wls@astrovax.UUCP (11/01/83)
Does anyone out there understand the Vax CRC instruction? I have found the desciption in the VAX Architecture manual difficult (i.e. impossible) to understand. It would seem that this instruction could be used for very fast implementations of these CRC calculations. I remember seeing assembler variants of the uucp CRC calculation posted to the net last May or June. Is the CRC calculation in uucp of such a type that this instruction could be used? -- Bill Sebok Princeton Univ. Astrophysics {allegra,akgua,burl,cbosgd,decvax,ihnp4,knpo,princeton}!astrovax!wls
thomas@utah-gr.UUCP (Spencer W. Thomas) (11/02/83)
When we first got our VAX, I rewrote sin and cos to use the POLY instruction. Got maybe 10% out of it. Not worth it. Now, I'm not claiming the same problem for the CRC, but you never can tell. =Spencer
grunwald@uiuccsb.UUCP (11/08/83)
#R:sri-arpa:-1308600:uiuccsb:14900004:000:682 uiuccsb!grunwald Nov 6 17:37:00 1983 re: computing sin and cos using the POLY op-code Wouldn't it be better to compute the sin and cos using the CORDIC method? I would think that this would give you much better results than using the taylor series expansion (which is what I assume you're using the POLY for). Or is there some problem with this (I've just done CORDIC routines for fixed point fractions, not floating point -- so the range arctangents you need to store is much smaller. However, combining the CORDIC with a small-angle approxim- ation would seem to work). Keep those letters and cards coming... Dirk Grunwald University of Illinois USENET : ihnp4 ! uiucdcs ! grunwald CSNET : grunwald.uiuc@Rand-Relay