Dynamically Reconfigurable Networks Using Programmable Integrated Circuits

The programmable IC with static and dynamic regions addresses the inefficiency of recompilation by allowing dynamic reconfiguration of user-specified circuits, ensuring uninterrupted network operation and reduced compilation times.

JP7592066B2Active Publication Date: 2024-11-29XILINX INC
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2022507696
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2019-08-11
Filing Date
2020-08-10
Publication Date
2024-11-29
Estimated Expiration
2040-08-10

AI Technical Summary

Technical Problem

Existing programmable ICs require recompilation and downtime for adding or changing user-specified circuits, which is time-consuming and inefficient, especially in dynamic network environments.

Method used

A programmable IC with a static and dynamic region, where the static region provides basic NIC functionality and the dynamic region allows for dynamic reconfiguration of user-specified circuits without interrupting network traffic, enabling runtime insertion and removal of user-specified circuitry.

Benefits of technology

Enables efficient and uninterrupted operation of programmable ICs by allowing dynamic reconfiguration of user-specified circuits, reducing compilation time and maintaining network communication links, thus enhancing flexibility and performance in network-attached acceleration tasks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007592066000001
    Figure 0007592066000001
  • Figure 0007592066000002
    Figure 0007592066000002
  • Figure 0007592066000003
    Figure 0007592066000003
Patent Text Reader

Abstract

The integrated circuit IC may include a platform (218) at least partially implemented in a static region (202) of the programmable circuit, the platform including a network interface card controller NIC coupled to a host interface (212) and a network interface (206), and a dynamic region (204) of the programmable circuit configured to implement a user-specified circuit (216) that communicates with the platform, the user-specified circuit implementing one or more packet processing functions. The NIC includes multiple communication interfaces connected to the dynamic region. The platform can establish and maintain a first communication link with a host data processing system using the host interface and a second communication link with a network using the network interface. The platform can continue to transfer data between the network and the host data processing system while at least a portion of the dynamic region of the programmable circuit is dynamically reconfigured to implement different user-specified circuits therein.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] Technical Field The present disclosure relates to integrated circuits (ICs), and more particularly to dynamically reconfigurable networks using programmable ICs. [Background technology]

[0002] background A programmable integrated circuit (IC) refers to a type of device that contains programmable logic. One example of a programmable IC is a field programmable gate array (FPGA). FPGAs are characterized by the inclusion of programmable circuit blocks. Examples of programmable circuit blocks include, but are not limited to, input / output blocks (IOBs), configurable logic blocks (CLBs), dedicated random access memory blocks (BRAMs), digital signal processing blocks (DSPs), processors, clock managers, and delay locked loops (DLLs).

[0003] Each programmable tile typically includes both programmable interconnect circuitry and programmable logic circuitry. The programmable interconnect circuitry typically includes a number of interconnect lines of various lengths interconnected by programmable interconnect points (PIPs). The programmable logic circuitry implements the logic of a user design using programmable elements that may include, for example, function generators, registers, arithmetic logic, and the like.

[0004] Programmable interconnects and programmable logic circuits are typically programmed by loading a stream of configuration data into internal configuration memory cells that define how the programmable elements are configured. The configuration data can be read from memory (e.g., an external PROM) or written to the FPGA from an external device. The collective state of the individual memory cells then determines the functionality of the FPGA.

[0005] Modern programmable ICs have evolved to include programmable logic in combination with one or more other subsystems. For example, some programmable ICs have evolved into systems-on-chips or "SoCs" that include both programmable logic and hardwired processors. Other types of programmable ICs include additional and / or different subsystems. The phrase "programmable IC" refers to a device that includes at least a portion of programmable logic and is partially or fully programmable.

[0006] In general, the functionality of a programmable IC is controlled by data bits provided to the device for that purpose. The data bits may be stored in volatile memory (e.g., static memory cells as in FPGAs and some other programmable ICs such as complex programmable logic devices (CPLDs)), non-volatile memory (e.g., flash memory as in some CPLDs), or any other type of memory cell. Summary of the Invention [Means for solving the problem]

[0007] overview The integrated circuit may include a platform at least partially implemented in a static region of the programmable circuit. The platform may include a network interface card controller coupled to a host interface and a network interface. The IC may include a dynamic region of the programmable circuit configured to implement a user-specified circuit that communicates with the platform. The user-specified circuit may implement one or more packet processing functions. The network interface card controller may include a plurality of communication interfaces connected to the dynamic region. The platform may be configured to establish and maintain a first communication link with a host data processing system using the host interface and a second communication link with a network using the network interface. The platform may be configured to continue to transfer data between the network and the host data processing system while at least a portion of the dynamic region of the programmable circuit is dynamically reconfigured to implement different user-specified circuits therein.

[0008] This summary section is provided merely to introduce certain concepts and is not intended to identify key or essential features of the claimed subject matter. Other features of the inventive subject matter will be apparent from the accompanying drawings and the detailed description that follows.

[0009] BRIEF DESCRIPTION OF THE DRAWINGS Arrangements of the present invention are illustrated by way of example in the accompanying drawings. The drawings should not, however, be construed as limiting the arrangements of the present invention to only the particular embodiments shown. Various aspects and advantages will become apparent upon consideration of the following detailed description and upon reference to the drawings. [Brief description of the drawings]

[0010] [Figure 1] 1 illustrates an example of a host data processing system and a hardware acceleration card. [Figure 2A]1 illustrates an exemplary platform implemented on a programmable integrated circuit (IC). [Figure 2B] 1 illustrates another exemplary platform implemented in a programmable IC. [Diagram 3] 1 illustrates another exemplary platform implemented in a programmable IC. [Figure 4] 1 illustrates another exemplary platform implemented in a programmable IC. [Diagram 5] 1 illustrates another exemplary platform implemented in a programmable IC. [Figure 6] 1 illustrates another exemplary platform implemented in a programmable IC. [Figure 7] 1 illustrates another exemplary platform implemented in a programmable IC. [Figure 8] 1 illustrates another exemplary platform implemented in a programmable IC. [Figure 9] 1 illustrates an exemplary implementation of a network interface card (NIC) controller in a platform. [Figure 10] 1 illustrates another exemplary platform implemented in a programmable IC. [Figure 11] 1 illustrates another exemplary platform implemented in a programmable IC. [Figure 12] 1 illustrates another exemplary platform implemented in a programmable IC. [Figure 13] 1 illustrates an exemplary method of operating a programmable IC that implements the smart NIC functionality described herein. [Figure 14] 1 illustrates an exemplary method for implementing a platform on a programmable IC. [Figure 15] 1 illustrates an exemplary architecture of an IC. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Detailed Description Although the present disclosure is completed by the claims that define the novel features, it is believed that the various features described in the present disclosure will be better understood by considering the description in conjunction with the drawings. The process or processes, the machine or machines, the product or products, and any variations thereof described herein are provided for illustrative purposes. The specific structural and functional details described in the present disclosure should not be interpreted as limiting, but merely as a representative basis for teaching those skilled in the art to use the described features in various ways in substantially any appropriately detailed structure. Moreover, the terms and phrases used in the present disclosure are not intended to be limiting, but rather to provide an understandable description of the described features.

[0012] The present disclosure relates to integrated circuits (ICs), and more particularly to dynamically reconfigurable networks using programmable ICs. In accordance with the inventive arrangements described herein, a programmable IC is used to implement and deploy dynamically reconfigurable networked acceleration functions. The described functions may be provided within a single device. For example, a platform implemented in a programmable IC as described herein may provide computation, storage, and networking acceleration functions within a single programmable IC. The platform supports the implementation and deployment of user-specified circuits for performing one or more networked acceleration functions. These user-specified circuits may be dynamically inserted and / or removed from the platform on the fly without affecting the network traffic processed by the platform of the programmable IC. Thus, the platform may be functionally expanded by including, removing, and / or replacing user-specified circuits over time during operation of the programmable IC.

[0013] Network-attached acceleration refers to processing that is typically performed on a network interface card (NIC) plugged into a data processing system, such as a server (e.g., a host system), via a communications bus. The NIC may include a field programmable gate array (FPGA) connected to an application specific integrated circuit (ASIC). The ASIC typically receives data via a network port and performs basic packet processing functions, while the FPGA is used to perform other enhanced network acceleration functions on the data, referred to as "smart NIC" functions. The ASIC can then deliver the processed data to the host system.

[0014] According to the inventive configurations described within this disclosure, a single device, e.g., a single programmable IC such as an FPGA, can be used to provide both ASIC functionality and enhanced or smart NIC functionality. The platform described herein, when implemented in a programmable IC, supports the described data processing and the ability to implement different custom or user-specified circuits over time without interfering with the ability of the programmable IC to continue performing basic NIC functions. Thus, portions of the programmable IC can be dynamically reconfigured (e.g., partially reconfigured) to perform packet processing operations related to computation, storage, and / or networking acceleration functions, while other areas on the programmable IC can continue to operate uninterrupted so as not to lose data or a communication link to the network. The programmable IC can also maintain a communication link to a host system. In this manner, the programmable IC can continue to pass data between the network and the host system without losing data while one or more different user-specified circuits are dynamically inserted, removed, and / or substituted at run-time to process network data.

[0015] The inventive configuration provides a platform that allows additional functionality to be added to the basic NIC functionality provided by the NIC controller. The additional functionality may be related to networking, computation, and / or storage acceleration. Developers can create specific functionality without having to recreate the entire NIC infrastructure provided by the platform. Use of the platform provides various advantages. These advantages include reduced compilation time when developing hardware for implementation in a programmable IC, since the platform does not need to be reimplemented for each different user-specified circuit that is developed. These advantages also include reduced time to download configuration data to a programmable IC to implement user-specified circuits at run time, since the platform does not need to be reimplemented in the programmable IC each time a different user-specified circuit is added, removed, and / or changed.

[0016] Further aspects and / or advantages of the inventive arrangement are described in more detail below with reference to the drawings. For simplicity and clarity of illustration, elements shown in the figures have not necessarily been drawn to scale. For example, dimensions of some elements may be exaggerated relative to other elements for clarity. Moreover, where considered appropriate, reference numerals have been repeated among the figures to indicate corresponding, similar or similar features.

[0017] 1 illustrates an example of a host data processing system (host system) 100 and a hardware acceleration card (card) 104. Components of the host system 100 may include, but are not limited to, one or more processors 106 (e.g., central processing units), memory 108, and a bus 110 coupling various system components including the memory 108 to the processor 106. The processor 106 may include any of a variety of processors capable of executing program code. Exemplary processor types include, but are not limited to, processors having x86-type architectures (IA-32, IA-64, etc.), Power architectures, ARM processors, etc.

[0018] Bus 110 represents one or more of any of several types of communication bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, and a processor or local bus using any of a variety of available bus architectures. By way of example and not limitation, such architectures include an Industry Standard Architecture (ISA) bus, a MicroChannel Architecture (MCA) bus, an Enhanced ISA (EISA) bus, a Video Electronics Standards Association (VESA) local bus, a Peripheral Component Interconnect (PCI) bus, and a PCI Express (PCIe) bus.

[0019] Host system 100 typically includes a variety of computer readable media. Such media may be any available media that is accessible by host system 100 and may include any combination of volatile media, nonvolatile media, removable media, and / or non-removable media.

[0020] Memory 108 may include computer readable media in the form of volatile memory such as random access memory (RAM) 112 and / or cache memory 114. Host system 100 may also include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example, storage system 116 may be provided for reading from and writing to a non-removable, non-volatile magnetic medium (not shown, typically referred to as a "hard drive"). Although not shown, a magnetic disk drive may be provided for reading from and writing to a removable, non-removable, non-volatile magnetic disk (e.g., a "floppy disk"), and an optical disk drive may be provided for reading from and writing to a removable, non-volatile optical disk, such as a CD-ROM, DVD-ROM, or other optical media. In such a case, each may be connected to bus 110 by one or more data media interfaces. As further shown and described below, memory 108 may include at least one computer program product having a set (e.g., at least one) of program modules (e.g., program code) configured to perform the functions and / or operations described within the present disclosure.

