A single-particle upset-resistant storage and computing architecture FPGA

By forming a computing unit inside the FPGA and using parity registers and verification circuits, the resource consumption and delay problems caused by data transmission between the storage unit and the computing unit are solved, data processing efficiency and computing speed are improved, and operation reliability is enhanced.

CN116755923BActive Publication Date: 2024-11-08WUXI ESIONTECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310774573.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-06-27
Publication Date
2024-11-08
Estimated Expiration
2043-06-27

AI Technical Summary

Technical Problem

When performing computing tasks, existing FPGAs frequently transmit data between storage units and computing units, resource consumption, delay and energy loss increase, limiting data processing efficiency.

Method used

Using a memory and computing architecture that resists single-particle flip, the memory and computing units are formed inside the FPGA. Each unit includes a processor and local memory units, and the data verification is performed using parity registers and verification circuits to ensure that when single-particle flip fails Re-execute operations in a timely manner to reduce data transmission and improve reliability.

Benefits of technology

It improves the data processing efficiency and computing speed of FPGA, enhances operation reliability, reduces the impact of transient errors caused by single-particle flips, and is suitable for environments with less electromagnetic radiation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116755923B_ABST
    Figure CN116755923B_ABST
Patent Text Reader

Abstract

The present application discloses a single-particle upset-resistant storage and computing architecture FPGA, which relates to the field of FPGA technology. In the storage and computing architecture FPGA, resource modules located in the same sub-area are connected through interconnection resources inside the FPGA to realize a storage and computing unit, and registers in the storage and computing unit are replaced with parity registers. The resource module is used to realize a verification circuit. The parity register is verified by the verification circuit during the storage and computing operation. When the verification determines that the register data is wrong, the storage and computing unit is triggered in time to re-execute the storage and computing operation, thereby avoiding operation errors caused by transient errors caused by single-particle upset failures, thereby ensuring the accuracy of operation when implementing parallel multi-core storage and computing operations through multiple storage and computing units, and having outstanding data processing efficiency, operation speed and operation reliability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of FPGA technology, and in particular to a single-particle upset-resistant storage and computing architecture FPGA. Background Art

[0002] With the development of high-tech technologies such as the Internet of Things, cloud computing, and artificial intelligence, various data-intensive applications based on hardware platforms have also grown dramatically. FPGA (Field Programmable Gate Array) is a commonly used hardware platform due to its advantages of parallelism and reconfigurability.

[0003] The existing FPGA, like most traditional hardware platforms, adopts the von Neumann architecture. The FPGA contains a large number of resource modules such as CLB, DSP and BRAM. These resource modules are connected by highly configurable interconnects to form computing units and storage units in the FPGA. BRAM is an important on-chip storage unit. On-chip data is mainly stored in BRAM, and on-chip data must be transferred to the computing unit during calculation.

[0004] Therefore, existing FPGAs, like other hardware platforms that use the von Neumann architecture, need to repeatedly transfer large amounts of data between storage units and computing units when performing various computing tasks. Frequent and large-scale data movement not only consumes a large amount of wiring resources and interconnection resources, but also causes huge delays and energy losses, thereby limiting the data processing efficiency of FPGAs. Summary of the invention

[0005] In response to the above problems and technical requirements, the applicant has proposed a single-particle upset-resistant storage and computing architecture FPGA. The technical solution of this application is as follows:

[0006] A single-particle upset-resistant storage and computing architecture FPGA, the storage and computing architecture FPGA includes a plurality of resource modules arranged according to a predetermined structure, interconnection resources arranged around the resource modules, and a global read-write channel realized by additional hardware resources. Resource modules located in the same sub-area are connected through the interconnection resources inside the FPGA to realize a storage and computing unit. Each storage and computing unit includes a processor and a local storage unit. The local storage unit of each storage and computing unit is used to store instructions and data of the storage and computing unit. The processor in each storage and computing unit is connected to the local storage unit in the same storage and computing unit via the interconnection resources and performs operations according to the instructions and data stored in the local storage unit. Resource modules in multiple different sub-areas correspondingly realize multiple storage and computing units, and the local storage unit in each storage and computing unit transmits data with the outside of the FPGA chip through the global read-write channel.

[0007] There is at least one storage and calculation unit that also includes a verification circuit, and the storage and calculation unit includes at least one register which is a parity register, and the parity register includes a data bit and a check bit; when the storage and calculation unit executes a storage and calculation operation, the verification circuit in the storage and calculation unit performs data verification on the parity register in the storage and calculation unit, and determines that the register data of the parity register is wrong when the verification fails, triggering the storage and calculation unit to re-execute the current storage and calculation operation.

[0008] A further technical solution is that when the storage-computing unit performs a storage-computing operation, the storage-computing instruction and the initial data are written into the local storage unit of the storage-computing unit via the global read-write channel, and the processor in the storage-computing unit performs the storage-computing operation based on the initial data according to the storage-computing instruction, and in the process of performing the storage-computing operation, the storage-computing instruction and the initial data stored in the local storage unit remain unchanged;

[0009] When it is determined that the register data of the parity register is erroneous, the processor in the storage-calculation unit re-executes the storage-calculation operation based on the initial data according to the storage-calculation instruction.

