CN109240972A - A kind of GPU board and the VPX signal processing cabinet using the board - Google Patents

A kind of GPU board and the VPX signal processing cabinet using the board Download PDF

Info

Publication number
CN109240972A
CN109240972A CN201811304660.3A CN201811304660A CN109240972A CN 109240972 A CN109240972 A CN 109240972A CN 201811304660 A CN201811304660 A CN 201811304660A CN 109240972 A CN109240972 A CN 109240972A
Authority
CN
China
Prior art keywords
module
srio
board
vpx
pcie
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201811304660.3A
Other languages
Chinese (zh)
Inventor
程彩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Zhongke Haixun Digital Technology Co Ltd
Original Assignee
Beijing Zhongke Haixun Digital Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Zhongke Haixun Digital Technology Co Ltd filed Critical Beijing Zhongke Haixun Digital Technology Co Ltd
Priority to CN201811304660.3A priority Critical patent/CN109240972A/en
Publication of CN109240972A publication Critical patent/CN109240972A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F13/00Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
    • G06F13/38Information transfer, e.g. on bus
    • G06F13/42Bus transfer protocol, e.g. handshake; Synchronisation
    • G06F13/4282Bus transfer protocol, e.g. handshake; Synchronisation on a serial bus, e.g. I2C bus, SPI bus
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00Data switching networks
    • H04L12/02Details
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/40Constructional details, e.g. power supply, mechanical construction or backplane
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2213/00Indexing scheme relating to interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
    • G06F2213/0002Serial port, e.g. RS232C
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2213/00Indexing scheme relating to interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
    • G06F2213/0026PCI express
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Small-Scale Networks (AREA)

Abstract

A kind of GPU board towards VPX and the VPX signal processing cabinet using the board, which includes: PCIE to SRIO conversion module, CPU module, ethernet switching module and VPX connector.The present invention can satisfy high performance data transmission, process demand, have the advantages that reinforcing, miniaturization, good environmental adaptability simultaneously, it is particularly adapted to higher to size, power consumption, environmental suitability and reinforcement type index request, and needs to realize the application scenarios of high performance parallel information processing.

