Re: crc-32Q

[julian_satran@il.ibm.com]

Sanjay,

Please have a look at

http://www.haifa.il.ibm.com/satran/ips/draft-sheinwald-iSCSI-CRC-00.txt

The reference (a paper by Jack Wolf from 1994) has this polynomial as the
best performing between those explored and 4 orderds of magnitude better on
bursts longer than 32 bits.  Given the stature of the author and the fact
the the polynomial is quoted in several papers we did not check that after
division by x+1 the polynomial is irreducible.

CCITT-CRC32 is x^32+x^31+x^4+1 (our previous choice)

We are looking now at several other codes. IEEE802  hex 104C11DB7 - the one
you quoted is among them but there are several more promising.

Julo

Sanjay Goyal <sanjay_goyal@ivivity.com> on 02/03/2001 03:11:55

Please respond to Sanjay Goyal <sanjay_goyal@ivivity.com>

To:   ips@ece.cmu.edu
cc:
Subject:  crc-32Q




Hi
 the generator polynomial for the crc-32Q (32 31 24 22 16 14 8 7 5 3 1 0)
as
given in Appendix A is reducable, as I tried it on the internet on a
CRC-calculator tool.
The ethernet polynomial which is
  (32 26 23 22 16 12 11 10 8 7 5 4 2 1 0)   is not reducable.
 Now as per my understanding reducable polynomial is not better than the
irreducable one.
 If reducable one is not good, why did we choose crc-32Q polynomial instead
of crc-32 polynomial(used for ethernet)?


Can somebody please explain it to me.


Regards and Thanks
Sanjay Goyal