[0010] A further technical solution is that the verification circuit includes a verification code generator and a verification code verifier, and in the process of the storage and calculation unit performing a storage and calculation operation:

[0011] When writing register data to the data bit in the parity register, the check code generator generates a corresponding check code according to the register data and updates the write check bit;

[0012] The check code checker performs data verification according to the data bits and check bits of the parity register, and reads the register data from the data bits of the parity register when the verification is determined to be successful, and triggers the storage and calculation unit to re-execute the current storage and calculation operation when the verification is determined to be failed.

[0013] A further technical solution is that a storage-calculation unit includes a plurality of parity registers, and when the checking circuit determines that there is a register data error in at least one parity register, the storage-calculation unit is triggered to re-execute the current storage-calculation operation.

[0014] Its further technical solution is that when a storage-computing unit includes multiple parity registers, the verification circuit includes multiple checker groups, each parity register corresponds to a checker group, and each checker group includes a check code generator and a check code checker; when there is at least one check code checker that determines that the verification fails, it triggers the storage-computing unit to re-execute the current storage-computing operation.

[0015] A further technical solution is that when a storage unit includes multiple parity registers:

[0016] There are at least two parity registers that share the same check code generator, the check code generator generates respective check codes according to register data of the two parity registers and updates and writes corresponding check bits, and the parity registers that share the same check code generator do not write register data at the same time;

[0017] And / or, there are at least two parity registers that share the same check code checker, the check code checker performs data check on the register data of the corresponding parity register according to the data bit and check bit content of each parity register, and the parity registers that share the same check code checker do not read out the register data at the same time; there is at least one check code checker that triggers the storage and calculation unit to re-execute the current storage and calculation operation when it determines that the check has failed.

[0018] A further technical solution is that the parity register includes K data bits and 1 check bit, and the check circuit performs data check on the parity register based on the parity check method:

[0019] When writing register data to the data bit in the parity register, when the binary K-bit register data includes an even number of 1s, the check code generator generates a check code 1 and updates the write check bit; when the binary K-bit register data includes an odd number of 1s, the check code generator generates a check code 0 and updates the write check bit; then the check code checker determines that the check succeeds when it detects that the K+1-bit data of the parity register includes an odd number of 1s, and determines that the check fails when it detects that the K+1-bit data of the parity register includes an even number of 1s;

[0020] Alternatively, when writing register data to the data bits in the parity register, when the binary K-bit register data includes an even number of 1s, the check code generator generates a check code 0 and updates the write check bit; when the binary K-bit register data includes an odd number of 1s, the check code generator generates a check code 1 and updates the write check bit; then the check code verifier determines that the check is successful when it detects that the K+1-bit data of the parity register includes an even number of 1s, and determines that the check has failed when it detects that the K+1-bit data of the parity register includes an odd number of 1s.

[0021] A further technical solution is that each check code checker has a corresponding failure flag, and each check code checker keeps the failure flag as an invalid signal when determining that the check is successful, and sets the failure flag to a valid signal when determining that the check fails;

[0022] When the fail flag switches from an invalid signal to a valid signal, an interrupt signal is generated and sent to the processor in the same storage and calculation unit. When the processor receives the interrupt signal, it re-executes the current storage and calculation operation and resets all fail flags to invalid signals.

[0023] A further technical solution is that the valid signal of the failure flag generates an interrupt signal after a predetermined delay and sends it to a processor in the same storage and computing unit.

[0024] A further technical solution is that all registers in the storage and calculation unit are parity registers, or some registers are parity registers.

[0025] The beneficial technical effects of this application are:

[0026] The present application discloses a storage and computing architecture FPGA that is resistant to single-particle flips. In the FPGA, resource modules located in the same sub-area are connected through interconnection resources inside the FPGA to realize a storage and computing unit. When the registers in the storage and computing unit are replaced with parity registers, the resource modules are used to implement the verification circuit to verify the parity registers. When the verification determines that the register data is wrong, the storage and computing unit is triggered in time to re-execute the storage and computing operation to avoid the operation error caused by the transient error caused by the single-particle flip failure. Therefore, the operation reliability can be improved when parallel multi-core storage and computing operations are realized through multiple storage and computing units, and it has outstanding data processing efficiency, operation speed and operation reliability. The storage and computing architecture FPGA of the present application is mainly suitable for situations where the electromagnetic irradiation particles are small and the duration is short. The transient errors caused by the single-particle flip failure do not occur intensively. In the process of the storage and computing architecture FPGA executing parallel storage and computing operations of multiple storage and computing units, when a register data error is detected, an interrupt is generated to stop the operation to eliminate the influence, which has little effect on the overall process of the storage and computing operation.

[0027] The processor and local data storage in each storage and computing unit formed inside the FPGA are in the same sub-area. When executing, the processor only needs to read and write data in the adjacent local data storage, which reduces the wiring resources and interconnection resources consumed by transmitting data between the processor and the local data storage, reduces data transmission delay and energy loss, and improves the data processing efficiency of the FPGA. In addition, the FPGA can realize multi-core parallel storage and computing operations through multiple storage and computing units. The processors of each storage and computing unit will not cause conflicts with each other. Each storage and computing unit can perform operations at full speed, which is not affected by the storage bandwidth, and can further improve the overall data processing efficiency and computing speed of the FPGA. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic diagram of the arrangement of resource modules inside the storage-computing architecture FPGA and a schematic diagram of the sub-areas obtained by division in an example of the present application.