Description

A kind of GPU board and the VPX signal processing cabinet using the board
Technical field
The present invention relates to a kind of GPU board, especially a kind of GPU board towards VPX framework and the VPX using the board Signal processing cabinet.
Background technique
In fields such as booming sonar, radar and electronic countermeasures, signal handling equipment is usually by several functional modules It is integrated in VPX to reinforce in cabinet, including computing module, SRIO Switching Module, ETH Switching Module etc., passes through the SRIO of VPX back panel High-speed data exchange inside universal serial bus realization.
And the conventional means of signal processing is broadly divided into two classes at present: one kind is the processing platform using CPU as representative, another Class is the array signal processing platform based on large scale integrated chips such as DSP and FPGA.The former is time-consuming serious, real-time Difference;Although the latter can complete real time signal processing, also has that the development cycle is long, board is numerous and at high cost etc. many lack Point.
Summary of the invention
The disclosure provides a kind of GPU board towards VPX structure and the VPX signal processing cabinet using the board, Can be realized it is a kind of calculate real-time, cost, structure, in terms of the VPX signal that more optimizes of comprehensive performance Processing equipment.
The one side of the disclosure provides a kind of GPU board, comprising: PCIE to SRIO conversion module, CPU module, ether Net Switching Module and VPX connector, in which:
PCIE to SRIO conversion module is connected with external SRIO bus, by externally input SRIO data conversion at PCIE data, output to CPU module;
Ethernet switching module is used to provide the communication interface between board and external ethernet;
VPX connector is used to be inserted into the VPX connector in VPX case back plate load slot.
Preferably, PCIE to SRIO conversion module uses Tsi721 chip, the PCIe interface and GPU mould of Tsi721 chip Block connection, SRIO interface are connect with external SRIO bus.
Preferably, above-mentioned CPU module is Jetson TX2GPU module.
Preferably, include on board all the way or the parallel CPU module of two-way, and with the matched PCIE to of each CPU module SRIO conversion module.
Preferably, when board includes two-way CPU module, ethernet switching module provides logical between the two-way CPU module Letter.
It preferably, further include one or more of boundary scan module, IPMI module and forward direction I/O interface, In:
Boundary scan module is used to carry out meeting IEEE1149.1 specification to PCIE to SRIO conversion module and CPU module Jtag boundary sweep test;
IPMI module is monitored the parameters such as the temperature of board, voltage, electric current by IPMB bus, and support excess temperature, Over-voltage, overcurrent protection;
Forward direction I/O interface is for providing the interface of board and external debugging device, status monitor device etc..
Another aspect of the present disclosure provides a kind of VPX signal processing cabinet using the GPU board, comprising: power slot, SRIO exchange groove, network exchange slot and load slot, in which:
Power module is installed in power slot, for being each module for power supply in cabinet;
SRIO Switching Module is equipped in SRIO exchange groove, for the data exchange between SRIO bus in cabinet;
Ethernet switching module is equipped in network exchange slot, for provide in cabinet the data exchange of gigabit Ethernet and Ten thousand external mbit ethernet interfaces;
Common calculation module is equipped in load slot or above-mentioned GPU board, GPU board provide accelerometer for common calculation module Calculate service;
Each load slot draws 2 road 4x SRIO high speed data links to SRIO exchange groove.
Preferably, above-mentioned common calculation module is DSP module or CPU module.
The disclosure make full use of GPU in recent years floating-point operation, in terms of the potentiality that show and GPU open The reduction for degree of raising difficult questions provides and a kind of turns what GPU was used towards VPX framework, by the SRIO bus interface used in VPX cabinet The GPU board of PCIE interface, so that successfully GPU is added in existing VPX signal handling equipment, VPX framework and GPU is parallel Calculating advantage combines, and improves the performance of existing VPX signal handling equipment.Compared with prior art, the beneficial effect of the disclosure Fruit is: can satisfy the transmission of high performance data, process demand, while have reinforcing, miniaturization, good environmental adaptability it is excellent Point, it is higher to size, power consumption, environmental suitability and reinforcement type index request to be particularly adapted to sonar, radar, electronic countermeasure etc., And need to realize the application scenarios of high performance parallel information processing, practical application value with higher.
Detailed description of the invention
Disclosure exemplary embodiment is described in more detail in conjunction with the accompanying drawings, the disclosure it is above-mentioned and other Purpose, feature and advantage will be apparent, wherein in disclosure exemplary embodiment mode, identical reference label Typically represent same parts.
Fig. 1 is the signal flow schematic diagram of the GPU board of exemplary embodiment;
Fig. 2 is the concrete structure diagram of the GPU board of exemplary embodiment;
Fig. 3 is the structural schematic diagram of the Tsi721 chip in the GPU board of exemplary embodiment;
Fig. 4 is the topological schematic diagram of the VPX signal processing cabinet using above-mentioned GPU board of exemplary embodiment.
Specific embodiment
Preferred embodiment of the present disclosure is more fully described below with reference to accompanying drawings.Although showing the disclosure in attached drawing Preferred embodiment, however, it is to be appreciated that may be realized in various forms the disclosure without that should be limited by embodiments set forth here System.On the contrary, thesing embodiments are provided so that the disclosure is more thorough and complete, and can be complete by the scope of the present disclosure Ground is communicated to those skilled in the art.
