Abstract: In one embodiment, a temperature management system comprises a plurality of thermal sensors at different locations on a chip, and a temperature manager. The temperature manager is configured to receive a plurality of temperature readings from the thermal sensors, to fit a quadratic temperature model to the received temperature readings, and to estimate a hotspot temperature on the chip using the fitted quadratic temperature model.

Abstract: Various aspects of this disclosure describe measuring timing slack using an endpoint criticality sensor on a chip. A sensor circuit is attached to sensitive endpoints on the chip (e.g., logical gates in a timing critical path) so that the sensor circuit receives the endpoint's data signal and clock signal. The sensor circuit introduces skew between the data signal and the clock signal by delaying the data signal more than the clock signal, and compares skewed data signals to determine if an error occurs because of the induced skew. By delaying the data signal with different delay amounts and monitoring what delays cause errors, an amount of timing slack in the data signal and clock signal (e.g., margin to criticality) is measured during operation of the chip for relevant circuitry to the system implemented on the chip, compared to test circuitry operating while the chip is in a test mode.

Abstract: Techniques for power-density-based clock cell spacing and resulting integrated circuits (ICs) are disclosed herein. In one example, the techniques determine power-usage density for different types of clock cells, as power-usage density relates to heat and IR droop. With the power-usage density for each type of clock cell determined, the techniques assign a keep-out region for each type of clock cell that is not fixed for all types of clock cells. These regions are instead based on the heat and IR droop corresponding to estimated power-usage density for each type of clock cell. Clock cells are then placed in a layout of an IC. The resulting IC has clock cells spaced sufficiently to reduce heat and IR droop while concurrently having excellent timing closure and performance.

Abstract: Systems and methods for power distribution network (PDN) droop/overshoot mitigation are provided. In certain embodiments, overshoot is mitigated by ramping down a frequency of a clock signal to a processor when the processor is switching clock frequencies and/or the processor is transitioning from an active mode to an idle mode. In certain embodiments, droop is mitigated by ramping up a frequency of a clock signal to a processor when the processor is switching clock frequencies and/or the processor is transitioning from an idle mode to an active mode.

Abstract: Systems and methods for dynamic clock and voltage scaling can switch integrated circuits between frequency-voltage modes with low latency. These systems include a resource power manager that can control a power management integrated circuit (PMIC), phase locked loops (PLLs), and clock dividers. The resource power manager controls transitions between frequency-voltage modes. The systems and methods provide dynamic clock and voltage scaling where the transitions between frequency-voltage modes are an atomic operation. Additionally, the resource power manager can control many modules, for example, clock dividers, in parallel. The invention can, due to lower latency between frequency-voltage modes, can provide improved system performance and reduced system power.

Abstract: A method of setting a supply voltage in a device is disclosed. The method includes receiving a first plurality of inputs from a plurality of sensors that are representative of a gate delay of a signal path on the device, and receiving a second plurality of inputs from a plurality of temperature sensors. The method further includes estimating a plurality of interconnect delays for the signal path based on the second plurality of inputs, and determining the supply voltage for the signal path based on the first plurality of inputs and the plurality of interconnect delays.

Abstract: A flip-flop is provided that includes a master latch clocked according to a first delay during a normal mode of operation and clocked by a smaller second delay during a scan mode of operation.

Abstract: In one embodiment, a temperature management system comprises a plurality of thermal sensors at different locations on a chip, and a temperature manager. The temperature manager is configured to receive a plurality of temperature readings from the thermal sensors, to fit a quadratic temperature model to the received temperature readings, and to estimate a hotspot temperature on the chip using the fitted quadratic temperature model.

Abstract: Techniques for power-density-based clock cell spacing and resulting integrated circuits (ICs) are disclosed herein. In one example, the techniques determine power-usage density for different types of clock cells, as power-usage density relates to heat and IR droop. With the power-usage density for each type of clock cell determined, the techniques assign a keep-out region for each type of clock cell that is not fixed for all types of clock cells. These regions are instead based on the heat and IR droop corresponding to estimated power-usage density for each type of clock cell. Clock cells are then placed in a layout of an IC. The resulting IC has clock cells spaced sufficiently to reduce heat and IR droop while concurrently having excellent timing closure and performance.

Abstract: Systems and methods for power distribution network (PDN) droop/overshoot mitigation are provided. In certain embodiments, overshoot is mitigated by ramping down a frequency of a clock signal to a processor when the processor is switching clock frequencies and/or the processor is transitioning from an active mode to an idle mode. In certain embodiments, droop is mitigated by ramping up a frequency of a clock signal to a processor when the processor is switching clock frequencies and/or the processor is transitioning from an idle mode to an active mode.

Abstract: A first apparatus includes at least one scan chain. Each of the at least one scan chain includes scan cells coupled together. Each scan cell in the at least one scan chain includes a first type of scan cell when a reset state of the scan cell is a first state, and a second type of scan cell when the reset state of the scan cell is a second state. One or more scan chains of the at least one scan chain includes at least one of the first type of scan cell and at least one of the second type of scan cell. A second apparatus includes first and second sets of scan chains including flip-flops without both set and reset functionality. Each of the flip-flops in the first and second sets of scan chains has a reset state of a first state and a second state, respectively.