[0029] Figure 2 Taking the example of the instruction registers in each storage unit as parity registers, based on Figure 1 Schematic diagram of the circuit structure of each storage and computing unit implemented.

[0030] Figure 3 It is a connection diagram of the check circuit, the parity checker and the processor in one embodiment of the present application.

[0031] Figure 4 It is a connection diagram of a check circuit in one embodiment of the present application with multiple parity checkers and a processor.

[0032] Figure 5 This is another connection diagram of the check circuit in one embodiment of the present application with multiple parity checkers and a processor.

[0033] Figure 6 yes Figure 2 Another circuit structure diagram in which the local storage units of each storage and computing unit are connected to the outside of the FPGA chip.

[0034] Figure 7 yes Figure 2 Another circuit structure diagram in which the local storage units of each storage and computing unit are connected to the outside of the FPGA chip.

[0035] Figure 8 yes Figure 2 Another circuit structure diagram in which the local storage units of each storage and computing unit are connected to the outside of the FPGA chip. DETAILED DESCRIPTION

[0036] The specific implementation of the present application is further described below in conjunction with the accompanying drawings.

[0037] The present application discloses a storage and computing architecture FPGA that is resistant to single-particle flipping. The storage and computing architecture FPGA, like a conventional FPGA, includes several resource modules arranged in a predetermined structure and interconnected resources arranged around the resource modules. The predetermined structure can be arranged according to actual conditions. For example, taking the currently mainstream Column-Based structure as an example, the resource modules are arranged in a row and column structure to form a two-dimensional array architecture, and each column is arranged with resource modules of the same type. Interconnected resources are arranged around each resource module. The types of resource modules contained in the FPGA include configurable logic modules CLB, configurable storage modules BRAM, and configurable multiplication and addition modules DSP. The configurable logic module CLB contains units such as lookup tables and registers, which can be configured to implement various logical functions. The configurable multiplication and addition module DSP contains arithmetic units that can perform multiple operations such as multiplication, addition, and multi-bit logic gates. The configurable storage module BRAM contains block storage units that can implement single- and double-port storage and FIFO storage of various widths and depths. This part of the structure follows the structure inside the existing FPGA, and this application will not elaborate on it.

[0038] All resource modules arranged in a predetermined structure in the storage and computing architecture FPGA of the present application are divided into several sub-areas, each of which contains a certain number of resource modules. Resource modules in the same sub-area are connected through the interconnection resources inside the FPGA to realize a storage and computing unit, and resource modules in multiple different sub-areas correspondingly realize multiple storage and computing units.

[0039] Each storage and computing unit includes its own processor and local storage unit. The processor in each storage and computing unit is connected to the local storage unit in the storage and computing unit via interconnection resources. The local storage unit in each storage and computing unit is used to store the instructions and data of the storage and computing unit, and the processor in each storage and computing unit is used to operate according to the instructions and data stored in the connected local storage unit.

[0040] In one embodiment, the instruction sets used by the two storage and computing units implemented by the resource modules in any two sub-areas are the same or different, and the bit widths of the two storage and computing units using the same instruction set are the same or different. In the present application, the storage and computing units formed inside the storage and computing architecture FPGA can form various typical general-purpose 8-bit processors including RISC, 8051 and 6802.

[0041] In one embodiment, the number and type of resource modules contained in any two sub-areas obtained by division are the same, or in another embodiment, the number and type of resource modules contained in any two sub-areas obtained by division are at least one different, that is, the multiple sub-areas obtained by division may be completely the same or not necessarily the same. In addition, each resource module in the storage-computing architecture FPGA belongs to a sub-area, all resource modules in the storage-computing architecture FPGA are divided into corresponding sub-areas, or some resource modules in the storage-computing architecture FPGA are divided into corresponding sub-areas, that is, the storage-computing architecture FPGA can be divided into sub-areas globally or locally. For example, Figure 1 In the example shown, the resource modules inside the FPGA are divided into 8 sub-areas, each of which contains the same number and type of resource modules, and each area framed by a dotted box represents a sub-area obtained by the division.

[0042] This application does not limit the number and type of resource modules contained in each sub-area, but the number and type of resource modules contained in each sub-area must be sufficient to implement the devices required for a storage and computing unit. The processor in a storage and computing unit is mainly implemented based on DSP, and the storage unit in a storage and computing unit is mainly implemented based on BRAM. In addition, a certain number of CLBs are also required.

[0043] The storage and computing unit implemented by the resource module in the sub-area actually includes many registers, such as instruction registers and data registers: when the processor reads instructions from the local storage unit, the instructions in the local storage unit are first written into the instruction register IR, and then written into the processor by the instruction register IR. When the processor reads data from the local storage unit, the data in the local storage unit is first written into the data register DR, and then written into the processor by the data register DR. Correspondingly, when the processor writes data to the local storage unit, it first writes the data into the data register DR, and then writes it into the local storage unit by the data register DR. In addition to the above-mentioned instruction registers and data registers, the storage and computing unit also includes many general registers necessary for the circuit structure. In a complex electromagnetic environment, the registers will be affected by single-particle upsets due to radiation exposure and other reasons. Although the single-particle upset effect is transient and non-destructive, it will change the content of the register, thereby causing persistent functional errors and causing errors in the operation of the storage and computing unit.

