that, where there is no conflict, the features in the embodiments of this application can be combined with each other.

[0023] Figure 1 A flowchart illustrating a flow control method provided in an embodiment of this application is shown below. Figure 1 As shown, the method may include: S101. Based on the cache queue status of the first chip, selectively generate first flow control information. The first flow control information includes at least the first congestion level corresponding to the cache queue status.

[0024] For example, a flow control information sending unit can be independently set on the first chip, and the flow control information sending unit is used to perform steps S101-S102. Figure 2 This is a schematic block diagram of a flow control information sending unit provided in an embodiment of this application, as shown below. Figure 2 As shown, the flow control information sending unit may include: a first buffer unit 11, a capacity monitoring unit 12, and a flow control information generation unit 13. The capacity monitoring unit 12 is communicatively connected to the first buffer unit 11, and the flow control information generation unit 13 is communicatively connected to the capacity monitoring unit 12. Alternatively, a flow control information sending unit can be set on the first chip, or it can be set on each interconnect channel of the first chip.

[0025] In this embodiment, a first cache unit 11 is provided in the first chip. The first cache unit 11 is used to cache all messages sent to the first chip. The messages sent to the first chip can be requests and / or data. Requests can be read requests, write requests, control requests, etc. The messages sent to the first chip are first cached in the first cache unit 11 so that the processing unit in the first chip can process the messages in the first cache unit 11 one by one. For example, the messages that need to be flow controlled are write requests and related data. Write requests are for writing data to the target chip and carry corresponding data, which can easily cause congestion in the cache queue, requiring fine-grained flow control.

[0026] The capacity monitoring unit 12 monitors the cache queue status in the first cache unit 11 of the first chip. The first cache unit 11 has multiple preset capacity thresholds, each corresponding to a different congestion level. A smaller capacity threshold corresponds to a lower congestion level, and a larger capacity threshold corresponds to a higher congestion level. The unit monitors the used capacity of the first cache unit 11 in real time or periodically, compares the used capacity of the first cache unit 11 with the multiple capacity thresholds, determines the capacity threshold corresponding to the used capacity of the first cache unit 11, and obtains the first congestion level.

[0027] The flow control information generation unit 13 receives the capacity threshold corresponding to the used capacity / the target area corresponding to the used capacity from the capacity monitoring unit 12. If the cache queue status indicates that the first cache unit 11 has not experienced message congestion, the first chip will not generate the first flow control information. If the cache queue status indicates that the first cache unit 11 has experienced message congestion, the first congestion level corresponding to the target area is determined, and the first flow control information is generated according to the congested interconnect channel (such as PCIe channel) or the target address range of the congested party and the first congestion level.

[0028] In some embodiments, the first cache unit 11 is a high-speed buffer, typically a fast first-in-first-out (FIFO) storage unit. The used capacity of the first cache unit 11 is an indicator of whether the downstream processing capability matches the upstream transmission rate. When the used capacity is close to full, it means that the downstream processing rate is difficult to meet the upstream transmission rate. When the used capacity is low, it means that the downstream processing rate can meet the upstream transmission rate. The system efficiency is highest when the two are matched.

[0029] Furthermore, the first cache unit 11 can be divided into multiple regions. The capacity monitoring unit 12 is pre-configured with multiple regions and multiple congestion levels. The capacity monitoring unit 12 monitors the target region corresponding to the used capacity of the first cache unit 11. When a region change event is detected, flow control information is generated.

[0030] Furthermore, the first flow control information may also include: the channel identifier of the interconnect channel where the congested buffer queue in the first chip is located, or the target address range where the first chip is congested. The channel identifier / target address range is used to instruct the second chip to perform backpressure control on messages sent to the interconnect channel / target address range.

[0031] S102. Send first flow control information to a plurality of second chips that can communicate with the first chip, so that each second chip performs flow control on messages sent to the first chip according to a first congestion level.

[0032] In this embodiment, to prevent data overflow in the first cache unit 11 of the first chip, when message congestion occurs in the first cache unit 11 of the first chip, the flow control information generation unit 13 can notify multiple, or even all, multiple second chips that can communicate with the first chip to perform flow control. That is, the second chips can include both the current message sender and potential message senders who have not yet sent a message.

[0033] Specifically, multiple second chips capable of communicating with the first chip are identified. These second chips can include: chips directly connected to the first chip on the same computing card; chips directly connected to the first chip on different computing cards; chips on different computing cards but capable of cross-card communication via a relay chip; and chips on the same computing card but not directly connected to the first chip, but capable of communication via a relay chip. For example, the multiple second chips can be the N most recent chips that have communicated with the first chip, where N ≥ 2 and is less than the total number of second chips. In large interconnected systems, this can reduce the amount of flow control information transmitted and data processed, and allow for flow control by comparing potential sources of congestion.

[0034] For example, Figure 3 A system architecture diagram provided for one embodiment of this application, such as Figure 3 As shown, chips 1, 2, and 3 can be mounted on the same computing card and communicate with each other. Chips 4 and 5 can also be mounted on the same computing card and communicate with each other. Chips 1 and 4 communicate directly, while chip 5 communicates with chip 1 across cards through chip 4. The first chip can be chip 1, and the second chip can be any one of chips 2, 3, 4, and 5.