[0021] Stored in memory 108 is a program / utility 118 having a set (at least one) of program modules 120, which may include, for example, but are not limited to, an operating system, one or more application programs, other program modules, and / or program data. The program modules 120 generally perform at least the functions and / or methodologies described herein with respect to operations performed by host system 100. For example, program modules 120 may include one or more applications and drivers or daemons for communicating with hardware acceleration card 104 and / or programmable IC 132.

[0022] The programs / utilities 118 are executable by the processor 106. The programs / utilities 118, as well as any data items used, generated, and / or manipulated by the processor 106, are functional data structures that impart functionality when used by the processor 106. As defined within this disclosure, a "data structure" is a physical implementation of the data configurations of a data model in physical memory. Thus, a data structure is formed from specific electrical or magnetic structural elements in memory. The data structure imposes a physical organization on the data stored in memory for use by application programs executed using the processor.

[0023] The host system 100 may include one or more input / output (I / O) interfaces 128 communicatively linked to the bus 110. The I / O interfaces 128 enable the host system 100 to communicate with external devices, to couple to external devices that allow a user to interact with the host system 100, to couple to external devices that allow the host system 100 to communicate with other computing devices, and the like. For example, the host system 100 may be communicatively linked to a hardware acceleration card 104 via the I / O interfaces 128. The host system 100 may also be coupled to other external devices, such as a keyboard and / or a display (not shown), via the I / O interfaces 128. Examples of the I / O interfaces 128 may include, but are not limited to, a network card, a modem, a network adapter, a hardware controller, and the like.

[0024] In an exemplary implementation, the I / O interface 128 through which the host system 100 communicates with the hardware acceleration card 104 is a PCIe adapter. The hardware acceleration card 104 may be implemented as a circuit board that couples to the host system 100. The hardware acceleration card 104 may be inserted, for example, into a card slot, for example, an available bus and / or PCIe slot, of the host system 100.

[0025] The hardware acceleration card 104 includes a programmable IC 132. The hardware acceleration card 104 also includes a volatile memory 134 coupled to the programmable IC 132 and a non-volatile memory 136 also coupled to the programmable IC 132. The volatile memory 134 may be implemented as RAM and is considered to be "local memory" of the programmable IC 132, while the memory 108 in the host system 100 is not considered to be local to the programmable IC 132. The memory 108 is local to the host system 100, for example. In some implementations, the volatile memory 134 may include multiple gigabytes of RAM, for example 64 GB of RAM. The non-volatile memory 136 may be implemented as flash memory. The non-volatile memory 136 is considered to be local to the programmable IC 132.

[0026] In the example of FIG. 1, the programmable IC 132 is connected to a network 138. In one example, the network 138 is an Ethernet type network. The network 138 can operate at any of a variety of different speeds. For example, the network 138 can be a 10G, 25G, 50G, 100G, 200G, 400G, or other speed network. In certain implementations, the network 138 can be, include, or be coupled to a 5G network. The programmable IC 132 includes an Ethernet interface (not shown) that is used to connect, e.g., communicatively link, the programmable IC 132 to the network 138. For example, the programmable IC 132 can be connected to an Ethernet switch or one or more other network connection devices via the network 138. For purposes of illustration, the term "network" is used herein to refer to the network 138, e.g., an Ethernet network.

[0027] 1 is not intended to suggest any limitation as to the scope of use or functionality of the examples described herein. Host system 100 is an example of computer hardware (e.g., a system) capable of performing various operations described within this disclosure in connection with hardware acceleration card 104 and / or programmable IC 132.

[0028] The host system 100 is merely one example of a computer that may be used with a hardware acceleration card. The host system 100 is shown in the form of a computing device, such as a computer or server. The host system 100 may be implemented as a standalone device, as a bare metal server, in a cluster, or in a distributed cloud computing environment. In a distributed cloud computing environment, tasks are performed by remote processing devices that are linked through a communications network. In a distributed cloud computing environment, program modules may be located in both local and remote computer system storage media, including memory storage devices.

[0029] As used herein, the term "cloud computing" refers to a computing model that facilitates convenient, on-demand network access to a shared pool of configurable computing resources, such as networks, servers, storage, applications, ICs (e.g., programmable ICs), and / or services. These computing resources can be rapidly provisioned and released with minimal management effort or service provider interaction. Cloud computing promotes availability and may be characterized by on-demand self-service, pervasive network access, shared use of resources, rapid elasticity, and measurable services.

[0030] Some computing environments, e.g., cloud computing environments and / or edge computing environments using host system 100 or other suitable data processing systems, generally support an FPGA-as-a-Service (FaaS) model in which user functions are hardware accelerated as user-specified or custom circuit designs and implemented within programmable ICs operating under the control of a host computer system. Another example of a cloud computing model is described by the National Institute of Standards and Technology (NIST), and more specifically, by NIST's Information Technology Laboratory.

[0031] Host system 100 is operational with numerous other general purpose or special purpose computing system environments or configurations. Examples of computing systems, environments, and / or configurations that may be suitable for use with host system 100 include, but are not limited to, personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable appliances, network PCs, minicomputer systems, mainframe computer systems, distributed cloud computing environments that include any of the above systems or devices, and the like.

[0032] Figure 2A illustrates an exemplary platform 218 implemented in the programmable IC 132 of Figure 1. The programmable IC 132 includes a static region of programmable circuitry 202 and a dynamic region of programmable circuitry 204. It should be understood that while both the static region 202 and the dynamic region 204 are implemented using programmable circuitry, either or both regions may include one or more hardwired circuit blocks, as described in more detail herein in connection with the remaining figures.

[0033] The static area 202 represents an area of ​​the programmable circuitry of the programmable IC 132 that does not change over time once configured in the circuit design. The static area 202 may be reconfigured, but such an operation takes the programmable IC 132 offline (e.g., offline from the host system 100 and / or the network 138). For illustrative purposes, the static area 202 is shown shaded and is used to implement a platform 218 that provides networking operations. The platform 218 is specified by a circuit design that is loaded into the programmable IC 132, for example, as a configuration bitstream. In the example of FIG. 2A, the platform 218 includes various circuit blocks included in the shaded area. The platform 218, once implemented in the programmable IC 132, provides functional networking, a host system interface, and optionally a memory interface during operation. The platform 218 can pass network data from an Ethernet interface to the host system 100 (e.g., the processor 106) via a communication bus that has basic network interface controller (NIC) functionality.

[0034] The dynamic region 204 represents a region of the programmable circuitry of the programmable IC 132 that may be dynamically reconfigured over time to implement different user-specified (e.g., custom) circuits. The user-specified circuits implemented in the dynamic region 204 are designed to connect to circuit blocks implemented in the platform 218 in the static region 202. In general, the user-specified circuits implemented in the dynamic region 204 are used to extend the functionality and / or operability of the platform 218.

[0035] Within the static region 202, circuit designs are loaded to form a platform 218. For example, a complete or partial configuration bitstream-specified platform 218 is loaded into the programmable IC 132. In the example of FIG. 2A, the platform 218 includes an Ethernet interface 206, a NIC controller 208, a direct memory access (DMA) circuit 210, and a peripheral component interconnect express (PCIe) controller 212. Thus, the Ethernet interface 206, the NIC controller 208, the DMA circuit 210, and the PCIe controller 212 are implemented in the static region 202. The Ethernet interface 206 is connected to the NIC controller 208. The NIC controller 208 connects to the DMA circuit 210. The DMA circuit 210 connects to the PCIe controller 212. The Ethernet interface 206 can establish a communication link to the network 138 (e.g., a node or switch therein). The PCIe controller 212 can establish a communication link with the host system 100.

[0036] In other examples, platform 218 can use other types of interfaces to connect to network 138. Additionally, platform 218 can use other types of interfaces to connect to host system 100. The particular examples described within this disclosure are provided for purposes of illustration and not limitation. Other alternative types of interfaces may be used.

[0037] 2A, each of Ethernet interface 206, NIC controller 208, DMA circuit 210, and PCIe controller 212 are implemented using programmable circuitry. In one or more other exemplary implementations, one or more or all (e.g., any combination) of Ethernet interface 206, DMA circuit 210, and / or PCIe controller 212 are implemented as hardwired circuit blocks.

[0038] The NIC controller 208 is implemented with one or more communication interfaces 214 that connect to the dynamic region 204. In certain implementations, the communication interfaces 214 and / or other communication interfaces between the NIC controller 208 and the dynamic region 204 may be implemented as stream interfaces. One example of a stream interface is one that complies with the AMBA® 4 AXI4-Stream protocol (e.g., “AXI”). It should be understood that other types of stream interfaces may be used to connect the NIC controller 208 to the dynamic region 204, with AXI being provided only as a non-limiting example. Another example of a stream interface may be a network-on-chip.

[0039] In one aspect, the communication interface 214 is software controllable. For example, the configuration bitstream used to implement the user-specified circuitry may be stored with associated metadata. The metadata may specify the state of the communication interface 214, e.g., on or off. The metadata may specify which communication interface 214 is available and should be used, if there are multiple such interfaces. In this regard, the metadata describes the connectivity between the platform 218 and the dynamic area 204 and / or the user-specified circuitry 216.

[0040] As an example, the communications interface 214 may be implemented using one or more multiplexers that may be provided with control signals (e.g., select signals) from a control interface. The select signals may be specified by metadata. When a configuration bitstream is loaded into the programmable IC 132 to implement the user-specified circuit 216, the host system 100 may obtain metadata corresponding to the configuration bitstream and write the metadata to the programmable IC 132. The metadata may be provided to the control interface that generates appropriate select signals to the multiplexers of the communications interface 214 to activate and / or deactivate the communications interface 214 as needed to communicate with the user-specified circuit 216.

[0041] Through the communication interface 214, data is transmitted from the platform 218 to the dynamic domain 20 4 Move to Dynamic Area 20 4The NIC controller 208 may have multiple communication interfaces, which may be software controllable as described. For example, with respect to input, the NIC controller 208 may receive data from the network 138 via the Ethernet interface 206. The NIC controller 208 may or may not operate on the data before providing it to the user-specified circuit 216 via the communication interface 214. The user-specified circuit 216 may operate on the received data and provide the processed data to the NIC controller 208 via the communication interface 214. The user-specified circuit 216 may, for example, perform one or more packet processing, i.e., smart NIC functions. The NIC controller 208 may or may not perform further processing on the data and provide the data to the host system 100 via the DMA circuit 210 and the PCIe controller 212. In another example, the NIC controller 208 may be configured to route data received from the Ethernet interface 206 back to the Ethernet interface 206 to the network 138, either after processing performed by the NIC controller 208 and / or the user-specified circuitry 216, or without such processing.

[0042] With respect to output, the NIC controller 208 may receive data from the host system 100 via the PCIe controller 212 and the DMA circuitry 210. The NIC controller 208 may or may not operate on the data before providing it to the user-specified circuitry 216 via the communications interface 214. The user-specified circuitry 216 may operate on the received data and provide the processed data to the NIC controller 208 via the communications interface 214. The NIC controller 208 may or may not perform further processing on the data before providing it to the network 138 via the Ethernet interface 206. In another example, the NIC controller 208 may be configured to route data received from the host system 100 back to the host system 100, either after processing performed by the NIC controller 208 and / or the user-specified circuitry 216 or without such processing.

