Network controller and method of network control
US-2025219896-A1 · Jul 3, 2025 · US
US12487960B2 · US · B2
| Field | Value |
|---|---|
| Publication number | US-12487960-B2 |
| Application number | US-202218571192-A |
| Country | US |
| Kind code | B2 |
| Filing date | Jun 1, 2022 |
| Priority date | Dec 9, 2021 |
| Publication date | Dec 2, 2025 |
| Grant date | Dec 2, 2025 |
A practical reading order for non-experts. Skip the full description unless you need deep technical detail.
What the patent document calls the invention.
A short plain-language summary of the technical disclosure.
Who owns or filed the patent and who is credited as inventor.
Filing, priority, publication, and grant dates set the timeline.
The legal scope of protection — read this for what is actually claimed.
Technology tags used to group this patent with similar filings.
Prior art links and similar publications in this corpus.
Official abstract text for this publication.
Provided are a communication method and system for a distributed heterogeneous acceleration platform, a device and a medium. The method includes: after starting a collaborative acceleration task, determining, by a first target heterogeneous acceleration card in a distributed heterogeneous acceleration platform, a second target heterogeneous acceleration card from the distributed heterogeneous acceleration platform by querying an information table corresponding to the collaborative acceleration task; generating, by the first target heterogeneous acceleration card, a target data packet according to a predefined data packet format, and sending the target data packet to the second target heterogeneous acceleration card via a PCIE interface; and parsing, by the second target heterogeneous acceleration card, the target data packet according to the data packet format, and executing a corresponding read operation or write operation according to a parsing result, so as to complete the collaborative acceleration task.
Opening claim text (preview).
What is claimed is: 1 . A communication method for a distributed heterogeneous acceleration platform, comprising: after starting a collaborative acceleration task, determining, by a first target heterogeneous acceleration card in a distributed heterogeneous acceleration platform, a second target heterogeneous acceleration card from the distributed heterogeneous acceleration platform by querying an information table corresponding to the collaborative acceleration task; generating, by the first target heterogeneous acceleration card, a target data packet according to a predefined data packet format, and sending the target data packet to the second target heterogeneous acceleration card via a PCIE interface; and parsing, by the second target heterogeneous acceleration card, the target data packet according to the data packet format, and executing a corresponding read operation or write operation according to a parsing result, so as to complete the collaborative acceleration task. 2 . The communication method for a distributed heterogeneous acceleration platform according to claim 1 , wherein before determining, by the first target heterogeneous acceleration card in a distributed heterogeneous acceleration platform, a second target heterogeneous acceleration card from the distributed heterogeneous acceleration platform by querying an information table corresponding to the collaborative acceleration task, the method further comprises: configuring a respective information table for each heterogeneous acceleration card in the distributed heterogeneous acceleration platform according to task information corresponding to the collaborative acceleration task, wherein the information table comprises a board card lookup table and a collaborative acceleration parameter table; and after the information table is configured, initiating a collaborative acceleration task startup command to the distributed heterogeneous acceleration platform. 3 . The communication method for a distributed heterogeneous acceleration platform according to claim 2 , wherein configuring a board card lookup table for each heterogeneous acceleration card in the distributed heterogeneous acceleration platform comprises: acquiring identity information allocated by a PCIE system to each heterogeneous acceleration card, wherein the identity information comprises a bus number, a device number and a function number; setting a unique identification number for each heterogeneous acceleration card in the distributed heterogeneous acceleration platform, and obtaining parameter information corresponding to each heterogeneous acceleration card on the basis of the identity information and the unique identification number; and configuring the board card lookup table for each heterogeneous acceleration card on the basis of the parameter information corresponding to all heterogeneous acceleration cards. 4 . The communication method for a distributed heterogeneous acceleration platform according to claim 3 , wherein configuring a collaborative acceleration parameter table for each heterogeneous acceleration card in the distributed heterogeneous acceleration platform comprises: configuring a corresponding collaborative acceleration information table for each heterogeneous acceleration card according to task information corresponding to the collaborative acceleration task, wherein the collaborative acceleration information table comprises participation information corresponding to the current collaborative acceleration task, startup information and a unique identification number of an associated heterogeneous acceleration card. 5 . The communication method for a distributed heterogeneous acceleration platform according to claim 1 , wherein generating a target data packet according to a predefined data packet format comprises: performing packaging according to a processing layer data packet format of PCIE, and filling payload fields in the processing layer data packet format according to the predefined data packet format to configure a source ID field, a target ID field, an operation type field, an operation address field, a data size field and a data field, so as to obtain the target data packet; wherein the source ID field is used for storing a unique identification number of the first target heterogeneous acceleration card, the target ID field is used for storing a unique identification number of the second target heterogeneous acceleration card, and the operation type comprises a read operation and a write operation. 6 . The communication method for a distributed heterogeneous acceleration platform according to claim 5 , wherein parsing the target data packet according to the data packet format, and executing a corresponding read operation or write operation according to a parsing result comprise: parsing the payload fields of the target data packet according to the data packet format; when parameter information stored in the target ID field in the payload fields is consistent with a local unique identification number, determining an operation type corresponding to the operation type field in the payload fields; and according to the operation address field, the data size field and the data field in the payload fields, executing a corresponding operation according to an operation mode corresponding to the operation type. 7 . The communication method for a distributed heterogeneous acceleration platform according to claim 6 , wherein according to the operation address field, the data size field and the data field in the payload fields, executing a corresponding operation according to an operation mode corresponding to the operation type comprises: when the operation type is a write operation, determining a first starting address of write according to the operation address field, determining a first data length according to the data size field, and then locally writing the content contained in the data field according to the first starting address and the first data length; and when the operation type is a read operation, determining a second starting address of read according to the operation address field, determining a second data length according to the data size field, then reading locally corresponding data according to the second starting address and the second data length to generate a processing layer data packet, and sending the processing layer data packet to the first target heterogeneous acceleration card. 8 . A communication system for a distributed heterogeneous acceleration platform, comprising: a first target heterogeneous acceleration card, configured to determine a second target heterogeneous acceleration card from a distributed heterogeneous acceleration platform by querying an information table corresponding to a collaborative acceleration task after starting the collaborative acceleration task, generate a target data packet according to a predefined data packet format, and send the target data packet to the second target heterogeneous acceleration card via a PCIE interface; and the second target heterogeneous acceleration card, configured to parse the target data packet according to the data packet format, and execute a corresponding read operation or write operation according to a parsing result, so as to complete the collaborative acceleration task. 9 . A computer device, comprising a memory and one or more processors, wherein the memory stores computer readable instructions which, when executed by the one or more processors, cause the one or more processors to: after starting a collaborative acceleration task, determine, by a first target heterogeneous acceleration card in a distributed heterogeneous acceleration platform, a second target heterogen
PCI express · CPC title
the resource being a machine, e.g. CPUs, Servers, Terminals · CPC title
on a serial bus, e.g. I2C bus, SPI bus (on daisy chain buses G06F13/4247) · CPC title
being an input/output bus, e.g. ISA bus, EISA bus, PCI bus, SCSI bus · CPC title
Live connection to bus, e.g. hot-plugging (current or voltage limitation during live insertion H02H9/004) · CPC title
Related publications grouped by family.
Answers are generated from the same data shown on this page.