Abstract: Certain aspects of the present disclosure are directed to methods and apparatus for configuring a multiply-accumulate (MAC) block in an artificial neural network. A method generally includes receiving, at a neural processing unit comprising one or more logic elements, at least one input associated with a use-case of the neural processing unit; obtaining a set of weights associated with the at least one input; selecting a precision for the set of weights; modifying the set of weights based on the selected precision; and generating an output based, at least in part, on the at least one input, the modified set of weights, and an activation function.

Abstract: A true random number generator (TRNG) for generating a sequence of random numbers of bits is disclosed. The TRNG includes a TRNG cell configured to generate a sequence of bits logically alternating with a mean frequency and with substantially random period jitter; a period monitor configured to generate a first sequence of random bits based on a set of periods of the sequence of logically alternating bits; and a sampling circuit configured to sample the first sequence of random bits in response to a sampling clock to generate a second sequence of random bits.

Abstract: A true random number generator (TRNG) for generating a sequence of random numbers of bits is disclosed. The TRNG includes a TRNG cell configured to generate a sequence of bits logically alternating with a mean frequency and with substantially random period jitter; a period monitor configured to generate a first sequence of random bits based on a set of periods of the sequence of logically alternating bits; and a sampling circuit configured to sample the first sequence of random bits in response to a sampling clock to generate a second sequence of random bits.

Abstract: Embodiments of the present invention provide an authenticating service of a chip having an intrinsic identifier (ID). In a typical embodiment, an authenticating device is provided that includes an identification (ID) engine, a self-test engine, and an intrinsic component. The intrinsic component is associated with a chip and includes an intrinsic feature. The self-test engine retrieves the intrinsic feature and communicates it to the identification engine. The identification engine receives the intrinsic feature, generates a first authentication value using the intrinsic feature, and stores the authentication value in memory. The self-test engine generates a second authentication value using an authentication challenge. The identification engine includes a compare circuitry that compares the first authentication value and the second authentication value and generates an authentication output value based on the results of the compare of the two values.

Abstract: Certain aspects of the present disclosure are directed to methods and apparatus for configuring a multiply-accumulate (MAC) block in an artificial neural network. A method generally includes receiving, at a neural processing unit comprising one or more logic elements, at least one input associated with a use-case of the neural processing unit; obtaining a set of weights associated with the at least one input; selecting a precision for the set of weights; modifying the set of weights based on the selected precision; and generating an output based, at least in part, on the at least one input, the modified set of weights, and an activation function.

Abstract: Embodiments of the present invention provide an authenticating service of a chip having an intrinsic identifier (ID). In a typical embodiment, an authenticating device is provided that includes an identification (ID) engine, a self-test engine, and an intrinsic component. The intrinsic component is associated with a chip and includes an intrinsic feature. The self-test engine retrieves the intrinsic feature and communicates it to the identification engine. The identification engine receives the intrinsic feature, generates a first authentication value using the intrinsic feature, and stores the authentication value in memory. The self-test engine generates a second authentication value using an authentication challenge. The identification engine includes a compare circuitry that compares the first authentication value and the second authentication value and generates an authentication output value based on the results of the compare of the two values.

Abstract: Embodiments of the present invention provide an authenticating service of a chip having an intrinsic identifier (ID). In a typical embodiment, an authenticating device is provided that includes an identification (ID) engine, a self-test engine, and an intrinsic component. The intrinsic component is associated with a chip and includes an intrinsic feature. The self-test engine retrieves the intrinsic feature and communicates it to the identification engine. The identification engine receives the intrinsic feature, generates a first authentication value using the intrinsic feature, and stores the authentication value in memory. The self-test engine generates a second authentication value using an authentication challenge. The identification engine includes a compare circuitry that compares the first authentication value and the second authentication value and generates an authentication output value based on the results of the compare of the two values.

Abstract: Embodiments of the present invention provide an authenticating service of a chip having an intrinsic identifier (ID). In a typical embodiment, an authenticating device is provided that includes an identification (ID) engine, a self-test engine, and an intrinsic component. The intrinsic component is associated with a chip and includes an intrinsic feature. The self-test engine retrieves the intrinsic feature and communicates it to the identification engine. The identification engine receives the intrinsic feature, generates a first authentication value using the intrinsic feature, and stores the authentication value in memory. The self-test engine generates a second authentication value using an authentication challenge. The identification engine includes a compare circuitry that compares the first authentication value and the second authentication value and generates an authentication output value based on the results of the compare of the two values.

Abstract: An authenticating service of a chip having an intrinsic identifier (ID) is provided. The authenticating device includes an identification (ID) engine, a self-test engine, and an intrinsic component. The intrinsic component is associated with a chip and includes an intrinsic feature. The self-test engine retrieves the intrinsic feature and communicates it to the identification engine. The identification engine receives the intrinsic feature, generates a first authentication value using the intrinsic feature, and stores the authentication value in memory. The self-test engine generates a second authentication value using an authentication challenge. The identification engine includes a compare circuitry that compares the first authentication value and the second authentication value and generates an authentication output value based on the results of the compare of the two values.

Abstract: Embodiments of an sequential state element (SSE) capable of providing triple modular redundant (TMR) correction is disclosed. The SSE has a setup stage and a feedback stage. The setup stage is configured to generate an output bit signal having an output bit state while a clock signal is in the first clock state. The setup stage also generates a feedback input bit signal as feedback of the output bit state. However, the feedback stage is capable of providing TMR correction without this feedback signal. Instead, the feedback stage utilizes the second feedback input bit signal and a third feedback input bit signal from two other SSEs. Since TMR correction can be provided with just the second feedback input bit signal and the third feedback input bit signal, the power and area consumed by the SSE is reduced.