[0035] If the second chip is located on the same computing card as the first chip, the flow control information generation unit 13 directly sends the first flow control information as a local flow control message (FCMsg) to the second chip. If the second chip is located on a different computing card than the first chip, the flow control information generation unit 13 generates a remote flow control message (RFCMsg) based on the first flow control information and sends the remote flow control information to the second chip.

[0036] For the second chip that receives the first flow control information, it determines the first congestion level by parsing the first flow control information, and determines the flow control duration or backpressure cycle proportion based on the first congestion level. Based on the flow control duration or backpressure cycle proportion, it performs flow control on the messages sent to the first chip, that is, it stops sending messages to the first chip within the flow control duration, or stops sending messages to the first chip within the corresponding backpressure cycle proportion within the preset data transmission cycle X.

[0037] In some embodiments, the chips are interconnected via PCIe, and a flow control information transmitting unit and / or a flow control information receiving unit can be provided on the first chip. Alternatively, a flow control information transmitting unit and / or a flow control information receiving unit can be provided on each interconnect channel of the first chip.

[0038] For example, a PCIe channel of the first chip, such as PCIe0, is connected to a PCIe channel of the second chip, such as PCIe1. The flow control information sending unit of the PCIe0 of the first chip generates first flow control information and sends the first flow control information to the flow control information sending unit of the PCIe1 of the second chip. The flow control information sending unit of the PCIe1 of the second chip can broadcast the first flow control information to all flow control information receiving units in the second chip.

[0039] The flow control method provided in the above embodiments generates first flow control information based on the cache queue state of the first chip, and sends the first flow control information to multiple second chips that can communicate with the first chip. This ensures that multiple second chips can perform backpressure control on messages to be sent in advance, reducing the occurrence of data overflow in the first chip. Based on the cache queue state, a first congestion level is determined to generate flow control information that matches the cache queue state, thereby enabling fine-grained flow control of messages sent to the first chip by multiple second chips.

[0040] In one possible implementation, the process of selectively generating the first flow control information based on the cache queue state of the first chip in step S101 may include: Based on the used capacity of the cache queue of the first chip and the first capacity change trend, first flow control information is generated. The first flow control information includes: first congestion level and first capacity change trend.

[0041] In this embodiment, in addition to monitoring the used capacity of the first cache unit 21, the capacity monitoring unit 12 can also monitor the first capacity change trend of the first cache unit 21. Specifically, the first capacity change trend can be determined by judging the used capacity of two adjacent monitoring times, or it can be determined by the level before and after the congestion level change. For example, if the level before the change is 5 and the level after the change is 4, then the first capacity change trend is decreasing.

[0042] The flow control information generation unit 13 generates first flow control information based on the target address range of the congested interconnection channel or the congested party, the first congestion level, and the first capacity change trend.

[0043] After receiving the first flow control information, multiple second chips determine the flow control duration or backpressure cycle based on the first congestion level and the first capacity change trend. Based on the flow control duration or backpressure cycle, they perform flow control on the messages sent to the first chip, that is, they stop sending messages to the first chip within the flow control duration, or stop sending messages to the first chip for the corresponding number of backpressure cycles within a preset data transmission cycle.

[0044] The flow control method provided in the above embodiments generates first flow control information based on the first congestion level and the first capacity change trend of the cache queue. This allows multiple second chips to determine the backpressure period according to the congestion level and capacity change trend. This enables the messages sent to the first chip to be matched not only with the congestion level of the first chip's cache queue but also with the capacity change trend of the first chip's cache queue, achieving fine-grained flow control. On the one hand, this reduces the occurrence of data overflow in the first chip, and on the other hand, it ensures that the message sending rate of the second chip matches the message processing rate of the first chip, thereby improving message transmission efficiency.

[0045] In one possible implementation, Figure 4 A flowchart illustrating a flow control method provided in another embodiment of this application is shown below. Figure 4 As shown, the method may further include: S201, Receive the second flow control information sent by the third chip.

[0046] For example, a flow control information receiving unit can be independently set on the first chip, and the flow control information receiving unit is used to perform steps S201-S203. Figure 5 This is a schematic block diagram of a flow control information receiving unit provided in an embodiment of this application, as shown below. Figure 5 As shown, the flow control information receiving unit may include: an information receiving buffer unit 21, a back pressure information judgment unit 22, and a flow back pressure unit 23. The back pressure information judgment unit 22 is communicatively connected to the information receiving buffer unit 21, and the flow back pressure unit 23 is communicatively connected to the back pressure information judgment unit 22.

[0047] During the process of the first chip sending data to the third chip, if the third chip experiences data congestion due to bandwidth limitations or the first chip sending messages too quickly, the third chip sends a second flow control message to the first chip. Alternatively, when the third chip experiences congestion at the receiving end due to message transmission between the third chip and other chips, the third chip can also send a second flow control message to the first chip that is able to send messages to itself.

[0048] Specifically, if the third chip is located on the same computing card as the first chip, the second flow control information can be local flow control information; if the third chip is located on a different computing card than the first chip, the second flow control information can be remote flow control information. This allows for targeted expansion settings for different types of flow control information, achieving more efficient flow control.

[0049] To prevent multiple chips from sending a large amount of flow control information to the first chip, which could cause a message storm and result in message loss, the information receiving buffer unit 21 can be equipped with a buffer to cache multiple flow control messages and prevent message loss.

