On Sun, 19 Jan 2003, Kyrre Aalerud wrote:

*>Only practical table would be 8x8 bits.
*>But a 8x8 I can do in very short time...
*>Problem is a 64x64 byte (512 bit) calculation...  Even with a mul command
*>I'd need 64*64 multiplications! (Worst case...)

With 4x4 bits you need 128*128 m. worst case but each m. is a vector
mapping (lookup) operation that takes about 6 T cycles. Ripple adding 64
8-bit numbers aligned to byte and nybble boundaries should cost less than
132T cycles (each) on the average. A PIC with T=1usec should do this
operation in under 13 seconds. These calculations are probably wrong and
unchecked, it is 3AM here.

Peter

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics