Help for lpaq8 !

[frankdi]

I'm a graduate student in China. In recent days ,I'm studying the lpaq8 algorithm. My tutor asked me to implement the lpaq8 data compressor on FPGA boards. The destination is to get a rapid data compressor on FPGA , and the speed should achieve 20~30MB/s.
But now i'm even not very understand the lpaq8 algorithm .and i can't find much useful information through Google . So i hope find some useful information in the forum.
Please help me . Thank you!

[m^3]

Repeat your request without shouting at us and you'll receive your help.

[Matt Mahoney]

The speed of large memory algorithms like lpaq8 depends on random memory access times, so a hardware implementation will probably not help you much. You can pipeline compression (bit prediction, arithmetic coding, model update) but not decompression because there is a serial dependency between decoding a bit and updating the model.

lpaq8 is obsolete, but very similar to zpaq -method 4. You might find description of zpaq compression algorithm here useful.
Archive specification: http://mattmahoney.net/dc/zpaq204.pdf
Compression algorithm: http://mattmahoney.net/dc/zpaq_compression.pdf
Context mixing algorithms in general: http://mattmahoney.net/dc/dce.html#Section_43

lpaq8 uses a fixed model, not a self describing algorithm like zpaq. However an interesting project might be to implement the ZPAQL virtual machine in hardware. The current implementation either translates to x86 (32 or 64 bit) or interprets the byte code on other hardware.

[frankdi]

Thank you for you remind. And I'll do as you said next time...

[frankdi]

I'm very grateful for your suggestions. And I'll earnestly read the three documents. And my tutor said maybe one day we will go to Florida to visit you.....

[frankdi]

I'm very grateful for your suggestions. And I'll earnestly read the three documents. And my tutor said maybe one day we will go to Florida to visit you.....