[0044] Therefore, in order to improve reliability, in the storage and computing architecture FPGA of the present application, there is at least one storage and computing unit including a parity register, and in the storage and computing unit including the parity register, a check circuit is also implemented through the resource module in the corresponding sub-area, and the check circuit is used to connect the parity register and check the parity register. In addition, the check circuit is also connected to the processor in the same storage and computing unit. The parity register here can be any one of a data register, an instruction register, and a general register according to the function type. The parity register corresponds to the traditional register. The traditional register only contains a number of data bits, and the parity register includes a number of check bits in addition to the data bits. It can be considered as a replacement and upgrade of the traditional register, and the number of check bits is configured according to the actual situation. There is at least one parity register in the storage and computing unit. Generally, according to actual needs, the traditional register in the circuit structure of the storage and computing unit that is easily affected by single-particle upsets is replaced with a parity register. However, when the parity register is used to replace the traditional register, the logical function of the storage and computing unit is not affected. In one embodiment, all registers in a storage and computing unit are replaced and upgraded to use parity registers, or only some registers use parity registers, and the remaining registers that are not easily affected by single-particle upsets still use traditional registers. In one embodiment, all storage and computing units formed by the storage and computing architecture FPGA have parity registers and include verification circuits. Or in another embodiment, some storage and computing units formed by the storage and computing architecture FPGA have parity registers and include verification circuits.

[0045] For example, based on Figure 1For the 8 sub-areas divided in the storage-computing architecture FPGA, the storage-computing architecture FPGA implements 8 storage-computing units through the resource modules in the 8 sub-areas. Each storage-computing unit includes a processor and a local storage unit. The instruction register IR in each storage-computing unit is implemented using a parity register. Each storage-computing unit includes a check circuit. The circuit structure diagram of the 8 storage-computing units is as follows: Figure 2 As shown, Figure 2 Since the complete circuit structure is not shown in detail, the general registers in the storage and calculation unit are not shown.

[0046] In addition to resource modules and interconnection resources, the storage and computing architecture FPGA of this application also adds additional hardware resources to implement global read and write channels compared to conventional FPGAs. The local storage unit in each storage and computing unit transmits data to the FPGA chip through the global read and write channels, such as Figure 2 shown.

[0047] The storage and computing architecture of the present application forms multiple storage and computing units inside the FPGA, and each storage and computing unit supports independent storage and computing operations in the form of a soft core. Therefore, the FPGA can jointly implement a multi-core storage and computing operation through multiple storage and computing units formed inside, including:

[0048] 1. Write the storage and computing instructions and initial data of each storage and computing unit into the local storage unit of the corresponding storage and computing unit from outside the FPGA chip through the global read and write channel.

[0049] 2. Start each storage and computing unit to perform parallel calculations simultaneously. The processor in each storage and computing unit performs storage and computing operations based on the initial data according to the storage and computing instructions in the local storage unit of the storage and computing unit, and stores the generated calculation results in the local storage unit of the storage and computing unit.

[0050] For any storage unit including a parity register and a check circuit, during the process of the storage unit performing a storage operation, the check circuit in the storage unit performs data check on the parity register in the storage unit, and determines that the register data of the parity register is wrong when the check fails. If the operation continues, it will cause an operation error, thereby triggering the storage unit to re-execute the current storage operation. Since the single particle upset failure is transient and non-destructive, the register data error caused by the single particle upset can be corrected when the current storage operation is re-executed to ensure operational reliability.

[0051] In order to enable the storage and computing unit to re-execute the current storage and computing operation, in the present application, after the storage and computing instructions and initial data of the storage and computing unit are written into the local storage unit through step 1, during the process of the processor executing the storage and computing operation, the storage and computing instructions and initial data stored in the local storage unit remain unchanged, so that the processor can re-execute the current storage and computing operation by re-executing the storage and computing instructions in the local storage unit based on the initial data.

[0052] In one embodiment, please refer to Figure 3 , the verification circuit in any storage and calculation unit includes a verification code generator and a verification code verifier. In the process of the storage and calculation unit performing a storage and calculation operation:

[0053] When the register data Data_reg is written to the data bit in the parity register, the check code generator generates the corresponding check code Data_par according to the register data Data_reg and updates the check bit written to the parity register, so that the data bit of the parity register is written with the register data Data_reg and the check bit is written with the check code Data_par. When the register data needs to be read from the parity register, the check code verifier performs data verification according to the data bit and the check bit of the parity register, and reads the register data Data_reg from the data bit of the parity register when it is determined that the verification is successful, and triggers the storage and calculation unit to re-execute the current storage and calculation operation when it is determined that the verification fails.

[0054] The method of performing data verification by the check code verifier according to the data bit and the check bit of the parity register corresponds to the method of generating the check code Data_par according to the register data Data_reg by the check code generator. In one embodiment, the parity register includes K data bits and 1 check bit, and only one check bit needs to be added compared to the traditional register. Then the above-mentioned verification circuit performs data verification on the parity register based on the parity check method, including:

[0055] (1) When odd check is performed on the parity register, when register data is written to the data bit in the parity register, when the binary K-bit register data includes an even number of 1s, the check code generator generates a check code 1 and updates the written check bit. When the binary K-bit register data includes an odd number of 1s, the check code generator generates a check code 0 and updates the written check bit.

