Fault mapping apparatus for memory

• US 5,233,614 A
• Filed: 01/07/1991
• Issued: 08/03/1993
• Est. Priority Date: 01/07/1991
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Singh, Shanker
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Tu, Trinh
• Application Number:
  US07/637,731
• Time in Patent Office: 939 Days
• Field of Search: 371/21.6, 371/21.1, 371/2.1, 371/2.2, 371/5.1, 371/10.1, 371/51.1, 371/40.1, 371/40.2, 371/40.3, 371/5.5, 371/13, 364/243, 364/246.3, 364/200
• US Class Current: 714/723
• CPC Class Codes: G06F 11/1044 with specific ECC/EDC distribution
• CPC Class Codes: G06F 2201/88 Monitoring involving counting
• CPC Class Codes: G11C 29/88 with partially good memories

Abstract

A memory fault mapping apparatus detects faults generated in a memory array during on-line operation. As the memory array is randomly accessed, single bit error are detected, corrected, and mapped into an error memory. The errors may be mapped in an error memory having a memory location for each memory of the memory array or alternatively, by grouping memories together and when the errors generated by any one group exceeds a predetermined threshold of errors, testing only the memories in that group off-line. By grouping the memories a substantial reduction in the amount of error memory required can be achieved. A SEC/DED syndrome generator detects single and double bit errors, correcting the single bit errors while providing an indication of which memory generated the error. An error memory stores error counts for the memory array, each error count indicating the number of errors for a specific memory or a group of memories. The error counts are incremented by loading the error count into a counter for incrementing then writing the incremented error count back to the error memory location from which it was read.

Claims

1. A memory fault mapping apparatus for monitoring randomly accessed data from a plurality of memory chips arranged in rows and columns and addressed by at least a row select address, comprising:

detecting means coupled to the plurality of memory chips for checking the accessed data from said plurality of memory chips and providing an error indication and an error syndrome if an error is detected in the accessed data, the error syndrome indicating the column from which the error was detected;

error memory, coupled to said detecting means and addressable by the error syndrome and the row select address, for storing a count of detected errors generated by each memory chip of said plurality of memory chips in corresponding predetermined locations of said error memory; and

counting means coupled to said error memory and to said detecting means for receiving an error count from a selected one of said predetermined locations of said error memory, incrementing the error count if an error indication is provided by said detecting means, and rewriting the incremented error count to said selected one of said predetermined locations of said error memory.

2. The memory fault mapping apparatus according to claim 1 wherein said detecting means is an SEC/DED syndrome generator providing the error syndrome when a single bit error is detected and corrected.

3. The memory fault mapping apparatus according to claim 1 wherein said counting means comprises a single counter.

4. The memory fault mapping apparatus according to claim 1 wherein the error count is accessed, incremented, and rewritten to said selected one of said predetermined locations of said error memory within a single data access cycle.

5. The memory fault mapping apparatus according to claim 1 wherein said error memory is logically divided into two arrays, a first array for storing the count of detected errors generated by each memory chip of said plurality of memory chips and a second array for storing a status word associated with each of the error counts.

6. The memory fault mapping apparatus according to claim 5 wherein each error count and corresponding status word form a fault status having an address corresponding to a memory chip of the plurality of memory chips.

7. The memory fault mapping apparatus according to claim 6 wherein the status word receives cell kill, line kill, and chip kill indications from said counting means when said counting means reaches predetermined first, second and third thresholds, respectfully.

8. The memory fault mapping apparatus according to claim 1 wherein said counting means comprises:

a first input coupled to an output of said detecting means for receiving the error indication;

a second input coupled to an output of said error memory for receiving the error count from said selected one of said predetermined locations; and

an output coupled to an input of said error memory for providing the incremented error count to said selected one of said predetermined locations.

9. A method of mapping detected errors from a plurality of memory chips, comprising the steps of:

a) resetting a plurality of error counts to in predetermined locations of an error memory zeros indicating no errors detected for a corresponding plurality of memory chips organized in rows and columns;

b) randomly accessing said plurality of memory chips for reading data therefrom;

c) checking data read from the plurality of memory chips for errors during each random access;

d) if a single bit error is detected;

correcting the detected single bit error and providing an indication of the existence of the detected error and a location thereof;

reading a first error count from a predetermined location of the error memory corresponding to a first memory chip of the plurality of memory chips that generated the single bit error;

incrementing the first error count; and

writing the incremented first error count back to the predetermined location of the error memory; and

e) repeating steps b) through d) until an error count of the plurality of error counts reaches a predetermined threshold or another predetermined event occurs.

10. The method according to claim 9 wherein the step of reading a first error count further comprises the step of providing an address to the error memory indicating the column and the row in which the first memory chip generating the error is located.

11. The method according to claim 10 further comprising the step of providing a status word for each error count indicating when a predetermined threshold or other predetermined event has occurred.

12. The method according to claim 9 wherein steps a) through d) occur during a single data access cycle.

13. The method according to claim 9 wherein said steps of reading, incrementing and writing occur in a single counter for all of the plurality of memory chips.