[0050] In some embodiments, two buffers can be set up for local flow control information and remote flow control information respectively. One buffer is used to buffer local flow control information, and the other buffer is used to cache remote flow control information, so as to facilitate the setting of different data structures and / or backpressure rules for the two types of flow control information.

[0051] S202. Based on the second flow control information, determine the flow restriction channel and back pressure rules.

[0052] In this embodiment, the second flow control information includes at least: a source identifier, an information type, and a backpressure field. The source identifier is used to indicate the source location of the second flow control information, such as the chip local number / interconnect channel local number, and is used to indicate which chip / interconnect channel is sending the second flow control information. The information type is used to indicate whether the second flow control information is local flow control information or remote flow control information. The backpressure field is used to indicate the congestion level of the chip that sends the second flow control information.

[0053] In some embodiments, the second flow control information may include a header portion (MsgHeader) and a frame portion (MsgFrame), wherein only the remote flow control information has a header portion.

[0054] The frame portion may include a Bus, Device and Function number information (BDF) bit, a type bit, a source identifier bit, and a backpressure field.

[0055] The BDF bit is used by the flow control information receiving unit to identify and determine the computing card number information in the second flow control information, and the type bit is used to indicate whether the second flow control information is remote flow control information or local flow control information. In some embodiments, the flow control information is defaulted to local flow control information when it is first generated. When the flow control module enables forwarding and determines that the flow control information needs to be forwarded, it will generate remote flow control information based on the local flow control information. For example, remote flow control information is generated when forwarding is done across cards.

[0056] The source identifier is used to identify the chip number of the sender of the flow control information in the local computing card. For example, if a computing card contains two expansion I / O chips and one GPGPU, they can be numbered 0, 1, and 2 respectively. The source identifier can be represented by two bits.

[0057] The backpressure field can include the second congestion level.

[0058] The backpressure information judgment unit 22 can have multiple built-in logic processing channels. Each logic processing channel corresponds to a set of processing circuits in hardware. The backpressure information judgment unit 22 obtains the second flow control information from the information receiving buffer unit 21. The backpressure information judgment unit 22 can determine the processing channel for the second flow control information based on the source identifier of the second flow control information, combined with the pre-configured mapping relationship between the source identifier and the processing channel, or the order, etc. This facilitates efficient processing of multiple sets of flow control information and also facilitates specific processing of flow control information from different sources.

[0059] After determining the processing channel, the second flow control information is processed within that channel. If the second flow control information is local flow control information, the processing channel determines the flow limiting channel based on this local flow control information. For example, the interconnect channel / chip that requires backpressure flow control can be obtained by parsing the second flow control information. For instance, the PCIe channel PCIech connected between the chip sending the flow control information and the first chip can be used as the flow limiting channel requiring backpressure. The interconnect channel can be determined based on the interconnect topology between the sender of the flow control information and the first chip. When the interconnect channels of the two chips are unique, the interconnect channel corresponding to the flow limiting is equivalent to the chip corresponding to the flow limiting.

[0060] The processing channel can also determine the corresponding backpressure rule register based on the second flow control information and mapping relationship. The backpressure rule register pre-stores the corresponding flow limiting channel and backpressure rule.

[0061] If the second flow control information is remote flow control information, the processing channel determines the specific address range of the remote memory corresponding to the chip that needs backpressure flow control based on the remote flow control information. Through address mapping conversion, it converts the remote address into the destination equivalent address used when this card sends data to the remote card, thus using it as the flow-limiting address. This avoids flow limiting the entire channel and restricting message interaction between non-congested parties. Furthermore, the backpressure information judgment unit 22 also parses the second congestion level of the data receiving end from the second flow control information and determines the backpressure rules based on the second congestion level.

[0062] For example, refer to Figure 3The connection architecture shown has a corresponding interconnect channel on each chip, such as PCIech (not shown). The PCIech can be used for cross-card connection expansion. Each interconnect channel number corresponds to the chip it belongs to. There is also a noc (not shown) which can be used for interconnection of chips within the card. For example, chip 2 generates local flow control information due to congestion and sends it to chips 1 and 3. The local flow control information includes information type (local), interconnection channel number (corresponding to the fourth chip), congestion level, and forwarding card limit information (2, used to limit the number of forwarding cards; forwarding is allowed when the number is greater than or equal to 1). The corresponding chip that needs flow control can be obtained through the interconnection channel number. Chip 3 can determine the message to be sent by chip 2 that requires flow control based on the interconnection channel number. After receiving the local flow control information, chip 1 performs flow control on the message sent to chip 2 based on the local flow control information. On the other hand, when forwarding is enabled, forwarding generates first remote flow control information. The first remote flow control information includes card number (first computing card), flow-limiting address range (stored in the forwarding content register, the equivalent address range of chip 2 accessed from the perspective of the second computing card), forwarding card limit information (2-1=1, used to limit the number of forwarding cards; forwarding is allowed when the number is greater than or equal to 1), information type (remote), interconnection channel number (which changes and corresponds to chip 1), and congestion level. Congestion level (unchanged); After receiving the first remote flow control information, chip 4 generates and broadcasts the second remote flow control information to chip 5 within the second computing card. The second remote flow control information includes the card number (unchanged, first computing card), the rate-limiting address range (unchanged, the equivalent address range for accessing chip 2 from the perspective of the second computing card), the forwarding card limit information (1, used to limit the number of forwarding cards; forwarding is allowed when the number is greater than or equal to 1), the information type (remote), the interconnection channel number (changed, corresponding to chip 4), and the congestion level (unchanged); If chip 5 is connected to a third computing card, it generates the third remote flow control information based on the second remote flow control information and sends it to the third computing card. The third remote flow control information includes the card number (changed, second computing card), the rate-limiting address range (changed, the equivalent address range for chip 4 from the perspective of the third computing card), the forwarding card limit information (1-1=0, used to limit the number of forwarding cards; cross-card forwarding is not allowed), the information type (remote), the interconnection channel number (changed, corresponding to chip 5), and the congestion level (unchanged). For example, the chip number can also be used directly instead of the interconnection channel number.

