Decompression accelerator for flash memory

Abstract

A hardware accelerator for improving the decompression performance when decompressing data in Lempel-Ziv-Huffman compressed data format. The use of a Huffman encoding second stage in the popular and widely-used Lempel-Ziv-Huffman standard improves the compression ratio but complicates the decompression, because the Huffman encoding is applied selectively only to certain parts of the Lempel-Ziv tokens, and thus Huffman decoding must also be applied selectively during decompression. The present invention features a variable-length token decoder which is able to selectively decode the Huffman-encoded portions of the compressed data, and therefore enables high-performance decompression for compressed data having a very good compression ratio. Such an accelerator is well-suited for use in data processors which are to be loaded with pre-compressed data and software applications, particularly those employing flash memory.

Description

FIELD OF THE INVENTION [0001] The present invention relates to lossless data compression and decompression, and, more particularly, to optimizing data decompression performance for data which is stored in a flash memory device. BACKGROUND OF THE INVENTION [0002] It is often desirable to compress data in order to reduce the consumption of storage resources and / or transmission overhead. Through the use of lossless compression techniques, it is possible to compress and decompress the data exactly, without any loss of information during the compression / decompression process. [0003] Generally, there is a tradeoff between the compression ratio and the processing performance achieved when performing the compression and decompression. The term “compression ratio” commonly denotes a measure of the effectiveness of the compression, and is widely defined as the percentage of the original data volume that has been eliminated by the compression. The more effective compression is, the higher the ...

AI Technical Summary

Benefits of technology

[0018] The present invention is of a hardware decompression accelerator and an associated method for separately optimizing lossless data decompression performance and lossless data compression ratio for data stored in memory, such as flash memory.

[0019] The present invention also provides a hardware decompression accelerator for efficiently decompressing data that has been compressed using Lempel-Ziv-Huffman lossless compression utilizing Dynamic Huffman encoding, and a method for using such a hardware decompression accelerator to attain a reasonably-high data compression ratio and very high decompression performance for an important class of applications.

[0033] Furthermore, according to the present invention there is provided a method for efficiently storing and retrieving data for use, including: (a) providing a data processor having data storage and having a decompression accelerator; (b) compressing the data according to a lossless data compression format, thereby providing compressed data; (c) writing the compressed data to the processor data storage; and (d) decompressing the compressed data stored in the processor data storage using the processor decompression accelerator.

Problems solved by technology

Generally, there is a tradeoff between the compression ratio and the processing performance achieved when performing the compression and decompression.

Lempel-Ziv alone, however, does not achieve optimum compression ratios.

The selective use of Huffman encoding with Lempel-Ziv improves the compression ratio but complicates the decompression process and lowers the currently-attainable decompression performance.

Static Huffman encoding is usually employed only for short blocks of data, where it would be counterproductive to use Dynamic Huffman encoding, which requires writing information necessary to reconstruct the encoding tables into the compressed data blocks.

The higher the compression ratio, the lower will be the communication overhead, which usually justifies additional compression / decompression processing, because communication costs are always much higher than local processing costs.

It is emphasized that in these implementations, Dynamic Huffman encoding (as described above) is normally utilized, because Static Huffman encoding does not attain as good a compression ratio.

Not all applications, however, can justify increased processing overhead when decompressing compressed data.

Unfortunately, however, the complexity of Lempel-Ziv-Huffman decompression with Dynamic Huffman decoding currently requires a software implementation for practical applications, and software cannot decompress the data at sufficient speed for fast loading of compressed data from flash memory into CPU RAM.

Thus, although a good compression ratio can be attained by using Lempel-Ziv-Huffman compression using Dynamic Huffman encoding, the decompression performance is too low, thereby rendering Lempel-Ziv-Huffman compression using Dynamic Huffman encoding currently unsatisfactory for this and certain other important applications.

This imposes unnecessary limitations on prior art solutions, because in certain applications of data decompression, it may not be necessary for the compression to be performed by the same system or device that does the decompression.

Neither Mayers nor other prior art provides a means of separately optimizing lossless data compression and lossless data decompression in applications where compressed data stored in data storage or memory, such as flash memory, must be efficiently decompressed for loading into executable memory.

Images