home *** CD-ROM | disk | FTP | other *** search
- Path: news.eunet.fi!fipnet!kone!jsaarinen
- Newsgroups: comp.sys.amiga.programmer
- X-NewsReader: IntuiNews 1.2b (31.7.94)
- References: <38231816@kone.fipnet.fi> <judas.0h8u@tomtec.abg.sub.org>
- From: "Jyrki Saarinen" <jsaarinen@kone.fipnet.fi>
- Date: Tue, 9 Jan 96 16:00:35 UT
- Comments: Illegal date header - new date added by quicknews
- X-Original-Date: Tue, 09 Jan 96 09:36:44
- MIME-Version: 1.0
- Content-Type: text/plain; charset=iso-8859-1
- Content-Transfer-Encoding: binary
- Distribution: world
- Subject: Re: >>>>>>> The Future: Amiga goes PC (...aeh: "PowerMsPenti...?) <<<<<
- Message-ID: <38231946@kone.fipnet.fi>
-
-
- > >> 4MB/sec are not possible with 256 colors. On doublemodes, certain
- > >> VGA-users need, you can expect less than 2MB from the 32bit-chipset
- > >
- > >Bullshit.
- >
- > You never tried to play anims on a DBLPAL screen with deepth 8 ???
- > The serial port also collapses, when running more than 9600 Baud on
- > such a screen, due to the heavy dma access.
- >
- > I have an A4000 and an A1200, but none of them went faster than 2MB.
- > (Just move a window on a WB with such a resolution, this sucks .. )
-
- That is not same as bandwidth. Make a test program that moves
- memory from fast to chip and test it in DblPAL: Lores 256c.
-
- > >Slower CPUs can do that 20ms also, just less passes depending
- > >on the CPU speed. The rest c2p passes are done with the blitter.
- >
- > Well, I just wonder, why none of the doomengines really gives a
- > smooth performance. Is an A4000/040 or A1200/030 too slow ???
-
- A4000/040 has a crap memory system, and tmapping is largely
- memory moving. Chunky2planar is only small part of time
- consumed in these engines..
-
- A4000/040 should be able to do full 320x256x256 DOOM about
- 15fps according to my calculations.
-
- > >Because it does still not have a 040/060-optimized c2p!
- > >Only 030 optimized, which is 2x slower on 040!
- >
- > No, it`s not the 030 code. On the 040 or 060, there are more fps.
- > But as fastram benefits directly from a faster cpu and chipram does
- > not, the percentage you spend accessing chipram increases.
-
- It is the chunky2planar routine that is done for 020/030 in
- TextDemo57 that slows down on 040/060. On my 040/40 TextDemo57
- spends about 50% doing c2p, do you think this is reasonable?
- With a proper 040 routine it would be 25% perhaps.
-
- -- _
- a Stellar programmer _ //
- "Amiga - back for the future" \X/
-