阅读说明：本技术 一种PCIe信号处理卡的级联方法 (Cascade method of PCIe signal processing cards ) 是由 杨庸 于 2021-08-11 设计创作，主要内容包括：本发明公开了一种PCIe信号处理卡的级联方法,将多张相同的PCIe信号处理卡通过板间数据高速级联通道级联起来,其中一张PCIe信号处理卡连接FMC子卡作为主卡,其余作为协处理卡,每张PCIe信号处理卡通过PCIe插槽与上位机通信连接；FMC子卡将采集的原始数据发送主卡进行分片和封装,封装头包含对应PCIe信号处理卡的板卡ID；主卡通过板间数据高速级联通道将全部数据传输到协处理卡,每张协处理卡处理与自身板卡ID匹配的数据,并将处理结果数据传输给上位机,上位机对数据进行重组。本发明解决了硬件加速时受限于FPGA资源的问题,实现了弹性扩展的级联通道；适用于采集场景和发射场景。(The invention discloses a cascade method of PCIe signal processing cards, which is characterized in that a plurality of same PCIe signal processing cards are cascaded through inter-board data high-speed cascade channels, wherein one PCIe signal processing card is connected with an FMC daughter card as a main card, the rest are used as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot; the FMC sub-card sends the acquired original data to the main card for fragmentation and packaging, and a packaging head comprises a card ID corresponding to a PCIe signal processing card; the main card transmits all data to the co-processing cards through the inter-board data high-speed cascade channel, each co-processing card processes data matched with the ID of the board card, the processed result data is transmitted to the upper computer, and the upper computer recombines the data. The invention solves the problem that hardware is limited by FPGA resources during acceleration, and realizes an elastically expanded cascade channel; the method is suitable for acquisition scenes and emission scenes.)

1. A method for cascading PCIe signal processing cards, comprising:

step S110, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S120, the FMC sub-card sends the collected original data to the main card, the main card fragments and encapsulates the original data to obtain encapsulated data, and an encapsulation head of the encapsulated data comprises a board card ID and a data block serial number corresponding to a PCIe signal processing card;

and S130, the main card transmits all data to the co-processing cards through the inter-board data high-speed cascade channel, each co-processing card processes data matched with the ID of the own board card and transmits processing result data to the upper computer, and the upper computer recombines the received processing result data.

2. The PCIe signal processing card cascading method of claim 1, wherein the step S130 is replaced with: the main card takes out the data corresponding to the ID of the own board card from the encapsulated data, then sends the rest data to the co-processing card of the next stage through the inter-board data high-speed cascade channel, and similarly, the co-processing card takes out the data corresponding to the ID of the own board card and then sends other data to the co-processing card of the next stage through the inter-board data high-speed cascade channel, and so on until each co-processing card receives the data corresponding to the ID of the own board card;

and each PCIe signal processing card uploads data corresponding to the own board card ID to the upper computer through a PCIe slot, and the upper computer recombines all received data according to the board card ID and the data block serial number.

3. The cascading method for PCIe signal processing cards as claimed in claim 1 or 2, wherein the FMC daughter card is an AD chip.

4. The method for cascading PCIe signal processing cards according to claim 1, wherein the step S120 is replaced by: the upper computer fragments and encapsulates the modulated data to obtain encapsulated data, and an encapsulation head of the encapsulated data comprises a board card ID and a data block serial number corresponding to the PCIe signal processing card;

the step S130 is replaced by: the upper computer sends all the packaged data to each PCIe signal processing card through a PCIe slot, an FPGA on each PCIe signal processing card processes the data according to respective setting and carries out fragment packaging on the modulation result data according to the position of the FPGA in a stacking scene, each co-processing card sends the data to the main card, and the data is sent out after being recombined by the main card.

5. The method for cascading PCIe signal processing cards according to claim 4, wherein the step S130 is replaced by: the host computer distributes the packaging data to corresponding PCIe signal processing cards through PCIe slots according to the board card ID contained in the packaging head of the packaging data, each co-processing card transmits the received data to the main card through the inter-board data high-speed cascade channel, and the main card conducts data recombination according to the board card ID and the data block serial number and then transmits the data.

6. The cascading method for PCIe signal processing cards as claimed in claim 4 or 5, wherein the FMC daughter card is a DA chip.

Technical Field

The invention relates to the technical field of signal processing, in particular to a cascade method of PCIe signal processing cards.