[0043] The dynamic region 204 is implemented to include one or more partial reconfiguration (PR) regions. A PR region of the programmable IC 132 refers to an area of ​​the programmable circuit that is reserved for implementing one or more different user-specified (e.g., custom) circuit designs dynamically over time. The dynamic region 204 may include multiple PR regions. In general, each of the different PR regions implemented in the dynamic region 204 is independently reconfigurable. In the example of FIG. 2A, the dynamic region 204 implements a single PR region.

[0044] In the example of Figure 2A, static domain 202 refers to a portion of programmable IC 132 that is outside of dynamic domain 204. Platform 218 implemented in static domain 202 does not change, e.g., remains static, while circuitry in dynamic domain 204 can be modified over time to implement other, different, user-specified circuit designs. In the example of Figure 2A, the user-specified circuit designs are implemented in dynamic domain 204 and are shown as user-specified circuits 216. Dynamic domain 204 can be reconfigured over time to implement different user-specified circuits (e.g., circuits other than user-specified circuit 216) while platform 218 in static domain 202 continues to operate without interruption.

[0045] The user-specified circuitry 216 may, for example, perform any of a variety of network acceleration functions. In this regard, the functionality of the platform 218 and, for example, the NIC controller 208 may itself be static, but may be augmented by the addition of the user-specified circuitry 216. In one aspect, the user-specified circuitry 216 may perform any of a variety of incremental network acceleration functions because the NIC controller 208 may pass metadata so that the user-specified circuitry 216 does not have to repeat operations already performed by, for example, the NIC controller 208.

[0046] The user-specified circuits 216 can be dynamically inserted, removed, and / or replaced according to data flowing through the platform 218 in the static region 202 between the network 138 and the host system 100 during operation of the programmable IC 132 (e.g., data from the network 138 (referred to as “network” or “input” data) and / or data from the host system 100 (referred to as “host” or “output” data)), without interruption or loss of such data. In another aspect, in the case of a virtual machine executing on the host system 100, the terms “output” and “input” may be used with respect to the virtual machine. For example, output data of a virtual machine executing on the host system 100 may be directed to the network 138 or to the input of another virtual machine also executing on the host system 100. In either case, the platform 218 can maintain a communication link with the network 138 via the Ethernet interface 206, and with the host system 100 via the PCIe controller 212, while the dynamic region 204 is being dynamically reconfigured. Thus, while the dynamic region 204 is being reconfigured, the programmable IC 132 appears to be “online” or operational to the host system 100 and other devices that connect to the programmable IC 132 via the network 138 .

[0047] 2, the NIC controller 208 is shown as including one communication interface 214, however, the NIC controller 208 may include multiple different communication interfaces. The communication interfaces may be included in one or more different stages included in the NIC controller 208, allowing the NIC controller 208 to exchange data with user-specified circuitry 216 at any of a variety of different processing stages therein.

[0048] FIG. 2B illustrates another example of a platform 218 implemented in the programmable IC 132 of FIG. 1. The platform 218 of FIG. 2B is substantially similar to the example of FIG. 2A. However, in the example of FIG. 2B, the platform 218 is partially implemented in the static region 202. For example, the PCIe controller 212 is implemented in the static region 202. The Ethernet interface 206, the NIC controller 208, and the DMA circuitry 210 are each implemented in a different PR region (e.g., PR regions 220, 222, 224, respectively). In the example of FIG. 2B, the dynamic region 204 may be implemented as another PR region that is separate and independent from the PR regions 220, 222, and 224.

[0049] The example of FIG. 2B allows different components of the platform 218 to be updated over time. In practice, the platform 218 is configurable. Updates to the circuitry in the static region 202 typically require a recompilation of the user circuitry. The implementation of FIG. 2B minimizes the portion of the platform 218 that is implemented in the static region 202. By minimizing the amount of the platform 218 that is implemented in the static region 202, the likelihood of having to recompile the user-specified circuitry is reduced. For example, the DMA circuitry 210, the NIC controller 208, and the Ethernet interface 206 can be changed without requiring a recompilation of the user-specified circuitry 216 (the PR region can be reconfigured with different implementations of such components). A change to the PCIe controller 212 requires a recompilation of the user-specified circuitry 216. Recompiling user circuits (e.g., the user-specified circuitry 216) is a time-consuming task that may be undesirable, especially in Faas applications. In the example of FIG. 2B, most of the platform 218 can be updated over time without requiring a recompilation of the user-specified circuitry. This reduces the likelihood that the circuitry in the static domain 202 will have to be updated, and therefore reduces the likelihood that the user circuit design will have to be recompiled.

[0050] FIG. 2B is provided for illustrative purposes. Other exemplary implementations may use a different number of PR regions to implement the platform 218. For example, the platform 218 may implement the PCIe controller 212 in the static region 202 and the Ethernet interface 206, the NIC controller 208, and the DMA circuitry 210 in a single PR region. In another example, two PR regions may be used, with any two of the Ethernet interface 206, the NIC controller 208, and the DMA circuitry 210 implemented in a first PR region and the remaining components implemented in a second PR region. Although not specifically shown in the various exemplary platforms 218 depicted in the remaining figures, it should be understood that the platform 218 may be implemented using one or more PR regions in each of the presented examples as described in connection with FIG. 2B.

[0051] Figure 3 illustrates another example of the platform 218 implemented in the programmable IC 132. In the example of Figure 3, the Ethernet interface 206 is shown in more detail. The Ethernet interface 206 includes a physical layer (PHY) 302 connected to a medium access control (MAC) 304. The PHY 302 connects to the network 138. The MAC 304 connects to the NIC controller 208. For example, the MAC 304 can be connected to an Ethernet interface (not shown) included in the NIC controller 208.

[0052] 4 illustrates another example of a platform 218 implemented in a programmable IC 132. In the example of FIG. 4, one or more memory controllers 402 are implemented in the static domain 202 as part of the platform 218. The memory controllers 402 are connected to a volatile memory (e.g., RAM) 134. The memory controllers 402 may be implemented using programmable circuits. However, in another example, one or more of the memory controllers 402 may be implemented as hardwired circuit blocks.

[0053] The memory controller 402 may be connected to the user-specified circuitry 216 implemented in the dynamic region 204. Additionally, the memory controller 402 may be coupled to the DMA circuitry 210. Thus, the user-specified circuitry 216 may access the volatile memory 134 during operation. Additionally, the host system 100 may read and / or write the volatile memory 134 via the memory controller 402. This allows the host system 100 to share data with the user-specified circuitry 216.

[0054] Within this disclosure, various exemplary platforms shown, whether or not shown with a memory controller, may be implemented including one or more memory controllers. A memory controller may be included optionally. Additionally, in implementations in which platform 218 is implemented using one or more PR regions, the memory controller may be implemented in its own PR region or may be included in another PR region with one or more other components of platform 218.

[0055] Figure 5 illustrates another example of the platform 218 implemented in the programmable IC 132. In the example of Figure 5, the platform 218 includes a software defined network (SDN) data plane 502 implemented in the NIC controller 208. The SDN data plane 502 can separate data and control planes by implementing the data plane as part of the platform 218. In this example, the data plane software is hardware accelerated and implemented as the SDN data plane 502 in the NIC controller 208, rather than implementing the data plane in software running on the host system 100.

[0056] In some example implementations, the SDN data plane 502 can be customized using the user-specified circuitry 504 (and / or other user-specified circuitry in the dynamic domain 204 connected to the SDN data plane 502 via the communication interfaces 506). For example, using a port connecting the SDN data plane 502 with the dynamic domain 204, the user-specified circuitry 504 can be configured to add custom GRE headers to the data and / or implement a custom parser. In the custom parser, for example, metadata can be passed and associated with the new parser data. Additionally, the GRE can be modified by the user-specified circuitry 504 with a network overlay. In either case, the availability of one or more communication interfaces 506 allows the SDN data plane 502 to be customized using programmable circuitry in the dynamic domain 204.

[0057] The SDN data plane 502 allows programmatic access for more flexible network management. For example, to the extent that the NIC controller 208 and / or the SDN data plane 502 utilizes large flow tables retrieved on a lookup interface, customized flow tables may be implemented as user-specified circuits 216 and / or 504 in the dynamic domain 204. The SDN data plane 502 reduces the processing load that would otherwise be imposed on the host system 100 to run data plane software.

[0058] In the example of FIG. 5, the SDN data plane 502 can be connected to the dynamic domain 204 via a communication interface 506. The SDN data plane 502 can have multiple communication interfaces connected to the dynamic domain 204. In other examples, the SDN data plane 502 may not include a communication interface to the dynamic domain 204. Although a memory controller 402 is shown, the memory controller 402 is optional and may be omitted. If included, the memory controller 402 enables the user-specified circuits 504 and 214 to read from and write to the volatile memory 134. Additionally, the memory controller 402 enables the host system 100 to read from and write to the volatile memory 134.

[0059] For purposes of illustration, the dynamic domain 204 includes additional user-specified circuitry 504. In one example, the user-specified circuitry 504 connects to the NIC controller 208 and / or the SDN data plane 502 via a communication interface 506. The user-specified circuitry 504 can implement custom flow tables that enable the NIC controller 208 to implement data flow diversion.

[0060] In another example, the user-specified circuitry 216 can implement a video decoder and scaling circuitry. The user-specified circuitry 504 can implement a machine learning object detection circuitry. This configuration allows the dynamic area 204 to perform real-time video processing and machine learning. This example illustrates how network-connected real-time video decoding followed by object detection and / or classification can be implemented using the platform 218.

[0061] For example, the video decoder and scaling circuitry may operate on multiple different video streams received over the network 138. The video streams may be from a security system and / or cameras communicatively linked to the network 138. The video decoder and scaling circuitry may decode the received video streams. The video decoder and scaling circuitry may extract one or more individual frames from the different video streams and scale the extracted frames for use by the machine learning object detection circuitry.

[0062] In the example of FIG. 5, the NIC controller 208 can perform basic NIC functions. Video decoding and scaling functions and machine learning inference for object detection and / or image classification functions are implemented as hardware accelerated kernels in the dynamic domain 204. By implementing the functions as kernels in the dynamic domain 204, the kernels can operate on the video stream in real time. Real-time operation reduces latency since processing can be performed without the video stream passing through the host system 100's network stack (e.g., the user-specified circuits 504 and 216 operate on the video before the data reaches the host system 100). This accelerates processing by the programmable IC 132. Additionally, the platform 218 can discard images without flagged objects while saving images (e.g., frames) with flagged objects for further analysis, thus reducing storage requirements. Additionally, the workload on the host system 100 is significantly reduced since the processor 106 cycles can be used for other operations.

[0063] In another example, the platform 218 of FIG. 5 can be used to perform inline malware detection on encrypted data traffic. In this example, the platform 218 can operate based on extracting certain information from the data flow. The information can include the duration of the data flow, the number of packets sent from the client, the number of packets sent from the server, the number of bytes sent from the client, and the number of bytes sent from the server. For example, the user-specified circuitry 216 can implement a classification and feature extraction circuit. The extracted information can be used along with metadata from the unencrypted portions of the packet headers, also extracted by the user-specified circuitry 216, to train a random forest classifier implemented by the user-specified circuitry 504. The trained classifier can then be used to detect malware.

[0064] In one embodiment, the classifier can run in real time on packets entering the programmable IC 132 from the network 138. The user-specified circuit 216 can extract features and classify the packets into data flows. The user-specified circuit 216 can provide the determined information to the user-specified circuit 504. The user-specified circuit 504 can be pre-loaded with trained models for the malware of interest. In response to the user-specified circuit 504 detecting the malware, the user-specified circuit 504 can send a message to set a filter in the SDN data plane 502 to prevent further network data sent from the attacker from reaching the host system 100. In this example, the NIC controller 208 and the SDN data plane 502 run as part of the platform 218, while the feature extraction functions and classification for malware detection run as kernels loaded into the dynamic area 204.

