Abstract: The present description concerns a circuit configured to perform a multiply and accumulate operation in a layer of an artificial neural network, the operation taking, as an input, an input data value and a weight, and wherein the weight only has a value within a limited set only formed of value 0, of a plurality of values equal to 2n, where n is an integer, and of a plurality of values, each equal to the product of 2n by an odd number greater than or equal to 3.

Abstract: A reception element receives an analog signal. The received analog signal is converted by a reception chain into a digital signal. Based on the digital signal and a first filtering operation, a correction chain generates a correction digital signal reconstituting dynamic nonlinearities generated by the reception chain. A corrected signal from which the reconstituted dynamic nonlinearities have been removed is then generated by subtracting the correction digital signal from the digital signal.

Abstract: An integrated circuit includes a computer unit configured to execute the neural network. Parameters of the neural network are stored in a first memory. Data supplied at the input of the neural network or generated by the neural network are stored in a second memory. A first barrel shifter circuit transmits data from the second memory to the computer unit. A second barrel shifter circuit delivers data generated during the execution of the neural network by the computer unit to the second memory. A control unit is configured to control the computer unit, the first and second barrel shifter circuits, and accesses to the first memory and to the second memory.

Abstract: In one embodiment, a semiconductor device includes a carrier substrate, a buried dielectric region overlying the carrier substrate, and a semiconductor film separated from the carrier substrate by the buried dielectric region. NMOS transistors and PMOS transistors are disposed at a surface of the semiconductor film and coupled together to form a static random access memory (SRAM) cell. The NMOS transistors and the PMOS transistors each include a gate dielectric layer having a thickness greater than three nanometers and an active region in the semiconductor film. The active region of the PMOS transistors are formed from a silicon-germanium alloy.

Abstract: The present description concerns a method that includes the compression, by a processor, of an image comprising first patterns by transforming the image into a first representation formed of two-point elements. The method also includes the execution, by a neural network, of an inference operation on the first representation to generate a second representation formed of two-point elements. The method further includes the generation of a lithographic mask based on the decompression of the second representation.

Abstract: A flip flop includes a data input, a clock input, a test chain input, a test chain output, a monitoring circuit, and an alert transmission circuit. The monitoring circuit is adapted to generate an alert if the time between arrival of a data bit and a clock edge is less than a threshold. The alert transmission circuit is adapted to apply during a monitoring phase an alert level to the test chain output in the event of an alert generated by the monitoring circuit, and to apply the alert level to the test chain output when an alert level is received at the test chain input.

Abstract: The synchronous retention flip-flop circuit comprises a first circuit module suitable for being powered by an interruptible power source and a second circuit module suitable for being powered by a permanent power source. The first circuit module includes first and second latch stages, which are configured to store at least one datum while said interruptible power source is supplying power, transmitting means suitable for being controlled by a second control signal and configured to deliver said at least one datum to the second circuit module before an interruption of said interruptible power source, the second circuit module being configured to preserve said at least one datum during said interruption, and restoring means suitable for being controlled by a first control signal and configured to restore said at least one datum at the end of said interruption. Only the second control signal remains active during interruption of the interruptible power source.

Abstract: A flip flop includes a data input, a clock input, a test chain input, a test chain output, a monitoring circuit, and an alert transmission circuit. The monitoring circuit is adapted to generate an alert if the time between arrival of a data bit and a clock edge is less than a threshold. The alert transmission circuit is adapted to apply during a monitoring phase an alert level to the test chain output in the event of an alert generated by the monitoring circuit, and to apply the alert level to the test chain output when an alert level is received at the test chain input.

Abstract: A synchronous retention flip-flop circuit includes a first circuit module powered by an interruptible power source and a second circuit module powered by a permanent power source. The first circuit module includes a first latch circuit and a second latch circuit which are configured to store at least one datum while the interruptible power source is supplying power. A transmission circuit operates to deliver the at least one datum to the second circuit module before an interruption of the interruptible power source. The second circuit module preserves the at least one datum during the interruption. Following an end of the interruption, a restoring circuit transfers the at least one datum from the second circuit module to the first circuit module via a single one of the first and second latch circuits.