[0056] Correspondingly, when register data needs to be read from the parity register, the check code checker determines that the check is successful when it detects that the K+1-bit data of the parity register includes an odd number of 1s, and determines that the check fails when it detects that the K+1-bit data of the parity register includes an even number of 1s.

[0057] In order to realize the above functions, in one implementation of this embodiment, the check code generator is implemented by a K-bit XNOR gate, and the check code checker is implemented by a K+1-bit XNOR gate.

[0058] (2) When the parity register is even-checked, when the register data is written into the data bit in the parity register, when the binary K-bit register data includes an even number of 1s, the check code generator generates a check code 0 and updates the written check bit. When the binary K-bit register data includes an odd number of 1s, the check code generator generates a check code 1 and updates the written check bit.

[0059] Correspondingly, when register data needs to be read from the parity register, the check code checker determines that the check is successful when it detects that the K+1-bit data of the parity register includes an even number of 1s, and determines that the check fails when it detects that the K+1-bit data of the parity register includes an odd number of 1s.

[0060] In order to realize the above functions, in one implementation of this embodiment, the check code generator is implemented by a K-bit XOR gate, and the check code checker is implemented by a K+1-bit XOR gate.

[0061] Regardless of which of the above implementations is adopted, in one embodiment, when the check code verifier determines that the verification fails, it does not directly trigger the storage and calculation unit to re-execute the current storage and calculation operation. Each check code verifier has a corresponding failure flag (Failure Flag), which defaults to an invalid signal. When the check code verifier determines that the verification is successful, the corresponding failure flag is kept as an invalid signal, and when the check code verifier determines that the verification fails, the corresponding failure flag is set to a valid signal. When the failure flag switches from an invalid signal to a valid signal, an interrupt signal is generated and sent to the processor in the same storage and calculation unit. When the processor receives the interrupt signal, the current storage and calculation operation is re-executed and all failure flags are reset to invalid signals. The continuous valid signal of the failure flag only generates an interrupt signal once. This approach can effectively avoid the situation of repeated interrupt signal generation. If there is no failure flag, when the check code verifier fails to verify continuously, before the processor re-executes the current storage and calculation operation, the check code verifier will still continuously generate an interrupt signal to the processor. And through the existence of the failure flag, when the check code verifier fails to verify continuously, the continuous valid signal of the failure flag also only generates an interrupt signal to the processor once.

[0062] In addition, based on the existence of the failure flag, the failure flag can immediately generate an interrupt signal and send it to the processor in the same storage and computing unit when it switches from an invalid signal to a valid signal. Alternatively, the failure flag generates an interrupt signal and sends it to the processor in the same storage and computing unit after switching from an invalid signal to a valid signal and delaying for a predetermined time, and the predetermined delay time can be adjusted according to actual needs.

[0063] Figure 3 The above-mentioned embodiments all introduce the verification process of a single parity register by the verification circuit in the storage and computing unit. In one embodiment, a storage and computing unit may include multiple parity registers, and the verification circuit in the storage and computing unit is used to perform data verification on the multiple parity registers in the storage and computing unit. When the verification circuit determines that there is a register data error in at least one parity register, it triggers the storage and computing unit to re-execute the current storage and computing operation.

[0064] When a storage unit includes multiple parity registers, one implementation method is that the verification circuit includes multiple checker groups, each parity register corresponds to a checker group, each checker group includes a check code generator and a check code checker, and the connection method and verification process of each checker group and the corresponding parity register are as described in the above embodiment, which will not be repeated in this embodiment. Figure 4 In the example, the verification circuit includes two checker groups, corresponding to two parity registers. When at least one check code checker determines that the verification fails, the storage and calculation unit is triggered to re-execute the current storage and calculation operation. Common implementation methods are as follows: Figure 4 As shown, the failure flags of the check code checkers in each checker group are respectively connected to an input end of the OR gate. When any failure flag switches from an invalid signal to a valid signal, the output end of the OR gate outputs an interrupt signal and sends it to the processor in the same storage and computing unit.

[0065] When a storage and computing unit includes multiple parity registers, another implementation method is that there are at least two parity registers that share the same check code generator, and the check code generator generates respective check codes based on the register data of the two parity registers and updates and writes the corresponding check bits, and the parity registers that share the same check code generator do not write register data at the same time. When multiple parity registers do not write data at the same time, the number of check code generators used can be reduced without affecting the function implementation. And / or, there are at least two parity registers that share the same check code checker, and the check code checker performs data verification on the register data of the corresponding parity register based on the data bit and check bit content of each parity register, and the parity registers that share the same check code checker do not read register data at the same time. When multiple parity registers do not read data at the same time, the number of check code checkers used can be reduced without affecting the function implementation. For example, Figure 5 In, corresponding to Figure 4, these two registers share the same check code generator and the same check code verifier. The check code generator generates the corresponding Data_par1 to write the check bit of the parity register according to Data_reg1 of one parity register, and generates the corresponding Data_par2 to write the check bit of the parity register according to Data_reg2 of another parity register. The check code generator performs data verification according to the content of the data bit and the check bit of each parity checker, and sets the corresponding failure flag to a valid signal when it is determined that the verification fails according to the content of the data bit and the check bit of any parity checker, triggering the failure flag to send an interrupt signal to the processor, which is similar to the implementation method of the above embodiment. In this case, when there are multiple check code generators in the storage and calculation unit, the same as the above embodiment, when there is at least one check code verifier that determines that the verification fails, the storage and calculation unit is triggered to re-execute the current storage and calculation operation. The implementation method is the same as the above embodiment, and this embodiment will not be repeated.

