Memory Device with adaptive capacity

• US 8,239,735 B2
• Filed: 05/10/2007
• Issued: 08/07/2012
• Est. Priority Date: 05/12/2006
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Apple Inc.
• Inventors: Shalvi, Ofir, Sokolov, Dotan, Maislos, Ariel, Cohen, Zeev, Gurgi, Eyal, Semo, Gil
• Primary Examiner(s): Baderman, Scott
• Assistant Examiner(s): Leibovich, Yair
• Application Number:
  US12/063,544
• Publication Number:
  US 20100157641A1
• Time in Patent Office: 1,916 Days
• Field of Search: 714/763, 714/774, 365/45, 365/230.03, 365/218
• US Class Current: 714/774
• CPC Class Codes: G06F 2212/7205 Cleaning, compaction, garbage collection, erase control
• CPC Class Codes: G06F 2212/7206 Reconfiguration of flash memory system
• CPC Class Codes: G11C 11/56 using storage elements with more than two stable states represented by steps, e.g. of voltage, current, phase, frequency
• CPC Class Codes: G11C 11/5607 using magnetic storage elements
• CPC Class Codes: G11C 11/5621 using charge storage in a floating gate
• CPC Class Codes: G11C 11/5628 Programming or writing circuits; Data input circuits
• CPC Class Codes: G11C 11/5642 Sensing or reading circuits; Data output circuits
• CPC Class Codes: G11C 11/565 using capacitive charge storage elements
• CPC Class Codes: G11C 11/5657 using ferroelectric storage elements
• CPC Class Codes: G11C 11/5671 using charge trapping in an insulator
• CPC Class Codes: G11C 11/5678 using amorphous/crystalline phase transition storage elements
• CPC Class Codes: G11C 16/349 Arrangements for evaluating degradation, retention or wearout, e.g. by counting erase cycles
• CPC Class Codes: G11C 2211/5621 Multilevel programming verification
• CPC Class Codes: G11C 2211/5641 Multilevel memory having cells with different number of storage levels
• CPC Class Codes: G11C 27/005 with non-volatile charge storage, e.g. on floating gate or MNOS

Abstract

A method for data storage in a memory (28) that includes a plurality of analog memory cells (32) includes estimating respective achievable storage capacities of the analog memory cells. The memory cells are assigned respective storage configurations defining quantities of data to be stored in the memory cells based on the estimated achievable capacities. The data is stored in the memory cells in accordance with the respective assigned storage configurations. The achievable storage capacities of the analog memory cells are re-estimated after the memory has been installed in a host system and used for storing the data in the host system. The storage configurations are modified responsively to the re-estimated achievable capacities.

Claims

1. A method for data storage in a memory that includes a plurality of analog memory cells, the method comprising:

estimating respective achievable storage capacities of the analog memory cells;

assigning the memory cells respective storage configurations defining quantities of data to be stored in the memory cells based on the estimated achievable capacities;

storing the data in the memory cells in accordance with the respective assigned storage configurations;

re-estimating the respective achievable storage capacities of the analog memory cells after the memory has been installed in a host system and used for storing the data in the host system, and modifying the storage configurations responsively to the re-estimated achievable capacities,

wherein assigning and modifying the storage configurations comprise maintaining a predetermined margin between the quantities of the data stored in the memory cells and the respective estimated achievable capacities.

2. The method according to claim 1, wherein storing the data comprises encoding the data using an Error Correcting Code (ECC), converting the encoded data to analog values selected from a set of nominal analog values and writing the analog values into the respective memory cells, and wherein each storage configuration specifies a respective ECC code rate and a size of a set of nominal analog values used for storing the data.

3. The method according to claim 1, wherein, for each analog memory cell, storing the data comprises converting the data to analog values selected from a set of nominal analog values and writing the analog values into the memory cell, and wherein each storage configuration specifies the nominal analog values used for storing the data in the analog memory cell.

