Abstract: Ferroelectric capacitor structures for integrated decoupling capacitors and the like. The ferroelectric capacitor structure includes two or more ferroelectric capacitors connected in series with one another between voltage nodes. The series connection of the ferroelectric capacitors reduces the applied voltage across each, enabling the use of rough ferroelectric dielectric material, such as PZT deposited by MOCVD. Matched construction of the series-connected capacitors, as well as uniform polarity of the applied voltage across each, is beneficial in reducing the maximum voltage across any one of the capacitors, reducing the vulnerability to dielectric breakdown.

Abstract: An FRAM device can comprise a sense amplifier, at least a first bitcell, a first control line, and a second control line. The first bitcell can have a bit line that connects to the sense amplifier via a first isolator and a complimentary bit line that connects to the sense amplifier via a second isolator that is different from the first isolator. The first control line can connect to and control the aforementioned first isolator. And the second control line can connect to and control the second isolator such that the second isolator is independently controlled with respect to the first isolator to facilitate testing the device.

Abstract: An FRAM device can comprise a sense amplifier and at least a first bitcell. The first bitcell can have a bit line and a complimentary bit line that connects to the sense amplifier. A first precharge circuit responds to a first control signal during a test mode of operation to precharge the bit line with respect to a first voltage while a second precharge circuit responds to a second control signal (that is different from the first control signal) during the test mode of operation to precharge the complimentary bit line with respect to a test voltage that is different than the first voltage (such as, but not limited to, a test voltage of choice such as a voltage that is greater than ground but less than the first voltage).

Abstract: A processing device using a plurality of volatile storage elements to execute a boot process for and stores in a plurality of non-volatile logic element arrays a boot state representing a state of the processing device after a given amount of the boot process is completed. When it is determined that the processing device needs to restart from a boot state, energy can be saved by restoring the machine state at that boot state instead of re-booting. The stored boot state will not change, and given the nature of certain non-volatile storage elements, the data read from the NVL storage elements needs to be re-written to the elements after read out. Accordingly, a round-trip data restoration operation is executed that automatically writes back data to an individual non-volatile logic element after reading data from the individual non-volatile logic element without completing separate read and write operations.

Abstract: Input power quality for a processing device is sensed. In response to detection of poor power quality, input power is disconnected, and the processing device backs up its machine state in non-volatile logic element arrays using available stored charge. When power is restored, the stored machine state is restored from the non-volatile logic element arrays to the volatile logic elements whereby the processing device resumes its process from the state immediately prior to power loss allowing seamless processing across intermittent power supply.

Abstract: A processing device handles two or more operating threads. A non-volatile logic controller stores first program data from a first program in a first set of non-volatile logic element arrays and second program data from a second program in a second set of non-volatile logic element arrays. The first program and the second program can correspond to distinct executing threads, and the storage can be completed in response to receiving a stimulus regarding an interrupt for the computing device apparatus or in response to a power supply quality problem for the computing device apparatus. When the device needs to switch between processing threads, the non-volatile logic controller restores the first program data or the second program data from the non-volatile logic element arrays in response to receiving a stimulus regarding whether the first program or the second program is to be executed by the computing device apparatus.

Abstract: A processing device includes a plurality of non-volatile logic element array domains having two or more non-volatile logic element arrays to store 2006 a machine state of the processing device stored in a plurality of volatile store elements. Configuration bits are read to direct which non-volatile logic element array domains are enabled first and to direct an order in which the first enabled non-volatile logic element array domains are restored or backed up in response to entering a wakeup or backup mode. Configuration bits can be read to direct an order of and a parallelism of how individual non-volatile logic element arrays in a first enabled non-volatile logic element array domain are restored or backed up. The order of restoration or backing up can be controlled by instructions from non-volatile arrays of the first enabled of the plurality of non-volatile logic element array domains.

