home *** CD-ROM | disk | FTP | other *** search
- Path: hydra.zrz.TU-Berlin.DE!rawneiha
- From: rawneiha@hydra.zrz.TU-Berlin.DE (Philipp Boerker)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: doubling pixels horizontally
- Date: 18 Mar 1996 12:18:29 GMT
- Organization: Technical University Berlin, Germany
- Message-ID: <4ijkal$3vu@brachio.zrz.TU-Berlin.DE>
- References: <576.6613T1070T1730@login.eunet.no> <4i6eql$qpo@brachio.zrz.TU-Berlin.DE> <38233047@kone.fipnet.fi>
- NNTP-Posting-Host: hydra.zrz.tu-berlin.de
-
- "Jyrki Saarinen" <jsaarinen@kone.fipnet.fi> writes:
-
-
- >> >> Yes, I did my own interrupthandler.
- >>
- >> >Is this a system interrupthandler??
- >>
- >> No, it is impossible to get in charge of the blitterint without some
- >> hardware banging.
-
- >How about QBlit()? It is the same you would do with HW blitter
- >interrupt.
-
- Yes, but if you can't do *everything* with the OS you don't need
- to do *anything* with the OS.
-
- >Btw, why Gomjabbar had very slow c2p routines, using BYTE
- >writes to CHIP-ram etc..
-
- No, it does not do byte writes to chip mem! But the c2p is slow, it is
- the usual cpu-only c2p from aminet. They were to busy to take mine into
- the demo, the demo was finished just some hours before the release...
- BTW, how did you like it?
-
- > 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!
-
- > Divs for calculating slopes etc.
-
- Yeah, we haven't optimized like we could have: c2p, divs and we even rotate/
- project every point even if it is removed because of not being visible...
- I'm not responsible for that, I was busy with uni at the time... :|
- And please keep in mind that Sharon doesn't even *own* an amiga or any other
- 68k machine!
-
- >I think the best way for slope calculation is a 1/x table
- >with 15 bit fraction, so that muls.w (28 cycles max.) can
- >be used, then add.l dn,dn on the result to get 16.16. Faster
- >than divs.w and more accurate, it is impossible to use
- >16.16 and divs.w because of the potential overflows..
-
- We hope to make a demo until asm96 but I'm busy with uni until end of April,
- Sharon is going to do a PCdemo for asm96. We will do the usual optimizations
- and I'm convinced that we can push the performance above 20000 polys/second,
- hopefuly enough for a poly-world.
-
-
- Greets,
- Phil.
-