home *** CD-ROM | disk | FTP | other *** search
- Path: w353zrz.zrz.TU-Berlin.DE!rawneiha
- From: rawneiha@w353zrz.zrz.TU-Berlin.DE (Philipp Boerker)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: TMapping again!
- Date: 17 Jan 1996 15:44:17 GMT
- Organization: Technical University of Berlin, Germany
- Message-ID: <4dj5gh$q5k@news.cs.tu-berlin.de>
- References: <4d0ou6$835@astfgl.idb.hist.no> <Z31Wx*zA0@mkmk.in-chemnitz.de> <4d42di$9e9@maureen.teleport.com> <4d5lvi$emc@brachio.zrz.TU-Berlin.DE> <4d6v0t$3dt@maureen.teleport.com> <4dg4jk$km@news.cs.tu-berlin.de> <4dhvd5$5r2@maureen.teleport.com>
- NNTP-Posting-Host: w353zrz.zrz.tu-berlin.de
- Mime-Version: 1.0
- Content-Type: text/plain; charset=iso-8859-1
- Content-Transfer-Encoding: 8bit
-
- sschaem@teleport.com (Stephan Schaem) writes:
-
- >Philipp Boerker (rawneiha@w255zrz.zrz.TU-Berlin.DE) wrote:
-
- >: Yes, I knew that one, but I never tried it because I doubt,
- >: that it will have the accuracy needed. Or can you correct me?
-
- > Acuracy of 6bit for Y. 1/64 is not that bad, it
- > basicly mean you will repeat 1 pixel from the source upto 64time.
- > 64x64 pixels are pretty huge, and for a 32x32 texture map 1/64
- > should be enought for reduction.
-
- > I never tryed myself,I use 256x256 tmap. and 16/16bit fixed point
-
- And so do we... The speed is ok too.
-
-
- >: >: Gee, how am I waiting for the 32 regs of the PPC... 8)
-
- >: The PPC is fast enough to do some more ops between mem accesses
- >: than 68k...
-
- > Then you can do operation in between read if the dest register is not
- > used by the folowing inst?
-
- Yes, of course!! Optimizing code for PPC means mainly to optimize
- mem accesses...
-
-
- >: The PPC isn't as RISC as most people fear. It's true
- >: that there are only four address modes but they are as all
- >: PPC instructions very powerful:
-
- >: [...]
-
- >: Are you missing something?
-
- > scaling of the index for any work on array that are not bytes...
-
- There are very powerful shift and rotation ops on PPC. If you want
- scaling you do something like (pseudo asm)
-
- lsl.l rsource2,rx,#2
- move.l rdest,(rsource1,rsource2)
-
- (remember RISC: operation dest,source1,source2)
-
- This takes 2 cycles execution time.
- Since mmu and int-unit work independently you can optimize that
- down to 1 cycle by inserting some ops in between...
- Is that ok for you ? ;-)
-
-
- > For non time critical part I like/use all the 680x0 adr mode.
-
- I never do. Not for speed reasons. I don't even know them...
- For non time critical parts on PPC you should use C.
-
- > pretty much what a 68000 is capable of, should be fun with a good
- > macro assembler to work on that chip.
-
- Anything should be fun on a chip (133 Mhz 604) that is faster than the
- not-yet-existing Pentium Pro 200Mhz...
-
- Greets,
- Phil.
- ----------------------------
- grond/matrix
- rawneiha@sp.zrz.tu-berlin.de
- ----------------------------
-
-