Background

PCIe-based signal processing cards typically use FPGAs to implement hardware acceleration of signal processing. The application scene is mainly inserted into a PCIe slot of a PC or a server, and after external medium radio frequency signal hardware is operated and processed, a data result is transmitted to a CPU of an upper computer for further processing. Such as hardware DDC or FFT operations for multiple channels within a particular FPGA device. But are limited by two limitations that result in the acceleration capability of a single PCIe card to the ceiling: (1) limited by FPGA resources. When logic resources in the FPGA are used up, hardware acceleration of more channels cannot be realized, and only larger-scale FPGA devices can be replaced. (2) Limited by the PCIe connection bandwidth. The bandwidth of the PCIe3.0X 8 or X16 of a single card is limited to the maximum, while some FPGAs or upper computers do not support the PCIe4.0, and when more hardware channels need to communicate with the upper computers, the bandwidth bottleneck always exists.

Disclosure of Invention

The invention aims to provide a cascade method of PCIe signal processing cards, which is used for solving the problem that in the prior art, when an FPGA is adopted to realize the acceleration of PCIe signal processing hardware, the acceleration of the PCIe signal processing hardware is limited by FPGA resources, and further solving the problem that the acceleration capability of a single PCIe card is limited due to the PCIe connection bandwidth limitation.

The invention solves the problems through the following technical scheme:

a cascading method of PCIe signal processing cards comprises the following steps:

step S110, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S120, the FMC sub-card sends the collected original data to the main card, the main card divides and packages the original data to obtain packaged data, and a packaging head of the packaged data comprises a board card ID corresponding to a PCIe signal processing card;

and S130, the main card transmits all data to the co-processing cards through the inter-board data high-speed cascade channel, each co-processing card processes all original data and transmits processing result data to the upper computer, and the upper computer recombines the received processing result data.

Each PCIe signal processing card is communicated with the upper computer. In a computer system, each PCIe signal processing card occupies one PCIe slot. In the acquisition scene, when the problem that FPGA hardware resources of the PCIe signal processing card are insufficient is solved, the main card needs to transmit all original data, and the downstream cascade board card can perform hardware accelerated calculation only by using all the original data. Each PCIe signal processing card uses all original data and carries out independent hardware accelerated computation according to the configuration of the PCIe signal processing card, and a plurality of PCIe signal processing cards carry out parallel processing and independently transmit processing result data to an upper computer.

A cascading method of PCIe signal processing cards comprises the following steps:

step S210, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S220, the FMC sub-card sends the collected original data to the main card, the main card fragments and encapsulates the original data to obtain encapsulated data, and an encapsulation head of the encapsulated data comprises information such as a board card ID and a data block serial number corresponding to a PCIe signal processing card;

step S230: the main card takes out the data corresponding to the ID of the own board card from the encapsulated data, then sends the rest data to the co-processing card of the next stage through the inter-board data high-speed cascade channel, and similarly, the co-processing card takes out the data corresponding to the ID of the own board card and then sends other data to the co-processing card of the next stage through the inter-board data high-speed cascade channel, and so on until each co-processing card receives the data corresponding to the ID of the own board card;

and each PCIe signal processing card uploads data corresponding to the own board card ID to the upper computer through a PCIe slot, and the upper computer recombines all received data according to the information such as the board card ID, the data block serial number and the like.

The method is applied to solving the problem of insufficient PCIe transmission bandwidth with an upper computer in an acquisition scene, under the condition, the main card can only transmit partial data, the main card takes out the data corresponding to the own board card ID from the packaging data, and uploads the data matched with the own board card ID to the upper computer through a PCIe slot, so that the partial data does not need to be transmitted to a downstream cascaded PCIe signal processing card, and only the residual data is transmitted to a next-stage co-processing card through an inter-board data high-speed cascade channel; similarly, after the data corresponding to the own board card ID is taken out by the co-processing card, the data is uploaded to the upper computer through the PCIe slot, other data is sent to the co-processing card of the next stage through the inter-board data high-speed cascade channel, and so on until each co-processing card receives the data corresponding to the own board card ID and independently uploads the data to the upper computer, and the upper computer recombines all the received data according to the information such as the board card ID, the data block serial number and the like.

Under the collection scene, the FMC daughter card is an AD chip or at least comprises the AD chip.