[0063] When performing flow control, local flow control can directly determine the chip corresponding to the flow control of the flow limiting channel based on local flow control information; remote flow control can perform flow control on messages within a specific equivalent address range based on the interconnection channel number and equivalent address range of the remote flow control information, thereby achieving fine-grained flow control and flexible adaptation to the interconnection architecture.

[0064] In some embodiments, the backpressure rule includes a backpressure cycle percentage, which is the number of cycles in which the message to be sent to the destination is blocked within a preset number X flow control cycles. The higher the congestion level, the larger the backpressure cycle percentage.

[0065] S203. Based on the current limiting channel and back pressure rules, perform back pressure control on the messages sent from the first chip to the third chip.

[0066] In this embodiment, after determining the current limiting channel and the backpressure rule, the entire current limiting channel or the corresponding part of the current limiting address in the current limiting channel and the backpressure rule are sent as the current limiting (throtting) value to the flow backpressure unit 23. The flow backpressure unit 23 monitors the data output terminal of the first chip. If the destination of the message to be sent at the data output terminal is consistent with the current limiting channel or current limiting address of the third chip, a backpressure signal is generated according to the backpressure rule to backpressure the message to be sent.

[0067] For example, if X is 20 cycles and the backpressure cycle ratio is 5, then when the first chip sends messages to the third chip, it can be controlled to stop sending for 5 cycles within every 20 cycles, and send normally for the remaining 15 cycles. This will gradually reduce the receiving pressure on the third chip. By reducing the receiving pressure on the third chip and continuing message processing, congestion can be alleviated. The backpressure cycle ratio can be completed continuously within a fixed number of cycles or intermittently.

[0068] The flow control method provided in the above embodiments, based on the flow control information sent by other chips, implements backpressure control on the sent messages according to the flow control information, which can achieve fine-grained flow control for messages on a specified channel, rather than flow control for all sent messages, thereby improving the fine-grained flow control capability of the chip and making it more suitable for multi-chip, especially multi-card interconnection communication scenarios.

[0069] In one possible implementation, Figure 6 A flowchart illustrating a flow control method provided in another embodiment of this application is shown below. Figure 6 As shown, the process of determining the flow-limiting channel and back pressure rule based on the second flow control information in S202 above may include: S301. Analyze the second flow control information to determine the flow restriction channel, the second congestion level, and the trend of second capacity changes.

[0070] S302. Determine the backpressure rules based on the second congestion level and the second capacity change trend. The backpressure rules include the backpressure cycle percentage.

[0071] In this embodiment, in addition to the flow-limiting channel and the second congestion level, the second flow control information may also include the second capacity change trend of the third chip. The second capacity change trend is used to indicate whether the congestion of the third chip is on an upward or downward trend. When determining the proportion of back pressure cycles for the third chip, not only the second congestion level but also the second capacity change trend should be considered.

[0072] Multiple congestion levels are predefined, as well as the mapping relationship between the proportion of backpressure cycles corresponding to the capacity increase trend and the capacity decrease trend under each congestion level. After parsing the second congestion level and the second capacity change trend from the second flow control information, the backpressure cycle corresponding to the second congestion level and the second capacity change trend is determined according to the predefined mapping relationship.

[0073] In some embodiments, the congestion level includes at least three levels, and under the same congestion level, the proportion of backpressure cycles with an upward trend in capacity change is greater than the proportion of backpressure cycles with a downward trend.

[0074] In this embodiment, multiple congestion levels correspond to different backpressure cycle proportions, with higher congestion levels resulting in higher backpressure cycle proportions. For example, the congestion levels can include at least three levels: level 0, level 1, and level 2. If the congestion level is 0, the backpressure cycle proportion is 0 cycles, meaning all messages are sent normally without backpressure. If the congestion level is level 2, the backpressure cycle proportion is the full cycle, meaning all messages sent to the rate-limiting channel within a fixed period are stopped. If the congestion level is level 1, the backpressure cycle proportion can be a partial cycle, such as 5 or 7 cycles out of 20 cycles.

[0075] Under the same congestion level, the proportion of backpressure cycles under the corresponding congestion level is adjusted according to the capacity change trend to ensure that the proportion of backpressure cycles with an upward trend is greater than the proportion of backpressure cycles with a downward trend under the same congestion level.

[0076] In some embodiments, under the same congestion level, the proportion of upward trend backpressure cycles remains unchanged, while the proportion of downward trend backpressure cycles decreases. For example, if the congestion level is level 2 and the backpressure cycle proportion is 20 cycles of full backpressure, then the proportion of upward trend backpressure cycles is 20 cycles, and the proportion of downward trend backpressure cycles can be 18 cycles, 15 cycles, etc. It should be noted that when the backpressure cycle proportion is 0 cycles, the proportion of downward trend backpressure cycles remains unchanged.

