英语论文网

We observed that our approach can operate at a much higher frequency and occupies only 60% area compared to original bitmask-based decompression engine. Since our approach has the identical bitmask decoding circuit of the original one, the improvement is due to our ABLSA as we expected. Compared with previous techniques, our approach achieves almost the same operating speed with less area and also achieved 15%-20% better compression which means we can decompress more configuration information during the same amount of time.

仿真结果与讨论-6. SIMULATION RESULTS AND DISCUSSIONS

6.1 SIMULATION RESULT OF DICTIONARY BASED COMPRESSION:

The compacted scan chain network, which is mentioned in this section, reduces the scan depth of the scan chain, and reduces the test cost. Then combine the dictionary-based compression scheme with the compacted scan chain network to achieve the high compression ratio and the fast application time.

6.2 SIMULATION RESULT OF DICTIONARY BASED DECOMPRESSION:

Here the bit stream which has been compressed is converted to its original size by subjecting it to the decompression engine. The simulation results shown here prove that the original data is retrieved from the compressed data.

6.3 SIMULATION RESULT OF BITMASK COMPRESSION AND DECOMPRESSION:

This have been developed using an efficient test data compression algorithm using bitmasks. A compressed code stores information regarding the mask type, mask location and the mask pattern itself. The mask can be applied on different places on a vector and th number of bits required for indicating the position varies depending on the mask type.

6.4 SIMULATION RESULT OF RLE BASED COMPRESSION:

A variation of Run-Length encoding perfectly meets the requirements for the address compression. A series of addresses with a common offset can be compressed into a codeword of the form base, offset and length.

6.5 SIMULATION RESULT OF RLE BASED DECOMPRESSION:

This strategy uses a more intensive decompression algorithm. It attempts to reorder the address data pairs in more optimal manner. Since the addresses must be decompressed and sent in addition to the data, our decompression strategies use both the address and data buses to send the decompressed code words.

结论-7.CONCLUSION:

The existing compression algorithms either provide good compression with slow decompression or fast decompression at the cost of compression efficiency. In this paper, we proposed a decoding-aware compression technique that tries to obtain both best possible compression and fast decompression performance.We also exploit run length encoding of consecutive repetitive patterns efficiently combined with bitmask-based compression to further improve both compression ratio and decompression efficiency. Our experimental results demonstrated that our technique improves the compression ratio by 10% to 15% while the decompression engine is capable of retrieving the compressed data without any loss. The configuration time is reduced by 15% to 20% compared to the best known decompression accelerator. In the future, we plan to investigate more placement algorithms that are compatible with other compression techniques such as Huffman coding, Goulomb coding and Arithmetic coding.