A cascading method of PCIe signal processing cards comprises the following steps:

step S310, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S320: the upper computer fragments and encapsulates the modulated data to obtain encapsulated data, and an encapsulation head of the encapsulated data comprises information such as a board card ID and a data block serial number corresponding to the PCIe signal processing card;

step S330: the upper computer sends all the packaged data to each PCIe signal processing card through a PCIe slot, an FPGA on each PCIe signal processing card processes the data according to respective setting and carries out fragment packaging on the modulation result data according to the position of the FPGA in a stacking scene, each co-processing card sends the data to the main card, and the main card carries out data recombination according to information such as the ID of the board card, the serial number of the data block and the like and then sends the data out.

The method is applied to the problem that the FPGA hardware resources of the PCIe signal processing card are insufficient in a transmitting scene, in this case, an upper computer needs to send out modulation data, the modulation function needs to be realized on the FPGA, different hardware acceleration channels are needed to process different modulation parameters at the moment, therefore, the upper computer sends all data to each PCIe signal processing card through a PCIe slot, the FPGA on each PCIe signal processing card processes the received data according to respective setting (each acceleration channel can have different parameter setting), and carries out fragment packaging on the modulation result data according to the position of the upper computer in a cascading and stacking scene, and the packaged data also contains information such as a board card ID and a serial number. On the inter-board high-speed cascade channel, the fragment data of a plurality of PCIe signal processing cards may be transmitted in an interleaving manner at the same time. After receiving the data, the master card classifies and recombines the data transmitted by each PCIe signal processing card, obtains complete modulation data from a plurality of processing cards after recombination, and transmits the complete modulation data through the FMC daughter card.

A cascading method of PCIe signal processing cards comprises the following steps:

step S410, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S420: the upper computer divides the modulated data into pieces and packages the modulated data to obtain packaged data, and a packaging head of the packaged data comprises a board card ID corresponding to the PCIe signal processing card;

step S430: the upper computer distributes the packaging data to corresponding PCIe signal processing cards through PCIe slots according to information such as board card IDs, serial numbers and the like contained in packaging heads of the packaging data, each co-processing card transmits the received data to the main card through the inter-board data high-speed cascade channel, and the main card conducts data recombination and then transmits the data.

When the upper computer needs to send the modulated data to the PCIe card for transmission, but the data volume is very large, and the bandwidth of a PCIe channel of a single PCIe card is insufficient, then the data slice needs to be packaged into a header, the header corresponds to the ID of each processing card and the serial number of the data block, as shown in FIG. 3.

The PCIe signal processing card is applied to a transmitting scene, the problem that PCIe transmission bandwidth is insufficient with an upper computer is solved, the upper computer needs to send modulated data out, but PCIe transmission bandwidth of a single PCIe signal processing card is not high enough and cannot be transmitted, at the moment, the co-processing card is used for sharing transmission, at the moment, the upper computer fragments data, each piece of data has information such as a board card ID and a serial number, the corresponding fragments are sent to the corresponding PCIe signal processing card through the corresponding PCIe slots, after the cards are received, the data are transmitted to a main card through a high-speed cascade channel between the boards, and the main card conducts data recombination according to the board card ID, the serial number and the like.

Under the transmitting scene, the FMC daughter card is a DA chip or at least comprises the DA chip.

When the device size or the PCIe transmission bandwidth is increased, in some scenarios, there is a waste when not using so many resources or bandwidths. According to the invention, the board card ID of the PCIe signal processing card is added in the packaging head, and a plurality of cards are needed in a use scene according to the number of the acceleration channels of the FPGA or the PCIe bandwidth, so that the plurality of cards are cascaded, and the whole cascade channel is elastically expanded.

Compared with the prior art, the invention has the following advantages and beneficial effects:

(1) the invention solves the problem that the PCIe signal processing hardware is limited by FPGA resources when the acceleration is realized by adopting FPGA in the prior art, and when more hardware acceleration channels are needed, the number of processing cards is only needed to be selected through the card ID of the packaging head; and a cascade channel with elastic expansion is realized; the method is suitable for acquisition scenes and emission scenes.

(2) The problem that the acceleration capability of a single PCIe card is limited due to the limitation of the connection bandwidth of PCIe is solved, each processing card only processes the database matched with the ID of the own card and independently uploads the database to a main card or an upper computer, and the bottleneck that the acceleration capability of the single PCIe card is limited is broken through; and a cascade channel with elastic expansion is realized; the method is suitable for acquisition scenes and emission scenes.

