Abstract: A novel method for performing replication of messages in a network that bridges one or more physical networks to an overlay logical network is provided. A physical gateway provides bridging between network nodes of a physical network and virtual machines in the overlay logical network by serving as an endpoint of the overlay logical network. The physical gateway does not replicate messages from the bridged physical network to destination endpoints in the overlay logical network directly, but instead tunnels the message-to-be-replicated to a designated tunnel endpoint in the overlay logical network. The designated tunnel endpoint in turn replicates the message that was tunneled to it to other endpoints in the overlay logical network.

Abstract: Techniques and mechanisms for an integrated clock gate (ICG) to selectively output a clock signal, and to provide frequency division functionality. In an embodiment, an ICG circuit comprises first circuitry which is coupled to receive a first clock signal, and second circuitry which is coupled to receive a control signal. The first circuitry provides a single edge triggered flip-flop functionality, and is coupled to communicate a feedback signal which the first circuitry is further coupled to receive. Based on the control signal and the feedback signal, the second circuitry performs an exclusive OR (XOR) operation to selectively enable the first circuitry to generate a second clock signal based on the first clock signal. In another embodiment, a frequency of the second clock signal is substantially equal to one half of a frequency of the first clock signal.

Abstract: An apparatus is described. The apparatus includes a long short term memory (LSTM) circuit having a multiply accumulate circuit (MAC). The MAC circuit has circuitry to rely on a stored product term rather than explicitly perform a multiplication operation to determine the product term if an accumulation of differences between consecutive, preceding input values has not reached a threshold.

Abstract: Examples herein relate to a memory device comprising an eDRAM memory cell, the eDRAM memory cell can include a write circuit formed at least partially over a storage cell and a read circuit formed at least partially under the storage cell; a compute near memory device bonded to the memory device; a processor; and an interface from the memory device to the processor. In some examples, circuitry is included to provide an output of the memory device to emulate output read rate of an SRAM memory device comprises one or more of: a controller, a multiplexer, or a register. Bonding of a surface of the memory device can be made to a compute near memory device or other circuitry. In some examples, a layer with read circuitry can be bonded to a layer with storage cells. Any layers can be bonded together using techniques described herein.

Abstract: A compute near memory (CNM) convolution accelerator enables a convolutional neural network (CNN) to use dedicated acceleration to achieve efficient in-place convolution operations with less impact on memory and energy consumption. A 2D convolution operation is reformulated as 1D row-wise convolution. The 1D row-wise convolution enables the CNM convolution accelerator to process input activations row-by-row, while using the weights one-by-one. Lightweight access circuits provide the ability to stream both weights and input rows as vectors to MAC units, which in turn enables modules of the CNM convolution accelerator to implement convolution for both [1×1] and chosen [n×n] sized filters.

Abstract: An apparatus is described. The apparatus includes a long short term memory (LSTM) circuit having a multiply accumulate circuit (MAC). The MAC circuit has circuitry to rely on a stored product term rather than explicitly perform a multiplication operation to determine the product term if an accumulation of differences between consecutive, preceding input values has not reached a threshold.

Abstract: A parasitic-aware single-edge triggered flip-flop reduces clock power through layout optimization, enabled through process-circuit co-optimization. The static pass-gate master-slave flip-flop utilizes novel layout optimization enabling significant power reduction. The layout removes the clock poly over notches in the diffusion area. Poly lines implement clock nodes. The poly lines are aligned between n-type and p-type active regions.

Abstract: An apparatus is described. The apparatus includes a circuit to process a binary neural network. The circuit includes an array of processing cores, wherein, processing cores of the array of processing cores are to process different respective areas of a weight matrix of the binary neural network. The processing cores each include add circuitry to add only those weights of an i layer of the binary neural network that are to be effectively multiplied by a non zero nodal output of an i?1 layer of the binary neural network.

Abstract: A fast Mux-D scan flip-flop is provided, which bypasses a scan multiplexer to a master keeper side path, removing delay overhead of a traditional Mux-D scan topology. The design is compatible with simple scan methodology of Mux-D scan, while preserving smaller area and small number of inputs/outputs. Since scan Mux is not in the forward critical path, circuit topology has similar high performance as level-sensitive scan flip-flop and can be easily converted into bare pass-gate version. The new fast Mux-D scan flip-flop combines the advantages of the conventional LSSD and Mux-D scan flip-flop, without the disadvantages of each.

Abstract: Embodiments herein describe techniques for a semiconductor device including a RRAM memory cell. The RRAM memory cell includes a FinFET transistor and a RRAM storage cell. The FinFET transistor includes a fin structure on a substrate, where the fin structure includes a channel region, a source region, and a drain region. An epitaxial layer is around the source region or the drain region. A RRAM storage stack is wrapped around a surface of the epitaxial layer. The RRAM storage stack includes a resistive switching material layer in contact and wrapped around the surface of the epitaxial layer, and a contact electrode in contact and wrapped around a surface of the resistive switching material layer. The epitaxial layer, the resistive switching material layer, and the contact electrode form a RRAM storage cell. Other embodiments may be described and/or claimed.

