Abstract: Systems, methods, and devices for isolating a misbehaving accelerator circuit, such as an accelerator function unit or an accelerated function context, are provided. An integrated circuit may include a region that includes an accelerator circuit. When the accelerator circuit issues a request, another region of the integrated circuit or a processor connected to the integrated circuit may determine whether there is a misbehavior associated with the request and, in response to determining that there is a misbehavior associated with the request, may perform a misbehavior response to mitigate a negative impact of the misbehavior of the accelerator circuit.

Abstract: A network interface controller can be programmed to direct write received data to a memory buffer via either a host-to-device fabric or an accelerator fabric. For packets received that are to be written to a memory buffer associated with an accelerator device, the network interface controller can determine an address translation of a destination memory address of the received packet and determine whether to use a secondary head. If a translated address is available and a secondary head is to be used, a direct memory access (DMA) engine is used to copy a portion of the received packet via the accelerator fabric to a destination memory buffer associated with the address translation. Accordingly, copying a portion of the received packet through the host-to-device fabric and to a destination memory can be avoided and utilization of the host-to-device fabric can be reduced for accelerator bound traffic.

Abstract: A system comprising an accelerator circuit comprising an accelerator function unit to implement a first function, and one or more device feature header (DFH) circuits to provide attributes associated with the accelerator function unit, and a processor to retrieve the attributes of the accelerator function unit by traversing a device feature list (DFL) referencing the one or more DFH circuits, execute, based on the attributes, an application encoding the first function to cause the accelerator function unit to perform the first function.

Abstract: Systems and devices can include a controller and a command queue to buffer incoming write requests into the device. The controller can receive, from a client across a link, a non-posted write request (e.g., a deferred memory write (DMWr) request) in a transaction layer packet (TLP) to the command queue; determine that the command queue can accept the DMWr request; identify, from the TLP, a successful completion (SC) message that indicates that the DMWr request was accepted into the command queue; and transmit, to the client across the link, the SC message that indicates that the DMWr request was accepted into the command queue. The controller can receive a second DMWr request in a second TLP; determine that the command queue is full; and transmit a memory request retry status (MRS) message to be transmitted to the client in response to the command queue being full.

Abstract: Systems and devices can include a controller and a command queue to buffer incoming write requests into the device. The controller can receive, from a client across a link, a non-posted write request (e.g., a deferred memory write (DMWr) request) in a transaction layer packet (TLP) to the command queue; determine that the command queue can accept the DMWr request; identify, from the TLP, a successful completion (SC) message that indicates that the DMWr request was accepted into the command queue; and transmit, to the client across the link, the SC message that indicates that the DMWr request was accepted into the command queue. The controller can receive a second DMWr request in a second TLP; determine that the command queue is full; and transmit a memory request retry status (MRS) message to be transmitted to the client in response to the command queue being full.

Abstract: A system comprising an accelerator circuit comprising an accelerator function unit to implement a first function, and one or more device feature header (DFH) circuits to provide attributes associated with the accelerator function unit, and a processor to retrieve the attributes of the accelerator function unit by traversing a device feature list (DFL) referencing the one or more DFH circuits, execute, based on the attributes, an application encoding the first function to cause the accelerator function unit to perform the first function.

Abstract: Systems, methods, and devices for isolating a misbehaving accelerator circuit, such as an accelerator function unit or an accelerated function context, are provided. An integrated circuit may include a region that includes an accelerator circuit. When the accelerator circuit issues a request, another region of the integrated circuit or a processor connected to the integrated circuit may determine whether there is a misbehavior associated with the request and, in response to determining that there is a misbehavior associated with the request, may perform a misbehavior response to mitigate a negative impact of the misbehavior of the accelerator circuit.

Abstract: There is disclosed in one example an apparatus, including: a plurality of interconnects to communicatively couple an accelerator device to a host device; and an address translation module (ATM) to provide address mapping between host-physical address (HPA) and guest-physical address (GPA) spaces for the accelerator device, wherein the plurality of devices share a common GPA domain and wherein address mapping is to be associated with only one of the plurality of interconnects.

Abstract: Systems, methods, and devices for isolating a misbehaving accelerator circuit, such as an accelerator function unit or an accelerated function context, are provided. An integrated circuit may include a region that includes an accelerator circuit. When the accelerator circuit issues a request, another region of the integrated circuit or a processor connected to the integrated circuit may determine whether there is a misbehavior associated with the request and, in response to determining that there is a misbehavior associated with the request, may perform a misbehavior response to mitigate a negative impact of the misbehavior of the accelerator circuit.

Abstract: Aspects of the embodiments are directed to systems, devices, and methods for accessing storage-as-memory. Embodiments include a microprocessor including a microprocessor system agent and a field programmable gate array (FPGA). The FPGA including an FPGA system agent to process memory access requests received from the microprocessor system agent across a communications link; a memory controller communicatively coupled to the system agent; and a high speed serial interface to link the system agent with a storage system. Embodiments can also include a storage device connected to the FPGA by the high speed serial interface.

Abstract: There is disclosed in one example an accelerator apparatus, including: a programmable region capable of being programmed to provide an accelerator function unit (AFU); and a platform interface layer (PIL) to communicatively couple to the AFU via an intra-accelerator protocol, and to provide multiplexed communication with a processor via a plurality of platform interconnect interfaces, wherein the PIL is to provide abstracted communication services for the AFU to communicate with the processor.

