Abstract: A system obtains a performance profile corresponding to times taken to perform an inferencing by a machine learning (ML) model using a different number of processing resources from a plurality of processing resources. The system determines one or more groupings of processing resources from the plurality of processing resources, each grouping includes one or more partitions. The system calculates performance speeds corresponding to each grouping based on the performance profile. The system determines a grouping having a best performance speed from the calculated performance speeds. The system partitions the processing resources based on the determined grouping to perform the inferencing.

Abstract: A system determines a timing matrix corresponding to inference times taken for a number of machine learning (ML) models to be executed by a number of processing resources of a computing device. The processing resources includes at least a first and a second type of processing resources. The system applies a service-specific model-first scheduling scheme or a service-specific hardware-first scheduling scheme to obtain corresponding service-specific mappings. The system determines a best mapping from the corresponding service-specific mappings. The system schedules each of the ML models to a corresponding processing resource from the processing resources according to the best mapping. The system executes the ML models using corresponding mapped processing resources.

Abstract: Implementations of the present disclosure relate to packet fragmentation in Generic Routing Encapsulation (GRE). A method comprises in accordance with a determination that the target packet is to be fragmented, generating at least two packets from the target packet. The at least two packets comprise an Internet Protocol (IP) header, a GRE header and a fragment of the payload of the target packet, respectively. The GRE header comprises a fragment indication indicating that the respective packet comprises a respective fragment of the payload of the target packet. The method further comprises transmitting the at least two packets to a second network device. In this way, the packet fragment dropping due to security policies will be avoided.

Abstract: Systems are methods are provided for implementing load balancing of IPsec security associations (SAs) SAs between different heterogeneous cores, for example in a multiple controller systems. The balancing of IPsec SAs are performed using software implementations of IPsec offloading and/or hardware implementations of IPsec offloading. By balancing the processing functions related to orchestration of IPsec SAs between hardware (e.g., IPsec hardware accelerator) and software (e.g., IPsec software accelerator), the methods realize increased optimization, improved user experience, and enhanced product performance in networks using IPsec. For example, an IPsec session is initiated, and it is determined whether a resource limitation for a hardware-based IPsec accelerator has been reached. If the limitation for the hardware-based IPsec accelerator has been reached, processing for the IPsec SA for the initiated IPsec session is offloaded to a software-based IPsec accelerator.

Abstract: A packet that includes a header and a payload can be acquired. A first portion of the payload can be selected such that the first portion that is smaller than the payload. The header and the first portion of the payload can be encrypted based on an encryption algorithm to generate an encrypted packet. The encrypted packet can be transmitted to a node on a network.

Abstract: An example device comprising: a processor to configure a user anchor controller (UAC) for each network controller when an access point (AP) is connected to at least two network controllers; assign each user of the AP to one of the UACs; and automatically select an uplink for each user based on the user's corresponding UAC to transmit the user's data traffic.

Abstract: An example device comprising: a processor to configure a user anchor controller (UAC) for each network controller when an access point (AP) is connected to at least two network controllers; assign each user of the AP to one of the UACs; and automatically select an uplink for each user based on the user's corresponding UAC to transmit the user's data traffic.