Abstract: A compute near memory (CNM) convolution accelerator enables a convolutional neural network (CNN) to use dedicated acceleration to achieve efficient in-place convolution operations with less impact on memory and energy consumption. A 2D convolution operation is reformulated as 1D row-wise convolution. The 1D row-wise convolution enables the CNM convolution accelerator to process input activations row-by-row, while using the weights one-by-one. Lightweight access circuits provide the ability to stream both weights and input rows as vectors to MAC units, which in turn enables modules of the CNM convolution accelerator to implement convolution for both [1×1] and chosen [n×n] sized filters.

Abstract: An apparatus is described. The apparatus includes a compute-in-memory (CIM) circuit for implementing a neural network disposed on a semiconductor chip. The CIM circuit includes a mathematical computation circuit coupled to a memory array. The memory array includes an embedded dynamic random access memory (eDRAM) memory array. Another apparatus is described. The apparatus includes a compute-in-memory (CIM) circuit for implementing a neural network disposed on a semiconductor chip. The CIM circuit includes a mathematical computation circuit coupled to a memory array. The mathematical computation circuit includes a switched capacitor circuit. The switched capacitor circuit includes a back-end-of-line (BEOL) capacitor coupled to a thin film transistor within the metal/dielectric layers of the semiconductor chip. Another apparatus is described. The apparatus includes a compute-in-memory (CIM) circuit for implementing a neural network disposed on a semiconductor chip.

Abstract: An apparatus is formed. The apparatus includes a stack of semiconductor chips. The stack of semiconductor chips includes a logic chip and a memory stack, wherein, the logic chip includes at least one of a GPU and CPU. The apparatus also includes a semiconductor chip substrate. The stack of semiconductor chips are mounted on the semiconductor chip substrate. At least one other logic chip is mounted on the semiconductor chip substrate. The semiconductor chip substrate includes wiring to interconnect the stack of semiconductor chips to the at least one other logic chip.

Abstract: An apparatus is described. The apparatus includes a circuit to process a binary neural network. The circuit includes an array of processing cores, wherein, processing cores of the array of processing cores are to process different respective areas of a weight matrix of the binary neural network. The processing cores each include add circuitry to add only those weights of an i layer of the binary neural network that are to be effectively multiplied by a non zero nodal output of an i?1 layer of the binary neural network.

Abstract: Examples described herein relate to a neural network whose weights from a matrix are selected from a set of weights stored in a memory on-chip with a processing engine for generating multiply and carry operations. The number of weights in the set of weights stored in the memory can be less than a number of weights in the matrix thereby reducing an amount of memory used to store weights in a matrix. The weights in the memory can be generated in training using gradients from back propagation. Weights in the memory can be selected using a tabulation hash calculation on entries in a table.

Abstract: A DIMA semiconductor structure is disclosed. The DIMA semiconductor structure includes a frontend including a semiconductor substrate, a transistor switch of a memory cell coupled to the semiconductor substrate and a computation circuit on the periphery of the frontend coupled to the semiconductor substrate. Additionally, the DIMA includes a backend that includes an RRAM component of the memory cell that is coupled to the transistor switch.

Abstract: Embodiments are directed to systems and methods of implementing an analog neural network using a pipelined SRAM architecture (“PISA”) circuitry disposed in on-chip processor memory circuitry. The on-chip processor memory circuitry may include processor last level cache (LLC) circuitry. One or more physical parameters, such as a stored charge or voltage, may be used to permit the generation of an in-memory analog output using a SRAM array. The generation of an in-memory analog output using only word-line and bit-line capabilities beneficially increases the computational density of the PISA circuit without increasing power requirements.

Abstract: An apparatus is provided which comprises: a multi-bit quad latch with an internally coupled level sensitive scan circuitry; and a combinational logic coupled to an output of the multi-bit quad latch. Another apparatus is provided which comprises: a plurality of sequential logic circuitries; and a clocking circuitry comprising inverters, wherein the clocking circuitry is shared by the plurality of sequential logic circuitries.

Abstract: The present disclosure is directed to systems and methods of implementing a neural network using in-memory, bit-serial, mathematical operations performed by a pipelined SRAM architecture (bit-serial PISA) circuitry disposed in on-chip processor memory circuitry. The on-chip processor memory circuitry may include processor last level cache (LLC) circuitry. The bit-serial PISA circuitry is coupled to PISA memory circuitry via a relatively high-bandwidth connection to beneficially facilitate the storage and retrieval of layer weights by the bit-serial PISA circuitry during execution. Direct memory access (DMA) circuitry transfers the neural network model and input data from system memory to the bit-serial PISA memory and also transfers output data from the PISA memory circuitry to system memory circuitry.

Abstract: A new family of shared clock single-edge triggered flip-flops that reduces a number of internal clock devices from 8 to 6 devices to reduce clock power. The static pass-gate master-slave flip-flop has no performance penalty compared to the flip-flops with 8 clock devices thus enabling significant power reduction. The flip-flop intelligently maintains the same polarity between the master and slave stages which enables the sharing of the master tristate and slave state feedback clock devices without risk of charge sharing across all combinations of clock and data toggling. Because of this, the state of the flip-flop remains undisturbed, and is robust across charge sharing noise. A multi-bit time borrowing internal stitched flip-flop is also described, which enables internal stitching of scan in a high performance time-borrowing flip-flop without incurring increase in layout area.