True true... I found my bug (as there always is one...) in a factor-handling fumble of proportions. My algo worked and was extreemely fast (O:n) but I found it only worked for a very small subset of input data. (1 of every 256 LOL) A possibility is to use a external storage for the table. It might still be faster for very large multiplications. KreAture ----- Original Message ----- From: "Spehro Pefhany" To: Sent: Sunday, January 19, 2003 2:26 AM Subject: Re: [PIC]: Fastest possible 512 bit multiply ??? > At 02:10 AM 1/19/03 +0100, you wrote: > >Only practical table would be 8x8 bits. > > Errr... no PIC has internal memory big enough for that 128K byte table and > anything else, so it would only be true for very small values of "practical". > > It does illustrate that brute-force LUT approaches don't work that well to > evaluate dyadic operations. > > Here's a book that looks pretty good for more sophisticated algorithms. > (kickback link). They have a free simulator so you can play with the > algorithms. > > http://www.amazon.com/exec/obidos/ASIN/1568811608/speffcom-20 > > For multiplication, there are these algorithms: > > Sequential > Booth's Algorithm > Modified Booth's Algorithm > Two's Complement Array Multiplier > Fused Multiplier-Adder > > Of course, if the simple algorithm is fast enough, no need to fool around, > that's how Intel got into trouble with the Pentium divide bug, by not > testing a very small percentage of possible operands with a sophisticated > algorithm. > > Best regards, > > Spehro Pefhany --"it's the network..." "The Journey is the reward" > speff@interlog.com Info for manufacturers: http://www.trexon.com > Embedded software/hardware/analog Info for designers: http://www.speff.com > > -- > 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 > -- 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