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 techniques for preventing a stall when transmitting data between a producer and a consumer in the same integrated circuit (IC). A stall can occur when there is a split point and a convergence point between the producer and consumer. To prevent the stall, the embodiments herein adjust the latencies of one of the paths (or both paths) such that a maximum latency of the shorter path is greater than, or equal to, the minimum latency of the longer path. When this condition is met, this means the shortest path has sufficient buffers (e.g., a sufficient number of FIFOs and registers) to queue/store packets along its length so that a packet can travel along the longer path and reach the convergence point before the buffers in the shortest path are completely full (or just become completely full).

The technology disclosed relates to interconnect-based resource allocation for reconfigurable processors. In particular, the technology disclosed relates to a runtime logic that is configured to receive target interconnect bandwidth and target interconnect latency, and rated interconnect bandwidth and rated interconnect latency. The runtime logic is further configured to respond by allocating, to configuration files defining an application graph, processing elements in a plurality of processing elements, and interconnects between the processing elements, and executing the configuration files using the allocated processing elements and the allocated interconnects.

In a method for accessing an extended memory, after receiving a first memory access request from a processor system in a computer, an extended memory controller sends a read request for obtaining to-be-accessed data to the extended memory and return, to the processor system, a first response message indicating the to-be-accessed data has not been obtained. The extended memory controller writes the to-be-accessed data into a data buffer after receiving the to-be-accessed data returned by the extended memory. After receiving, from the processor system, a second memory access request comprising a second access address, the extended memory controller returns, to the processor system, the to-be-accessed data in the data buffer in response to the second memory access request, wherein the second access address is different from the first access address and points to the physical address of the to-be-accessed data.

Racks and rack systems to support a plurality of sleds are disclosed herein. A rack comprises an elongated support post and a plurality of support chassis. The elongated support post extends vertically. The plurality of support chassis are coupled to the elongated support post. Each support chassis of the plurality of support chassis is sized to house a corresponding sled of the plurality of sleds.

A computer system may comprise a multi-chip package (MCP), which includes multi-core processor circuitry and hardware accelerator circuitry. The multi-core processor circuitry may comprise a plurality of processing cores, and the hardware accelerator circuitry may be coupled with the multi-core processor circuitry via one or more coherent interconnects and one or more non-coherent interconnects. A coherency domain of the MCP may be extended to encompass the hardware accelerator circuitry, or portions thereof An interconnect selection module may select an individual coherent interconnect or an individual non-coherent interconnect based on application requirements of an application to be executed and a workload characteristic policy. Other embodiments are described and/or claimed.

A system is described that has a node and runtime logic. The node has a plurality of processing elements operatively coupled by interconnects. The runtime logic is configured to receive target interconnect bandwidth, target interconnect latency, rated interconnect bandwidth and rated interconnect latency. The runtime logic responds by allocating to configuration files defined by the application graph: (1) processing elements in the plurality of processing elements, and (2) interconnects between the processing elements. The runtime logic further responds by executing the configuration files using the allocated processing elements and the allocated interconnects.

One example method includes receiving an IO request from an application, determining if an affinity policy applies to the application that transmitted the IO request, when an affinity policy applies to the application, directing the IO request to a specified site of a replication system, when no affinity policy applies to the application, determining if a lag in replication of the IO request from a primary site to a replication site is acceptable, if a lag in replication of the IO request is acceptable, processing the IO request using performance based parameters and/or load balancing parameters, and if a lag in replication of the IO request is not acceptable, either directing the IO request to a most up to date replica site, or requesting a clone copy of a volume to which the IO request was initially directed and directing the IO request to the cloned copy.

Technologies for application-specific network acceleration include a computing device including a processor and an accelerator device such as a field-programmable gate array (FPGA). The processor and the accelerator device are coupled via a coherent interconnect and may be included in a multi-chip package. The computing device binds a virtual machine executed by the processor with an application function unit of the accelerator device via the coherent interconnect. The computing device processes network application data with the virtual machine and the application function unit within a coherency domain maintained with the coherent interconnect. Processing the network data may include processing a packet of a network flow by the virtual machine and processing subsequent packets of the network flow by the application function unit. Other embodiments are described and claimed.

Technologies for allocating resources across data centers include a compute device to obtain resource utilization data indicative of a utilization of resources for a managed node to execute a workload. The compute device is also to determine whether a set of resources presently available to the managed node in a data center in which the compute device is located satisfies the resource utilization data. Additionally, the compute device is to allocate, in response to a determination that the set of resources presently available to the managed node does not satisfy the resource utilization data, a supplemental set of resources to the managed node. The supplemental set of resources are located in an off-premises data center that is different from the data center in which the compute device is located. Other embodiments are also described and claimed.