[0077] In other embodiments, under the same congestion level, the proportion of upward trend backpressure cycles increases, while the proportion of downward trend backpressure cycles remains unchanged. For example, if the congestion level is level 0 and the proportion of backpressure cycles is 0 cycles, then the proportion of upward trend backpressure cycles can be 3 cycles, 5 cycles, etc., while the proportion of downward trend backpressure cycles is 0 cycles. It should be noted that when the proportion of backpressure cycles is the full cycle, such as 20 cycles, the proportion of upward trend backpressure cycles remains unchanged.

[0078] In other embodiments, under the same congestion level, the proportion of upward trend backpressure cycles increases and the proportion of downward trend backpressure cycles decreases. For example, if the congestion level is level 1 and the proportion of backpressure cycles is 6 cycles, then the proportion of upward trend backpressure cycles can be 7 cycles and the proportion of downward trend backpressure cycles can be 5 cycles.

[0079] The flow control method provided in the above embodiments determines the backpressure period based on the congestion level and capacity change trend, so that the message sent from the first chip to the third chip can not only match the congestion level of the third chip, but also match the capacity change trend of the third chip, thereby achieving more refined flow control within the same level. On the one hand, it reduces the occurrence of data overflow in the first chip, and on the other hand, it can improve the matching effect between the message sending rate of the first chip and the message processing rate of the third chip, thereby improving message transmission efficiency.

[0080] In one possible implementation, Figure 7 A flowchart illustrating a flow control method provided in another embodiment of this application is shown below. Figure 7 As shown, the method may further include: S401. Determine whether to forward the second flow control information to the fourth chip according to the preset forwarding rules. The fourth chip and the first chip are located in different computing cards, and the fourth chip communicates with the third chip through the first chip.

[0081] In this embodiment, the third chip and the first chip are located in the unified computing card and are directly connected. The fourth chip needs to communicate with the third chip through the first chip. That is, the first chip acts as a relay chip between the third chip and the fourth chip. The second flow control information sent by the third chip to the first chip is local flow control information. When the third chip is experiencing message congestion, in addition to sending local flow control information to the first chip located in the unified computing card with the third chip so that the first chip can perform backpressure control on the messages sent to the third chip, it is also necessary to determine whether it is necessary to notify the fourth chip to perform backpressure control on the messages sent to the third chip, so as to avoid further aggravating the message congestion of the third chip due to the fourth chip continuing to send messages to the third chip.

[0082] For example, such as Figure 3As shown, the third chip can be chip 2 or chip 3, and the fourth chip can be chip 4.

[0083] like Figure 5 As shown, the flow control information receiving unit may further include: a flow control information forwarding unit 24, which is communicatively connected to the information receiving buffer unit 21 and the flow control information sending unit, respectively. The flow control information forwarding unit 24 is pre-configured with flow control information forwarding rules. The flow control information forwarding rules may include: when the second flow control information is local flow control information and the forwarding enable state of the flow control module is enabled, cross-card forwarding is performed.

[0084] S402. When the forwarding rules are met, remote flow control information is generated based on the second flow control information.

[0085] In this embodiment, if the second flow control information is local flow control information and the forwarding enable state is enabled, then remote flow control information is generated based on the computing card number information, the forwarding enable state information, the target address information of the flow control information sending unit of the forwarding fourth chip, the remote equivalent memory address information, and the source identifier and backpressure field.

[0086] S403: Forward remote flow control information to the fourth chip so that the fourth chip can perform backpressure control on the messages sent to the third chip according to the remote flow control information.

[0087] In this embodiment, the flow control information forwarding unit 24 can notify the flow control information sending unit to send remote flow control information to the fourth chip according to the target address information of the flow control information sending unit of the forwarding fourth chip.

[0088] In some embodiments, the chip associated with the source identifier of the second flow control information is determined from the flow control information forwarding rules. If the chip associated with the source identifier of the second flow control information only includes the first chip, then the second flow control information does not need to be forwarded. If the chip associated with the source identifier of the second flow control information includes other chips besides the first chip, and the other chips cannot communicate directly with the flow control information sender corresponding to the source identifier, and the first chip is located on the data forwarding link between the other chips and the flow control information sender, then the second flow control information needs to be forwarded.

[0089] If the chip associated with the source identifier of the second flow control information includes chips other than the first chip, and the flow control module enables forwarding and determines that the preset forwarding rules are met, then remote flow control information is generated based on the second flow control information, and the remote flow control information is sent to the fourth chip through the flow control information sending unit.

[0090] The flow control method provided in the above embodiments sends the second flow control information to a fourth chip located on a different computing card than the first chip according to a preset forwarding rule. This ensures that multiple chips that communicate with or may communicate with the message-congested third chip can perform backpressure control on the messages to be sent in advance, reducing the occurrence of data overflow on the third chip. By setting forwarding rules, the forwarding function of flow control information can be adaptively adjusted and configured in software, flexibly adapting to different interconnect architectures.

[0091] In one possible implementation, the method may further include: If the second flow control information is remote flow control information and the forwarding enable state of the first chip is enabled, the remote flow control information is broadcast to the fifth chip located on the same computing card as the first chip, so that the fifth chip can perform backpressure control on the messages sent to the third chip according to the remote flow control information.