It as depicted in figs. 1 and 2, include: that CPU module, PCIE turn SRIO on the GPU board in disclosure exemplary embodiment Chip, ethernet switching module and VPX Bussing connector, in which: PCIE to SRIO conversion module and external SRIO are total Line is connected, by the SRIO data conversion of input at PCIE data, output to CPU module;Ethernet switching module is for providing plate Communication interface between card and external ethernet;Backward interface of the VPX Bussing connector as board, for board and VPX machine The connection of case backboard.
Preferably, in the present embodiment, CPU module uses Jetson TX2 module, and PCIE turns SRIO chip and adopts With Tsi721 chip, in which:
Jetson TX2 module is the supercomputer module an of very-high performance, low-power consumption, and typical power consumption only has 7.5W, Size only has credit card-sized, integrates the ARMv8 processor of 256 core NVIDA Pascal GPU and 6 core 64, possesses 128 memories of 8GB LPDDR4, performance have been more than Intel Xeon server level CPU;
Tsi721 is dedicated SRIO, PCIE conversion chip, as shown in figure 3, it includes PCIe interface, SRIO interface, disappears Several main functional modules such as engine, mapping engine and block DMA engine are ceased, wherein connect by PCIe interface with CPU module, It is connected to the network through SRIO interface and SRIO;Message engine realizes the message communicating between PCIe interface and SRIO interface, includes 8 tunnels Outbound Message DMA channel and 8 road Inbound Message DMA channels;Mapping engine realizes that PCIe interface arrives SRIO interface IP address converts (PC2SR) and SRIO interface to PCIe interface address conversion (SR2PC);Block DMA engine realizes the side DMA Formula initiates the read and write access of data packet or maintenance packet;Each channel in 8 DMA channels work in a manner of chain type descriptor Make, initiates RapidIO read-write operation for master control.
Preferably, the Ethernet switching chip in the present embodiment on GPU board uses gigabit networking and exchanges core Piece ETH switch, PHY1, PHY2 function phase of the exchange chip through port physical layer chip PHY connection Ethernet interface, in figure Together, only contained port number is different, hangs on same switch, and PHY1 has 2 port, is connected on VPX connector P4, PHY2 has 4 port, in connecting plate and forward direction debugging serial interface.
Preferably, the parallel CPU module in 2 tunnels, each CPU module be placed on the GPU board in the present embodiment There is matching PCIE to turn SRIO chip, acceleration operation can be carried out to the 2 road SRIO data from VPX connector respectively. Preferably, interconnection is realized by ETH switch chip between 2 pieces of CPU modules.
Preferably, the GPU board in the present embodiment also includes baseboard management controller BMC, for realizing IPMI Function: BMC is acquired according to the order of host computer and (I2C i.e. on the VPX case back plate inserted of board is total by IPMB bus Line), the parameters such as the temperature of board, voltage, electric current are sent to host computer, thus realize the monitoring to board physical state, and Support excess temperature, over-voltage, overcurrent protection.
Preferably, the GPU board in the present embodiment further includes a variety of I/O interfaces and its interface chip, wherein before It include RJ45, serial ports etc. to interface, convenient for the monitoring of debugging and state to board.
Using the topological structure of the VPX signal processing cabinet of above-mentioned GPU board as shown in figure 4, including electricity in the present embodiment Source slot, SRIO exchange groove, network exchange slot and load slot, in which:
Power module is installed in power slot, for being each module for power supply in cabinet;
SRIO Switching Module is equipped in SRIO exchange groove, for the data exchange between SRIO bus in cabinet;
Ethernet switching module is equipped in network exchange slot, for provide in cabinet the data exchange of gigabit Ethernet and Ten thousand external mbit ethernet interfaces;
It loads and is equipped with common calculation module or above-mentioned GPU board in slot, each load slot draws 2 road 4x SRIO high speed numbers According to link to the SRIO exchange groove.
The quantity of various slots is all made of the regulation of VPX standard (VITA65).
Preferably, above-mentioned common calculation module includes DSP module or CPU module in the present embodiment;Load slot Middle insertion DSP, CPU, the type of CPU module and quantity are depending on the specific needs of user.
In addition, each load slot also draws 2 road I2C buses in the present embodiment, for each module temperature, voltage, electric current etc. Monitoring.
In the present embodiment, high speed SRIO data are passed to the GPU board by SRIO Switching Module by common calculation module, by Dedicated conversion chip Tsi721 by SRIO data conversion at PCIE data needed for CPU module, to enable the GPU board Acceleration enough, which is provided, for common calculation module calculates service.
As it can be seen that a kind of GPU board for turning PCIE towards VPX framework, by SRIO is present embodiments provided, so that GPU be added Enter in existing VPX signal handling equipment, provides acceleration for common calculation module and calculate service.The present embodiment can satisfy height The data transmission of performance, process demand, while having the advantages that reinforcing, miniaturization, good environmental adaptability, it is particularly adapted to sound Receive, radar, electronic countermeasure etc. it is higher to size, power consumption, environmental suitability and reinforcement type index request, and need to realize high-performance The application scenarios of parallel information processing, practical application value with higher.
Above-mentioned technical proposal is exemplary embodiment of the present invention, for those skilled in the art, at this On the basis of disclosure of the invention application method and principle, it is easy to make various types of improvement or deformation, be not limited solely to this Method described in above-mentioned specific embodiment is invented, therefore previously described mode is only preferred, and it is not restrictive Meaning.