Abstract: This disclosure relates generally to computerized systems and methods of producing a physical representation of an in silico Integrated Circuit (IC) having an in silico Multi-Mode Redundant (MMR) pipeline circuit. An IC layout of the in silico IC is initially generated with the electronic design automation (EDA) program. Multi-Mode Redundant Self-Correcting Sequential State Element (MMRSCSSE) layouts are then rendered immotile while initial redundant Combinational Logic Circuit (CLC) layouts are removed from the IC layout after the MMRSCSSE layouts have been rendered immotile. By first placing the MMRSCSSE layouts and then rendering them immotile, the remaining logic can be placed again and optimized without compromising critical node spacing. As such, the described method provides for a more efficient way to create the IC layout of the in silico IC while maintaining critical node spacing.

Abstract: Embodiments of the present invention provide an authenticating service of a chip having an intrinsic identifier (ID). In a typical embodiment, an authenticating device is provided that includes an identification (ID) engine, a self-test engine, and an intrinsic component. The intrinsic component is associated with a chip and includes an intrinsic feature. The self-test engine retrieves the intrinsic feature and communicates it to the identification engine. The identification engine receives the intrinsic feature, generates a first authentication value using the intrinsic feature, and stores the authentication value in memory. The self-test engine generates a second authentication value using an authentication challenge. The identification engine includes a compare circuitry that compares the first authentication value and the second authentication value and generates an authentication output value based on the results of the compare of the two values.

Abstract: Embodiments of the present invention provide an authenticating service of a chip having an intrinsic identifier (ID). In a typical embodiment, an authenticating device is provided that includes an identification (ID) engine, a self-test engine, and an intrinsic component. The intrinsic component is associated with a chip and includes an intrinsic feature. The self-test engine retrieves the intrinsic feature and communicates it to the identification engine. The identification engine receives the intrinsic feature, generates a first authentication value using the intrinsic feature, and stores the authentication value in memory. The self-test engine generates a second authentication value using an authentication challenge. The identification engine includes a compare circuitry that compares the first authentication value and the second authentication value and generates an authentication output value based on the results of the compare of the two values.

Abstract: Embodiments of an sequential state element (SSE) capable of providing triple modular redundant (TMR) correction is disclosed. The SSE has a setup stage and a feedback stage. The setup stage is configured to generate an output bit signal having an output bit state while a clock signal is in the first clock state. The setup stage also generates a feedback input bit signal as feedback of the output bit state. However, the feedback stage is capable of providing TMR correction without this feedback signal. Instead, the feedback stage utilizes the second feedback input bit signal and a third feedback input bit signal from two other SSEs. Since TMR correction can be provided with just the second feedback input bit signal and the third feedback input bit signal, the power and area consumed by the SSE is reduced.

Abstract: A local sorting module includes a set of storage elements storing binary vectors configured in a one-dimensional (1D) or two-dimensional (2D) array structure and separated by respective comparators configured to conditionally compare and sort the binary vectors. The comparators may perform a sort using a compare-and-flip or a compare-and-swap operation. Local sorting modules may be coupled with a global sorting module for enabling a tournament sort algorithm to output values stored in storage elements one at a time until all data is outputted in a predetermined sorting order.

Abstract: A local sorting module includes a set of storage elements storing binary vectors configured in a one-dimensional (1D) or two-dimensional (2D) array structure and separated by respective comparators configured to conditionally compare and sort the binary vectors. The comparators may perform a sort using a compare-and-flip or a compare-and-swap operation. Local sorting modules may be coupled with a global sorting module for enabling a tournament sort algorithm to output values stored in storage elements one at a time until all data is outputted in a predetermined sorting order.

Abstract: A local sorting module includes a set of storage elements storing binary vectors configured in a one-dimensional (1D) or two-dimensional (2D) array structure and separated by respective comparators configured to conditionally compare and sort the binary vectors. The comparators may perform a sort using a compare-and-flip or a compare-and-swap operation. Local sorting modules may be coupled with a global sorting module for enabling a tournament sort algorithm to output values stored in storage elements one at a time until all data is outputted in a predetermined sorting order.

Abstract: A memory having variable size blocks of failed memory addresses is connected to a TCAM storing data values of ranges of addresses in the memory. The ranges of addresses correspond to virtual addresses that, in combination with an offset, point away from failed memory addresses. A reduction circuit connected to the TCAM produces an output for each programmed range of addresses based on a virtual address. A priority encoder, connected to the reduction circuit, selects a first range from the reduction circuit and passes the first range to a random-access memory (RAM). Responsive to the virtual address bring an address in one of the ranges of addresses, the priority encoder passes the first range containing the virtual address to the RAM, which passes a corresponding offset value to the Adder based on the first range. The Adder calculates a physical memory address directing the virtual address to a functional memory location.

Abstract: A local sorting module includes a set of storage elements storing binary vectors configured in a one-dimensional (1D) or two-dimensional (2D) array structure and separated by respective comparators configured to conditionally compare and sort the binary vectors. The comparators may perform a sort using a compare-and-flip or a compare-and-swap operation. Local sorting modules may be coupled with a global sorting module for enabling a tournament sort algorithm to output values stored in storage elements one at a time until all data is outputted in a predetermined sorting order.

Abstract: A local sorting module includes a set of storage elements storing binary vectors configured in a one-dimensional (1D) or two-dimensional (2D) array structure and separated by respective comparators configured to conditionally compare and sort the binary vectors. The comparators may perform a sort using a compare-and-flip or a compare-and-swap operation. Local sorting modules may be coupled with a global sorting module for enabling a tournament sort algorithm to output values stored in storage elements one at a time until all data is outputted in a predetermined sorting order.