14. A multi-pass memory fault mapping apparatus for monitoring randomly accessed data from a plurality of memory chips arranged in rows and columns and addressed by at least a row select address, the plurality of memory chips logically divided into a plurality of groups and each group logically divided into a plurality of subgroups, said fault mapping apparatus providing a count of errors generated by the plurality of memory chips, comprising:

detecting means coupled to the plurality of memory chips for checking the accessed data from said plurality of memory chips for errors and providing an error indication and an error syndrome if an error is detected in the accessed data;

first pass decoder means coupled to said detecting means for receiving the error syndrome and for providing a group address indicating a first group from which the error was detected during a first pass of fault mapping;

first error memory coupled to said first pass decoder means and further coupled for receiving subgroup address signals such that, if an error is detected, a first predetermined location of said first error memory is accessed, said first predetermined location corresponding to a first subgroup from which the error was detected and having a first error count stored therein;

counting means coupled to said first error memory for receiving the first error count from said first predetermined location, incrementing the first error count, and returning the incremented first error count to said first predetermined location;

second pass decoder means coupled to said detecting means for receiving the error syndrome, said second pass decoder means being activated when any error count in said first error memory reaches a predetermined threshold to determine which memory chip from a subgroup of memory chips has generated a detected error during a second pass of fault mapping; and

second error memory, coupled to said second pass decoder means and to said counting means, for storing error counts for each memory chip in one of the plurality of subgroups error memory further coupled to said counter, for providing a second error count to said counting means for incrementing and for receiving the incremented error count from said counting means.

15. The multi-pass memory fault mapping apparatus according to claim 14 wherein the detecting means is a SEC/DED syndrome generator providing the error syndrome when a single bit error is detected and corrected.

16. The multi-pass memory fault mapping apparatus according to claim 14 wherein the error count is accessed, incremented, and rewritten to said first predetermined location of said first error memory within a single data access cycle.

17. The multi-pass memory fault mapping apparatus according to claim 14 wherein:

the first error memory is logically divided into two arrays, a first array for storing the count of first detected errors generated by each subgroup of said plurality of memories and a second array for storing a status word associated with each of the first detected error counts; and

the second error memory is logically divided into two arrays, a third array for storing the count of second detected errors generated by each memory chip in one of the subgroups and a fourth array for storing a status word associated with each of the second detected error counts.

18. The multi-pass memory fault mapping apparatus according to claim 17 wherein the status word receives cell kill, line kill, and chip kill indications from said counting means when said counting means reaches predetermined first, second and third thresholds, respectfully.

19. The multi-pass memory fault mapping apparatus according to claim 14 wherein the second pass decoder means further comprises:

a decoder for decoding the error syndrome into a plurality of second pass signals, each second pass signal corresponding to one column of memories chips of said plurality of memory chips;

a mask bit register providing one mask bit for each column of memory chips; and

a plurality of And gates each having a first input coupled to receive one of the second pass signals from said decoder and a second input coupled to receive a corresponding one of the mask bits from said mask bit register, said plurality of And gates providing a masked address to said second error memory.

20. The multi-pass memory fault mapping apparatus according to claim 14 wherein said counting means comprises a single counter.

21. A multi-pass memory fault mapping apparatus comprising:

a plurality of memory chips physically organized by cards, rows, and columns and logically organized into a plurality of groups each being further organized into a plurality of subgroups;

detecting means coupled to the plurality of memory chips for checking data accessed from said plurality of memory chips and providing an error indication and an error syndrome if an error is detected in the accessed data;

first pass decoder means coupled to said detecting means for receiving a first enable signal and the error syndrome and for providing a group address a first group from which the error was detected from during a first pass of fault mapping;

first error memory coupled to said first pass decoder means and further coupled for receiving card and row address signals such that if an error is detected, a first predetermined location of said first error memory is accessed, said first predetermined location corresponding a first subgroup from which the error was detected and having a first error count stored therein, said first error memory comprising;

a first memory array for retaining the count of errors generated by each subgroup of the plurality of memory chips; and

a second memory array for retaining a status word associated with each of the error counts; and

counting means coupled to said first error memory for receiving the first error count from said first predetermined location, incrementing the first error count and returning the incremented first error count to said first predetermined location;

second pass decoder means coupled to said detecting means for receiving a second enable signal and the error syndrome, said second pass decoder means being activated when an error count in said first error memory reaches a predetermined threshold to determine which memory chip from one of the plurality of subgroups has generated a detected error during a second pass of fault mapping; and

second error memory coupled to said second pass decoder means and to said counting means for storing error counts, for each memory chip in one of the plurality of subgroups for providing a second error count to said counting means for incrementing and for receiving the incremented error count from said counting means.

22. The multi-pass memory fault mapping apparatus according to claim 21 wherein the detecting means is a SEC/DED syndrome generator providing the error syndrome when a single bit error is detected and corrected.

23. The multi-pass memory fault mapping apparatus according to claim 21 wherein the error count is accessed, incremented, and rewritten to said first predetermined location of said first error memory within a single data access cycle.

24. The multi-pass memory fault mapping apparatus according to claim 23, wherein the status word receives cell kill, line kill, and chip kill indications from said counting means when said counting means reaches predetermined first, second and third thresholds, respectfully.

25. The multi-pass memory fault mapping apparatus according to claim 21 wherein said second pass decoder further comprises:

a decoder for decoding the error syndrome into a plurality of second pass signals, each second pass signal corresponding to one column of memory chips of said plurality of memory chips;

a mask bit register having one mask bit for each column of memory chips; and

a plurality of And gates each having an input coupled to one of the second pas signals from said decoder and a second input coupled to receive a corresponding one of the mask bits from said mask bit register, said plurality of And gates providing a masked address to said second error memory.

26. The multi-pass memory fault mapping apparatus according to claim 21 wherein said counting means comprises a single counter.