[0066] 3. The calculation results in the local storage unit of each storage unit are read out to the outside of the FPGA chip through the global read and write channel.

[0067] On this basis, according to actual needs, the storage-computing architecture FPGA can also realize multiple storage-computing operations in sequence through the internal storage-computing units until all operations are completed. And the storage-computing architecture FPGA can use all or part of the storage-computing units when realizing any storage-computing operation, and the storage-computing units used when realizing each storage-computing operation can be the same or different.

[0068] Since the processor and local storage unit of each storage unit are in the same sub-area, the processor only needs to read and write the storage instructions and initial data in the adjacent local storage unit during execution, thereby reducing the wiring resources and interconnection resources consumed by data transmission between the processor and the local storage unit, reducing data transmission delay and energy loss, and improving the data processing efficiency of the storage architecture FPGA. In addition, in a multi-core scenario, if multiple computing cores read and write the same main memory at the same time, the read and write bandwidth of the main memory will form a bottleneck, restricting the data processing efficiency. This problem will be particularly obvious in high-performance computing. For FPGAs that support artificial intelligence or machine learning applications, the FPGA generally has a large number of relatively simple computing cores, and the number of computing cores will exceed 100. The demand for the read and write bandwidth of the main memory is greater, and the problem is more prominent. The storage architecture FPGA of the present application is also a multi-core scenario. Since the processors in each storage unit read and write local storage units respectively, instead of all processors uniformly reading and writing the same memory, the processors of each storage unit will not cause conflicts with each other, and each storage unit can perform operations at full speed, which is not affected by the storage bandwidth, which is conducive to improving the overall data processing efficiency and computing speed. This makes the storage-computing architecture FPGA of the present application applicable to scenarios involving a large number of storage-computing units. Even if the number of storage-computing units exceeds 100, it can still have a high computing rate, thereby better supporting artificial intelligence and machine learning scenarios.

[0069] In one embodiment, when the predetermined structure adopted by the arrangement of resource modules inside the storage and computing architecture FPGA of the present application is a Column-Based structure, the BRAM module occupies one or more columns, and the local storage units in each storage and computing unit are implemented based on the BRAM module. Therefore, the BRAM module used to implement the local storage unit in each storage and computing unit is located in one or more columns, and the local storage units in the storage and computing units and several BRAM modules located in the same column are used to implement data transmission with the storage and computing architecture FPGA chip via the same global read-write channel. The storage and computing architecture FPGA chip can exchange data with any local storage unit connected to the global read-write channel via each global read-write channel. Based on this, the present application provides the following two methods to implement data transmission between the local storage unit in each storage and computing unit and the storage and computing architecture FPGA chip:

[0070] In one embodiment, several BRAM modules located in the same column and used to implement local storage units in the storage and computing units are connected to an on-chip read / write interface of the FPGA through the same global read / write channel to perform data transmission with the outside of the FPGA, and different global read / write channels are connected to different on-chip read / write interfaces of the FPGA. Figure 2In the example shown, eight storage and computing units are formed inside the storage and computing architecture FPGA. The local storage units in storage and computing units 1 to 4 are implemented by BRAM modules located in the same column. The local storage units of these four storage and computing units are connected to the on-chip read-write interface 1 of the storage and computing architecture FPGA through the global read-write channel 1 to realize data transmission with the outside of the storage and computing architecture FPGA. The local storage units in storage and computing units 5 to 8 are implemented by BRAM modules located in the same column. The local storage units of these four storage and computing units are connected to the on-chip read-write interface 2 of the storage and computing architecture FPGA through the global read-write channel 2 to realize data transmission with the outside of the FPGA.

[0071] The method provided in the above embodiment often needs to occupy multiple on-chip read / write interfaces of the FPGA, and the on-chip read / write interface resources of the FPGA are relatively precious. Therefore, in another embodiment, only one on-chip read / write interface of the storage-computing architecture FPGA is needed to realize the local storage unit in the soft-core processor, and several BRAM modules located in the same column are connected to an active end of the multiplexer MUX through the same global read / write channel, and the fixed end of the multiplexer MUX is connected to the on-chip read / write interface of the FPGA. By selecting each active end of the multiplexer MUX, each global read / write channel can be selected, so that data can be exchanged with any local storage unit connected to the global read / write channel outside the FPGA chip via the selected global read / write channel. For example, in Figure 6 In the structure diagram shown, it corresponds to Figure 2 , the local storage units in storage and computing units 1 to 4 are connected to an active end of the multiplexer MUX through the global read-write channel 1, and the local storage units in storage and computing units 5 to 8 are connected to the other active end of the multiplexer MUX through the global read-write channel 2. The fixed end of the multiplexer MUX is connected to the on-chip read-write interface of the storage and computing architecture FPGA to realize data transmission with the outside of the chip. Compared with Figure 2 The method of this embodiment only needs to occupy one on-chip read / write interface of the FPGA. It should be noted that Figure 6 Mainly to show the structure of the local storage unit connected to the FPGA chip outside the storage and computing architecture, the check circuit and parity register in each storage and computing unit are omitted, and the subsequent figures are similarly omitted.