[0092] In this embodiment, if the third chip and the first chip are located on different computing cards, the second flow control information sent by the third chip to the first chip is remote flow control information. After receiving the remote flow control information, the first chip broadcasts the remote flow control information within the same computing card, so that all fifth chips within the same computing card perform backpressure control on the messages sent to the third chip.

[0093] In some embodiments, remote flow control information can be broadcast within the card when the forwarding enable state of the first chip is enabled.

[0094] The flow control method provided in the above embodiments broadcasts the received remote flow control information within the card, solving the problem of synchronously distributing flow control commands from remote cards within multi-chip computing cards. This allows for backpressure control of multiple potential data sources, improving the efficiency of data transmission between multi-chip computing cards and minimizing the occurrence of data overflow.

[0095] In one possible implementation, Figure 8 A flow diagram illustrating a flow control method provided in another embodiment of this application, such as... Figure 8 As shown, the method may further include: S501. If the third flow control information sent by the third chip is received within the preset flow control period, the back pressure rule is updated according to the third flow control information after the preset flow control period ends.

[0096] S502. Based on the current limiting channel and the updated backpressure rules, perform backpressure control on the messages sent from the first chip to the third chip.

[0097] In this embodiment, the flow control information is generated when the target area corresponding to the used capacity changes. If the target area corresponding to the used capacity changes frequently, it means that multiple flow control information is generated in sequence. The first chip overwrites the old second flow control information with the new third flow control information when the third chip sends multiple flow control information that has not yet been parsed. For the second flow control information that has been parsed and started to execute backpressure control, the backpressure rule in the next third flow control information can only be used to perform backpressure control on the message sent to the third chip after the flow control cycle of the preset number X has ended. If the target area corresponding to the used capacity has not changed, flow control is performed according to the flow control information stored in the register until normal operation is restored.

[0098] The flow control method provided in the above embodiments uses a new flow control information backpressure rule after a flow control cycle ends to perform backpressure control, avoiding the situation where the backpressure effect decreases or even fails when the same sender continuously switches backpressure rules when sending multiple flow control information, thus ensuring that the backpressure process is executed normally.

[0099] In one possible implementation, the method may further include: Receive a write request sent by the host processor; if the target chip of the write request is the first chip, reduce the credit value of the first virtual channel between the host processor and the first chip so that the host processor can perform backpressure control on the write request sent to the first chip; if the target chip of the write request is another chip, forward the write request to the other chip through the second virtual channel between the host processor and the first chip.

[0100] In this embodiment, the host processor (HOST) does not have a flow control module. On the one hand, it cannot generate flow control information, and on the other hand, it cannot process the received flow control information. If the first chip becomes congested during the process of the host processor sending a data write request to the first chip or through the first chip to other chips, the first chip cannot send flow control information to the host processor so that the host processor can perform backpressure control on the data write request sent to the first chip.

[0101] There are two virtual channels between the host processor and the first chip. The first virtual channel is used for communication between the host processor and the first chip, and the second virtual channel is used for communication between the host processor and other chips through the first chip. When the first chip detects congestion in the cache queue, it determines whether the target chip of the write request is the first chip. If the target chip of the write request is the first chip, the flow control information generation unit of the first chip reduces the credit value of the first virtual channel. After the host processor detects the decrease in the credit value of the first virtual channel, it slows down the speed at which it sends write requests to the first chip. If the target chip of the write request is another chip, the write request is forwarded to the other chip normally.

[0102] Furthermore, during the process of the first chip sending messages to the host processor, because the host processor is unable to send flow control information to the first chip when there is message congestion, the host processor transmits the standard reverse pressure signal back step by step through the PCIe interface between the host processor and the first chip. The flow control information sending unit of the first chip monitors the congestion in the HOST direction and sends local flow control information to perform flow control within the system.

[0103] The flow control method provided in the above embodiments, for host processors without flow control functions, achieves backpressure control by adjusting the credit value of virtual channels. By setting up two different virtual channels, information sent to the local machine and information forwarded locally are distinguished. When local transmission is congested, the host processor only needs to adjust the credit value to apply backpressure to information sent to the local machine, while information forwarded locally can be sent normally, thus achieving fine-grained control over the information sent by the host processor.

[0104] In some embodiments, since there may be multiple data senders in the first chip, such as a general-purpose graphics processing unit (GPGPU) and a data transfer engine, both of which can send data, when the first chip receives flow control information, it cannot distinguish whether the congestion of the data receiver is caused by data sent by the data transfer engine or by data sent by other data senders (such as GPGPU). Therefore, a reverse pressure switch identifier for the data transfer engine is provided.

[0105] If the first chip communicates with the second chip via a relay chip, and the reverse pressure switch identifier is set as the first identifier, the flow control information receiving unit, upon receiving the flow control information, performs reverse pressure control on all data transmission requests sent by data senders within the first chip, including the data transfer engine, to the flow-limiting channel or flow-limiting address. If the first chip and the second chip communicate directly, and the reverse pressure switch identifier is set as the second identifier, the flow control information receiving unit, upon receiving the flow control information, performs reverse pressure control not only on data transmission requests sent by other data senders within the first chip to the flow-limiting channel or flow-limiting address, but also on read data requests sent by the data transfer engine to the storage unit in the first chip and write data requests sent outwards.

