Method and apparatus for using memory compression to enhance error correction
View Patent ↗A system for improving a memory's error detecting and error correcting capabilities. During operation, the system receives a data-word. Next, the system compresses the data-word into a compressed-word. If the amount of compression is greater than or equal to a compression-threshold, the system applies a strong error-correcting-code to the compressed-word to generate a coded-word. On the other hand, if the amount of compression is less than the compression-threshold, the system applies a weak error-correcting-code to the data-word to generate a coded-word. In either case the size of the coded-word is less than or equal to the size of a storage-word. The system then generates a flag that indicates the type of error-correcting code that was used to generate the coded-word. The system then stores the flag along with the coded-word in the memory.
1. A method for using data compression to improve error detecting and error correcting capabilities of a memory, wherein the memory supports error detection and error correction, and wherein the memory stores data in storage-words whose size is equal to a storage-word size, the method comprising:
receiving a data-word stored in a storage word;
compressing the data-word to obtain a compressed-word, whereby the data-word is compressed by an amount of compression;
responsive to determining that the amount of compression is greater than or equal to a compression-threshold, applying a strong error-correcting-code to the compressed-word to generate a coded-word in place of the data-word in the storage word, wherein the coded-word's size is less than or equal to the storage-word size;
generating a flag that is associated with the strong error-correcting code; and
storing the flag along with the coded-word in the memory;
thereby improving the overall reliability of the memory without requiring the memory to use storage-words whose size is greater than the storage-word size.
2. The method of claim 1 , further comprising:
receiving the coded-word along with the flag, which indicates the type of error-correcting code that was used to generate the coded-word; and
responsive to determining that the strong error-correcting-code was used to generate the coded-word,
using the strong error-correcting-code to decode the coded-word to obtain the compressed-word; and
decompressing the compressed-word to obtain the data-word.
3. The method of claim 1 , wherein the data-word is 64 bits long, the compression-threshold is equal to 28 bits, and the strong error-correcting-code is used only if the size of the compressed-word is less than or equal to 36 bits.
4. The method of claim 1 , wherein the strong error-correcting-code is a 3-error-correcting (23, 12) Golay code, which uses 23 bits to encode 12 bits of data.
5. The method of claim 1 , wherein applying the strong error-correcting-code to the compressed-word to generate the coded-word involves:
splitting the compressed-word into a plurality of sub-words;
applying the strong error-correcting-code to each sub-word; and
concatenating the coded sub-words to form the coded-word.
6. The method of claim 1 , wherein the flag is generated by applying a flag error-correcting-code to an indicator, which contains one or more bits that indicate the type of error-correcting code used to generate the coded-word.
7. The method of claim 6 , wherein applying the flag error-correcting-code to the indicator involves concatenating multiple copies of the indicator.
8. A computer-readable storage medium storing instructions that when executed by a computer cause the computer to perform a method for using data compression to improve error detecting and error correcting capabilities of a memory, wherein the memory supports error detection and error correction, and wherein the memory stores data in storage-words whose size is equal to a storage-word size, the method comprising:
receiving a data-word stored in a storage word;
compressing the data-word to obtain a compressed-word, whereby the data-word is compressed by an amount of compression;
responsive to determining that the amount of compression is greater than or equal to a compression-threshold, applying a strong error-correcting-code to the compressed-word to generate a coded-word in place of the data-word in the storage word, wherein the coded-word's size is less than or equal to the storage-word size;
generating a flag that is associated with the strong error-correcting code; and
storing the flag along with the coded-word in the memory;
thereby improving the overall reliability of the memory without requiring the memory to use storage-words whose size is greater than the storage-word size.
9. The computer-readable storage medium of claim 8 , the method further comprising:
receiving the coded-word along with the flag, which indicates the type of error-correcting code that was used to generate the coded-word; and
responsive to determining that the strong error-correcting-code was used to generate the coded-word,
using the strong error-correcting-code to decode the coded-word to obtain the compressed-word; and
decompressing the compressed-word to obtain the data-word.
10. The computer-readable storage medium of claim 8 , wherein the data-word is 64 bits long, the compression-threshold is equal to 28 bits, and the strong error-correcting-code is used only if the size of the compressed-word is less than or equal to 36 bits.
11. The computer-readable storage medium of claim 8 , wherein the strong error-correcting-code is a 3-error-correcting (23, 12) Golay code, which uses 23 bits to encode 12 bits of data.
12. The computer-readable storage medium of claim 8 , wherein applying the strong error-correcting-code to the compressed-word to generate the coded-word involves:
splitting the compressed-word into a plurality of sub-words;
applying the strong error-correcting-code to each sub-word; and
concatenating the coded sub-words to form the coded-word.
13. The computer-readable storage medium of claim 8 , wherein the flag is generated by applying a flag error-correcting-code to an indicator, which contains one or more bits that indicate the type of error-correcting code used to generate the coded-word.
14. The computer-readable storage medium of claim 13 , wherein applying the flag error-correcting-code to the indicator involves concatenating multiple copies of the indicator.
15. An apparatus for using data compression to improve error detecting and error correcting capabilities of a memory, wherein the memory supports error detection and error correction, and wherein the memory stores data in storage-words whose size is equal to a storage-word size, the apparatus comprising:
a receiving mechanism configured to receive a data-word stored in a storage word;
a compressing mechanism configured to compress the data-word to obtain a compressed-word, whereby the data-word is compressed by an amount of compression;
a strong-encoding mechanism, wherein responsive to determining that the amount of compression is greater than or equal to a compression-threshold, the strong-encoding mechanism is configured to apply a strong error-correcting-code to the compressed-word to generate a coded-word in place of the data-word in the storage word, wherein the coded-word's size is less than or equal to the storage-word size;
a flag-generating mechanism configured to generate a flag that is associated with the strong error-correcting code; and
a storing mechanism configured to store the flag along with the coded-word in the memory;
wherein the apparatus improves the overall reliability of the memory without requiring the memory to use storage-words whose size is greater than the storage-word size.
16. The apparatus of claim 15 , further comprising:
a second receiving mechanism configured to receive the coded-word along with the flag, which indicates the type of error-correcting code that was used to generate the coded-word; and
a strong-decoding mechanism, wherein responsive to determining that the strong error-correcting-code was used to generate the coded-word, the strong-decoding mechanism is configured to:
use the strong error-correcting-code to decode the coded-word to obtain the compressed-word; and
decompress the compressed-word to obtain the data-word.
17. The apparatus of claim 16 , wherein the data-word is 64 bits long, the compression-threshold is equal to 28 bits, and the strong error-correcting-code is used only if the size of the compressed-word is less than or equal to 36 bits.
18. The apparatus of claim 16 , wherein the strong error-correcting-code is a 3-error-correcting (23, 12) Golay code, which uses 23 bits to encode 12 bits of data.
19. The apparatus of claim 16 , wherein the strong-encoding mechanism is further configured to:
split the compressed-word into a plurality of sub-words;
apply the strong error-correcting-code to each sub-word; and
concatenate the coded sub-words to form the coded-word.
20. The apparatus of claim 16 , wherein the flag-generating mechanism is further configured to apply a flag error-correcting-code to an indicator to generate the flag, wherein the indicator contains one or more bits that indicate the type of error-correcting code used to generate the coded-word.
21. The apparatus of claim 20 , wherein the flag-generating mechanism is further configured to concatenate multiple copies of the indicator, thereby applying a flag error-correcting-code to the indicator.