Abstract: A computing device includes a first set of non-volatile logic element arrays associated with a first function and a second set of non-volatile logic element arrays associated with a second function. The first and second sets of non-volatile logic element arrays are independently controllable. A first power domain supplies power to switched logic elements of the computing device, a second power domain supplies power to logic elements configured to control signals for storing data to or reading data from non-volatile logic element arrays, and a third power domain supplies power for the non-volatile logic element arrays. The different power domains are independently powered up or down based on a system state to reduce power lost to excess logic switching and the accompanying parasitic power consumption during the recovery of system state and to reduce power leakage to backup storage elements during regular operation of the computing device.

Abstract: A processing device is operated using a plurality of volatile storage elements. N groups of M volatile storage elements of the plurality of volatile storage elements per group are connected to an N by M size non-volatile logic element array of a plurality of non-volatile logic element arrays using a multiplexer. The multiplexer connects one of the N groups to the N by M size non-volatile logic element array to store data from the M volatile storage elements into a row of the N by M size non-volatile logic element array at one time or to write data to the M volatile storage elements from a row of the N by M size non-volatile logic element array at one time. A corresponding non-volatile logic controller controls the multiplexer operation with respect to the connections between volatile storage elements and non-volatile storage elements.

Abstract: A processing device selectively backups only certain data based on a priority or binning structure. In one approach, a non-volatile logic controller stores the machine state by storing in non-volatile logic element arrays a portion of data representing the machine state less than all the data of the machine state. Accordingly, the non-volatile logic controller stores the machine state in the plurality of non-volatile logic element arrays by storing a first set of program data of the machine state according to a first category for backup and restoration and storing a second set of program data of the machine state according to a second category for backup and restoration.

Abstract: Design and operation of a processing device is configurable to optimize wake-up time and peak power cost during restoration of a machine state from non-volatile storage. The processing device includes a plurality of non-volatile logic element arrays configured to store a machine state represented by a plurality of volatile storage elements of the processing device. A stored machine state is read out from the plurality of non-volatile logic element arrays to the plurality of volatile storage elements. During manufacturing, a number of rows and a number of bits per row in non-volatile logic element arrays are based on a target wake up time and a peak power cost. In another approach, writing data to or reading data of the plurality of non-volatile arrays can be done in parallel, sequentially, or in any combination to optimize operation characteristics.

Abstract: A device's configuration is controlled through control of its pre-boot process. Protected non-volatile logic element arrays store a machine state configuration of a processing device configured to backup data from volatile storage elements in a plurality of non-volatile logic element arrays. The machine state configuration is read in response to the processing device's entering a pre-boot process. The processing device's configuration is then set to the machine state configuration. This setting of the device configuration can be done by receiving instructions from the protected non-volatile logic element arrays to direct an order in which data for individual device functions are restored from non-volatile logic element arrays in response to the processing device's entering a wakeup or recovery mode. In one approach, the instructions arrange configuration bits that direct operation of a non-volatile logic controller during the wakeup or recovery mode to control the order of data restoration.

Abstract: Individual first ones of a plurality of non-volatile logic element arrays are designated to restore first in response to entering a wakeup or restoration mode. These non-volatile logic element arrays include instructions for an order in which other non-volatile logic element arrays are to be restored next. So configured, the processing device can be set to have one or more NVL arrays restored first, which arrays are pre-configured to guide further wakeup of the device through directed restoration from particular NVL arrays. Certain NVL arrays can be skipped if the functions stored therein are not needed, and the order of restoration of others can be tailored to a particular wakeup time and power concern through restoration in parallel, serial, or combinations thereof.

Abstract: A processing device is operated using a plurality of volatile storage elements. Data in the plurality of volatile storage elements is stored in a plurality of non-volatile logic element arrays. A primary logic circuit portion of individual ones of the plurality of volatile storage elements is powered by a first power domain, and a slave stage circuit portion of individual ones of the plurality of volatile storage elements is powered by a second power domain. During a write back of data from the plurality of non-volatile logic element arrays to the plurality of volatile storage elements, the first power domain is powered down and the second power domain is maintained. In a further approach, the plurality of non-volatile logic element arrays is powered by a third power domain, which is powered down during regular operation of the processing device.

