home *** CD-ROM | disk | FTP | other *** search
- Path: nntp.teleport.com!sschaem
- From: sschaem@teleport.com (Stephan Schaem)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: PPC compilers
- Date: 2 Feb 1996 14:24:59 GMT
- Organization: Teleport - Portland's Public Access (503) 220-1016
- Message-ID: <4et6rr$o2l@maureen.teleport.com>
- References: <38232191@kone.fipnet.fi> <551.6602T1231T2646@ifi.uio.no> <38232238@kone.fipnet.fi>
- NNTP-Posting-Host: linda.teleport.com
- X-Newsreader: TIN [version 1.2 PL2]
-
- Jyrki Saarinen (jsaarinen@kone.fipnet.fi) wrote:
-
- : > Sorry, I have not followed this discusion closely.
- : >
- : > What problems will it cause to have 2 addx? Is there anything wrong with
- : > just exchanging the fraction parts like we are doing today?
-
- : For a normal loop with one addx you set the x-flag correctly like
- : this:
-
- : moveq #0,d0
- : add.l d1,d0
- : ...
- : .loop addx.l d1,d2
- : dbf d7,.loop
-
- : But, what we do if we have two addx (16.16 for V too) and still
- : want 100% accuracy.
-
- and again.. what is do for d0 for?
- just sub.l d0,d0 it will do the same as the moveq/add.
- But you might want to do it correctly VS just clearing X.
-
- move.l d2,d0
- sub.w d1,d0
- add.l d1,d0
-
- should be the right setup, with a loop count -1 and
- an addx.w d1,d0 for the last iteration.
-
- : You just cleared the x-flag before the loop, is that really
- : correct?
-
- Before the loop you need to set the x flag by adding the fractional
- part for X int or Y int or Z etc.... and at the end of the loop
- (count -1) you just add the int part.
-
- Check some back post , its more detailed there.
-
- You can actually use 3 addx if you wanted... it require the same setup.
-
- Stephan
-