Abstract:
Technologies are generally provided for methods and circuitry to rank a large number of cells in a timeframe of about one sense cycle. In some examples, an architecture may be implemented to rank memory cells such as volatile memories, non-volatile memories, and other types of data storage devices, where there may not be an equivalent to threshold voltage. In other examples, an arbitrary group of circuits, such as in neural networks where there may not be an equivalent control gate to set the timing resolution, may be ranked. Relative sense timing may be used to rank the cells having different current carrying abilities. A ramped gate voltage may be used to control the timing resolution and to reduce contention between close separate cells. Digital logic may be used to latch and/or record the rank information.
Abstract:
Technologies are generally described to store data in single-level memory using rank modulation. In some examples, data to be encoded to single-level memory may be represented with a bit ranking for a group of bits. A program vector may then be determined from the bit ranking and partial program characteristics associated with the memory group(s). The memory group(s) may then be programmed according to the program vector. The encoded data may be subsequently retrieved by performing a series of partial programming operations on the memory group(s) to recover the bit ranking and derive the data represented.
Abstract:
Technologies are generally described to store data in single-level memory using rank modulation. In some examples, data to be encoded to single-level memory may be represented with a bit ranking for a group of bits. A program vector may then be determined from the bit ranking and partial program characteristics associated with the memory group(s). The memory group(s) may then be programmed according to the program vector. The encoded data may be subsequently retrieved by performing a series of partial programming operations on the memory group(s) to recover the bit ranking and derive the data represented.
Abstract:
Technologies are generally described for partial programming of memory having physically unclonable functions for secure data storage. In some examples, a sender that wishes to securely send data to a recipient using a physical memory may measure a program threshold average and a program threshold variation for bits in the memory and group the bits into different bins based on the measured average and variation. The sender may partially program the data to a set of bits selected from one or more of the bins by applying partial program pulses to the bits based on the program threshold average and the program threshold variation. The sender may then provide the partially-programmed memory to the recipient. The recipient may then partially program the received memory based on the program threshold average and the program threshold variation to recover the programmed data.
Abstract:
Technologies are generally provided for methods and circuitry to rank a large number of cells in a timeframe of about one sense cycle. In some examples, an architecture may be implemented to rank memory cells such as volatile memories, non-volatile memories, and other types of data storage devices, where there may not be an equivalent to threshold voltage. In other examples, an arbitrary group of circuits, such as in neural networks where there may not be an equivalent control gate to set the timing resolution, may be ranked. Relative sense timing may be used to rank the cells having different current carrying abilities. A ramped gate voltage may be used to control the timing resolution and to reduce contention between close separate cells. Digital logic may be used to latch and/or record the rank information.
Abstract:
Technologies are generally described to store data in single-level memory using rank modulation. In some examples, data to be encoded to single-level memory may be represented with a bit ranking for a group of bits. A program vector may then be determined from the bit ranking and partial program characteristics associated with the memory group(s). The memory group(s) may then be programmed according to the program vector. The encoded data may be subsequently retrieved by performing a series of partial programming operations on the memory group(s) to recover the bit ranking and derive the data represented.
Abstract:
Technologies are generally described to store data in single-level memory using rank modulation. In some examples, data to be encoded to single-level memory may be represented with a bit ranking for a group of bits. A program vector may then be determined from the bit ranking and partial program characteristics associated with the memory group(s). The memory group(s) may then be programmed according to the program vector. The encoded data may be subsequently retrieved by performing a series of partial programming operations on the memory group(s) to recover the bit ranking and derive the data represented.
Abstract:
Technologies are generally described for partial programming of memory having physically unclonable functions for secure data storage. In some examples, a sender that wishes to securely send data to a recipient using a physical memory may measure a program threshold average and a program threshold variation for bits in the memory and group the bits into different bins based on the measured average and variation. The sender may partially program the data to a set of bits selected from one or more of the bins by applying partial program pulses to the bits based on the program threshold average and the program threshold variation. The sender may then provide the partially-programmed memory to the recipient. The recipient may then partially program the received memory based on the program threshold average and the program threshold variation to recover the programmed data.
Abstract:
Technologies are generally provided for methods and circuitry to rank a large number of cells in a timeframe of about one sense cycle. In some examples, an architecture may be implemented to rank memory cells such as volatile memories, non-volatile memories, and other types of data storage devices, where there may not be an equivalent to threshold voltage. In other examples, an arbitrary group of circuits, such as in neural networks where there may not be an equivalent control gate to set the timing resolution, may be ranked. Relative sense timing may be used to rank the cells having different current carrying abilities. A ramped gate voltage may be used to control the timing resolution and to reduce contention between close separate cells. Digital logic may be used to latch and/or record the rank information.
Abstract:
Technologies are generally described for partial programming of memory having physically unclonable functions for secure data storage. In some examples, a sender that wishes to securely send data to a recipient using a physical memory may measure a program threshold average and a program threshold variation for bits in the memory and group the bits into different bins based on the measured average and variation. The sender may partially program the data to a set of bits selected from one or more of the bins by applying partial program pulses to the bits based on the program threshold average and the program threshold variation. The sender may then provide the partially-programmed memory to the recipient. The recipient may then partially program the received memory based on the program threshold average and the program threshold variation to recover the programmed data.