[0065] By performing classification, extraction, and malware detection in real time, malware can be blocked before it physically traverses the PCIe bus from the programmable IC 132 to the host system 100. This provides more robust protection for the host system 100. The random forest classification technique enables detection of malware on encrypted traffic without the need for complex decryption and proxy techniques. Additionally, performing this function on the programmable IC 132 rather than on the appliance provides better scaling through workload distribution and protects against attacks from within the data center.

[0066] The particular user-specified circuitry described herein is provided for purposes of illustration and not limitation. These examples illustrate that any user-specified circuitry may be implemented in the dynamic area 204 to extend the capabilities of the platform 218. Additionally, different blocks of user-specified circuitry may be included to communicate with the NIC controller 208, but may or may not communicate with each other.

[0067] In the example of FIG. 5, the dynamic region 204 is implemented as a single PR region. In one or more other examples, each user-specified circuit block (e.g., 504 and 216) may be implemented within its own independently reconfigurable PR region. In such a configuration, the particular user-specified circuit implemented in each PR region is dynamically changeable (e.g., reconfigurable) without changing the other user-specified circuit implemented in the dynamic region 204. For example, if the user-specified circuit 504 is in a first PR region and the user-specified circuit 216 is in a second PR region, the user-specified circuit 504 can be removed and a different user-specified circuit can be added to the first PR region without interrupting the operation of the user-specified circuit 216. Similarly, the user-specified circuit 216 can be removed and a different user-specified circuit can be added to the second PR region without interrupting the operation of the user-specified circuit 504. In either case, the platform 218 continues to operate uninterrupted. The platform 218 of FIG. 5 may also be implemented using one or more PR regions as described herein.

[0068] FIG. 6 illustrates another example of the platform 218 implemented in the programmable IC 132. The platform 218 of FIG. 6 is substantially similar to the example of FIG. 5. However, in the example of FIG. 6, the SDN data plane 502 is replaced with an Open Virtual Switch or Open vSwitch (OVS) data plane 602. The OVS data plane 602 can connect to a virtualization server. The OVS data plane 602 implements a hardware accelerated version of a virtual switch. Rather than implementing a virtual switch in software running on the host system 100, the virtual switch software is hardware accelerated and implemented as the OVS data plane 602 in the platform 218. The OVS data plane 602 reduces the processing burden that would otherwise be imposed on the host system 100. In the example of FIG. 6, the OVS data plane 602 and / or the flow tables can be customized using user-specified circuitry 504 and / or user-specified circuitry 216 (e.g., using user-specified circuitry implemented in the dynamic area 204 and connected to the OVS data plane 602 and / or NIC controller 208 via one or more communication interfaces).

[0069] In the example of FIG. 6, the OVS data plane 602 can be connected to the dynamic domain 204 via a communication interface 604. The OVS data plane 602 can have multiple communication interfaces connected to the dynamic domain 204. In another example, the OVS data plane 602 may not have a communication interface to the dynamic domain 204. Although a memory controller 402 is shown, the memory controller 402 is optional. If included, the memory controller 402 enables the user-specified circuits 504 and 214 to read and write to the volatile memory 134. Additionally, the memory controller 402 enables the host system 100 to read and write to the volatile memory 134.

[0070] For purposes of illustration, the dynamic region 204 includes additional user-specified circuitry 504. The user-specified circuitry 504 may be connected to the NIC controller 208 and / or the OVS data plane 602 via an interface 604. In one example, the user-specified circuitry 216 may implement circuitry that may perform one or more of storage compression, encryption, and / or deduplication operations.

[0071] In the example of FIG. 6, the platform 218 can be used to implement an integrated smart NIC for networking and storage. The platform 218 can perform both network and storage offload and acceleration. The platform 218 provides a NIC controller 208 with an OVS data plane 602, an Ethernet interface 206, a PCIe controller 212, and a memory controller 402. The user-specified circuits 216 and 504 implemented in the dynamic region 204 can perform operations such as compression and / or encryption. As described, data flow between the platform 218 and the user-specified circuits 216 and 504 can be implemented using, for example, a stream interface.

[0072] Control plane and slow-path processing may occur in host system 100. However, in another exemplary implementation, an embedded hardwired and / or soft processor within programmable IC 132 (see, e.g., FIG. 7), or an externally attached processor other than a processor within host system 100 (see, e.g., FIG. 8), may perform the storage offload. The latter model, involving an embedded or external processor, may be used in a "bare metal" implementation, where the embedded or external processor manages network functions, storage, etc., as opposed to a host system processor performing such functions.

[0073] The particular user-specified circuits described herein are provided for purposes of illustration and not limitation. These examples illustrate that any user-specified circuitry may be implemented in the dynamic region 204 to extend the capabilities of the platform 218. Furthermore, the illustrated user-specified circuit blocks may be included in the same PR region or different PR regions depending on the implementation. Similarly, the platform 218 of FIG. 6 may be implemented using one or more PR regions as described herein.

[0074] FIG. 7 illustrates another example of the platform 218 implemented in the programmable IC 132. In the example of FIG. 7, the static area 202 is implemented to include a processor 702. The processor 702 is an embedded processor in the programmable IC 132. In one aspect, the processor 702 is a soft processor implemented using programmable circuitry of the programmable IC 132. In an exemplary implementation, the soft processor 702 is implemented as a MicroBlaze™ processor available from Xilinx, Inc. of San Jose, Calif. Furthermore, the processor 702 may be implemented as any of a variety of processors capable of executing program code. In another example, the processor 702 is a hardwired processor. For example, the processor 702 may be implemented as an ARM type processor or an x86 type processor.

[0075] As shown, the processor 702 is connected to an on-chip interconnect 704. In one example, the on-chip interconnect 704 is implemented as a memory-mapped interface. One example of a memory-mapped interface is a memory-mapped AXI interface. AXI is provided as an illustrative example of an interface and is not intended as a limitation of the inventive configurations described within this disclosure. Other examples of on-chip interconnects include, but are not limited to, other types of buses, networks-on-chip, crossbars, or other types of switches.

[0076] The processor 702 may perform any of a variety of different functions. Examples of functions that may be performed by the processor 702 include, but are not limited to, OVS control plane, storage, management, etc. In the example of Figure 7, the platform 218 may pass network data from the Ethernet interface 206 to the embedded processor 702 through a local interface (e.g., via the NIC controller 208, the DMA circuit 210, and the memory mapped interface 704) as shown.

[0077] 7, a memory controller is not included in the platform 218. In one or more other examples, a memory controller may be included in the platform 218. The memory controller may be connected to the processor 702, the user-specified circuits 216 and / or 504 in the dynamic region 204, and / or the host system 100. As described, the platform 218 may be implemented using one or more PR regions.

[0078] 8 illustrates another example of the platform 218 implemented on the programmable IC 132. In the example of FIG. 8, the platform 218 is implemented to include an additional (e.g., second) PCIe controller 802. The PCIe controller 802 is implemented using programmable circuitry. The PCIe controller 802 is connected to an external processor 804. The external processor 804 can execute program code and can be located on the hardware acceleration card 104 or on a different circuit board or system coupled to the hardware acceleration card 104.

[0079] The external processor 804 may perform any of a variety of different functions. Examples of functions that may be performed by the processor 804 include, but are not limited to, OVS control plane, storage, management, etc.

[0080] 8, the memory controller is not included in the static region 202. In one or more other examples, the memory controller may be included in the platform 218. The memory controller may be connected to the DMA circuit 210, the user-specified circuits 216 and / or 504 in the dynamic region 204, and / or the host system 100. As described, the platform 218 may be implemented with one or more PR regions. In one aspect, the PCIe controller 212 and the PCIe controller 802 may be implemented in the static region 202, while other components are included in one or more PR regions.

[0081] With reference to the exemplary embodiment of FIG. 7 and FIG. 8, the host system can implement one or more virtual machines. Each virtual machine has two separate virtual interfaces. One virtual interface is for networking and the other is for storage. Network traffic is attached to a multi-queue DMA interface and passes through a processing pipeline including network offloads such as receive side scaling (RSS), LSO / TSO (large send offload / TCP segmentation offload), virtual extension (VX) Lan / NVGRE (Network Virtualization using Generic Routing Encapsulation) tunnel offload, and SDN match / action processing such as OVS. Storage traffic is offloaded via an embedded processor 702 or an external processor 804, optionally programmed with RDMA capabilities. Storage traffic can pass through a processing pipeline including NVMe and storage offload operations such as compression and / or encryption performed in the dynamic area 204. The processed data can then be merged with the data pipeline of the NIC controller 208.

[0082] Figure 9 illustrates an exemplary implementation of NIC controller 208. In the example of Figure 9, NIC controller 208 includes an Ethernet interface 902, a bump-in-the-wire (BITW) arbitration circuit 904, a NIC core 906, an accelerator arbitration circuit 908, and a DMA interface 910. For purposes of illustration and explanation, NIC core 906 and accelerator arbitration circuit 908 may each be referred to as a "stage" of NIC controller 208. In some exemplary implementations, NIC core 906 itself may be formed from or include multiple different stages of circuitry.

[0083] 9, Ethernet interface 902 is connected to Ethernet interface 206 (e.g., MAC 304). DMA interface 910 is connected to DMA circuitry 210. DMA interface 910, accelerator arbitration circuitry 908, NIC core 906, BITW arbitration circuitry 904, and Ethernet interface 902 each support an input data flow (e.g., data flows from Ethernet interface 206 to NIC controller 208) and an output data flow (e.g., data flows from NIC controller 208 to Ethernet interface 206).

[0084] As shown, each of the BITW arbitration circuitry 904, the NIC core 906, the accelerator arbitration circuitry 908, and the DMA interface 910 includes an input communication interface to the dynamic region 204 and an output communication interface to the dynamic region 204. In one aspect, the communication interfaces can include an input communication interface and an output communication interface. For example, with reference to FIG. 9, each input / output pair can represent one communication interface. However, in other examples, the communication interfaces can include only input communication interface(s) or only output communication interface(s). Each of the circuit blocks described as having a communication interface(s) to the dynamic region 204 can transmit output data and / or input data via the appropriate input or output communication interface for further processing within any of the various user circuits 912, 914, 916, 918, 920, and / or 922 shown in the dynamic region 204.

[0085] In another exemplary implementation, the BITW arbitration circuit 904 may be implemented as one or more multiplexers (e.g., switches) that connect input and / or output data paths to the dynamic region 204. The BITW arbitration circuit 904 enables the NIC controller 208 to include one or more interfaces between the Ethernet interface 902 and the NIC core 906. The BITW arbitration circuit 904 can transmit data received from the Ethernet interface 902 directly to one or more user circuits, such as the user circuit 914, via the input communication interface for packet processing. The user circuit 914 can perform packet processing on the received data and return the processed data to the BITW arbitration circuit 904 via the input communication interface. The BITW arbitration circuit 904 can then forward the processed data to the NIC controller core 906.

[0086] In certain implementations, the BITW arbitration circuitry 904 may be configured to rotate data. For example, the BITW arbitration circuitry 904 may receive data from the Ethernet interface 902 and send the data back to the Ethernet interface 902 for output to the Ethernet interface 206. The BITW arbitration circuitry 904 may or may not provide the data to a user circuit in the dynamic domain 204 before rotating the data. In either case, data received by the NIC controller 208 from the network 138 may be optionally processed, rotated, and sent back from the NIC controller 208 to the network 138 using the BITW arbitration circuitry 904.

[0087] The BITW arbitration circuit 904 can further transmit data received from the NIC core 906 directly to one or more user circuits, such as the user circuit 912, via an output communication interface for packet processing. The user circuit 912 can perform packet processing on the received data and return the processed data to the BITW arbitration circuit 904 via an output communication interface. The BITW arbitration circuit 904 can then forward the processed data to the Ethernet interface 902. In another example, once sufficient data has been exchanged between the user circuits 912 and / or the user circuits 914 via the BITW arbitration 904, the NIC core 906 can perform operations such as subsequent upper protocol layer analysis.