Abstract: A method of setting a supply voltage in a device is disclosed. The method includes receiving a first plurality of inputs from a plurality of sensors that are representative of a gate delay of a signal path on the device, and receiving a second plurality of inputs from a plurality of temperature sensors. The method further includes estimating a plurality of interconnect delays for the signal path based on the second plurality of inputs, and determining the supply voltage for the signal path based on the first plurality of inputs and the plurality of interconnect delays.

Abstract: Circuits and methods for reducing leakage are provided. In one example, a system includes circuitry to reset a particular logic circuit to a state of reduced leakage. The state of reduced leakage would be known beforehand for the logic circuit. In this example, the logic circuit includes the combinational logic as well as flip flops that output a state to the combinational logic. Some of the flip flops are “SET” flip flops (assuming a 1 output value when a reset input is asserted) and some of the flip flops are “RESET” flip flops (assuming a 0 value when a reset input is asserted). The flip flops are chosen as inputs to the combinational logic so that the particular combination of zeros and ones output to the combinational logic puts the logic circuit in a state that is correlated with a desired level of leakage.

Abstract: A flip-flop is provided that includes a master latch clocked according to a first delay during a normal mode of operation and clocked by a smaller second delay during a scan mode of operation.

Abstract: A duty cycle adjustment apparatus includes a duty cycle adjustment determination module configured to determine an adjustment to a duty cycle of a clock signal, and includes a clock delay module configured to receive the clock signal, to delay the clock signal through first and second delay stage modules (with a first and a second plurality of delay paths, respectively) based on the duty cycle adjustment determined by the duty cycle adjustment determination module, and to output the delayed clock signal. The second plurality of delay paths have a greater delay difference between each of the corresponding delay paths than the first plurality of delay paths. The apparatus further includes a duty cycle adjustment module configured to receive the clock signal and the delayed clock signal, to adjust the duty cycle of the clock signal based on the delayed clock signal, and to output a duty cycle adjusted clock signal.

Abstract: Circuits and methods for reducing leakage are provided. In one example, a system includes circuitry to reset a particular logic circuit to a state of reduced leakage. The state of reduced leakage would be known beforehand for the logic circuit. In this example, the logic circuit includes the combinational logic as well as flip flops that output a state to the combinational logic. Some of the flip flops are “SET” flip flops (assuming a 1 output value when a reset input is asserted) and some of the flip flops are “RESET” flip flops (assuming a 0 value when a reset input is asserted). The flip flops are chosen as inputs to the combinational logic so that the particular combination of zeros and ones output to the combinational logic puts the logic circuit in a state that is correlated with a desired level of leakage.

Abstract: Systems and methods for efficiently generating electromigration reliability data for physical cells in an integrated circuit cell library are disclosed. Data for tables of electromigration susceptibility can be iteratively generated for multiple capacitive loadings on a cell output and for multiple transition times of a cell input. Each iteration can include simulating electrical performance of the physical cell and identifying a region with the largest ratio of current density to electromigration reliability limit. Between iterations, the data period of the input signal is updated using the ratio of current density to electromigration reliability limit and a relationship between currents and data period. Iterations may end when the ratio is close to one. The result can be used to evaluate electromigration reliability of an integrated circuit design and modify the design accordingly.

Abstract: A first apparatus includes at least one scan chain. Each of the at least one scan chain includes scan cells coupled together. Each scan cell in the at least one scan chain includes a first type of scan cell when a reset state of the scan cell is a first state, and a second type of scan cell when the reset state of the scan cell is a second state. One or more scan chains of the at least one scan chain includes at least one of the first type of scan cell and at least one of the second type of scan cell. A second apparatus includes first and second sets of scan chains including flip-flops without both set and reset functionality. Each of the flip-flops in the first and second sets of scan chains has a reset state of a first state and a second state, respectively.

Abstract: Systems and methods for dynamic clock and voltage scaling can switch integrated circuits between frequency-voltage modes with low latency. These systems include a resource power manager that can control a power management integrated circuit (PMIC), phase locked loops (PLLs), and clock dividers. The resource power manager controls transitions between frequency-voltage modes. The systems and methods provide dynamic clock and voltage scaling where the transitions between frequency-voltage modes are an atomic operation. Additionally, the resource power manager can control many modules, for example, clock dividers, in parallel. The invention can, due to lower latency between frequency-voltage modes, can provide improved system performance and reduced system power.

Abstract: A scannable sequential element is provided. The scannable sequential element includes a master stage that includes a data path configured to receive a data input. The master stage also includes a pass gate located on the data path and configured to selectively pass the data input, in which the data path has only one pass gate. The master stage also includes a test path coupled to the data path and configured to receive a test input. The master stage also includes pass gates located on the test path and configured to selectively pass the test input.