[0106] Based on the flow control method provided in the above embodiments, this application also provides a chip, which includes a flow control module, the flow control module including as follows: Figure 2 The flow control information sending unit shown and such Figure 5 The flow control information receiving unit shown is used to execute the flow control method provided in the above embodiments.

[0107] In this embodiment, the chip can be a main computing chip or a slave expansion I / O chip, and each chip can be equipped with a flow control module. The slave expansion I / O chip can be interconnected with the main computing chip via an interconnect channel with high bandwidth, and has multiple storage units to expand the bandwidth and storage capacity of the entire chip system formed with the main computing chip; the main computing chip is equipped with multiple computing cores for parallel data processing.

[0108] The interconnect channels between the main computing chip and the slave expansion I / O chip can be divided into Master ports and Slave ports according to the master-slave relationship. When the main computing chip or the slave expansion I / O chip acts as the initiator, if it is performing a write operation on its own internal storage unit, no backpressure is required. However, if it is performing a write operation on an interconnected device (such as other GP / GPU chips, CPUs, etc.), the interconnected device can perform backpressure when congestion occurs. Therefore, a flow control information receiving unit can be set up on the Master port of the interconnect interface for flow control; when performing read operations on internal storage units or external devices through the Master port, flow control information receiving unit is used for flow control.

[0109] When the interconnect interface of the main computing chip or the expansion I / O chip is used as a slave port, the interconnect device performs write operations through the slave port. The slave port is connected to a flow control information sending unit and a buffer unit. The buffer unit is used to cache data Date and request req. External devices perform read operations through the slave port and cache request req through the buffer unit. The flow control information sending unit monitors the used capacity of the buffer unit and is used to generate flow control messages.

[0110] The chip provided in the above embodiments, through its configured flow control module, can achieve data flow control between chips within the card based on local flow control information, and data flow control between chips in different cards based on remote flow control information. By processing data through a matched processing channel based on the identifier of the flow control information, the parallel processing capability of the flow control information can be improved. Determining the flow-limiting channel based on the flow control information enables fine-grained flow control for a specified channel, rather than flow control for all transmitted data, thus improving the chip's fine-grained flow control capability and making it more suitable for multi-card interconnection communication scenarios.

[0111] In one possible implementation, this application embodiment also provides a computing card, which may include multiple chips, and the multiple chips perform inter-chip backpressure control through a flow control module.

[0112] Specifically, the multiple chips include a main computing chip and at least one slave expansion I / O chip. The interconnection channel between the main computing chip and the slave expansion I / O chip communicates through a control network-on-chip (CNoC). If the first chip and the second chip are located on the same computing card, the flow control information sending unit of the first chip sends the first flow control information to the flow control information receiving unit of the second chip through the CNoC. That is, it performs hardware broadcasting instead of software broadcasting, thereby reducing the latency problem in transmitting the first flow control information.

[0113] If the first chip and the second chip are slave expansion I / O chips located in different computing cards, the first chip and the second chip are connected through a dedicated physical link to realize the communication connection between the two computing cards. The flow control information sending unit of the first chip sends the first flow control information to the flow control information sending unit of the second chip through the dedicated physical link. After receiving the first flow control information, the flow control information sending unit of the second chip broadcasts the first flow control information to the flow control information receiving units of each chip in the same computing card through CNoC.

[0114] During data transmission between the main computing chip and the slave expansion I / O chip, if data congestion occurs due to slow processing speed of the data receiver or excessively fast data transmission by the data sender, the flow control information sending unit of the flow control module of the data receiver generates local flow control information and broadcasts it to the flow control information receiving units of the flow control modules of all chips located on the same computing card. The flow control information receiving units of the flow control modules of all chips then apply back pressure to the data sent to the data receiver through their respective data sending ports to prevent the data congestion situation of the data receiver from worsening.

[0115] For example, the main computing chip sends data to the interconnect channel of the slave expansion I / O chip. The flow control information sending unit of the congested interconnect channel generates local flow control information based on the cache used capacity and sends it to the interconnect channel of the main computing chip and other slave expansion I / O chips. The flow control information receiving units of the interconnect channel of the main computing chip and other slave expansion I / O chips backpressure the data sent to the congested interconnect channel by their respective data sending ports.

[0116] The computing card provided in the above embodiments ensures, through the flow control module set on each chip, that all chips communicating with the target chip with slow message processing speed can perform backpressure control on the message to be sent in advance, thereby reducing the occurrence of data overflow in the target chip with slow message processing speed.

[0117] In one possible implementation, the computing system may include multiple computing cards, each located on the same PCB board, or the computing cards may be located on multiple PCB boards.

[0118] In this embodiment, multiple computing cards located on the same PCB board or different PCB boards can communicate with each other through a dedicated physical link between the expansion I / O chips on the computing cards.

[0119] In some embodiments, computing card A and computing card B are located on the same PCB board. Computing card B writes data to interconnect channel 1 of card A through interconnect channel 0. The flow control information sending unit of the Slave port of interconnect channel 1 of card A generates local flow control information and broadcasts it to each flow control information receiving unit inside card A. Interconnect channel 0 of card A and interconnect channel 0 of card B are connected by a dedicated physical link. The flow control information receiving unit of interconnect channel 0 of card A notifies the flow control information sending unit of interconnect channel 0 of card A to generate remote flow control information based on the local flow control information and forward the remote flow control information to the flow control information sending unit of interconnect channel 0 of card B. The flow control information sending unit of interconnect channel 0 of card B broadcasts the remote flow control information to each flow control information receiving unit inside card B.