4. The method according to claim 1, wherein storing the data comprises converting the data to analog values and writing the analog values into the respective memory cells, and wherein estimating the achievable storage capacities comprises estimating respective distortion levels affecting the analog values written into the memory cells and determining the achievable storage capacities responsively to the distortion levels.

5. The method according to claim 4, wherein estimating the distortion levels comprises reading the analog values from the memory cells and estimating the distortion levels based on the analog values read from the memory cells.

6. The method according to claim 5, wherein estimating the distortion levels comprises reconstructing the data from the analog values read from the memory cells and calculating the distortion levels based on the reconstructed data.

7. The method according to claim 5, wherein calculating the distortion levels comprises summing scalar functions of the analog values read from the memory cells.

8. The method according to claim 7, wherein summing the scalar functions comprises summing squares of differences between the analog values read from the memory cells and respective expected values of the data stored in the memory cells responsively to the data.

9. The method according to claim 6, wherein calculating the distortion levels comprises counting a number of memory cells in which a difference between the analog values read from the memory cells and respective expected values of the data stored in the memory cells responsively to the data exceeds a predetermined value.

10. The method according to claim 4, wherein storing the data comprises writing the analog values into the memory cells and verifying the written analog values using an iterative Program and Verify (P&

V) process, and wherein estimating the distortion levels comprises calculating the distortion levels based on the analog values verified by the P&

V process.

11. The method according to claim 4, wherein storing the data comprises writing the analog values into the memory cells and verifying the written analog values using an iterative Program and Verify (P&

V) process, which iteratively increments the analog values by a programming step, and wherein the storage configurations define a size of the programming step used by the iterative P&

V process.

12. The method according to claim 1, wherein storing the data comprises encoding the data to be stored in a group of the memory cells using an Error Correcting Code (ECC) and converting the encoded data to analog values for storage in the respective analog memory cells of the group, wherein estimating the achievable storage capacities comprises reading the analog values from the memory cells in the group and evaluating a distance metric between the read analog values and a valid codeword of the ECC that is nearest to the analog values, and wherein modifying the storage configurations comprises adapting the storage configuration of the group of the memory cells responsively to the evaluated distance metric.

13. The method according to claim 1, wherein re-estimating the achievable storage capacities comprises reading the data stored in the memory cells, and wherein modifying the storage configurations comprises adapting the storage configurations responsively to detecting errors in the read data.

14. The method according to claim 13, wherein storing the data comprises encoding the data using an Error Correcting Code (ECC), and wherein adapting the storage configurations comprises modifying a code rate of the ECC.

15. The method according to claim 1, wherein estimating the achievable capacities of the memory cells comprises tracking previous programming and erasure operations applied to the respective memory cells and estimating the achievable capacities responsively to the tracked previous programming and erasure operations.

16. The method according to claim 15, wherein estimating the achievable capacities comprises calculating the achievable capacities responsively to lengths of time periods that elapsed since the previous programming and erasure operations.

17. The method according to claim 1, wherein storing the data comprises converting an initial portion of the data to analog values and writing the analog values into the respective memory cells, and subsequently storing an additional portion of the data in at least some of the memory cells by increasing the analog values written to the memory cells without erasing the cells.

18. The method according to claim 1, wherein storing the data comprises encoding the data using an Error Correcting Code (ECC) that adds redundancy bits to the data and storing the redundancy bits in some of the analog memory cells, and wherein modifying the storage configurations comprises modifying a number of the redundancy bits added by the ECC without erasing the cells.

19. The method according to claim 18, wherein storing the redundancy bits comprises storing the redundancy bits separately from the data.

20. The method according to claim 1, wherein assigning the storage configurations comprises predefining a set of possible storage configurations, and wherein modifying the storage configurations comprises selecting updated storage configurations from the predefined set.

21. The method according to claim 1, wherein the quantity of the data defined by at least one of the storage configurations specifies a non-integer number of bits per cell.