[0072] According to the traditional von Neumann architecture, the instructions and data in each storage unit are mixed and stored. However, in order to improve the data processing efficiency, in one embodiment, according to the Harvard structure, the instructions and data of each storage unit are stored separately and independently. Then the local storage unit in a storage and computing unit includes an instruction local storage unit and a data local storage unit. The instruction local storage unit is used to store the instructions of the storage and computing unit, and the data local storage unit is used to store the data of the storage and computing unit, so that the instructions and data do not interfere with each other when the processor operates, thereby improving the data processing efficiency. The instruction local storage unit is implemented by at least one BRAM module, and the data local storage unit is implemented by at least one BRAM module. Therefore, based on this, a common sub-area includes at least one DSP, two BRAMs, and 300 CLBs.

[0073] When the local storage unit in each storage and computing unit includes an instruction local storage unit and a data local storage unit, the storage size of the instruction local storage units of any two storage and computing units is the same or different, and the storage size of the data local storage units of any two storage and computing units is the same or different.

[0074] When the local storage unit in each storage unit includes an instruction local storage unit and a data local storage unit: (1) In one embodiment, the instruction local storage unit and the data local storage unit in a storage unit share a global read / write channel for data transmission, and the instruction local storage unit and the data local storage unit in the storage unit and the BRAM modules located in the same column are used to transmit data with the storage unit architecture FPGA chip via the same global read / write channel. In this embodiment, the global read / write channel can be used as follows: Figure 2 The method of the embodiment shown is connected to the on-chip read / write interface of the storage computing architecture FPGA, or according to Figure 6 The method of the embodiment shown is connected to an active terminal of the multiplexer MUX. Figure 2 Taking the structure shown in the figure as an example, Figure 7As shown, the instruction local storage units and data local storage units in storage and computing units 1 to 4 are connected to the global read-write channel 1, and the instruction local storage units and data local storage units in storage and computing units 5 to 8 are connected to the global read-write channel 2. (2) Or in another embodiment, the instruction local storage units and data local storage units use their own global read-write channels respectively, and the instruction local storage units in a storage and computing unit transmit instructions via the instruction global read-write channel, and the data local storage units transmit data via the data global read-write channel. Then, the instruction local storage units in the storage and computing units and several BRAM modules in the same column transmit data to the outside of the storage and computing architecture FPGA chip via the same instruction global read-write channel, and the data local storage units in the storage and computing units and several BRAM modules in the same column transmit data to the outside of the storage and computing architecture FPGA chip via the same data global read-write channel. Similarly in this embodiment, each instruction global read-write channel and data global read-write channel can be further configured according to Figure 2 The method of the embodiment shown is connected to the on-chip read / write interface of the storage computing architecture FPGA, or according to Figure 6 The method of the embodiment shown is connected to an active terminal of the multiplexer MUX. Figure 6 Take the structure shown in the figure as an example. Figure 8 The instruction local storage units in storage and computing units 1 to 4 are all connected to the instruction global read and write channel 1, the data local storage units in storage and computing units 1 to 4 are all connected to the data global read and write channel 1, the instruction local storage units in storage and computing units 5 to 8 are all connected to the instruction global read and write channel 2, the data local storage units in storage and computing units 5 to 8 are all connected to the data global read and write channel 2, the instruction global read and write channel 1, the data global read and write channel 1, the instruction global read and write channel 2 and the data global read and write channel 2 are respectively connected to an active end of the multiplexer MUX.

[0075] In another embodiment, the resource module in each sub-area is also used to implement the control signal and status signal of a corresponding storage and computing unit, the control signal is used to control the operating status of the storage and computing unit, and the status signal is used to display the operating status of the storage and computing unit. The control signal and status signal of the storage and computing unit are connected to other circuit structures inside the FPGA, or the control signal and status signal of the storage and computing unit are connected to the outside of the FPGA. That is, the control signal and status signal required by each storage and computing unit when performing storage and computing are implemented by the resource module in the corresponding sub-area.

[0076] The control signal of each storage and computing unit includes at least one of an enable signal, a clock signal, a reset signal and an interrupt signal. The enable signal is used to control the corresponding storage and computing unit to execute instructions or stop executing instructions. The reset signal is used to reset the storage and computing unit. The interrupt signal provides external interrupt capability. The control signals of any two storage and computing units are the same or different.

[0077] The above is only a preferred embodiment of the present application, and the present application is not limited to the above embodiments. It is understood that other improvements and changes directly derived or associated by those skilled in the art without departing from the spirit and concept of the present application should be considered to be included in the protection scope of the present application.