Abstract: The synchronous retention flip-flop circuit comprises a first circuit module suitable for being powered by an interruptible power source and a second circuit module suitable for being powered by a permanent power source. The first circuit module includes first and second latch stages, which are configured to store at least one datum while said interruptible power source is supplying power, transmitting means suitable for being controlled by a second control signal and configured to deliver said at least one datum to the second circuit module before an interruption of said interruptible power source, the second circuit module being configured to preserve said at least one datum during said interruption, and restoring means suitable for being controlled by a first control signal and configured to restore said at least one datum at the end of said interruption. Only the second control signal remains active during interruption of the interruptible power source.

Abstract: A synchronous retention flip-flop circuit includes a first circuit module powered by an interruptible power source and a second circuit module powered by a permanent power source. The first circuit module includes a first latch circuit and a second latch circuit which are configured to store at least one datum while the interruptible power source is supplying power. A transmission circuit operates to deliver the at least one datum to the second circuit module before an interruption of the interruptible power source. The second circuit module preserves the at least one datum during the interruption. Following an end of the interruption, a restoring circuit transfers the at least one datum from the second circuit module to the first circuit module via a single one of the first and second latch circuits.

Abstract: A wireless unit includes a first motion sensitive device; communications circuitry for wirelessly communicating with a further wireless unit; and a processing device configured to compare at least one first motion vector received from the first motion sensitive device with at least one second motion vector received from a second motion sensitive device of the further wireless unit.

Abstract: A device can be used for compensating bandwidth mismatches of time interleaved analog to digital converters. A processor of the device determines, for each original sample stream, an estimated difference between the time constant of a low pass filter representative of the corresponding converter and a reference time constant of a reference low pass filter, and uses this estimated difference and a filtered stream to correct the original stream and deliver a corrected stream of corrected samples.

Abstract: A power management circuit including, between a first terminal intended to be connected to an electric power generation source and a second terminal intended to be connected to a load to be powered, a linear regulator and a circuit capable of activating the linear regulator when the power supplied by said source is greater than a first threshold.

Abstract: A device can be used for compensating bandwith mismatches of time interleaved analog to digital converters. A processor of the device determines, for each original sample stream, an estimated difference between the time constant of a low pass filter representative of the corresponding converter and a reference time constant of a reference low pass filter, and uses this estimated difference and a filtered stream to correct the original stream and deliver a corrected stream of corrected samples.

Abstract: A stand-alone device comprising a silicon wafer having its front surface including a first layer of a first conductivity type and a second layer of a second conductivity type forming a photovoltaic cell; first vias crossing the wafer from the rear surface of the first layer and second vias crossing the wafer from the rear surface of the second layer; metallization levels on the rear surface of the wafer, the external level of these metallization levels defining contact pads; an antenna formed in one of the metallization levels; and one or several chips assembled on said pads; the metallization levels being shaped to provide selected interconnects between the different elements of the device.

Abstract: Adaptive scaling digital techniques attempt to place the system close to the timing failure so as to maximize energy efficiency. Rapid recovery from potential failures is usually by slowing the system clock and/or providing razor solutions (instruction replay.) These techniques compromise the throughput. This application presents a technique to provide local in-situ fault resilience based on dynamic slack borrowing. This technique is non-intrusive (needs no architecture modification) and has minimal impact on throughput.

Abstract: A power management circuit including, between a first terminal intended to be connected to an electric power generation source and a second terminal intended to be connected to a load to be powered, a linear regulator and a circuit capable of activating the linear regulator when the power supplied by said source is greater than a first threshold.

Abstract: According to one mode of implementation, a method includes an estimation including on the one hand a correlation processing involving at least one part of the sampled signal, at least one part of at least one first signal gleaned from a derived signal representative of a temporal derivative of the sampled signal and at least one part of N partial filtered signals respectively representative of N weighted differences between N pairs of bracketing versions flanking the sampled signal, N being greater than or equal to 1. On the other hand, the estimation includes a matrix processing on the results of this correlation processing. Correction processing of the M?1 trains involves respectively M?1 second signals gleaned from the derived signal and the suite of M?1 shift coefficients.

Abstract: According to one mode of implementation, a method includes an estimation including on the one hand a correlation processing involving at least one part of the sampled signal, at least one part of at least one first signal gleaned from a derived signal representative of a temporal derivative of the sampled signal and at least one part of N partial filtered signals respectively representative of N weighted differences between N pairs of bracketing versions flanking the sampled signal, N being greater than or equal to 1. On the other hand, the estimation includes a matrix processing on the results of this correlation processing. Correction processing of the M?1 trains involves respectively M?1 second signals gleaned from the derived signal and the suite of M?1 shift coefficients.