22. The method according to claim 1, wherein storing the data comprises compressing the data before writing the data to the memory cells, and wherein the storage configurations define respective compression ratios by which the data is to be compressed.

23. The method according to claim 1, wherein assigning the storage configurations comprises defining the storage configurations for respective groups of the memory cells.

24. The method according to claim 1, wherein storing the data in the memory cells comprises accepting a data item for storage in the memory, selecting a subset of the memory cells in which to store the data item responsively to the assigned storage configurations, and storing the data item in the selected subset of the memory cells.

25. The method according to claim 1, wherein re-estimating the achievable storage capacities is performed during idle time periods in which the data is not stored and read.

26. The method according to claim 1, wherein assigning and modifying the storage configurations comprise storing the storage configurations in a configuration table.

27. The method according to claim 26, and comprising storing initial values of the storage configurations in the configuration table before the memory has been installed in the host system.

28. The method according to claim 1, wherein at least some of the distortion is caused by leakage current in the analog memory cells, and wherein estimating the distortion comprises estimating the leakage current.

29. A method for data storage in a memory that includes a plurality of analog memory cells, the method comprising:

tracking respective achievable storage capacities of the analog memory cells while the memory is in use in a host system;

accepting data for storage in the memory;

selecting a subset of the memory cells for storing the data based on the tracked achievable capacities; and

storing the data in the memory cells of the subset,

wherein accepting the data for storage comprises accepting the data in fixed-capacity blocks from the host system and storing the data in variable-capacity groups of the memory cells based on the tracked achievable capacities of variable-size groups, sequentially and irrespective of boundaries between the variable-capacity groups.

30. The method according to claim 29, wherein selecting the subset comprises selecting the memory cells whose aggregate achievable capacity best matches a size of the accepted data.

31. The method according to claim 29, wherein accepting the data comprises accepting a requested reliability level for storing the data, and wherein selecting the subset comprises selecting the memory cells responsively to the requested reliability level.

32. The method according to claim 29, wherein accepting the data comprises accepting a requested retention period for storing the data, and wherein selecting the subset comprises selecting the memory cells responsively to the requested retention period.

33. The method according to claim 29, wherein selecting the subset comprises selecting the memory cells having low distortion levels with respect to other memory cells.

34. The method according to claim 29, wherein tracking the achievable storage capacities comprises tracking previous programming and erasure operations applied to the memory cells, and wherein selecting the subset comprises selecting the memory cells responsively to the previous programming and erasure operations.

35. The method according to claim 34, wherein selecting the subset comprises distributing a number of programming and erasure operations evenly over the memory cells in the memory by selecting the memory cells having a smaller number of the previous programming and erasure operations with respect to other memory cells.

36. The method according to claim 35, wherein distributing the number of programming and erasure operations comprises selecting the memory cells having the smaller number of the previous programming and erasure operations performed during a predetermined recent time period with respect to the other memory cells.

37. The method according to claim 29, wherein tracking the achievable storage capacities comprises calculating and reporting to the host system a size of an available memory space by summing the achievable capacities of the memory cells that are available for storing the data.

38. The method according to claim 29, wherein the memory is divided into multiple erasure blocks, each erasure block comprising a group of the memory cells that are erased in a single erasure operation, and comprising erasing a data item from the memory by:

identifying one or more of the erasure blocks in which the data item is stored;

when the erasure blocks in which the data item is stored contain stored data of another data item, identifying alternative memory cells outside the erasure blocks in which the data item is stored based on the tracked achievable capacities of the memory cells and on a size of the other data item, and copying the data of the other data item to the alternative memory cells; and

erasing the erasure blocks in which the data item is stored.

39. The method according to claim 29, wherein storing the data in the variable-capacity groups comprises identifying two or more variable-capacity groups having capacities that are smaller than a capacity of the fixed-capacity block, and storing the data accepted in one or more of the fixed-capacity blocks in the identified two or more variable-capacity groups.