Drawings

FIG. 1 is a schematic block diagram of a first embodiment of the present invention;

FIG. 2 is a schematic block diagram of a first embodiment of the present invention;

FIG. 3 is a flow chart of an acquisition scenario implementing hardware acceleration.

Detailed Description

The present invention will be described in further detail with reference to examples, but the embodiments of the present invention are not limited thereto.

Example 1:

referring to fig. 1 and fig. 3, a method for cascading PCIe signal processing cards includes:

step S110, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S120, the FMC daughter card sends the acquired original data to the master card, the master card fragments and encapsulates the original data to obtain encapsulated data, an encapsulating head of the encapsulated data comprises information such as a board card ID and a serial number corresponding to the PCIe signal processing card, and a data format of the encapsulating head is shown in the following table:

and S130, the main card transmits all data to the co-processing cards through the inter-board data high-speed cascade channel, each co-processing card processes all original data and performs independent operation, processing result data is transmitted to the upper computer, and the upper computer recombines the received processing result data.

Each PCIe signal processing card is communicated with the upper computer. In a computer system, each PCIe signal processing card occupies one PCIe slot. In the acquisition scene, when the problem that FPGA hardware resources of the PCIe signal processing card are insufficient is solved, the main card needs to transmit all original data, and the downstream cascade board card can perform hardware accelerated calculation only by using all the original data. Each PCIe signal processing card only processes hardware acceleration tasks related to the own card configuration information (each project or application has own calculation method, such as FFT, DDC and the like in the field of digital signal processing, parameters of each FFT or DDC channel are different, and calculation results are different), a plurality of PCIe signal processing cards perform parallel processing, and processing result data are independently transmitted to an upper computer. For a hardware acceleration scenario, the work done by each card is independent, for example, the work done by the FPGA of each card is independent for the primary card to complete FFT channels 1-16, the first co-processor card to complete channels 17-32, the second manuscript co-processor card to complete channels 33-48, and … …. Since these tasks consume hardware resources, a single card may not be able to do so with multiple channels, and thus the tasks are distributed to multiple cards. Because the work of each card is independent and the result is complete, the operation result transmitted to the upper computer is independent.

Example 2:

a cascading method of PCIe signal processing cards comprises the following steps:

step S210, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S220, the FMC sub-card sends the collected original data to the main card, the main card fragments and encapsulates the original data to obtain encapsulated data, and an encapsulation head of the encapsulated data comprises information such as a board card ID and a data block serial number corresponding to a PCIe signal processing card;

step S230: the main card takes out the data corresponding to the ID of the own board card from the encapsulated data, then sends the rest data to the co-processing card of the next stage through the inter-board data high-speed cascade channel, and similarly, the co-processing card takes out the data corresponding to the ID of the own board card and then sends other data to the co-processing card of the next stage through the inter-board data high-speed cascade channel, and so on until each co-processing card receives the data corresponding to the ID of the own board card;

and each PCIe signal processing card uploads data corresponding to the own board card ID to the upper computer through a PCIe slot, and the upper computer recombines all received data according to the board card ID and the data block serial number.

The method is applied to solving the problem of insufficient PCIe transmission bandwidth with an upper computer in an acquisition scene, under the condition, the main card can only transmit partial data, the main card takes out the data corresponding to the ID of the own board card from the encapsulated data, and uploads the data matched with the ID of the own board card to the upper computer through a PCIe slot, so that the partial data does not need to be transmitted to a downstream cascaded PCIe co-processing card, and only the residual data is transmitted to a next-stage co-processing card through an inter-board data high-speed cascade channel; similarly, after the data corresponding to the own board card ID is taken out by the co-processing card, the data is uploaded to the upper computer through the PCIe slot, and other data is sent to the co-processing card of the next stage through the inter-board data high-speed cascade channel, and so on until each co-processing card receives the data corresponding to the own board card ID and independently uploads the data to the upper computer, and the upper computer reassembles all the received data.

In the acquisition scene, the FMC daughter card is an AD chip (analog-to-digital chip) or at least includes an AD chip.

Example 3:

referring to fig. 2, a method for cascading PCIe signal processing cards includes:

step S310, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S320: the upper computer fragments and encapsulates the modulated data to obtain encapsulated data, and an encapsulation head of the encapsulated data comprises information such as a board card ID and a data block serial number corresponding to the PCIe signal processing card;

step S330: the upper computer sends all the packaged data to each PCIe signal processing card through a PCIe slot, an FPGA on each PCIe signal processing card processes the data according to respective setting and carries out fragment packaging on the modulation result data according to the position of the FPGA in a stacking scene, each co-processing card sends the data to the main card, and the data is sent out after being recombined by the main card.