[0088] In certain implementations, the BITW arbitration circuitry 904 may be configured to rotate data. For example, the BITW arbitration circuitry 904 may receive data from the NIC controller 906 and send the data back to the NIC controller 906 for output to the DMA circuitry 210. The BITW arbitration circuitry 904 may or may not provide the data to user circuits in the dynamic area 204 before rotating the data. In either case, data received by the NIC controller 208 from the host system may be optionally processed, rotated, and sent from the NIC controller 208 back to the host system using the BITW arbitration circuitry 904.

[0089] Thus, the BITW arbitration circuitry 904 functions as a switch or interface by which input data can be provided to user-specified circuitry in the dynamic domain 204 before entering the processing stages of the NIC controller 208. Similarly, the BITW arbitration circuitry 904 functions as a switch or interface by which output data can be provided to user-specified circuitry in the dynamic domain 204 just before exiting the NIC controller 208. As explained, the BITW arbitration circuitry 904 can also rotate input data and / or output data.

[0090] In one aspect, the input and output communication interfaces are implemented as stream interfaces. In one implementation, the user circuit 912 and the user circuit 914 may implement Internet Protocol Security (IPsec) circuit blocks. For example, after processing by the user circuit 914, input packets directed to the NIC core 906 may be fully decrypted so that higher layer protocols such as Transmission Control Protocol (TCP) can be further processed. Unencrypted output packets (e.g., clear text) from the NIC core 906 may be encrypted by the user circuit 912.

[0091] In another aspect, the user circuitry 912 can perform operations such as adding Ethernet IP and TCP headers or GRE. The user circuitry 914 may be used to implement GRE header removal, also known as decapsulation. The user circuitry 912 can perform GRE header insertion, also known as encapsulation. Again, the user circuitry 912 and / or 914 can perform operations on data at points in the data path between the Ethernet interface 206 and the NIC core 906.

[0092] The NIC core 906 can transmit the data received from the BITW arbitration circuit 904 directly to one or more user circuits, such as a user circuit 918, via an input communication interface for packet processing. The user circuit 918 can perform packet processing on the received data and return the processed data to the NIC core 906 via an input communication interface. The NIC core 906 can then forward the processed data to the acceleration arbitration circuit 908.

[0093] The NIC core 906 can transmit the data received from the acceleration arbitration circuit 908 directly to one or more user circuits, such as a user circuit 916, via an output communication interface for packet processing. The user circuit 916 can perform packet processing on the received data and return the processed data to the NIC core 906 via an output communication interface. The NIC core 906 can then forward the processed data to the BITW arbitration circuit 904.

[0094] In one aspect, the input / output communication interfaces of NIC core 906 are implemented as ternary content addressable memory (TCAM) interfaces (e.g., at least one for input and at least one for output). User circuits 916 and / or user circuits 918 can implement any of a variety of different circuits including, but not limited to, access control lists / flow tables, algorithmic TCAMs, MAC filters, and / or algorithmic content addressable memories (CAMs).

[0095] In another example, for video processing, operations such as UDP and checksum offloading may occur in the NIC core 906. User circuits 918 and / or 916 may be used to perform flow lookups to identify where to direct the data. The data may be directed to a location indicated by user circuits 918 and / or 916 implementing the flow tables. For example, a user may wish to use a larger table than can be implemented in the NIC core 906. Thus, the user circuits 918 and / or 916 may use dynamic area 20 to extend the tables contained in the NIC core 906. 4 Additional RAM in the NI PXI Express 8038 can be used.

[0096] The acceleration arbitration circuit 908 can transmit data received from the NIC core 906 directly to one or more user circuits, such as a user circuit 922, via an input communication interface for packet processing. The user circuit 922 can perform packet processing on the received data and return the processed data to the acceleration arbitration circuit 908 via an input communication interface. The acceleration arbitration circuit 908 can then transfer the processed data to a DMA interface 910.

[0097] The acceleration arbitration circuit 908 can transmit data received from the DMA interface 910 directly to one or more user circuits, such as the user circuit 920, via an output communication interface for packet processing. The user circuit 920 can perform packet processing on the received data and return the processed data to the acceleration arbitration circuit 908 via an output communication interface. The acceleration arbitration circuit 908 can then forward the processed data to the NIC core 906.

[0098] In an exemplary implementation, the accelerator arbitration circuit 908 is located in the data path after TCP / IP, checksum offload, and other high level functions have been performed. Packets and associated metadata are passed to the dynamic region 20 where user circuits (e.g., user circuits 920 and / or 922) can perform incremental acceleration. 4 That is, when user circuitry 920 and / or 922 is performing a high-level operation, the circuitry does not need to newly perform a previous operation that was performed elsewhere in the data pipeline of the NIC controller 208.

[0099] In one aspect, the input and output communication interfaces are implemented as stream interfaces (e.g., at least one for input and at least one for output). In an exemplary implementation, user circuitry 920 and user circuitry 922 can implement circuitry for performing deep packet inspection of Layer 4 or above.

[0100] The DMA interface 910 can transmit data received from the acceleration arbitration circuit 908 to the DMA circuit 210. The DMA interface 910 can transmit data received from the DMA circuit 210 to the accelerator arbitration circuit 908.

[0101] In the example of FIG. 9, the user circuits are shown as being implemented in a single PR region. It should be understood that more than one PR region may be implemented in the dynamic region 204 containing various combinations of user circuits. In another example, user circuits 912 and 914 are implemented in one PR region, user circuits 916 and 918 are implemented in another PR region, and user circuits 920 and 922 are implemented in yet another PR region. Furthermore, the various examples of different user circuits and interfaces are for illustration purposes and not limitation. In another example, each user circuit may be implemented in a different PR region.

[0102] As shown in the example of FIG. 9, the platform 218 can provide a number of different communication interfaces to the dynamic area 204. These communication interfaces, e.g., connection points, allow data to move between the platform 218 and dynamically loaded functionality implemented in the dynamic area 204 at any of the various supported connection points. As mentioned above, one or more or all of the described communication interfaces connecting the platform 218 to the dynamic area 204 (e.g., communication interfaces of the NIC controller 208) are software controllable. Thus, an embedded processor in the programmable IC 132 (whether soft or hardwired as shown in FIG. 7) or an external device or processor (e.g., as shown in FIG. 8), and / or a processor of a host system can individually control whether each of the communication interfaces is turned on to exchange data between the static platform 218 and the dynamic area 204, or turned off to prevent exchange of data between the platform 218 and the dynamic area 204.

[0103] Figure 10 illustrates another example of a platform 218 implemented on a programmable IC 132. In the example of Figure 10, the platform 218 includes a stream traffic manager 1002. The stream traffic manager 1002 is connected to the accelerator arbitration circuit 908 and the DMA circuit 210. Although not shown in Figure 10, the stream traffic manager 1002 may include one or more connections to the dynamic region 204.

[0104] Stream traffic manager 1002 can coordinate the flow of data streams / packets between host system 100 and circuitry implemented in dynamic region 204. In the example of Figure 10, stream traffic manager 1002 coordinates the flow of data between host system 100 and user circuit 1010 and / or user circuit 1012. In other examples, dynamic region 204 can include additional user circuitry, e.g., kernel circuitry, that can be connected to stream traffic manager 1002 via an interconnect (not shown).

[0105] In the example of FIG. 10, the dynamic region 204 is divided into PR regions 1006 and 1008. The PR region 1006 includes a user circuit 1010. The PR region 1008 includes a user circuit 1012. Each of the PR regions 1006 and 1008 may be dynamically reconfigured independently of one another. Thus, since the user circuit 1010 is in the PR region 1006 and the user circuit 1012 is in the PR region 1008, the user circuit 1010 can be removed or replaced without affecting the operation of the user circuit 1012 or the platform 218. Similarly, since the user circuit 1012 is implemented in the PR region 1008, the user circuit 1012 can be removed or replaced without affecting the operation of the user circuit 1010 or the platform 218. As described, one or both of the PR regions 1006, 1008 (e.g., the entirety of the dynamic region 204) can be reconfigured without interrupting the operation of the platform 218. Although dynamic region 204 is shown as including two PR regions, it should be understood that in another example, user circuit 1010 and user circuit 1012 may be implemented in a single PR region.

[0106] In one aspect, the user circuit 1012 implements an algorithmic TCAM capable of performing high speed packet processing. The user circuit 1010 implements an IPsec circuit capable of performing operations between two communicating points on an IP network, including but not limited to data authentication, integrity, and / or confidentiality.

[0107] In the example of FIG. 10, the user circuit 1010 can receive data from the NIC core 906, process the data, and return the processed data to the NIC core 906. The user circuit 1012 illustrates an example in which the user circuit 1012 (e.g., an IPsec circuit) can be programmed by the host system 100. In this case, the host system 100 (e.g., or another trusted processor, whether internal or external) can send post-configuration data to the user circuit 1012 via the PCIe controller 212 and a control interface of the user circuit 1012 coupled thereto. An example control interface is one that conforms to the AXI-Light protocol, although other control interfaces may be used.

[0108] 11 illustrates another example of a platform 218 implemented in a programmable IC 132. In the example of FIG. 11, a stream traffic manager 1002 is connected to an interconnect 1102 and an interconnect 1106 in the dynamic domain 204. The interconnect 1102 may be a memory mapped interconnect and the interconnect 1106 may be a stream interconnect. The interconnect 1102 is connected to one or more network kernels 1104. The interconnect 1106 is connected to one or more compute kernels 1108. The stream traffic manager 1002 and / or accelerator arbitration circuit 908 can manage (e.g., direct) the flow of data between the host system 100 and the network kernel 1104 and the compute kernels 1108. In the example of Figure 11, the accelerator arbitration circuitry 908 can communicate with user circuits implemented in the dynamic domain 204 via the stream traffic manager 1002.

[0109] Although the dynamic region 204 is shown as including a single PR region, in alternative embodiments, the dynamic region 204 may include multiple PR regions. For example, the user circuit 1012 may be implemented in a first PR region, and the user circuit 1010 is implemented in a second PR region. Additionally, the interconnects 1102, 1106 , Ne Network kernel 1104, and the computation kernel In another aspect, the interconnect 1102 and the network kernel 1104 are implemented in a third PR region, and the interconnect 1106 and Computational Kernel 1108 is implemented in the fourth PR region. In the example of Figures 10 and 11, the stream traffic manager 1002 may be implemented in a PR region alone or together with one or more other components of the platform 218. Alternatively, the stream traffic manager 1002 may be implemented in the static region 202.

[0110] A stream traffic manager 1002, which may operate at least in part as a switch, is used to route data from a host system to an interconnect 1102 and to one or more network kernels 1104, to an interconnect 1106 and to one or more compute kernels 1108, or to an accelerator arbitration. circuit 908 to the NIC controller 208. For example, the stream traffic manager 1002 can receive packets having a specified flow (e.g., route or destination). The stream traffic manager 1002 can identify the flow of the received packets and deliver the packets to a destination corresponding to the identified flow. This allows the dynamic region 204 to perform a computational function (e.g., hardware acceleration of a task offloaded from the host system 100) and simultaneously perform a network function.

[0111] As another example, the architecture shown in FIG. 11 allows data (e.g., packets) to flow from the host system 100 to the dynamic region 204 for processing (e.g., by a compute kernel and / or a network kernel) and back to the host system 100, or to flow from the host system 100 to the dynamic region 204 for processing (e.g., by a compute kernel and / or a network kernel) and out to the Ethernet via the NIC controller 208, which may or may not further process the data using the dynamic region 204 (e.g., user circuits 1010 and / or 1012). Thus, the host system 100 can send data to the dynamic region 204 (e.g., the compute kernel 1108) for processing. The stream traffic manager 1002 can determine whether to send the resulting data output from the dynamic region 204 back to the host system 100 or the NIC controller 208.

