Xref: utzoo comp.sys.amiga.advocacy:1488 comp.sys.amiga.hardware:7819 Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!snorkelwacker.mit.edu!ira.uka.de!smurf!flatlin!tpki!kris From: kris@tpki.toppoint.de (Kristian Koehntopp) Newsgroups: comp.sys.amiga.advocacy,comp.sys.amiga.hardware Subject: Re: Amiga OS *IS* state of the art Keywords: Amiga OS Message-ID: <3075@tpki.toppoint.de> Date: 1 Apr 91 15:11:36 GMT References: <1003@cbmger.UUCP> <7827@jhunix.HCF.JHU.EDU> <8806@gollum.twg.com> <3058@tpki.toppoint.de> Followup-To: comp.sys.amiga.advocacy, comp.sys.amiga.hardware Organization: Toppoint Mailbox e.V., Kiel, BRD Lines: 60 [ Please feel free to adjust the Followup-To:-line as appropiate ] Well, now I even following up to my own articles. I had a talk with a friend (He is an A3000 owner) on Amiga and its future. Being more the suit type than me, he said "What is going to hurt future Amiga sales more than a lack of 1280 * 1024 in true color is the lack of 320 * 200 in 256 colors." The availability of this video graphics mode makes it easy to port VGA games to the amiga without having the entire graphics repainted by an artist. In his opinion we are going to see less ports of games from IBM to Amiga, if Amiga graphics does not improve in this area. Has anyone got an idea, what the average commodore customer looks like? Is it the 15-25 year old "grown up quality-games player", if such species does indeed exist? I consider his argument quite striking, since 320 * 200 in 256 colors can be done relatively easy with only minor redesigns of the current custom chips, as far as I know. My experience in VLSI design is very limited and only theoretical, but I suggest the following (I have already tried to express this in another article "Why no 320 * 200 * 8 ?"): This is what I know: The three big Amiga custom chips are in fact one single big VLSI IC, that had to be split up due to size limitations of silicium chips. These three chips communicate via the register bus, an 8 addressbits and 16 databits wide specialized bus. This bus is the reason, why the copper can only write to custom chip locations and not into normal memory: The copper can only generate addresses on that register bus. Addresses of 8 bits, each 16 bits wide, give you 512 bytes of custom chips area. From the RKM, Hardware reference manual, I know that in low resolution mode there are 2 unused time slots per pixel. If they were used, one had 8 bitplanes available per pixel, loading the bus as if it were in hi-res/4 bitplane mode. I guess, hardware for this additional dma channels is missing. To make 320 * 200 * 8 usable one had to do at least the following: - add 2 additional dma channels (same functionality as the current 6 video dma channels) - add 2 more address lines to the internal register bus - add 256 32 bit wide color registers covering the address range from register $200 to register $400 (the back 512 register adresses) - redirect the current 32 16 bit wide color registers to the appropiate bits in the new color register bank for compatibility - fiddle with the copper instruction format and the sprite and dual playfield hardware to make it consistent. - replace the current 12 bit video d/a converters with 24 bit converters Well, rereading the list, I come to the conclusion that might not be as easy as I considered it at first thought. Perhaps one of you techies can shade some light on the particular problems one should exspect doing such an chipset upgrade. I really want to know, if it could be done! Kristian Kristian Koehntopp, Harmsstrasse 98, 2300 Kiel, +49 431 676689 Jeder Mann kann eine Frau dorthin bringen, wo sie ihn haben moechte. -- kruemel@citymail.zer.sub.org