40. The method according to claim 29, wherein storing the data in the variable-capacity groups comprises allocating one or more of the variable-capacity groups to serve as an excess memory storage area, storing the data accepted in the fixed-capacity blocks in respective variable-capacity groups, and, when the achievable capacity of a variable-capacity group is smaller than a capacity of the fixed-capacity blocks, storing some of the data accepted in the respective fixed-capacity block in the allocated excess memory storage area.

41. The method according to claim 29, wherein storing the data comprises initially storing the data accepted in each fixed-capacity block in a respective variable-capacity group, and subsequently re-programming the initially-stored data sequentially, irrespective of the boundaries between the variable-capacity groups.

42. The method according to claim 29, and comprising retrieving the data from the variable-size groups, reconstructing the fixed-size blocks and outputting the data using the fixed-size blocks to the host system.

43. The method according to claim 29, and comprising reducing a number of memory access operations applied to the memory by caching at least some of the data.

44. The method according to claim 43, wherein caching the at least some of the data comprises transferring the cached data to the memory upon detecting an immediate storage event.

45. A method for data storage in a memory that includes a plurality of analog memory cells, the method comprising:

tracking respective achievable storage capacities of the analog memory cells while the memory is in use in a host system;

accepting data for storage in the memory;

selecting a subset of the memory cells for storing the data based on the tracked achievable capacities;

storing the data in the memory cells of the subset; and

reducing a number of memory access operations applied to the memory by caching at least some of the data, wherein caching the at least some of the data comprises transferring the cached data to the memory upon detecting an immediate storage event, comprising at least one event selected from a group of events consisting of an approaching power failure, a time out and an End Of File (EOF) command accepted from the host system.

46. Data storage apparatus, comprising:

an interface, which is arranged to communicate with a memory that includes a plurality of analog memory cells; and

a memory signal processor (MSP), which is arranged to estimate respective achievable storage capacities of the memory cells, to assign the memory cells respective storage configurations defining quantities of data to be stored in the memory cells based on the estimated achievable capacities, to store the data in the memory cells in accordance with the respective assigned storage configurations, to re-estimate the respective achievable storage capacities of the analog memory cells after the memory has been installed in a host system and used for storing the data in the host system, and to modify the storage configurations responsively to the re-estimated achievable capacities.

47. The apparatus according to claim 46, wherein the MSP is arranged to encode the data using an Error Correcting Code (ECC), to convert the encoded data to analog values selected from a set of nominal analog values and to write the analog values into the respective memory cells, and wherein each storage configuration specifies a respective ECC code rate and a size of a set of nominal analog values used for storing the data.

48. The apparatus according to claim 46, wherein, for each analog memory cell, the MSP is arranged to store the data by converting the data to analog values selected from a set of nominal analog values and writing the analog values into the memory cell, and wherein each storage configuration specifies the nominal analog values used for storing the data in the analog memory cell.

49. The apparatus according to claim 46, wherein the MSP is arranged to convert the data to analog values and to write the analog values into the respective memory cells, to estimate respective distortion levels affecting the analog values written into the memory cells and to determine the achievable storage capacities responsively to the distortion levels.

50. The apparatus according to claim 49, wherein the MSP is arranged to read the analog values from the memory cells and to estimate the distortion levels based on the analog values read from the memory cells.

51. The apparatus according to claim 50, wherein the MSP is arranged to reconstruct the data from the analog values read from the memory cells, and to calculate the distortion levels based on the reconstructed data.

52. The apparatus according to claim 51, wherein the MSP is arranged to calculate the distortion levels by summing scalar functions of the analog values read from the memory cells.

53. The apparatus according to claim 52, wherein the MSP is arranged to sum squares of differences between the analog values read from the memory cells and respective expected values of the data stored in the memory cells responsively to the data.