[0112] Although not shown in FIG. 11, in another example, data may be exchanged directly between the computation kernel and the network kernel (or user circuitry connected to the NIC controller 208).

[0113] Figure 12 illustrates another exemplary platform implemented in a programmable IC 132. In the example of Figure 12, a portion of the static domain 202 is shown, including the NIC controller 208 and the stream traffic manager 1002. Other elements of the static domain 202 are not shown for ease of illustration. Additionally, the communication interfaces between components of the NIC controller 208 are shown more generally, as opposed to showing separate input and output communication interfaces.

[0114] The architecture shown in Figure 12 is for illustrative purposes to show more complex and powerful processing that can be used to extend the capabilities of the platform 218 and the NIC controller 208. In this example, the dynamic area 204 is subdivided into multiple domains. Each domain is dedicated to a specific function. For example, the dynamic area 2044 includes a NIC extension domain 1202, a storage domain 1204, a stream domain 1206, and a memory mapped domain 1208. In one aspect, each domain is implemented as a different PR region.

[0115] The NIC extension domain 1202 may include a stream interface 1210 connected to the acceleration arbitration circuit 908. The stream interface 1210 is connected to the RDMA circuit 1212. The stream interface 1210 is further connected to a stream interconnect 1230 and a stream-memory mapped interface 1228 in the stream domain 1206. The stream-memory mapped interface 1228 bridges the domains 1206 and 1208. For example, the stream-memory map circuit of the interface 1228 may be located in the domain 1206 and the memory-to-stream map circuit of the interface 1228 may be located in the domain 1208. The NIC extension domain 1202 further includes an algorithm search circuit 1214 connected to the NIC core 906 and a layer 4 processing and search circuit 1216 connected to the BITW interface 904.

[0116] The storage domain 1204 includes an NVMe EP 1218 connected to the stream traffic manager 1002. The NVMe EP 1218 further connects to an NVMe-NVMeoF 1220 and an NVMe / aggregation / encryption circuit 1222. The NVMe-NVMeoF 1220 further connects to an RDMA 1212 of the NIC extension domain 1202. The NVMe / aggregation / encryption circuit 1222 connects to a PCIe RP 1224 and a PCIe 1226 also included in the storage domain 1204. The PCIe RPs 1224, 1226 can be connected to external non-volatile storage (e.g., a non-volatile storage device 136 such as an SSD).

[0117] The stream domain 1206 includes a stream interconnect 1230 connected to the stream traffic manager 1002. The stream domain 1206 also includes one or more stream kernels 1232 connected to the stream interconnect 1230. The memory mapped domain 1208 includes a memory mapped interconnect 1234 connected to the stream traffic manager 1002. The memory mapped domain 1208 further includes one or more memory mapped kernels 1236 connected to the memory mapped interconnect 1234. The memory mapped interconnect 1234 can be connected to one or more external DDR memories (e.g., volatile memory 134) via one or more memory controllers (not shown). The stream-memory mapped interface 1228 can also be connected to one or more external DDR memories (e.g., volatile memory 134) via one or more memory controllers (not shown).

[0118] 12 is provided for purposes of illustration and not limitation. In other examples, one or more of the domains or any combination may be a dynamic region 20 4 At any given time, the programmable IC 132 may be implemented within the programmable IC 132. Additionally, dynamic partial reconfiguration may be used to add and / or remove specific domains as needed during operation of the programmable IC 132.

[0119] For example, when the platform is implemented in the programmable IC 132, a first partial configuration bitstream is loaded into the programmable IC 132 to reconfigure the dynamic region 20 to include only the stream domain 1206 and the memory mapped domain 1208. 4 At another point in time, a second partial configuration bitstream may be loaded into the programmable IC 132 to configure the dynamic region 20 to implement the storage domain 1204, the stream domain 1206, and the memory mapped domain 1208. 4At another point in time, a third partial configuration bitstream may be loaded into programmable IC 132 to configure dynamic region 20 to implement NIC expansion domain 1202, storage domain 1204, stream domain 1206, and memory mapped domain 1208. 4 can be constructed.

[0120] In another aspect, each domain is implemented as a different PR region. Thus, the circuitry implemented in each domain can be modified over time without disrupting the circuitry of other domains. When domains interface with each other, data flow can be temporarily halted until the connecting PR region has undergone a partial reconfiguration.

[0121] 13 illustrates an example method 1300 of operating a programmable IC that implements the smart NIC functionality described herein. In one aspect, the operations described in connection with FIG. 13 relating to the configuration of the programmable IC may be performed or initiated by a host system.

[0122] In block 1302, the programmable IC implements the platform in the device. The platform is implemented at least partially in the static circuitry of the programmable IC. For example, the host system can write a configuration bitstream (e.g., a full or partial configuration bitstream) that specifies the platform to the non-volatile memory 136. The programmable IC can bootstrap itself and load the platform's configuration bitstream at boot time. The platform's configuration bitstream can remain resident in the programmable IC during operation. As described, the configuration bitstream defines a static region and the platform is at least partially physically implemented in the static region. The loading of the configuration bitstream in block 1302 can also define a dynamic region of the device as a portion of the programmable IC that is not used to implement the platform.

[0123] In implementations in which a platform is implemented using one or more PR regions, the platform can be specified by loading multiple partial configuration bitstreams. For example, one partial configuration bitstream can be loaded that implements a portion of the platform in the static region. For each PR region used to implement the platform, a separate partial configuration bitstream is loaded for that PR region.

[0124] In block 1304, the programmable IC establishes a communication link with a host system and a network. For example, a PCIe controller of the platform establishes a communication link with the host system. An Ethernet interface of the platform establishes a communication link with a network, for example, a node or a switch in the network. It should be understood that when the platform is implemented in the programmable IC, the programmable IC can transmit data to and from the host system and the network.

[0125] At block 1306, the programmable IC implements the user-specified circuitry in the dynamic region of the device. For example, the host system may initiate a partial reconfiguration of the programmable IC by loading a partial configuration bitstream therein. The partial configuration bitstream implements the user-specified circuitry in the dynamic region of the device. If the dynamic region includes a single PR region, a single partial configuration bitstream may be loaded. If the dynamic region includes multiple PR regions, the host system may initiate loading multiple partial configuration bitstreams into the programmable IC. For example, one partial configuration bitstream may be loaded per PR region. In either case, one or more kernels may be implemented in each PR region of the dynamic region to implement the user-specified circuitry.

[0126] In one aspect, as part of the implementation of the user-specified circuitry, the communication interfaces are selectively enabled and / or disabled based on metadata associated with the loaded partial configuration bitstream. If the dynamic region includes multiple PR regions, the partial configuration bitstream for a given PR region may include the metadata necessary to program the communication interfaces of the NIC controller that couples to that particular PR region.

[0127] At block 1308, the programmable IC may perform one or more operations using the user-specified circuitry as described within this disclosure. At block 1308, the user-specified circuitry is operable to process data asymmetrically for both output data and input data. That is, the input data may be processed by one or more kernels in the dynamic domain or not processed at all using such kernels, while the output data may be processed by one or more different kernels in the dynamic domain independent of the processing performed on the input data or not processed at all by such kernels.

[0128] In block 1308, the left block shows the processing of input data, and the right block shows the processing of output data. In block 1310, the programmable IC receives data from the network. The Ethernet interface receives the data and provides the data to the NIC controller. In block 1312, the NIC controller in the platform can exchange data with user-specified circuits in the dynamic domain. The data is sent from one or more of the software-controlled communication interfaces, processed by the user-specified circuits, and returned to the platform. In block 1314, the platform sends the processed data to the host system. For example, the data is provided to a DMA circuit and a PCIe controller. The PCIe controller provides the processed data to the host system.

[0129] In block 1316, the programmable IC receives data from the host system. The PCIe controller receives data from the host system and provides the data to a DMA circuit. The DMA circuit provides the data to one or more other circuits in the platform. In block 1318, the NIC controller can exchange data with user-specified circuits in the dynamic domain. The data is sent from one or more of the software-controlled communication interfaces, processed by the user-specified circuits, and returned to the platform. In block 1320, the platform sends the processed data to a network. For example, the data is provided to an Ethernet interface. The Ethernet interface provides the processed data to the network.

[0130] In one aspect, one or more of the software-controlled communication interfaces of the platform can be turned on or off as part of block 1308 or at any time under the control of the processor described herein. For example, each of the software-controlled communication interfaces can be in a particular state (on or off) upon implementation of the dynamic region. As the programmable IC continues to operate, the software-controlled communication interfaces can be turned on or off independently of each other and independent of partial reconfiguration of the dynamic region.

[0131] For example, a given software-controlled communications interface may be turned on and then turned off while the dynamic region continues to operate uninterrupted. In some cases, a software-controlled communications interface may be deactivated when a portion of the dynamic region connected to it (e.g., a particular PR region) is undergoing a partial reconfiguration. Once the partial reconfiguration of the PR region is successful, the software-controlled communications interface may be turned on again. In either case, dynamic control over communications interfaces as described allows for the data flow through the dynamic region to be dynamically altered without initiating a partial reconfiguration within the dynamic region.

[0132] At block 1322, a determination is made as to whether a trigger event is detected. In one aspect, the host system may receive a request for a different type of user-specified circuit to be implemented within the programmable IC, and the request is considered a trigger event for partial reconfiguration. In another example, the trigger event may be a signal or combination of signals detected by the programmable IC itself that causes the programmable IC to notify the host system that a partial reconfiguration is required. In either case, in response to detecting a trigger event, the method 1300 loops back to block 1306. In response to not detecting a trigger event, the method 1300 proceeds to block 1324, where the programmable IC continues to operate in the current configuration for both the platform and the dynamic domain. The method 1300 may iterate between blocks 1322 and 1324 until a trigger event is detected.

[0133] In block 1306, the programmable IC is partially reconfigured. The host system, for example, loads one or more other partial configuration bitstreams into the programmable IC. The programmable IC can implement completely new user-specified circuitry in the dynamic region (e.g., if the dynamic region includes a single PR region) or only new / different user-specified circuitry in one or more selected or all PR regions of the dynamic region. Additionally, the number of PR regions implemented in the dynamic region can be changed compared to a conventional configuration of the dynamic region.

[0134] The inventive arrangements described herein depict a platform implemented in a programmable IC that supports dynamic, e.g., run-time, insertion and / or replacement of functions according to network traffic. The programmable IC enables the platform to continue passing network traffic between the network and the host system without data loss. The inserted functions can be inserted at any of a variety of insertion points in the data flow through the platform.

[0135] 14 illustrates an exemplary method 1400 of implementing a platform on a programmable IC. At block 1402, a platform is provided on a device, at least partially implemented in a static region of a programmable circuit. At block 1404, a dynamic region of the programmable circuit is provided on the device, the dynamic region being configured to implement a user-specified circuit that communicates with the platform. The platform is configured to establish and maintain a first communication link with a host data processing system and a second communication link with a network while at least a portion of the dynamic region of the programmable circuit is being dynamically reconfigured.

[0136] FIG. 15 illustrates an exemplary architecture 1500 of an IC. The architecture 1500 can be used to implement a programmable IC described herein, such as the programmable IC 132. As an example, the architecture 1500 can be used to implement a field programmable gate array (FPGA). The architecture 1500 can also represent a system-on-chip (SoC) type of IC. An SoC is an IC that includes a processor that executes program code and one or more other circuits. The other circuits may be implemented as hardwired circuits, programmable circuits, and / or combinations thereof. The circuits can operate in cooperation with each other and / or with the processor.