Claims

1. A single-particle upset resistant storage and computing architecture FPGA, characterized in that: The storage and computing architecture FPGA includes a plurality of resource modules arranged in a predetermined structure, interconnection resources arranged around the resource modules, and a global read-write channel realized by additional hardware resources. The resource modules located in the same sub-area are connected through the interconnection resources inside the FPGA to realize a storage and computing unit. Each storage and computing unit includes a processor and a local storage unit. The local storage unit of each storage and computing unit is used to store instructions and data of the storage and computing unit. The processor in each storage and computing unit is connected to the local storage unit in the same storage and computing unit via the interconnection resources and operates according to the instructions and data stored in the local storage unit. The resource modules in multiple different sub-areas correspondingly realize multiple storage and computing units, and the local storage unit in each storage and computing unit transmits data with the outside of the FPGA chip through the global read-write channel. There is at least one storage and calculation unit that also includes a verification circuit, and the storage and calculation unit includes at least one register that is a parity register, and the parity register includes a data bit and a check bit; when the storage and calculation unit executes a storage and calculation operation, the verification circuit in the storage and calculation unit performs data verification on the parity register in the storage and calculation unit, and determines that the register data of the parity register is wrong when the verification fails, triggering the storage and calculation unit to re-execute the current storage and calculation operation.

2. The storage-computing architecture FPGA according to claim 1, characterized in that: When the storage and computing unit performs a storage and computing operation, the storage and computing instruction and the initial data are written into the local storage unit of the storage and computing unit via the global read and write channel, and the processor in the storage and computing unit performs the storage and computing operation based on the initial data according to the storage and computing instruction, and in the process of performing the storage and computing operation, the storage and computing instruction and the initial data stored in the local storage unit remain unchanged; When it is determined that the register data of the parity register is erroneous, the processor in the storage-calculation unit re-executes the storage-calculation operation based on the initial data according to the storage-calculation instruction.

3. The storage-computing architecture FPGA according to claim 1, characterized in that: The verification circuit includes a verification code generator and a verification code verifier. During the process of the storage and calculation unit performing a storage and calculation operation: When writing register data into the data bit in the parity register, the check code generator generates a corresponding check code according to the register data and updates and writes the check bit; The check code checker performs data check according to the data bit and the check bit content of the parity register, and reads the register data from the data bit of the parity register when it is determined that the check is successful, and triggers the storage and calculation unit to re-execute the current storage and calculation operation when it is determined that the check fails.

4. The storage-computing architecture FPGA according to claim 3, characterized in that: A storage-calculation unit includes a plurality of parity registers. When the checking circuit determines that there is a register data error in at least one parity register, the storage-calculation unit is triggered to re-execute the current storage-calculation operation.

5. The storage-computing architecture FPGA according to claim 4, characterized in that: When a storage-computing unit includes multiple parity registers, the verification circuit includes multiple checker groups, each parity register corresponds to a checker group, and each checker group includes a check code generator and a check code checker; when there is at least one check code checker that determines that the verification fails, the storage-computing unit is triggered to re-execute the current storage-computing operation.

6. The storage-computing architecture FPGA according to claim 4, characterized in that: When a storage unit includes multiple parity registers: There are at least two parity registers that share the same check code generator, the check code generator generates respective check codes according to register data of the two parity registers and updates and writes corresponding check bits, and the parity registers that share the same check code generator do not write register data at the same time; And / or, there are at least two parity registers that share the same check code checker, and the check code checker performs data check on the register data of the corresponding parity register according to the data bit and check bit content of each parity register, and the parity registers that share the same check code checker do not read out the register data at the same time; there is at least one check code checker that triggers the storage and calculation unit to re-execute the current storage and calculation operation when it determines that the check fails.

7. The storage-computing architecture FPGA according to claim 3, characterized in that: The parity register includes K data bits and 1 check bit, and the check circuit performs data check on the parity register based on a parity check method: When writing register data to the data bit in the parity register, when the binary K-bit register data includes an even number of 1s, the check code generator generates a check code 1 and updates and writes the check bit; when the binary K-bit register data includes an odd number of 1s, the check code generator generates a check code 0 and updates and writes the check bit; then the check code checker determines that the check succeeds when it detects that the K+1-bit data of the parity register includes an odd number of 1s, and determines that the check fails when it detects that the K+1-bit data of the parity register includes an even number of 1s; Alternatively, when writing register data to the data bits in the parity register, when the binary K-bit register data includes an even number of 1s, the check code generator generates a check code 0 and updates the check bit; when the binary K-bit register data includes an odd number of 1s, the check code generator generates a check code 1 and updates the check bit; then the check code verifier determines that the check is successful when it detects that the K+1-bit data of the parity register includes an even number of 1s, and determines that the check has failed when it detects that the K+1-bit data of the parity register includes an odd number of 1s.

8. The storage-computing architecture FPGA according to claim 3, characterized in that: Each of the verification code checkers has a corresponding failure flag, and each of the verification code checkers keeps the failure flag as an invalid signal when determining that the verification is successful, and sets the failure flag to a valid signal when determining that the verification fails; When the failure flag switches from an invalid signal to a valid signal, an interrupt signal is generated and sent to the processor in the same storage and calculation unit. When the processor receives the interrupt signal, it re-executes the current storage and calculation operation and resets all the failure flags to invalid signals.

9. The storage-computing architecture FPGA according to claim 8, characterized in that: The valid signal of the failure flag generates an interrupt signal after a predetermined delay and is sent to the processor in the same storage and computing unit.

10. The storage-computing architecture FPGA according to claim 1, characterized in that: All registers in the storage and calculation unit are parity registers, or some registers are parity registers.