54. The apparatus according to claim 52, wherein the MSP is arranged to calculate the distortion levels by counting a number of memory cells in which a difference between the analog values read from the memory cells and respective expected values of the data stored in the memory cells responsively to the data exceeds a predetermined value.

55. The apparatus according to claim 49, wherein the MSP is arranged to write the analog values into the memory cells and verify the written analog values using an iterative Program and Verify (P&

V) process, and to calculate the distortion levels based on the analog values verified by the P&

V process.

56. The apparatus according to claim 49, wherein the MSP is arranged to write the analog values into the memory cells and verify the written analog values using an iterative Program and Verify (P&

V) process, which iteratively increments the analog values by a programming step, and wherein the storage configurations define a size of the programming step used by the iterative P&

V process.

57. The apparatus according to claim 46, wherein the MSP is arranged to encode the data to be stored in a group of the memory cells using an Error Correcting Code (ECC), to convert the encoded data to analog values for storage in the respective analog memory cells of the group, to read the analog values from the memory cells in the group, to evaluate a distance metric between the read analog values and a valid codeword of the ECC that is nearest to the analog values and to adapt the storage configuration of the group of the memory cells responsively to the evaluated distance metric.

58. The apparatus according to claim 46, wherein the MSP is arranged to read the data stored in the memory cells and to modify the storage configurations responsively to detecting errors in the read data.

59. The apparatus according to claim 58, wherein the MSP is arranged to encode the data using an Error Correcting Code (ECC), and to modify the storage configurations by modifying a code rate of the ECC.

60. The apparatus according to claim 46, wherein the MSP is arranged to track previous programming and erasure operations applied to the memory cells and to estimate the respective achievable capacities responsively to the tracked previous programming and erasure operations.

61. The apparatus according to claim 60, wherein the MSP is arranged to calculate the achievable capacities responsively to lengths of time periods that elapsed since the previous programming and erasure operations.

62. The apparatus according to claim 46, wherein the MSP is arranged to convert an initial portion of the data to analog values and write the analog values into the respective memory cells, and to subsequently store an additional portion of the data in the memory cells by increasing the analog values written to at least some of the memory cells without erasing the cells.

63. The apparatus according to claim 46, wherein the MSP is arranged to encode the data using an Error Correcting Code (ECC) that adds redundancy bits to the data and storing the redundancy bits in some of the analog memory cells, and to modify the storage configurations by modifying a number of the redundancy bits added by the ECC without erasing the cells.

64. The apparatus according to claim 63, wherein the MSP is arranged to store the redundancy bits separately from the data.

65. The apparatus according to claim 46, wherein the MSP is arranged to modify the storage configurations by selecting updated storage configurations from a predefined set of possible storage configurations.

66. The apparatus according to claim 46, wherein the quantity of the data defined by at least one of the storage configurations specifies a non-integer number of bits per cell.

67. The apparatus according to claim 46, wherein the MSP is arranged to compress the data before writing the data to the memory cells, and wherein the storage configurations define respective compression ratios by which the data is to be compressed.

68. The apparatus according to claim 46, wherein the MSP is arranged to define the storage configurations for respective groups of the memory cells.

69. The apparatus according to claim 46, wherein the MSP is arranged to accept a data item for storage in the memory, to select a subset of the memory cells in which to store the data item responsively to the assigned storage configurations, and to store the data item in the selected subset of the memory cells.

70. The apparatus according to claim 46, wherein the MSP is arranged to re-estimate the achievable storage capacities during idle time periods in which the data is not stored and read.

71. The apparatus according to claim 46, wherein the MSP is arranged to store the storage configurations in a configuration table.

72. The apparatus according to claim 46, wherein the MSP is arranged to maintain a predetermined margin between the quantities of the data stored in the memory cells and the respective estimated achievable capacities.

73. The apparatus according to claim 46, wherein at least some distortion affecting the analog values written into the memory cells is caused by leakage current in the analog memory cells, and wherein the MSP is arranged to estimate the leakage current.