[0137] As shown, architecture 1500 includes several different types of programmable circuits, e.g., logic blocks. For example, architecture 1500 may include a number of different programmable tiles including multi-gigabit transceivers (MGTs) 1501, configurable logic blocks (CLBs) 1502, random access memory blocks (BRAMs) 1503, input / output blocks (IOBs) 1504, configuration and clocking logic (CONFIG / CLOCKS) 1505, digital signal processing blocks (DSPs) 1506, specialized I / O blocks 1507 (e.g., configuration and clock ports), and other programmable logic 1508, such as digital clock managers, analog-to-digital converters, system monitoring logic, etc.

[0138] In some ICs, each programmable tile includes a programmable INT 1511 with standardized connections between corresponding interconnect elements (INTs) 1511 in each adjacent tile. Together, the INTs 1511 thus implement the programmable interconnect structure for the illustrated IC. Each INT 1511 also includes connections with programmable logic elements in the same tile, as shown by the example included at the top of FIG.

[0139] For example, the CLB 1502 may include a configurable logic element (CLE) 1512 that can be programmed to implement user logic in addition to a single INT 1511. The BRAM 1503 may include a BRAM logic element (BRL) 1513 in addition to one or more INTs 1511. Typically, the number of INTs 1511 included in a tile depends on the height of the tile. As shown, the BRAM tile has the same height as five CLBs, although other numbers (e.g., four) may be used. The DSP tile 1506 may include a DSP logic element (DSPL) 1514 in addition to an appropriate number of INTs 1511. The IOB 1504 may include, for example, one instance of an INT 1511 in addition to two instances of an I / O logic element (IOL) 1515. The actual I / O pads connected to the IOL 1515 may not be limited to the area of ​​the IOL 1515.

[0140] 15, a horizontal region near the center of the die formed, for example, from regions 1505, 1507, and 1508 can be used for configuration, clocks, and other control logic. Vertical region 1509 extending from this horizontal region can be used to distribute clock and configuration signals across the width of the programmable IC.

[0141] Some ICs utilizing the architecture shown in Figure 15 include additional logic blocks that break up the regular columnar structure that makes up the bulk of the IC. The additional logic blocks may be programmable blocks and / or dedicated circuits. For example, the processor block shown as PROC1510 spans several columns of CLBs and BRAMs.

[0142] In one aspect, PROC 1510 may be implemented as a dedicated circuit, e.g., a hardwired processor, fabricated as part of a die that implements the programmable circuitry of an IC. PROC 1510 may represent any of a variety of different processor types and / or systems ranging in complexity from an individual processor, e.g., a single core capable of executing program code, to an entire processor system having one or more cores, modules, coprocessors, interfaces, etc.

[0143] In another aspect, PROC 1510 may be omitted from architecture 1500 and replaced with one or more of the other types of programmable blocks described. Furthermore, such blocks may be utilized to form a "soft processor," in that various blocks of programmable circuitry may be used to form a processor that is capable of executing program code, as in the case of PROC 1510.

[0144] The phrase "programmable circuitry" refers to programmable circuit elements within an IC, such as the various programmable or configurable circuit blocks or tiles described herein, as well as the interconnect circuitry that selectively couples the various circuit blocks, tiles, and / or elements according to configuration data loaded into the IC. For example, the circuit blocks shown in FIG. 15 that are external to PROC 1510, such as CLB 1502 and BRAM 1503, are considered programmable circuitry of the IC.

[0145] Generally, the functionality of a programmable circuit is not established until configuration data is loaded into the IC. A set of configuration bits may be used to program the programmable circuit of an IC, such as an FPGA. The configuration bits are typically referred to as a "configuration bitstream." Generally, a programmable circuit will not operate or function without first loading a configuration bitstream into the IC. The configuration bitstream effectively implements a particular circuit design in the programmable circuit. The circuit design specifies, for example, the functional aspects of the programmable circuit blocks and the physical connections between the various programmable circuit blocks.

[0146] "Hardwired" or "hardened", i.e., non-programmable, circuits are fabricated as part of an IC. Unlike programmable circuits, hardwired circuits or circuit blocks are not implemented by the loading of a configuration bitstream after the IC is fabricated. Hardwired circuits are generally considered to have dedicated circuit blocks and interconnects that function without first loading a configuration bitstream into an IC, e.g., PROC1510.

[0147] In some cases, a hardwired circuit may have one or more operating modes that can be set or selected according to register settings or values ​​stored in one or more memory elements within the IC. The operating mode may be set, for example, by loading a configuration bitstream into the IC. Despite this capability, hardwired circuits are not considered programmable circuits because they are operational and have a specific function when they are fabricated as part of an IC.

[0148] In the case of an SoC, the configuration bitstream may specify circuits to be implemented within the programmable circuitry and program code to be executed by PROC 1510 or a soft processor. In some cases, architecture 1500 includes a dedicated configuration processor that loads the configuration bitstream into appropriate configuration memory and / or processor memory. The dedicated configuration processor does not execute user-specified program code. In other cases, architecture 1500 may utilize PROC 1510 to receive the configuration bitstream, load the configuration bitstream into appropriate configuration memory, and / or extract the program code for execution.

[0149] FIG. 15 is intended to illustrate an exemplary architecture that may be used to implement a programmable circuit, e.g., an IC including a programmable fabric. For example, the number of logic blocks in a column, the relative width of the columns, the number and order of columns, the types of logic blocks included in the columns, the relative sizes of the logic blocks, and the interconnect / logic implementation included in the upper portion of FIG. 15 are purely exemplary. In an actual IC, for example, two or more adjacent columns of CLBs will typically be included wherever a CLB appears to facilitate efficient implementation of a user circuit design. However, the number of adjacent CLB columns may vary depending on the overall size of the IC. Additionally, the size and / or placement of blocks such as PROC1510 within an IC are for illustrative purposes only and are not intended as limitations.

[0150] For purposes of explanation, specific nomenclature is set forth in order to provide a thorough understanding of the various inventive concepts disclosed herein. However, the terminology used herein is for the purpose of describing particular aspects of the inventive configurations only and is not intended to be limiting.

[0151] As defined herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly dictates otherwise.

[0152] As defined herein, the terms "at least one," "one or more," and "and / or" are open-ended expressions that are both conjunctive and disjunctive in operation unless otherwise specified. For example, the phrases "at least one of A, B, and C," "at least one of A, B, or C," "one or more of A, B, and C," "one or more of A, B, or C," and "A, B, and / or C" mean A only, B only, C only, A and B together, A and C together, B and C together, or A, B, and C together, respectively.

[0153] As defined herein, the term "automatically" means without human intervention.

[0154] As defined herein, the term "computer-readable storage medium" means a storage medium that contains or stores program code for use by or in connection with an instruction execution system, apparatus, or device. As defined herein, a "computer-readable storage medium" is not a transitory propagating signal itself. A computer-readable storage medium may be, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. The various forms of memory described herein are examples of computer-readable storage media. A non-exhaustive list of more specific examples of computer-readable storage media may include portable computer diskettes, hard disks, RAM, read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), electronically erasable programmable read-only memory (EEPROM), static random access memory (SRAM), portable compact disk read-only memory (CD-ROM), digital versatile disk (DVD), memory stick, floppy disk, and the like.

[0155] As defined herein, the term "when" means "when" or "in the course of" or "in response to" or "reacting," depending on the context. Thus, the phrase "when determined" or "when [a described condition or event] is detected" can be interpreted to mean "upon determining" or "in response to determining" or "upon detecting [a described condition or event]" or "in response to detecting [a described condition or event]" or "in response to detecting [a described condition or event]," depending on the context.

[0156] As defined herein, the term "responsive to" and similar terms such as "when," "when," or "in the course of" refer to readily responding to or reacting to an action or event. The response or reaction is automatic. Thus, when a second action is performed "in response" to a first action, there is a causal relationship between the occurrence of the first action and the occurrence of the second action. "In response" indicates a causal relationship.

[0157] As defined herein, the term "processor" means at least one hardware circuit. The hardware circuit may be configured to execute instructions contained in a program code. The hardware circuit may be an integrated circuit. Examples of processors include, but are not limited to, a central processing unit (CPU), an array processor, a vector processor, a digital signal processor (DSP), an FPGA, a programmable logic array (PLA), an ASIC, a programmable logic circuit, and a controller.

[0158] As defined herein, the term "real-time" means a level of processing responsiveness that a user or system perceives as immediate enough for a particular process or decision to occur, or that allows the processor to keep up with some external process.

[0159] As defined herein, the term "substantially" means that the recited property, parameter, or value need not be achieved exactly, but that deviations or variations including, for example, tolerances, measurement errors, measurement accuracy limits, and other factors known to those of ordinary skill in the art, may occur in an amount that does not eliminate the effect that the property is intended to provide.

[0160] Terms such as first, second, etc. may be used herein to describe various elements. Unless otherwise specified or the context clearly indicates otherwise, these elements should not be limited by these terms as they are only used to distinguish one element from another.

[0161] The computer program product may include a computer-readable storage medium having computer-readable program instructions for causing a processor to execute aspects of the inventive arrangements described herein. Within this disclosure, the term "program code" is used interchangeably with the term "computer-readable program instructions." The computer-readable program instructions described herein may be downloaded from the computer-readable storage medium to the respective computing / processing device or to an external computer or external storage device via a network, such as the Internet, a LAN, a WAN, and / or a wireless network. The network may include edge devices including copper transmission cables, optical transmission fibers, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in the respective computing / processing device.

[0162] The computer readable program instructions for performing the operations for the inventive arrangements described herein may be either assembler instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, or source or object code written in any combination of one or more programming languages, including object-oriented and / or procedural programming languages. The computer readable program instructions may include state setting data. The computer readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer, partially on a remote computer, or entirely on a remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer via any type of network, including a LAN or WAN, or may be connected to an external computer (e.g., via the Internet using an Internet Service Provider). In some cases, for example, an electronic circuit, including a programmable logic circuit FPGA, or PLA, may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuit to perform aspects of the inventive arrangements described herein.

[0163] Certain aspects of the inventive arrangement are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer readable program instructions, e.g., program code.

[0164] These computer-readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions executing via the processor of the computer or other programmable data processing apparatus create means for performing the functions / operations specified in one or more blocks of the flowcharts and / or block diagrams. These computer-readable program instructions may also be stored on a computer-readable storage medium capable of directing a computer, programmable data processing apparatus, and / or other device to function in a particular manner, such that a computer-readable storage medium having instructions stored thereon includes a product including instructions for performing aspects of the operations specified in one or more blocks of the flowcharts and / or block diagrams.

[0165] The computer readable program instructions may also be loaded into a computer, other programmable data processing apparatus, or other device to cause the computer, other programmable apparatus, or other device to perform a sequence of operations to generate a computer-implemented process, such that the instructions executing on the computer, other programmable apparatus, or other device perform the functions / operations specified in one or more blocks of the flowcharts and / or block diagrams.

[0166] The flowcharts and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various aspects of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions that includes one or more executable instructions for performing the specified operations.

[0167] In some alternative implementations, the operations noted in the blocks may occur in a different order than that depicted in the figures. For example, two blocks shown in succession may be executed substantially simultaneously, or the blocks may sometimes be executed in the reverse order, depending on the functionality involved. In other examples, the blocks may generally be executed in ascending order, while in still other examples, one or more blocks may be executed in various orders, with the results stored and utilized in a subsequent block or in another block that does not immediately follow. It should also be noted that each block of the block diagrams and / or flowchart illustrations, and combinations of blocks in the block diagrams and / or flowchart illustrations, may be implemented by a dedicated hardware-based system that performs the specified functions or operations or executes a combination of dedicated hardware and computer instructions.

