Abstract: Systems, apparatuses and methods may provide for technology that processes an inference workload in a first subset of layers of a neural network that prevents or inhibits data dependent branch operations, conducts an exit determination as to whether an output of the first subset of layers satisfies one or more exit criteria, and selectively bypasses processing of the output in a second subset of layers of the neural network based on the exit determination. The technology may also speculatively initiate the processing of the output in the second subset of layers while the exit determination is pending. Additionally, when the inference workloads include a plurality of batches, the technology may mask one or more of the plurality of batches from processing in the second subset of layers.

Abstract: Systems, apparatuses and methods may provide for technology that processes an inference workload in a first subset of layers of a neural network that prevents or inhibits data dependent branch operations, conducts an exit determination as to whether an output of the first subset of layers satisfies one or more exit criteria, and selectively bypasses processing of the output in a second subset of layers of the neural network based on the exit determination. The technology may also speculatively initiate the processing of the output in the second subset of layers while the exit determination is pending. Additionally, when the inference workloads include a plurality of batches, the technology may mask one or more of the plurality of batches from processing in the second subset of layers.

Abstract: Systems, apparatuses and methods may provide for technology that processes an inference workload in a first subset of layers of a neural network that prevents or inhibits data dependent branch operations, conducts an exit determination as to whether an output of the first subset of layers satisfies one or more exit criteria, and selectively bypasses processing of the output in a second subset of layers of the neural network based on the exit determination. The technology may also speculatively initiate the processing of the output in the second subset of layers while the exit determination is pending. Additionally, when the inference workloads include a plurality of batches, the technology may mask one or more of the plurality of batches from processing in the second subset of layers.

Abstract: Systems, apparatuses and methods may provide for technology that processes an inference workload in a first subset of layers of a neural network that prevents or inhibits data dependent branch operations, conducts an exit determination as to whether an output of the first subset of layers satisfies one or more exit criteria, and selectively bypasses processing of the output in a second subset of layers of the neural network based on the exit determination. The technology may also speculatively initiate the processing of the output in the second subset of layers while the exit determination is pending. Additionally, when the inference workloads include a plurality of batches, the technology may mask one or more of the plurality of batches from processing in the second subset of layers.

Abstract: Systems, apparatuses and methods may provide for technology that processes an inference workload in a first subset of layers of a neural network that prevents or inhibits data dependent branch operations, conducts an exit determination as to whether an output of the first subset of layers satisfies one or more exit criteria, and selectively bypasses processing of the output in a second subset of layers of the neural network based on the exit determination. The technology may also speculatively initiate the processing of the output in the second subset of layers while the exit determination is pending. Additionally, when the inference workloads include a plurality of batches, the technology may mask one or more of the plurality of batches from processing in the second subset of layers.

Abstract: Technologies for multi-core wireless data transmission include a computing device having a processor with multiple cores and a wireless network interface controller (NIC). The computing device establishes multiple transmission queues that are each associated with a processor core. A driver receives a packet for transmission from an application in the execution context of the application, determines a current processor core of the execution context, adds metadata to the packet indicative of the current core, and enqueues the packet in the transmission queue associated with the current core. The wireless NIC merges the packet with packet data from the other transmission queues, adds a sequence number to each packet, and transmits each packet. The wireless NIC may determine the current processor core based on the metadata of the packet and raise an interrupt to the current processor core in response to transmitting the packet. Other embodiments are described and claimed.

Abstract: Technologies for multi-core wireless data transmission include a computing device having a processor with multiple cores and a wireless network interface controller (NIC). The computing device establishes multiple transmission queues that are each associated with a processor core. A driver receives a packet for transmission from an application in the execution context of the application, determines a current processor core of the execution context, adds metadata to the packet indicative of the current core, and enqueues the packet in the transmission queue associated with the current core. The wireless NIC merges the packet with packet data from the other transmission queues, adds a sequence number to each packet, and transmits each packet. The wireless NIC may determine the current processor core based on the metadata of the packet and raise an interrupt to the current processor core in response to transmitting the packet. Other embodiments are described and claimed.

Abstract: A method of managing connectivity between an end-device and one of a plurality of bridging-devices connected to the end-device. The method includes transmitting test messages by each of the bridging-devices, transmitting test messages by the end-device, determining, in each of the bridging-devices, with which of the other devices there is an operative connection, based on the test messages the bridging-device receives and determining, in each of the bridging-devices, whether a port leading to the end-device should be in a forwarding or blocking state responsive to the determination of the operative connections.

Abstract: A method of managing connectivity between an end-device and one of a plurality of bridging-devices connected to the end-device. The method includes transmitting test messages by each of the bridging-devices, transmitting test messages by the end-device, determining, in each of the bridging-devices, with which of the other devices there is an operative connection, based on the test messages the bridging-device receives and determining, in each of the bridging-devices, whether a port leading to the end-device should be in a forwarding or blocking state responsive to the determination of the operative connections.

Abstract: A method of managing connectivity of a cluster of bridging-devices including a plurality of links. The method includes pre-designating at least one of the links as a redundant link which is blocked when all the other links are operative and connected in accordance with a predetermined scheme, determining connectivity data on the connectivity of the links of the cluster, and activating or blocking the pre-designated redundant at least one link responsive to the connectivity data.

Abstract: A method of managing connectivity of a cluster of bridging-devices including a plurality of links. The method includes pre-designating at least one of the links as a redundant link which is blocked when all the other links are operative and connected in accordance with a predetermined scheme, determining connectivity data on the connectivity of the links of the cluster, and activating or blocking the pre-designated redundant at least one link responsive to the connectivity data.