74. Data storage apparatus, comprising:

an interface, which is arranged to communicate with a memory that includes a plurality of analog memory cells; and

a memory signal processor (MSP), which is arranged to track respective achievable storage capacities of the memory cells while the memory is in use in a host system, to accept data for storage in the memory, to select a subset of the memory cells for storing the data based on the tracked achievable capacities, and to store the data in the memory cells of the subset,

wherein the MSP is arranged to accept the data from the host system in fixed-capacity blocks and to store the data in variable-capacity groups of the memory cells based on the tracked achievable capacities of variable-size groups, sequentially and irrespective of boundaries between the variable-capacity groups.

75. The apparatus according to claim 74, wherein the MSP is arranged to select the memory cells whose aggregate achievable capacity best matches a size of the accepted data.

76. The apparatus according to claim 74, wherein the MSP is arranged to accept a requested reliability level for storing the data and to select the memory cells responsively to the requested reliability level.

77. The apparatus according to claim 74, wherein the MSP is arranged to accept a requested retention period for storing the data and to select the memory cells responsively to the requested retention period.

78. The apparatus according to claim 74, wherein the MSP is arranged to select the memory cells having low distortion levels with respect to other memory cells.

79. The apparatus according to claim 74, wherein the MSP is arranged to track previous programming and erasure operations applied to the memory cells and to select the memory cells responsively to the previous programming and erasure operations.

80. The apparatus according to claim 79, wherein the MSP is arranged to distribute a number of programming and erasure operations evenly over the memory cells in the memory by selecting the memory cells having a smaller number of the previous programming and erasure operations with respect to other memory cells.

81. The apparatus according to claim 80, wherein the MSP is arranged to select the memory cells having the smaller number of the previous programming and erasure operations performed during a predetermined recent time period with respect to the other memory cells.

82. The apparatus according to claim 74, wherein the MSP is arranged to calculate and report to the host system a size of an available memory space by summing the achievable capacities of the memory cells that are available for storing the data.

83. The apparatus according to claim 74, wherein the memory is divided into multiple erasure blocks, each erasure block comprising a group of the memory cells that are erased in a single erasure operation, and wherein the MSP is arranged to erase a data item from the memory by:

identifying one or more of the erasure blocks in which the data item is stored;

when the erasure blocks in which the data item is stored contain stored data of another data item, identifying alternative memory cells outside the erasure blocks in which the data item is stored based on the tracked achievable capacities of the memory cells and on a size of the other data item, and copying the data of the other data item to the alternative memory cells; and

erasing the erasure blocks in which the data item is stored.

84. The apparatus according to claim 74, wherein the MSP is arranged to identify two or more variable-capacity groups having capacities that are smaller than a capacity of the fixed-capacity block, and to store the data accepted in one or more of the fixed-capacity blocks in the identified two or more variable-capacity groups.

85. The apparatus according to claim 74, wherein the MSP is arranged to allocate one or more of the variable-capacity groups to serve as an excess memory storage area, to store the data accepted in the fixed-capacity blocks in respective variable-capacity groups, and, when the achievable capacity of a variable-capacity group is smaller than a capacity of the fixed-capacity blocks, to store some of the data accepted in the respective fixed-capacity block in the allocated excess memory storage area.

86. The apparatus according to claim 74, wherein the MSP is arranged to initially store the data accepted in each fixed-capacity block in a respective variable-capacity group, and to subsequently re-program the initially-stored data sequentially, irrespective of the boundaries between the variable-capacity groups.

87. The apparatus according to claim 74, wherein the MSP is arranged to retrieve the data from the variable-size groups, to reconstruct the fixed-size blocks and to output the data using the fixed-size blocks to the host system.

88. The apparatus according to claim 74, wherein the MSP is arranged to reduce a number of memory access operations applied to the memory by caching at least some of the data.

89. The apparatus according to claim 88, wherein the MSP is arranged to transfer the cached data to the memory upon detecting an immediate storage event.