[0168] The corresponding structure, materials, acts, and equivalents of all means-plus-function or step-plus-function elements that may be found in the following claims are intended to include any structure, material, or act for performing a function in combination with other claimed elements that are specifically claimed.

[0169] The device includes a platform implemented at least partially in a static region of the programmable circuitry and a dynamic region of the programmable circuitry configured to implement a user-specified circuit in communication with the platform, the platform configured to establish and maintain a first communication link with a host data processing system and a second communication link with a network while at least a portion of the dynamic region of the programmable circuitry is being dynamically reconfigured.

[0170] Each of the above and other embodiments can optionally include one or more of the following features, either alone or in combination. Some exemplary embodiments include all of the following features in combination.

[0171] In another aspect, the platform includes an Ethernet interface coupled to the NIC controller, a DMA circuit coupled to the network interface card controller, and a PCIe interface coupled to the DMA circuit.

[0172] In another aspect, at least one of the Ethernet interface, the DMA circuitry, or the network interface card controller is implemented in a partially reconfigurable domain.

[0173] In another aspect, the user-specified circuitry is configured to perform packet processing. In another aspect, the platform includes a NIC controller having a communication interface connected to the dynamic domain.

[0174] In another aspect, the communication interface is disposed between the MAC and the first stage of the NIC controller.

[0175] In another aspect, the communication interface is software-activatable. In another aspect, the NIC controller includes a plurality of interconnected stages, at least one of the plurality of interconnected stages having a communication interface.

[0176] In another aspect, a first stage and a second stage of the plurality of interconnected stages each have an independent communication interface configured to connect to the dynamic domain.

[0177] In another aspect, each communication interface connects to a different sub-reconfigurable domain within the dynamic domain, each sub-reconfigurable domain being independently reconfigurable to implement a different user-specified circuit.

[0178] In another aspect, the platform continues to operate uninterrupted while the dynamic regions are reconfigured to implement different user-specified circuitry in communication with the platform.

[0179] The method includes providing a platform on the device at least partially implemented in a static region of a programmable circuit, and providing a dynamic region of the programmable circuit on the device configured to implement a user-specified circuit that communicates with the platform, the platform configured to establish and maintain a first communication link with a host data processing system and a second communication link with a network while at least a portion of the dynamic region of the programmable circuit is being dynamically reconfigured.

[0180] Each of the above and other embodiments can optionally include one or more of the following features, either alone or in combination. Some exemplary embodiments include all of the following features in combination.

[0181] In another aspect, a platform includes an Ethernet interface coupled to the NIC controller, a DMA circuit coupled to the NIC controller, and a PCIe interface coupled to the DMA circuit.

[0182] In another aspect, at least one of the Ethernet interface, the DMA circuitry, or the network interface card controller is implemented in a partially reconfigurable domain.

[0183] In another aspect, the user-specified circuitry is configured to perform packet processing. In another aspect, a method includes providing, in a platform, a NIC controller having a communication interface connected to a dynamic domain.

[0184] In another aspect, the communication interface is disposed between the MAC and the first stage of the NIC controller.

[0185] In another aspect, the method includes selectively activating a communication interface under software control.

[0186] In another aspect, a method includes providing a network interface card controller with a plurality of interconnected stages and providing a communication interface for at least one of the plurality of interconnected stages.

[0187] In another aspect, a method includes providing a first stage and a second stage of a plurality of interconnected stages, the first stage and the second stage each having an independent communication interface configured to connect to a dynamic domain.

[0188] In another aspect, each communication interface connects to a different sub-reconfigurable domain within the dynamic domain, each sub-reconfigurable domain being independently reconfigurable to implement a different user-specified circuit.

[0189] In another aspect, the method includes reconfiguring the dynamic region to implement different user-specified circuitry in communication with the platform while the platform continues to operate uninterrupted.

[0190] The IC may include a platform at least partially implemented in a static region of the programmable circuit, the platform including a network interface card controller coupled to a host interface and a network interface. The IC may also include a dynamic region of the programmable circuit configured to implement a user-specified circuit that communicates with the platform. The user-specified circuit may implement one or more packet processing functions. The network interface card controller includes a plurality of communication interfaces connected to the dynamic region. The platform may be configured to establish and maintain a first communication link with a host data processing system using the host interface and a second communication link with a network using the network interface. The platform may be configured to continue to transfer data between the network and the host data processing system while at least a portion of the dynamic region of the programmable circuit is dynamically reconfigured to implement different user-specified circuits therein.

[0191] Each of the above and other embodiments can optionally include one or more of the following features, either alone or in combination. Some exemplary embodiments include all of the following features in combination.

[0192] In another aspect, the network interface includes an Ethernet interface. In another aspect, the network interface is coupled to an arbitration circuit located in a static domain between a first stage of the network interface card controller and the network interface, and a first communication interface of the plurality of communication interfaces connects a user-specified circuit in the dynamic domain to the arbitration circuit.

[0193] In another aspect, the arbitration circuit is configured to route data received from the network interface back to the network interface to the network without providing the data to a network interface card controller.

[0194] In another aspect, the arbitration circuit routes the data back to the network interface after processing by the user-specified circuitry.

[0195] In another aspect, the arbitration circuit is configured to route data received from the network interface card controller back to the network interface card controller without providing the data to the network interface.

[0196] In another aspect, the arbitration circuit routes the data back to the network interface card controller after processing by the user-specified circuitry.

[0197] In another aspect, the multiple communication interfaces are selectively enabled by the host data processing system.

[0198] In another aspect, the network interface card controller includes multiple stages, and each communication interface of the multiple communication interfaces connects to a different stage of the multiple stages.

[0199] In another aspect, the dynamic domain includes a plurality of partially reconfigurable domains, each communication interface connecting to a different one of the plurality of partially reconfigurable domains, each of the plurality of partially reconfigurable domains being independently reconfigurable to implement different user-specified circuitry therein.

[0200] In another embodiment, the host interface is a PCIe interface. In another aspect, the IC can include a DMA circuit implemented in the static domain, the DMA circuit being connected to the network interface card controller and the host interface. The memory controller may be implemented in the static domain. The memory controller may be coupled to the DMA circuit and user-specified circuitry in the dynamic domain.

[0201] In another aspect, a network interface card controller includes a network interface core connected to a network interface and an accelerator arbitration circuit connected to a DMA circuit. The acceleration arbitration circuit couples to one or more of a plurality of partial reconfiguration regions of a dynamic region via a stream interface. Each partial reconfiguration region can implement a different function.

[0202] In another aspect, the stream interface is implemented in a first of the plurality of partial reconfiguration domains and connects to the accelerator arbitration circuitry of the network interface controller, and the user-specified circuitry may be implemented in one or more other of the plurality of partial reconfiguration domains.

[0203] In another aspect, one or more of the multiple partial reconfiguration domains include one or more kernel circuits configured to perform computational hardware accelerated tasks offloaded from a host data processing system.

[0204] The description of the inventive configurations provided herein is for illustrative purposes and is not intended to be exhaustive or limited to the disclosed forms and examples. The terms used herein are selected to explain the principles of the inventive configurations, practical applications or technical improvements to the technology found in the market, and / or to enable those skilled in the art to understand the inventive configurations disclosed herein. Modifications and variations may be apparent to those skilled in the art without departing from the scope and spirit of the inventive configurations described. Accordingly, reference should be made to the following claims, rather than the foregoing disclosure, as indicating the scope of such features and embodiments.

Claims

1. 1. An integrated circuit comprising: a platform implemented at least partially in a static domain of programmable circuitry, the platform including a network interface card controller coupled to a host interface and a network interface; a dynamic region of programmable circuitry configured to implement user-specified circuitry in communication with the platform, the user-specified circuitry implementing one or more packet processing functions; the network interface card controller includes a plurality of communication interfaces coupled to the dynamic area; the platform is configured to establish and maintain a first communication link with a host data processing system using the host interface and a second communication link with a network using the network interface; the platform is configured to continue transferring data between the network and the host data processing system while at least a portion of the dynamic region of the programmable circuit is being dynamically reconfigured to implement different user-specified circuits therein; the network interface is coupled to an arbitration circuit located in the static area between a core circuit of the network interface card controller and the network interface; An integrated circuit, wherein a first communication interface of the plurality of communication interfaces connects the user-specified circuitry in the dynamic domain to the arbitration circuitry.

2. The integrated circuit of claim 1 , wherein the network interface comprises an Ethernet interface.

3. 2. The integrated circuit of claim 1, wherein the arbitration circuit is configured to route data received from the network interface back to the network interface to the network.

4. The integrated circuit of claim 3 , wherein the arbitration circuit routes the data back to the network interface after processing by the user-specified circuitry.

5. The integrated circuit of claim 1 , wherein the arbitration circuit is configured to route data received from the network interface card controller back to the network interface card controller.

6. 5. The integrated circuit of claim 4, wherein the arbitration circuit routes the data back to the network interface card controller after processing by the user-specified circuitry.

7. 2. The integrated circuit of claim 1, wherein said multiple communication interfaces are selectively enabled by said host data processing system.

8. The integrated circuit of claim 1 , wherein the network interface card controller includes multiple stages, and each communication interface of the multiple communication interfaces connects to a different stage of the multiple stages.

9. the dynamic domain includes a plurality of partial reconfiguration domains, each communication interface being connected to a different one of the plurality of partial reconfiguration domains; each of the plurality of partial reconfigurable regions is independently reconfigurable to implement different user-specified circuitry therein; 9. The integrated circuit of claim 8.

10. 10. The integrated circuit of claim 1, wherein the host interface is a Peripheral Component Interconnect (PCIe) interface.

11. a direct memory access circuit implemented in the static domain, the direct memory access circuit being connected to the network interface card controller and the host interface; a memory controller implemented in the static domain, the memory controller being coupled to the direct memory access circuitry and the user specified circuitry in the dynamic domain; 10. The integrated circuit of claim 1.

12. An integrated circuit comprising: a platform implemented at least partially in a static domain of programmable circuitry, the platform including a network interface card controller coupled to a host interface and a network interface; a dynamic region of programmable circuitry configured to implement user-specified circuitry in communication with the platform, the user-specified circuitry implementing one or more packet processing functions; a direct memory access circuit implemented in the static domain, the direct memory access circuit being connected to the network interface card controller and the host interface; a memory controller implemented in the static domain, the memory controller coupled to the direct memory access circuitry and the user specified circuitry in the dynamic domain; the network interface card controller includes a plurality of communication interfaces coupled to the dynamic area; the platform is configured to establish and maintain a first communication link with a host data processing system using the host interface and a second communication link with a network using the network interface; the platform is configured to continue transferring data between the network and the host data processing system while at least a portion of the dynamic region of the programmable circuit is being dynamically reconfigured to implement different user-specified circuits therein; an accelerator arbitration circuit coupled to the direct memory access circuitry, the accelerator arbitration circuit coupled to one or more of a plurality of partial reconfigurable regions of the dynamic domain via a stream interface, each partial reconfigurable region implementing a different function.

13. 13. The integrated circuit of claim 12, wherein the stream interface is implemented in a first of the plurality of partial reconfiguration domains and connects to the accelerator arbitration circuitry of the network interface card controller and the user-specified circuitry implemented in one or more other of the plurality of partial reconfiguration domains.

14. 13. The integrated circuit of claim 12, wherein one or more of the plurality of partial reconfigurable domains includes one or more kernel circuits configured to perform computational hardware accelerated tasks offloaded from the host data processing system.

Citation Information

Patent Citations

  • Programmable logic device, card, and transmission apparatus

    JP2010074461A

  • Communication processing device, communication method, and communication system

    JP2017069777A

  • Integration of a programmable device and a processing system in an integrated circuit package

    US20190096813A1

  • Partial reconfiguration using configuration transaction layer packets

    US9257987B1