[0120] In other embodiments, Figure 9 An architecture diagram of a computing system provided in one embodiment of this application, such as... Figure 9As shown, computing cards A and B are located on the same PCB board, while card C is located on another PCB board. Card C is directly connected to card B, and card C reads and writes data to card A via card B. When card A experiences congestion, it generates local flow control information. The flow control information receiving unit of the chip in card A, connected to card B, notifies the flow control information sending unit of the chip in card A to forward remote flow control information to card B. After receiving the remote flow control information, card B performs internal backpressure and forwards the remote flow control information to card C. After receiving the remote flow control information, card C performs backpressure control on the messages sent from card C to card A. In the illustration, cards A, B, and C are interconnected through an extended I / O chip. In some embodiments, they can also be interconnected through a main computing chip, or the main computing chips and I / O chips of different cards can be connected together.

[0121] In some embodiments, a host and at least one Peripheral Component Interconnect Express (PCIe) slot are provided on the server motherboard. Each PCIe slot can be used to connect a printed circuit board (PCB). Each PCB has at least one computing card, and each computing card can have one or more computing chips. In some embodiments, each computing card has a master computing chip and at least one slave expansion I / O chip. The first chip can be either the master computing chip or the slave expansion I / O chip, and they can be packaged together or separately. The master computing chip can include a general-purpose graphics processing unit (GPGPU), a host direct memory access (HDMA) controller, and on-chip memory, etc. Each slave expansion I / O chip can provide at least one interconnect channel (such as a PCIe channel). The master computing chip can interact with the slave expansion I / O chips through the interconnect channels, and can also interact with other master computing chips, CPUs, network cards, and other devices through the interconnect channels.

[0122] The computing system provided in the above embodiments ensures that all chips communicating with target chips with slow message processing speeds can perform backpressure control on the messages to be sent in advance through the flow control module set on each computing card chip, thereby reducing the occurrence of data overflow in target chips with slow message processing speeds.

[0123] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A flow control method, characterized in that, Applied to a first chip, the method includes: Based on the used capacity of the cache queue of the first chip and the first capacity change trend, first flow control information is selectively generated. The first flow control information includes: a flow limiting channel, the first congestion level of the cache queue, and the first capacity change trend. The first capacity change trend is determined by judging the used capacity of two adjacent monitoring times, or by judging the level before and after the congestion level change. The first flow control information is sent to multiple second chips that can communicate with the first chip, so that each second chip determines the corresponding backpressure cycle proportion based on the first congestion level and the first capacity change trend, and performs backpressure control on messages sent to the first chip based on the flow limiting channel and the backpressure cycle proportion; the first congestion level includes at least three levels, and under the same congestion level, the backpressure cycle proportion with an upward capacity change trend is greater than the backpressure cycle proportion with a downward capacity change trend. The higher the first congestion level, the larger the backpressure cycle proportion; the backpressure cycle proportion is the number of cycles that prevent the transmission of messages to be sent to the destination within a preset flow control cycle; the second chip includes: the current message sender, and / or, a potential message sender that has not yet sent a message.

2. The method as described in claim 1, characterized in that, The method further includes: Receive the second flow control information sent by the third chip; The second flow control information is analyzed to determine the flow restriction channel, the second congestion level, and the second capacity change trend; the second capacity change trend is determined by judging the used capacity of two adjacent monitoring times, or by the level before and after the congestion level change. The proportion of backpressure cycles is determined based on the second congestion level and the second capacity change trend; Based on the current limiting channel and the back pressure cycle ratio, back pressure control is applied to the messages sent from the first chip to the third chip.

3. The method as described in claim 2, characterized in that, The method further includes: The system determines whether to forward the second flow control information to the fourth chip according to the preset forwarding rules. The fourth chip and the first chip are located in different computing cards, and the fourth chip communicates with the third chip through the first chip. When the forwarding rule is satisfied, remote flow control information is generated based on the second flow control information; The remote flow control information is forwarded to the fourth chip so that the fourth chip can perform backpressure control on the messages sent to the third chip according to the remote flow control information.

4. The method as described in claim 2, characterized in that, The method further includes: If the second flow control information is remote flow control information and the forwarding enable state of the first chip is enabled, the remote flow control information is broadcast to the fifth chip located on the same computing card as the first chip, so that the fifth chip performs backpressure control on the messages sent to the third chip according to the remote flow control information.

5. The method as described in claim 2, characterized in that, The method further includes: If the third flow control information sent by the third chip is received within the preset flow control period, the backpressure rule is updated according to the third flow control information after the preset flow control period ends. Based on the current limiting channel and the updated backpressure rules, backpressure control is applied to the messages sent from the first chip to the third chip.

6. The method as described in claim 1, characterized in that, The method further includes: Receive write requests sent by the host processor; If the target chip of the write request is the first chip, the credit value of the first virtual channel between the host processor and the first chip is reduced so that the host processor can perform backpressure control on the write request sent to the first chip. If the target chip of the write request is another chip, the write request is forwarded to the other chip through the second virtual channel between the host processor and the first chip.

7. A chip, characterized in that, The chip includes a flow control module, which is used to perform the flow control method as described in any one of claims 1-6.

Citation Information

Patent Citations

  • Internal flow control method of distributed switch system

    CN106533962A