NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/sys/amiga/misc-part2/13534

Internet Message Format | 1996-08-05 | 5.2 KB

Path: news.ios.com!usenet
From: [e-mail] (Pacarana)
Newsgroups: comp.sys.amiga.misc
Subject: Re: Future Amigas
Date: 21 Apr 1996 00:26:20 GMT
Organization: Internet Online Services
Message-ID: <[e-mail]>
References: <[e-mail]> <[e-mail]>
    <[e-mail]> <[e-mail]>
NNTP-Posting-Host: ppp-33.ts-1.hck.idt.net
X-Newsreader: THOR 2.22 (Amiga;TCP/IP) *UNREGISTERED*

>The newer HP-PA chips have one bad problem that I am sure some people
>would frown upon: their off-chip caches MUST run at FULL CPU speed (200
>MHz or higher) using fast SRAM, which is expensive as all hell.  Such a
>system may not be very cost:perofmance effective.
 Well the current ones still seemto wallop the PPC in floating point and
nearly match the very top of the line PPC in integer performance and
are supposed to be cheap enough to stick into a game console. If
what you say is true for all future PA- RISC, then perhaps after some time
they would have become less cost effective and put us in a fix, although with
the horror stories about the sluggishness of the PPC620 I don't know (then
again they may have been disinformation).

>: >> then PA-RISC or PPC plus a new more powerful Amiga
>: >>custom chipset,
>: >We didn't say this either. Where do you get your claims?
>:   But a few of the official AT releases posted here and elsewhere mentioned
>: development of a new and more powerful custom chipset.

>Custom chipset doesn't have to mean "AT developed custom chipset".  It
>could be anything that AT likes and is willing to slap on a motherboard
>and then write a driver for.
  Well I guess so, but it sure didn't sounds like it. I don't know a
single person on Fido who took it that way. Poorly or sneakily phrased if
that's the case.

>Remember that the next generation chipset could be made of both off the
>shelf chips and in-house custom chips.  If we look at AAA, the grfx chips
>were a lot faster than the AGA ones, but alas, it would take a lot of
>optimized and custom code to get graphics to run at a good rate.  In a
>chip like S3's 968, what it lacks in fancy hardware (vs. AAA) it just
>makes up in raw speed.  However, if we look at the AAA version of Paula
>(the Mary chip), it is comparable to many current PC floppy, sound, and
>I/O chips, and has a few things that beat them out - it has 8 channel
>stereo 16-bit percision sample playback with rates up to 100KHz, support
>for 880KB, 1.76MB, and 3.52MB floppy disks [compatable with trackloaders,
>unlike Generic PC floppy controllers], and support for two high speed
>buffered serial ports, ect...  not too bad.  The only problem is that the
>Mary just happened to have the most bugs of all of the chips in the AAA
>chipset.
 Hmm. Ironically, I actually tend to go the other way. Fast large floppies
sound neat, but with 4x CD-ROMS and ZIP drives, etc. I don't know how much of
big deal that would be anymore. As for sound, a nice tooled for audio DSP
seems better. Say 32 voices with all sorts of programmable realtime effects,
and possible more general uses as well, still AAA's sound wasn't bad at all.
What MB/s transfer rate can the S3 blitter handle when performing moderate
effects (and what special effects does it have?)? Does it have mutiple
playfields with fade, 3D support with a million polys a second or 700,000
with texture mapping and MIP mapping etc., copper, 2D rotation and scaling?
I'd hope it a least has one or two of those features, anyway it would need a
truly remarkable bandwidth to make up for most of that, and nothing could make
up for the lack of some of those features. How well with all of this stuff
integrate? The A1000 had a many channel DMA system so everything, audio,
video, blitter, drives, sprites, etc. could run at once with no interference
(later on there was some overlap and stuff like non-CPU disk decoding became
too slow with the lack of chipset updates but an update could do away with
that again).










>: Many months ago when some people on Fido were afraid that the chipst
>: would be dropped, someone posted that you had said that you thought very
>: highly of custom Amiga chipsets and that that was the way to go even
>: though some others at AT didn't seem to think so. Maybe they
>: misinterpreted what you had said, or got the info from someone else, who
>: got it from someone else and meanings changed along the way. Anyway,
>: that's were I got this.

>Custom chipsets *CAN* be very costly to make, especially if you make them
>to be compatable wholly/partially with an older chipset (much like how AGA
>was to OCS/ECS, and how in many ways AAA was to AGA).  If AT were to make
>a new Grfx chipset from the ground up with people who are experienced at
>doing such a task, it wouldn't be quite as bad.  But why do that, when you
>can get a very good custom chip already designed from another company?
>Remember that the chip world has advanced a lot in the past 10 years while
>Commodore had its thumbs up its ass.

>                       Greg Baldwin ([e-mail])
>     Amiga junkie and user since 1987    Computer Science & DTV Student
>     Commodore64 fan since about 1983    http://www.eskimo.com/~drizzit
>               Tyranical EFNet #Amiga Channel Operator "Drizzit"