NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/sys/amiga/programmer/6229

Internet Message Format | 1996-08-05 | 3.6 KB

Path: informatik.tu-muenchen.de!fischerj
From: [e-mail] (Juergen "Rally" Fischer)
Newsgroups: comp.sys.amiga.programmer
Subject: Re: doubling pixels horizontally
Date: 25 Mar 1996 19:48:49 GMT
Organization: Technische Universitaet Muenchen, Germany
Distribution: world
Message-ID: <4j6tb1$[e-mail]>
References: <[e-mail]> <4i6eql$[e-mail]> <[e-mail]> <4ijkal$[e-mail]> <4ivvlu$[e-mail]> <4j65qk$[e-mail]>
NNTP-Posting-Host: hphalle5.informatik.tu-muenchen.de
Originator: [e-mail]


In article <4j65qk$[e-mail]>, [e-mail] (Philipp Boerker) writes:
|> Organization: Technical University Berlin, Germany
|> Lines: 49
|> Message-ID: <4j65qk$[e-mail]>
|> References: <[e-mail]> <4i6eql$[e-mail]> <[e-mail]> <4ijkal$[e-mail]> <4ivvlu$[e-mail]>
|> NNTP-Posting-Host: w350zrz.zrz.tu-berlin.de
|>
|> [e-mail] (Juergen "Rally" Fischer) writes:
|>
|>
|> >: Yes, but if you can't do *everything* with the OS you don't need
|> >: to do *anything* with the OS.
|>
|> >;) I get what you mean, but it's not really true.
|>
|> >The only thing you need to overtake for special fx is the copper.
|>
|> >But, with copper overtaken you still can do the rest with OS and
|> >keep the system happily alive and stable.
|>
|> Yes, but overtaking the blitter should be sure enough. Usually I tend
|> to do as much as possible with the OS.

We must be careful with terminology. I mean takeover as illegal using.
Test for blitter present, call OwnBlitter()/gfx.lib, then
WaitBlitter()/gfx.lib, and you got full legal hardware acess to it.

For Async blitter action use Qblit(), it also gives your function
full acess to blitter, just OS-controlled.
avoid busy-waiting for the flag you change in cleanup-routine to
avoid blitter locking.

full blitter speed with OS is possible for 99% of cases.

|>
|> >: > Also the polygon loops did not
|> >: >look very fast..
|>
|> >: Codewise or performancewise? The code isn't optimized to hell but it
|> >: shows better performance than every other routine until now!
|>
|> >mhm ? imho only optimized code will perform well ;)
|>
|>
|> No, algorithms make the speed, implementations waste it!

aah. ;)
|>
|>
|> >: and I'm convinced that we can push the performance above 20000 polys/second,
|> >: hopefuly enough for a poly-world.
|>
|> >oh yeah, those X00000 poly/second numbers ;)
|> >Fastest I have seen is estimated 20 polygons in 50hz in ARTE (A1201).
|> >would be 1000 polygons/sec.
|>
|> Get a 50MHz 030 with 8MB fast (GJ unfortunately does not run on 4 MB)
why ? what does GJ make with those 8mb ? fill with random numbers ? :)

|> and watch GomJabbar/Matrix. You will see an 3xxx-poly object doing an average
|> of 5 fps! There is also a duck running smoothly in 1*1 (ca. 20 fps).

how much is it on A1200 + fastmem ? ah, forgot, can't test it because
it needs 8mb. well, if it would run 8fps 1x1 it would be good.

|>
|>
|> >you got PPC 620 + AAA or how you get factor 20 faster routines ? ;) ;)
|>
|>
|> It's only factor 2 compared to ZIF/parallax...

ZIF was imho very slow, sorry. At least it was not running ideal on
020-14+fast.

|>
|>
|> Greets,
|> Phil.
|> grond/matrix
------------------------------------------------------------------------
   [e-mail] (Juergen "Rally" Fischer)   =:)