Abstract: A fused dot-product multiply-accumulate (MAC) circuit may support variable precisions of floating-point data elements to perform computations (e.g., MAC operations) in deep learning operations. An operation mode of the circuit may be selected based on the precision of an input element. The operation mode may be a FP16 mode or a FP8 mode. In the FP8 mode, product exponents may be computed based on exponents of floating-point input elements. A maximum exponent may be selected from the one or more product exponents. A global maximum exponent may be selected from a plurality of maximum exponents. A product mantissa may be computed and aligned with another product mantissa based on a difference between the global maximum exponent and a corresponding maximum exponent. An adder tree may accumulate the aligned product mantissas and compute a partial sum mantissa. The partial sum mantissa may be normalized using the global maximum exponent.

Abstract: A FPMAC operation has two operands: an input operand and a weight operand. The operands may have a format of FP16, BF16, or INT8. Each operand is split into two portions. The two portions are stored in separate storage units. Then operands are transferred to register files of a PE, with each register file storing bits of an operand sequentially. The PE performs the FPMAC operation based on the operands. The PE may include an FPMAC unit configured to compute an individual partial sum of the PE. The PE may also include an FP adder to accumulate the individual partial sum with other data, such as an output from another PE or an output form another PE array. The FP adder may be fused with the FPMAC unit in a single circuit that can do speculative alignment and has separate critical paths for alignment and normalization.

Abstract: In one embodiment, a processor comprises a first neuro-synaptic core comprising first circuitry to configure the first neuro-synaptic core as a neuron core responsive to a first value specified by a configuration parameter; and configure the first neuro-synaptic core as a synapse core responsive to a second value specified by the configuration parameter.

Abstract: Apparatus and method for a scalable, free running neuromorphic processor. For example, one embodiment of a neuromorphic processing apparatus comprises: a plurality of neurons; an interconnection network to communicatively couple at least a subset of the plurality of neurons; a spike controller to stochastically generate a trigger signal, the trigger signal to cause a selected neuron to perform a thresholding operation to determine whether to issue a spike signal.

Abstract: In one embodiment, a method comprises receiving a selection of a neural network topology type; identifying a synapse memory mapping scheme for the selected neural network topology type from a plurality of synapse memory mapping schemes that are each associated with a respective neural network topology type; and mapping a plurality of synapse weights to locations in a memory based on the identified synapse memory mapping scheme.

Abstract: Some embodiments include apparatus and methods using an input stage and an output stage of a circuit. The input stage operates to receive an input signal and a clock signal and to provide an internal signal at an internal node based at least in part on the input signal. The input signal has levels in a first voltage range. The internal signal has levels in a second voltage range greater than the first voltage range. The output stage operates to receive the internal signal, the clock signal, and an additional signal generated based on the input signal. The output stage provides an output signal based at least in part on the input signal and the additional signal. The output signal has a third voltage range greater than the first voltage range.

Abstract: In one embodiment, a method comprises determining that a membrane potential of a first neuron of a first neuron core exceeds a threshold; determining a first plurality of synapse cores that each store at least one synapse weight associated with the first neuron; and sending a spike message to the determined first plurality of synapse cores.

Abstract: An apparatus is provided which comprises: a first inverter to receive a clock; a pass-gate coupled to the first inverter; a second inverter coupled to the pass-gate and to provide an output clock; and a device coupled to the second inverter and the pass-gate, wherein the transistor and the pass-gate are controllable by a logic that depends on logic values of at least two signals (e.g., an enable and the clock).

Abstract: In accordance with some embodiments, the complexity of motion estimation algorithms that use Haar, SAD and Hadamard transforms may be reduced. In some embodiments, the number of summations may be reduced compared to existing techniques and some of the existing summations may be replaced with compare operations. In some embodiments, additions are replaced with compares in order to balance delay and area or energy or power considerations.