Abstract: Ferroelectric capacitor structures for integrated decoupling capacitors and the like. The ferroelectric capacitor structure includes two or more ferroelectric capacitors connected in series with one another between voltage nodes. The series connection of the ferroelectric capacitors reduces the applied voltage across each, enabling the use of rough ferroelectric dielectric material, such as PZT deposited by MOCVD. Matched construction of the series-connected capacitors, as well as uniform polarity of the applied voltage across each, is beneficial in reducing the maximum voltage across any one of the capacitors, reducing the vulnerability to dielectric breakdown.

Abstract: A ferroelectric memory device includes a shunt switch configured to short both sides of the ferroelectric capacitor of the ferroelectric memory device. The shunt switch is configured therefore to remove excess charge from around the ferroelectric capacitor prior to or after reading data from the ferroelectric capacitor. By one approach, the shunt switch is connected to operate in reaction to signals from the same line that controls accessing the ferroelectric capacitor. So configured, the high performance cycle time of the ferroelectric memory device is reduced by eliminating delays used to otherwise drain excess charge from around the ferroelectric capacitor, for example by applying a precharge voltage. The shunt switch also improves reliability of the ferroelectric memory device by ensuring that excess charge does not affect the reading of the ferroelectric capacitor during a read cycle.

Abstract: A self-timed sense amplifier read buffer pulls down a pre-charged high global bit line, which then feeds data into a tri state write back buffer that is connected directly to the bit line. The bit line provides charge to a ferroelectric capacitor to write a logical “one” or “zero” while by-passing an isolator switch disposed between the sense amplifier and the ferroelectric capacitor. Because the sense amplifier uses grounded bit line sensing, the read buffer will not start pulling down the global bit line until after the sense amplifier signal amplification, which makes the timing of the control signal for this read buffer non-critical. The write-back buffer enable timing is also self-timed off of the sense amplifier. Therefore, the read data write-back to a ferroelectric memory cell is locally controlled and begins quickly after reading data from the ferroelectric memory cell, thereby allowing a quick cycle time.

Abstract: Non-volatile latch circuits, such as in memory cells and flip-flops, that are constructed for reliability screening. The non-volatile latch circuits each include ferroelectric capacitors coupled to storage nodes, for example at the outputs of cross-coupled inverters. Separate plate lines are connected to the ferroelectric capacitors of the complementary storage nodes. A time-zero test of the latch stability margin is performed by setting a logic state at the storage nodes, then programming the state into the ferroelectric capacitors by polarization. After power-down, the plate lines are biased with a differential voltage relative to one another, and the latch is then powered up to attempt recall of the programmed state. The differential voltage disturbs the recall, and provides a measure of the cell margin and its later-life reliability.

Abstract: Non-volatile latch circuits, such as in memory cells and flip-flops, that are constructed for reliability screening. The non-volatile latch circuits each include ferroelectric capacitors coupled to storage nodes, for example at the outputs of cross-coupled inverters. Separate plate lines are connected to the ferroelectric capacitors of the complementary storage nodes. A time-zero test of the latch stability margin is performed by setting a logic state at the storage nodes, then programming the state into the ferroelectric capacitors by polarization. After power-down, the plate lines are biased with a differential voltage relative to one another, and the latch is then powered up to attempt recall of the programmed state. The differential voltage disturbs the recall, and provides a measure of the cell margin and its later-life reliability.

Abstract: Non-volatile latch circuits, such as in memory cells and flip-flops, that are constructed for reliability screening. The non-volatile latch circuits each include ferroelectric capacitors coupled to storage nodes, for example at the outputs of cross-coupled inverters. Separate plate lines are connected to the ferroelectric capacitors of the complementary storage nodes. A time-zero test of the latch stability margin is performed by setting a logic state at the storage nodes, then programming the state into the ferroelectric capacitors by polarization. After power-down, the plate lines are biased with a differential voltage relative to one another, and the latch is then powered up to attempt recall of the programmed state. The differential voltage disturbs the recall, and provides a measure of the cell margin and its later-life reliability.