US 5,459,742 AGranted

Solid state disk memory using storage devices with defects

AssigneeQuantum Corporation

InventorsCharles Cassidy, Paul Kemp, Donald Smelser

Life of this patent

Estimating the term…

Loading patent…

Abstract

A computer system includes a main memory that is able to make use of DRAM memory devices having a relatively high level of bad cells (hard faults). An EDC circuit is provided which uses combinatorial logic to perform a BCH code type of error detection and correction. A primary feature is the recognition that due to use of high density integrated circuits--gate arrays--it is no longer necessary to use sequential logic to decode the multiple-bit error correcting codes. An EDC with 128-bits of data and a check bit field 41-bits wide, using a BCH code, constructed in ASIC sea-of-gates technology using about 87,000 logic gates, can correct 5-bits in error and can detect 6-bits in error. By using multiple-bit EDC in the controller for main memory, it is no longer necessary that all DRAM devices be ostensibly "perfect." A certain density of non-functional memory cells can be tolerated, yet the memory system will still return perfect data. The added cost of multiple-bit EDC, including the added cost of extra storage for the check bits and the EDC circuit itself, is more than compensated by reduced cost of the DRAMs. In a addition the computer system includes a solid-state disk type memory for a computer system is able to make use of DRAM memory devices having a relatively high level of bad cells (hard faults). An EDC circuit is employed to perform a Reed-Solomon code type of error detection and correction. A primary feature is the recognition that it is no longer necessary that all DRAM devices be ostensibly "perfect." A certain density of non-functional memory cells can be tolerated, yet the memory system will still return perfect data. The added cost of multiple-bit EDC, including the added cost of extra storage for the check bits and the EDC circuit itself, is more than compensated by reduced cost of the DRAMs. A preferred data formatter circuit to convert between symbol and word data is also described.

Claims5

Dependent claims

A computer system comprising:

a central processing unit (CPU) generating read and write requests;
a system bus comprised of data and address lines coupled to said CPU and receiving read and write requests from said CPU;
a main memory for storing data for use by said CPU;
a solid state disk memory apparatus comprising:
(a) a plurality of random access memory devices each of said devices arranged in a regular array of memory cells with some of said memory cells of at least some of said arrays known to have defects, with said memory devices coupled to address, data, and control lines to receive read and write requests to said memory apparatus;
(b) controller means for coupling said address and data lines to a bus for conveying read and write data between said bus and said memory devices, said controller means comprising:
(i) converter means for converting a first plurality of M-bit data words to a first plurality of N-bit data symbols;
(ii) an error correction code generator circuit to provide a first plurality of N-bit error correction code check symbols in response to said plurality of data symbols for each write request to said memory apparatus;
(iii) second converter means for converting said first plurality of N-bit data symbols and first plurality of check symbols to a first plurality of K-bit data words for storage in said array of memory cells;
(iv) third converter means for converting a second plurality of K-bit data words read from said memory array to a second plurality of N-bit data symbols and second plurality of associated check symbols;
(v) an error detection and correction circuit, fed by said second plurality of data symbols and said second plurality of associated check bit symbols, for detecting and correcting up to N symbol errors in said plurality of read data and check symbols where N is an integer greater than 1;
and
(vi) fourth converter means for converting said second plurality of N-bit data symbols to a second plurality of M-bit data words read from said array of memory cells,
wherein the memory cells of each of said arrays of memory cells known to have defects are not excluded from an address of said CPU by mapping read requests or write requests around said memory cells having said known defects.

11 more elements

A solid state disk memory apparatus for connection via a bus structure to a central processing unit (CPU) generating read and write requests, the solid state disk memory apparatus comprising:

a plurality of random access memory circuits each of said circuit arranged in a regular array of memory cells with some of said memory cells of at least some of said arrays known to have defects, with said memory circuits coupled to address, data and control lines of the bus structure to receive read and write requests from the CPU upon connection to said memory apparatus, the memory cells of each of said arrays of memory cells known to have defects not being excluded from an address by mapping read requests or write requests around said memory cells having said known defects,
controller means for coupling said address and data lines to an internal bus for conveying read and write data between said bus and said memory devices, said controller means comprising:
(i) converter means for converting a first plurality of M-bit data words to a first plurality of N-bit data symbols;
(ii) an error correction code generator to provide a first plurality of N-bit error correction code check symbols in response to said first plurality of N-bit data symbols for each write request to said memory apparatus;
(iii) second converter means for converting said first plurality of N-bit data symbols and check symbols to a second plurality of K-bit data words for storage in said regular array of memory cells;
(iv) third converter means for converting a plurality of K-bit data words read from said regular memory array to a second plurality of N-bit data symbols and a second plurality of associated check symbols;
(v) an error detection and correction circuit, fed by said second plurality of data symbols and said second plurality of associated check symbols, that can correct up to n symbol errors in said plurality of read data and associated check symbols, where n is an integer greater than 1;
and
(vi) fourth converter means for converting said second plurality of N-bit data symbols to a second plurality of M-bit data words read from said array of memory cells.

6 more elements

Dependencies are inferred from the claim text.

Facts

Application: 08/212,334Filed 14 Mar 1994 · continuation
Priority: 11 Jun 1992
Granted: 17 Oct 1995
Term ends: Not estimatedThe source records don’t support an estimate.

Classification6 codes

G06F 11/1044
In individual solid state devices: with specific ECC/EDC distribution
H03M 13/152
Using error location or error correction polynomials: bose-Chaudhuri-Hocquenghem [BCH] codes
H03M 13/1575
Using error location or error correction polynomials: direct decoding, e.g. by a direct determination of the error locator polynomial from…
H03M 13/6575
Purpose and implementation aspects: implementations based on combinatorial logic, e.g. Boolean circuits