Abstract: An apparatus is provided which comprises a clock inverter having an input coupled to a clock node, the clock inverter having an output, wherein the clock inverter has an N-well which is coupled to a first power supply; and a plurality of sequential logics coupled to the output of the clock inverter and also coupled to the clock node, wherein at least one sequential logics of the plurality of the sequential logics has an N-well which is coupled to a second power supply, wherein the second power supply has a voltage level lower than a voltage level of the first power supply.

Abstract: Some embodiments include apparatus and methods using an input stage and an output stage of a circuit. The input stage operates to receive an input signal and a clock signal and to provide an internal signal at an internal node based at least in part on the input signal. The input signal has levels in a first voltage range. The internal signal has levels in a second voltage range greater than the first voltage range. The output stage operates to receive the internal signal, the clock signal, and an additional signal generated based on the input signal. The output stage provides an output signal based at least in part on the input signal and the additional signal. The output signal has a third voltage range greater than the first voltage range.

Abstract: An apparatus is provided which comprises: a multiplexer which is gated by a clock; and a flip-flop coupled to the multiplexer, wherein the flip-flop has a chain of at least four inverters one of which has an input to receive the clock.

Abstract: In one embodiment, a processor comprises a first neuromorphic core to implement a plurality of neural units of a neural network, the first neuromorphic core comprising a memory to store a current time-step of the first neuromorphic core; and a controller to track current time-steps of neighboring neuromorphic cores that receive spikes from or provide spikes to the first neuromorphic core; and control the current time-step of the first neuromorphic core based on the current time-steps of the neighboring neuromorphic cores.

Abstract: An apparatus is provided which comprises: a first inverter to receive a clock; a pass-gate coupled to the first inverter; a second inverter coupled to the pass-gate and to provide an output clock; and a device coupled to the second inverter and the pass-gate, wherein the transistor and the pass-gate are controllable by a logic that depends on logic values of at least two signals (e.g., an enable and the clock).

Abstract: In one embodiment, a processor comprises a first neuro-synaptic core comprising first circuitry to configure the first neuro-synaptic core as a neuron core responsive to a first value specified by a configuration parameter; and configure the first neuro-synaptic core as a synapse core responsive to a second value specified by the configuration parameter.

Abstract: An apparatus is provided which comprises: a clock node; a first inverter having an input coupled to the clock node; a data node; a master latch with a shared p-type keeper coupled to an output of the first inverter, the master latch coupled to the data node; and a slave latch coupled to an output of the master latch, the slave latch having a shared p-type keeper and a shared n-type footer, wherein the shared p-type keeper and the shared n-type footer of the slave latch are coupled to the clock node and the input of the first inverter.

Abstract: An apparatus is provided which comprises: a clock node; a test node; an enable node; and an AND-OR-INVERT (AOI) static latch coupled to the clock node, test node, and enable node, wherein the AOI static latch has embedded NOR functionality. Another apparatus comprises: a critical timing path having a pass-gate based integrated clock gate; and a non-critical timing path electrically coupled to the critical timing path, wherein the non-critical timing path includes an AND-OR-Inverter (AOI) based integrated clock gate with embedded NOR functionality.

Abstract: An apparatus is provided which comprises: a multiplexer which is gated by a clock; and a flip-flop coupled to the multiplexer, wherein the flip-flop has a chain of at least four inverters one of which has an input to receive the clock.

Abstract: Apparatus and method for a scalable, free running neuromorphic processor. For example, one embodiment of a neuromorphic processing apparatus comprises: a plurality of neurons; an interconnection network to communicatively couple at least a subset of the plurality of neurons; a spike controller to stochastically generate a trigger signal, the trigger signal to cause a selected neuron to perform a thresholding operation to determine whether to issue a spike signal.

Abstract: In one embodiment, a method comprises determining that a membrane potential of a first neuron of a first neuron core exceeds a threshold; determining a first plurality of synapse cores that each store at least one synapse weight associated with the first neuron; and sending a spike message to the determined first plurality of synapse cores.