Claims (8)

1. a kind of GPU board characterized by comprising PCIE to SRIO conversion module, CPU module, Ethernet exchanging mould Block and VPX connector, in which:
The PCIE to SRIO conversion module is connected with external SRIO bus, by externally input SRIO data conversion at PCIE data, output to the CPU module;
The ethernet switching module is used to provide the described the communication interface between board and external ethernet;
The VPX connector is used to be inserted into the VPX connector in VPX case back plate load slot.
2. GPU board according to claim 1, which is characterized in that the PCIE to SRIO conversion module uses The PCIe interface of Tsi721 chip, the Tsi721 chip is connect with the CPU module, SRIO interface and external SRIO bus Connection.
3. GPU board according to claim 1, which is characterized in that the CPU module is Jetson TX2GPU module.
4. GPU board according to claim 1, which is characterized in that including all the way or the parallel CPU module of two-way, Yi Jiyu Each matched PCIE to SRIO conversion module of CPU module.
5. GPU board according to claim 4, which is characterized in that the board include two-way CPU module when, it is described with Too net Switching Module provides the communication between the two-way CPU module.
6. GPU board according to claim 1, which is characterized in that further include boundary scan module, IPMI module and One or more of forward direction I/O interface, in which:
The boundary scan module to the PCIE to SRIO conversion module and CPU module for carrying out meeting IEEE1149.1 The jtag boundary sweep test of specification;
The IPMI module is monitored the parameters such as the temperature of the board, voltage, electric current by IPMB bus, and supports Temperature, over-voltage, overcurrent protection;
The forward direction I/O interface is used to provide the described the interface of board and external debugging device, status monitor device etc..
7. a kind of VPX signal processing cabinet using GPU board described in any one of claims 1 to 6, including power slot, SRIO exchange groove, network exchange slot and load slot, in which:
It is installed with power module in the power slot, for being each module for power supply in cabinet;
SRIO Switching Module is equipped in the SRIO exchange groove, for the data exchange between SRIO bus in cabinet;
Ethernet switching module is equipped in the network exchange slot, for provide in cabinet the data exchange of gigabit Ethernet and Ten thousand external mbit ethernet interfaces;
Common calculation module or the GPU board are equipped in the load slot, the GPU board provides for common calculation module to be added Speed calculates service;
Each load slot draws 2 road 4x SRIO high speed data links to the SRIO exchange groove.
8. VPX signal processing cabinet according to claim 7, which is characterized in that the common calculation module is DSP module Or CPU module.
CN201811304660.3A 2018-11-04 2018-11-04 A kind of GPU board and the VPX signal processing cabinet using the board Pending CN109240972A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811304660.3A CN109240972A (en) 2018-11-04 2018-11-04 A kind of GPU board and the VPX signal processing cabinet using the board

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811304660.3A CN109240972A (en) 2018-11-04 2018-11-04 A kind of GPU board and the VPX signal processing cabinet using the board

Publications (1)

Publication Number Publication Date
CN109240972A true CN109240972A (en) 2019-01-18

Family

ID=65076718

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811304660.3A Pending CN109240972A (en) 2018-11-04 2018-11-04 A kind of GPU board and the VPX signal processing cabinet using the board

Country Status (1)

Country Link
CN (1) CN109240972A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110708185A (en) * 2019-09-03 2020-01-17 中国科学院计算技术研究所 Data interconnection method, system, chip and device for artificial intelligence processor
CN113076066A (en) * 2021-04-14 2021-07-06 湖南兴天电子科技有限公司 High-capacity high-speed storage device and operation method thereof

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110708185A (en) * 2019-09-03 2020-01-17 中国科学院计算技术研究所 Data interconnection method, system, chip and device for artificial intelligence processor
CN110708185B (en) * 2019-09-03 2021-06-29 中国科学院计算技术研究所 Data interconnection method, system, chip and device for artificial intelligence processor
CN113076066A (en) * 2021-04-14 2021-07-06 湖南兴天电子科技有限公司 High-capacity high-speed storage device and operation method thereof
CN113076066B (en) * 2021-04-14 2023-12-08 湖南兴天电子科技股份有限公司 High-capacity high-speed storage device and operation method thereof

Similar Documents

Publication Publication Date Title
CN103107960B (en) The method and system of the impact of exchange trouble in switching fabric is reduced by switch card
CN104348673B (en) A kind of method of commissioning, master control borad and business board
CN112285530A (en) Universal testing device for VPX high-speed signal board
CN208188815U (en) BMC module system
CN209044488U (en) A kind of super fusion calculation platform that can configure based on VPX framework
CN106649021A (en) Testing device for PCIe slave device
CN109240972A (en) A kind of GPU board and the VPX signal processing cabinet using the board
CN212135408U (en) Board card bus data transmission test system
CN104641593B (en) Web plate and communication equipment
CN203133691U (en) Server computation node based on CPCI framework
CN208834296U (en) A kind of GPU board and the VPX signal processing cabinet using the board
CN109407574A (en) Output-controlling device and its method may be selected in a kind of multibus
CN209072526U (en) Ethernet exchanging device
CN201876870U (en) Test system for internal integrated circuit bus and device for internal integrated circuit bus
CN209328011U (en) Fly control communication system
CN208000578U (en) A kind of blade type data processing equipment
CN110022255A (en) The ubiquitous electric power platform of internet of things of modularization based on mixing communication network data exchange
CN103544133B (en) Conversion device and conversion method
CN106484656B (en) A kind of management board of collectable multinode management information
CN102495554A (en) Program controlled matrix switch and manufacturing method thereof
CN207408820U (en) A kind of portable intelligent transformer station secondary system commissioning equipment
CN111490912A (en) FPGA-based multichannel signal transmission reliability detection system and method
CN216286659U (en) High-speed data processing equipment based on homemade VPX framework
CN2182427Y (en) Double-unit tolerant apparatus for computer
CN213094226U (en) CAN communication device based on CPCI bus

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20190118

WD01 Invention patent application deemed withdrawn after publication