Path: utzoo!attcan!uunet!lll-winken!lll-tis!ames!necntc!ima!cfisun!lakart!dg
From: dg@lakart.UUCP (David Goodenough)
Newsgroups: comp.protocols.misc
Subject: Table driven CRC evaluation
Keywords: CRC
Message-ID: <238@lakart.UUCP>
Date: 7 Sep 88 21:37:49 GMT
Reply-To: dg@lakart.UUCP (David Goodenough)
Followup-To: poster
Organization: Lake - The systems people
Lines: 20

I have recently implemented an XMODEM file transfer program on my CP/M machine
at home, and since I evaluate the CRC "on the fly" i.e. transmit a character
then update the CRC, I found that even at 9600 BPS I could keep up doing
in the "slow" way i.e. bit by bit (it takes a hair over 100 Us to update
ONE BYTE of CRC on a 4mhz z80, so I spend about 90% of my time idle. However
back to the question. I have seen table driven CRC software, that does each
byte in no time flat by looking in a table. However there is one major
difference: In my software I have to "finish" the crc evaluation by pushing
a pair of zero bytes into it, whereas the table version doesn't need this.
How does the table get away without the extra pair of bytes: because both
systems get it right for any number of bytes, 128 and 1024 being the
obvious pair, as an aside I've tried it with many other byte counts.

E-mail responses, as I don't usually read this newsgroup.

		Thanks in advance,
-- 
	dg@lakart.UUCP - David Goodenough		+---+
							| +-+-+
	....... !harvard!cca!lakart!dg			+-+-+ |
						  	  +---+