Abstract: Systems and devices can include a controller and a command queue to buffer incoming write requests into the device. The controller can receive, from a client across a link, a non-posted write request (e.g., a deferred memory write (DMWr) request) in a transaction layer packet (TLP) to the command queue; determine that the command queue can accept the DMWr request; identify, from the TLP, a successful completion (SC) message that indicates that the DMWr request was accepted into the command queue; and transmit, to the client across the link, the SC message that indicates that the DMWr request was accepted into the command queue. The controller can receive a second DMWr request in a second TLP; determine that the command queue is full; and transmit a memory request retry status (MRS) message to be transmitted to the client in response to the command queue being full.

Abstract: A network interface controller can be programmed to direct write received data to a memory buffer via either a host-to-device fabric or an accelerator fabric. For packets received that are to be written to a memory buffer associated with an accelerator device, the network interface controller can determine an address translation of a destination memory address of the received packet and determine whether to use a secondary head. If a translated address is available and a secondary head is to be used, a direct memory access (DMA) engine is used to copy a portion of the received packet via the accelerator fabric to a destination memory buffer associated with the address translation. Accordingly, copying a portion of the received packet through the host-to-device fabric and to a destination memory can be avoided and utilization of the host-to-device fabric can be reduced for accelerator bound traffic.

Abstract: A network interface controller can be programmed to direct write received data to a memory buffer via either a host-to-device fabric or an accelerator fabric. For packets received that are to be written to a memory buffer associated with an accelerator device, the network interface controller can determine an address translation of a destination memory address of the received packet and determine whether to use a secondary head. If a translated address is available and a secondary head is to be used, a direct memory access (DMA) engine is used to copy a portion of the received packet via the accelerator fabric to a destination memory buffer associated with the address translation. Accordingly, copying a portion of the received packet through the host-to-device fabric and to a destination memory can be avoided and utilization of the host-to-device fabric can be reduced for accelerator bound traffic.

Abstract: Systems and devices can include a controller and a command queue to buffer incoming write requests into the device. The controller can receive, from a client across a link, a non-posted write request (e.g., a deferred memory write (DMWr) request) in a transaction layer packet (TLP) to the command queue; determine that the command queue can accept the DMWr request; identify, from the TLP, a successful completion (SC) message that indicates that the DMWr request was accepted into the command queue; and transmit, to the client across the link, the SC message that indicates that the DMWr request was accepted into the command queue. The controller can receive a second DMWr request in a second TLP; determine that the command queue is full; and transmit a memory request retry status (MRS) message to be transmitted to the client in response to the command queue being full.

Abstract: In one embodiment, an apparatus includes an input/output virtualization (IOV) device comprising: at least one function circuit to be shared by a plurality of virtual machines (VMs); and a plurality of assignable device interfaces (ADIs) coupled to the at least one function circuit, wherein each of the plurality of ADIs is to be associated with one of the plurality of VMs and comprises a first process address space identifier (PASID) field to store a first PASID to identify a descriptor queue stored in a host address space and a second PASID field to store a second PASID to identify a data buffer located in a VM address space. Other embodiments are described and claimed.

Abstract: In accordance with embodiments disclosed herein, there is provided systems and methods for extending a root complex to encompass an external component. A processor includes a processor core and root complex circuitry coupled to the processor core. The processor core is to execute a basic input/output system (BIOS) and an operating system (OS). The root complex circuitry includes a coherent interface port and a downstream port. The root complex circuitry is to couple to an external component via the downstream port and the coherent interface port. The BIOS, to extend a root complex beyond the root complex circuitry to encompass the external component, is to obfuscate the downstream port from the OS, define a virtual root bridge for the external component, and enable a security check at the external component to provide protection for the coherent interface port and the downstream port.

Abstract: A system comprising an accelerator circuit comprising an accelerator function unit to implement a first function, and one or more device feature header (DFH) circuits to provide attributes associated with the accelerator function unit, and a processor to retrieve the attributes of the accelerator function unit by traversing a device feature list (DFL) referencing the one or more DFH circuits, execute, based on the attributes, an application encoding the first function to cause the accelerator function unit to perform the first function.

Abstract: Systems and devices can include a controller and a command queue to buffer incoming write requests into the device. The controller can receive, from a client across a link, a non-posted write request (e.g., a deferred memory write (DMWr) request) in a transaction layer packet (TLP) to the command queue; determine that the command queue can accept the DMWr request; identify, from the TLP, a successful completion (SC) message that indicates that the DMWr request was accepted into the command queue; and transmit, to the client across the link, the SC message that indicates that the DMWr request was accepted into the command queue. The controller can receive a second DMWr request in a second TLP; determine that the command queue is full; and transmit a memory request retry status (MRS) message to be transmitted to the client in response to the command queue being full.

Abstract: A network interface controller can be programmed to direct write received data to a memory buffer via either a host-to-device fabric or an accelerator fabric. For packets received that are to be written to a memory buffer associated with an accelerator device, the network interface controller can determine an address translation of a destination memory address of the received packet and determine whether to use a secondary head. If a translated address is available and a secondary head is to be used, a direct memory access (DMA) engine is used to copy a portion of the received packet via the accelerator fabric to a destination memory buffer associated with the address translation. Accordingly, copying a portion of the received packet through the host-to-device fabric and to a destination memory can be avoided and utilization of the host-to-device fabric can be reduced for accelerator bound traffic.