Devices for coordinating or establishing a direct memory access for a network interface card to a graphics processing unit, and for a network interface card to access a graphics processing unit via a direct memory access are disclosed. For example, a central processing unit may request a graphics processing unit to allocate a memory buffer of the graphics processing unit for a direct memory access by a network interface card and receive from the graphics processing unit a first confirmation of an allocation of the memory buffer. The central processing unit may further transmit to the network interface card a first notification of the allocation of the memory buffer of the graphics processing unit, poll the network interface card to determine when a packet is received by the network interface card, and transmit a second notification to the graphics processing unit that the packet is written to the memory buffer.

Technologies for managing errors in a remotely accessible memory pool include a memory sled. The memory sled includes a memory pool having one or more byte-addressable memory devices and a memory pool controller coupled to the memory pool. The memory sled is to write test data to a byte-addressable memory region in the memory pool. The memory region is to be accessed by a remote compute sled. The memory sled is also to read data from the memory region to which the test data was written, compare the read data to the test data to determine whether a threshold number of errors are present in the read data, and send, in response to a determination that the threshold number of errors are present in the read data, a notification to the remote compute sled that the memory region is faulty.

Embodiments of the invention may improve the performance of multi-processor systems in processing information received via a network. For example, some embodiments may enable configuration of a system such that information received via a network may be distributed among multiple processors for efficient processing. A user (e.g., system administrator) may select from among multiple configuration options, each configuration option being associated with a particular mode of processing information received via a network. By selecting a configuration option, the user may specify how information received via the network is processed to capitalize on the system's characteristics, such as by aligning processors on the system with certain NICs. As such, the processor(s) aligned with a NIC may perform networking-related tasks associated with information received by that NIC. If initial alignment causes one or more processors to become over-burdened, processing tasks may be dynamically re-distributed to other processors so as to achieve a more even distribution of the overall processing burden across the system.

Technologies for dividing resources across partitions include a compute sled. The compute sled is to determine partitions among sockets of the compute sled. Each socket is associated with a corresponding processor. The compute sled is also to establish a separate memory space for each determined partition, obtain, from an application executed in one of the sockets, a request to access a logical memory address, identify the partition associated with the memory access request, determine a corresponding physical memory address as a function of the identified partition and the logical memory address, and access a memory of the compute sled at the determined physical memory address. Other embodiments are also described and claimed.

Technologies for allocating data storage capacity on a data storage sled include a plurality of data storage devices communicatively coupled to a plurality of network switches through a plurality of physical network connections and a data storage controller connected to the plurality of data storage devices. The data storage controller is to determine a target storage resource allocation to be used by one or more applications to be executed by one or more sleds in a data center, determine data storage capacity available for each of a plurality of different data storage types on the data storage sled, wherein each data storage type is associated with a different level of data redundancy, determine an amount of data storage capacity for each data storage type to be allocated to satisfy the target storage resource allocation, and adjust the amount of data storage capacity allocated to each data storage type.

Technologies for dynamic accelerator selection include a compute sled. The compute sled includes a network interface controller to communicate with a remote accelerator of an accelerator sled over a network, where the network interface controller includes a local accelerator and a compute engine. The compute engine is to obtain network telemetry data indicative of a level of bandwidth saturation of the network. The compute engine is also to determine whether to accelerate a function managed by the compute sled. The compute engine is further to determine, in response to a determination to accelerate the function, whether to offload the function to the remote accelerator of the accelerator sled based on the telemetry data. Also the compute engine is to assign, in response a determination not to offload the function to the remote accelerator, the function to the local accelerator of the network interface controller.

Embodiments herein describe creating tag bindings that can be used to assign tags to data corresponding to different tenants using a data processing unit (DPU) such as a SmartNIC, Artificial Intelligence Unit, Network Storage Unit, Database Acceleration Units, and the like. In one embodiment, the DPUs include tag gateways at the interface between a host and network element (e.g., a switch) that recognize and tag the data corresponding to the tenants. These tags are then recognized by data processing engines (DPEs) in the DPU such as AI engines, cryptographic engines, encryption engines, Direct Memory Access (DMA) engines, and the like. These DPEs can be configured to perform tag policies that provide security isolation and performance isolation between the tenants.

Embodiments of the invention may improve the performance of multi-processor systems in processing information received via a network. For example, some embodiments may enable configuration of a system such that information received can be distributed among multiple processors for efficient processing. A user may select from among multiple configuration options, each configuration option being associated with a particular mode of processing information received. By selecting a configuration option, the user may specify how received information is processed to capitalize on the system's characteristics, such as by aligning processors on the system with certain NICs. As such, the processor(s) aligned with a NIC may perform networking-related tasks associated with information received by that NIC. If initial alignment causes one or more processors to become over-burdened, processing tasks may be dynamically re-distributed to other processors.

A computer system, comprising a plurality of computers, each of the plurality of computers including at least one processor chip each including a plurality of processor cores, the at least one processor chip constructing a plurality of regions each constructed by at least one processor core, each of the plurality of processor cores carries out calculation processing for executing a predetermined program and inter-core communication processing, which is communication between the plurality of processor cores, the computer system comprising: a regulation module which controls a voltage and a frequency that are supplied to each of the plurality of regions; and a determination module which determines a power mode of each of the plurality of regions, to output an instruction to the regulation module.

Technologies for providing remote access to a shared memory pool include a memory sled. The memory sled includes a memory pool having one or more byte-addressable memory devices and a memory pool controller coupled to the memory pool. The memory pool controller is to produce, for each of a plurality of compute sleds, address space data indicative of addresses of byte-addressable memory in the memory pool accessible to the compute sled, and corresponding permissions associated with the addresses. The memory pool controller is also to provide the address space data to each corresponding compute sled and receive, from a requesting compute sled of the plurality of compute sleds, a memory access request. The memory access request includes an address from the address space data to be accessed. The memory pool controller is also to perform, in response to receiving the memory access request, a memory access operation on the memory pool. Other embodiments are also described and claimed.