The method is applied to the problem that the FPGA hardware resources of the PCIe signal processing card are insufficient in a transmitting scene, in this case, an upper computer needs to send out modulation data, the modulation function needs to be realized on the FPGA, different hardware acceleration channels are needed to process different modulation parameters at the moment, therefore, the upper computer sends all data to each PCIe signal processing card through a PCIe slot, the FPGA on each PCIe signal processing card processes the received data according to respective setting (each acceleration channel can have different parameter setting), and carries out fragment packaging on the modulation result data according to the position of the upper computer in a cascading and stacking scene, and the packaged data also contains information such as a board card ID and a serial number. On the inter-board high-speed cascade channel, the fragment data of a plurality of PCIe signal processing cards may be transmitted in an interleaving manner at the same time. After receiving the data, the master card classifies and recombines the data transmitted by each PCIe signal processing card, obtains complete modulation data from a plurality of processing cards after recombination, and transmits the complete modulation data through the FMC daughter card.

Example 4:

a cascading method of PCIe signal processing cards comprises the following steps:

step S410, a plurality of same PCIe signal processing cards are cascaded through the inter-board data high-speed cascade channel, wherein one PCIe signal processing card is connected with the FMC daughter card to serve as a main card, the other PCIe signal processing cards serve as co-processing cards, and each PCIe signal processing card is in communication connection with an upper computer through a PCIe slot;

step S420: the upper computer fragments and encapsulates the modulated data to obtain encapsulated data, and an encapsulation head of the encapsulated data comprises information such as a board card ID and a data block serial number corresponding to the PCIe signal processing card;

step S430: the upper computer distributes the packaging data to the corresponding PCIe signal processing cards through PCIe slots according to the board card ID contained in the packaging head of the packaging data, each co-processing card transmits the received data to the main card through the inter-board data high-speed cascade channel, and the main card conducts data recombination and then transmits the data.

When the upper computer needs to send the modulated data to the PCIe card for transmission, but the data volume is very large, and the bandwidth of the PCIe channel of a single PCIe card is insufficient, then the data slice needs to be packaged into a header, information such as the ID of each processing card and the serial number of a data block corresponds to the header, as shown in FIG. 3.

The PCIe signal processing card is applied to a transmitting scene, the problem that PCIe transmission bandwidth is insufficient with an upper computer is solved, the upper computer needs to send modulated data out, but PCIe transmission bandwidth of a single PCIe signal processing card is not high enough and cannot be transmitted, at the moment, the co-processing card is used for sharing transmission, at the moment, the upper computer fragments data, each piece of data has information such as a board card ID and a serial number, the corresponding fragments are sent to the corresponding PCIe signal processing card through the corresponding PCIe slots, after the cards are received, the data are transmitted to a main card through a high-speed cascade channel between the boards, and the main card conducts data recombination according to the board card ID, the serial number and the like.

For the scenario of solving the PCIe transmission bandwidth, if the upper computer CPU completes the modulation of the signal channel through software, only the transmission problem needs to be solved, and as the PCIe bandwidth of a single card or several cards is insufficient, such much data cannot be transmitted, at this time, more co-processing cards are needed. The host computer just needs to carry out the fragmentation to data, encapsulates, and the sign is handled by which card this piece of data is, transmits to corresponding PCIe signal processing card through corresponding PCIe slot like this on, PCIe signal processing card is again with data through the main card of inter-plate high speed cascade channel transmission. Because the inter-board high-speed cascade channel is that the fragmented data of a plurality of boards may be transmitted at the same time, all the data are interlaced together, and the sequence of the data arriving at the main card may be out of order, the main card is required to recombine the data gathered by the cards.

In the transmitting scene, the FMC daughter card is a DA chip (analog-to-digital chip) or at least comprises the DA chip.

The acquisition and the transmission are two independent scenes and are inverse processes, when the FMC daughter card simultaneously comprises an AD chip and a DA chip, the acquisition and the transmission exist in the same product. The data flow direction can be selected according to the scene.

Although the present invention has been described herein with reference to the illustrated embodiments thereof, which are intended to be preferred embodiments of the present invention, it is to be understood that the invention is not limited thereto, and that numerous other modifications and embodiments can be devised by those skilled in the art that will fall within the spirit and scope of the principles of this disclosure.