90. Data storage apparatus,

comprising:

an interface, which is arranged to communicate with a memory that includes a plurality of analog memory cells; and

a memory signal processor (MSP), which is arranged to track respective achievable storage capacities of the memory cells while the memory is in use in a host system, to accept data for storage in the memory, to select a subset of the memory cells for storing the data based on the tracked achievable capacities, and to store the data in the memory cells of the subset,

wherein the MSP is arranged to reduce a number of memory access operations applied to the memory by caching at least some of the data, and to transfer the cached data to the memory upon detecting an immediate storage event comprising at least one event selected from a group of events consisting of an approaching power failure, a time out and an End Of File (EOF) command accepted from the host system.

91. Data storage apparatus, comprising:

a memory, which comprises a plurality of analog memory cells; and

a memory signal processor (MSP), which is coupled to the memory and is arranged to estimate respective achievable storage capacities of the memory cells, to assign the memory cells respective storage configurations defining quantities of data to be stored in the memory cells based on the estimated achievable capacities, to store the data in the memory cells in accordance with the respective assigned storage configurations, to re-estimate the respective achievable storage capacities of the analog memory cells after the memory has been installed in a host system and used for storing the data in the host system, and to modify the storage configurations responsively to the re-estimated achievable capacities.

92. The apparatus according to claim 91, wherein the memory resides in a first Integrated Circuit (IC) and wherein the MSP resides in a second IC different from the first IC.

93. The apparatus according to claim 91, wherein the memory and the MSP are integrated in a single device.

94. The apparatus according to claim 91, wherein the MSP is embodied in a processor of the host system.

95. The apparatus according to claim 91, wherein the memory cells comprise Flash memory cells.

96. The apparatus according to claim 91, wherein the memory cells comprise Dynamic Random Access Memory (DRAM) cells.

97. The apparatus according to claim 91, wherein the memory cells comprise Phase Change Memory (PCM) cells.

98. The apparatus according to claim 91, wherein the memory cells comprise Nitride Read-Only Memory (NROM) cells.

99. The apparatus according to claim 91, wherein the memory cells comprise Magnetic Random Access Memory (MRAM) cells.

100. The apparatus according to claim 91, wherein the memory cells comprise Ferroelectric Random Access Memory (FRAM) cells.

101. Data storage apparatus, comprising:

a memory, which comprises a plurality of analog memory cells; and

a memory signal processor (MSP), which is coupled to the memory and is arranged to track respective achievable storage capacities of the memory cells while the memory is in use in a host system, to accept data for storage in the memory, to select a subset of the memory cells for storing the data based on the tracked achievable capacities, and to store the data in the memory cells of the subset,

wherein the MSP is arranged to accept the data from the host system in fixed-capacity blocks and to store the data in variable-capacity groups of the memory cells based on the tracked achievable capacities of variable-size groups, sequentially and irrespective of boundaries between the variable-capacity groups.

102. The apparatus according to claim 101, wherein the memory resides in a first Integrated Circuit (IC) and wherein the MSP resides in a second IC different from the first IC.

103. The apparatus according to claim 101, wherein the memory and the MSP are integrated in a single device.

104. The apparatus according to claim 101, wherein the MSP is embodied in a processor of the host system.

105. The apparatus according to claim 101, wherein the memory cells comprise Flash memory cells.

106. The apparatus according to claim 101, wherein the memory cells comprise Dynamic Random Access Memory (DRAM) cells.

107. The apparatus according to claim 101, wherein the memory cells comprise Phase Change Memory (PCM) cells.

108. The apparatus according to claim 101, wherein the memory cells comprise Nitride Read-Only Memory (NROM) cells.

109. The apparatus according to claim 101, wherein the memory cells comprise Magnetic Random Access Memory (MRAM) cells.

110. The apparatus according to claim 101, wherein the memory cells comprise Ferroelectric Random Access Memory (FRAM) cells.