AMIGA DOOM... HERE IT IS!
Andrew Robson
> Umm, yeah, users of Intel processors (just now giving up on that 8088
> backwards compatibility) and MS-Dog (oh, but now they've got
> MULTITASKING, only 8 years after the Amiga...) can easily make fun of
> Amiga people for hanging onto the past :-)
>
> Still, I'm not sure that ECS graphics are quite the throwback you
> think they are as far as gameplay. Seems to me that the 1200's slow
> '020 holds us back at least as much as ECS (not to mention the lack of
> an MMU to get real memory protection in the OS). Anyway, it's not as
> if PC games are generally running in Super-VGA 32768 color modes...
> 256 colors are lovely, but they're also more work to produce.
What has work to do with it? btw Marathon on the Mac is in 32768 and very nice
it is too... Adding mem protection to AmigaOS is very non-trivial - due to
the way message ports work on AmigaOS... plus I've got a '1200 and I'm not
using the stock processor...
>
> And besides, the only way to get AGA, in the relatively narrow time
> window before the bankruptcy was to downgrade to a 1200, or make the
> pricey leap to a A4000 and throw away most of your existing hardware
> and software. Standards do not become widespread just because time
> passes, you actually need to sell some machines... I'm curious what
> the actual percentages are, but I'll bet the majority of Amiga users
> still have ECS systems.
>
> -- Jim
An A4000 means you have to throw away existing soft- and hardware? How so?
In man cases 3.0/3.1 are more compatible than 2.04. The 4000 has Zorro slots,
the same as any other big-box Amiga. If thats what you have, get a GFX-card.
If it's a 500, then face it, its not gonna handle much in the way of Doom games
anyway ... (discounting the Accelerated 500 market - which is even smaller than
the AGA market, I'd guess) if it's a 600 then you have my sympathy - at least
the 500 was upgradable (CPU-wise...)
BTW the 1200 wasn't a downgrade for me or many other users, since I only had a
'500 at the time (albeit with extra RAM, HD, etc.)
Sure the majority have OCS/ECS (I have an OCS '500 here collecting dust) but
how many of them are in use? How many of those owners have an accelerated
mahcine that can handle a decent 'doom' clone? Most OCS/ECS machiens will
still be 6...@7.14MHz ...
More to the point, the 4000 and 1200 will soon be actualyl on sale in teh shops
- which is more than I can say for any OCS/ECS system
AGA/GFX-Card software is the way forward. If the new (well, rereleased)
machiens have any hope at all to survive, the need games that at least LOOK
good compared to the PC stuff - which is hard enough anyway, having seen some
recent PC stuff... eitehr way, 32-col or EHB just won't cut it anymore...
A.
=============================================================================
Andrew G. Robson Email: a.g.r...@northumbria.ac.uk
"There should be somthing witty here... but I ain't"
=============================================================================
James E. Reich
: [...] Seems to me that the 1200's slow
: > '020 holds us back at least as much as ECS (not to mention the lack of
: > an MMU to get real memory protection in the OS). Anyway, it's not as
: > if PC games are generally running in Super-VGA 32768 color modes...
: > 256 colors are lovely, but they're also more work to produce.
: What has work to do with it? btw Marathon on the Mac is in 32768 and very nice
Yes, Marathon on the Mac is in 32768, but it's not relying
on a slow '020 (admittedly with some help from the blitter) to run --
The Mac standard these days is pretty much a 33 MHz '040, better than
even most A4000's and with faster RAM, and 90 MHz PPCs are starting to
get common.
[...]
: it is too... Adding mem protection to AmigaOS is very non-trivial - due to
: the way message ports work on AmigaOS... plus I've got a '1200 and I'm not
: using the stock processor...
Not denying that it's highly nontrivial-- even MacOS 8 (aka Copland)
won't implement memory protection except for specially-coded software
-- but it is a critical step, and I think AT is making a *big* mistake
by continuing to build new machines which will not support it out of
the box.
: > And besides, the only way to get AGA, in the relatively narrow time
: > window before the bankruptcy was to downgrade to a 1200, or make the
: > pricey leap to a A4000 and throw away most of your existing hardware
: > and software. [...] I'm curious what
: > the actual percentages are, but I'll bet the majority of Amiga users
: > still have ECS systems.
: An A4000 means you have to throw away existing soft- and hardware? How so?
: In man cases 3.0/3.1 are more compatible than 2.04.
Yes, but think back to the early days of AGA and even with KillAGA,
what a large percentage of games (which are, after all, what we're
discussing here) wouldn't run at all.
: the same as any other big-box Amiga. If thats what you have, get a GFX-card.
But alas, they're not all that well supported...
: Sure the majority have OCS/ECS (I have an OCS '500 here collecting dust) but
: how many of them are in use? How many of those owners have an accelerated
: mahcine that can handle a decent 'doom' clone? Most OCS/ECS machiens will
: still be 6...@7.14MHz ...
I don't know anyone with an A2000 or greater who have less than an '030/25.
You know, I think an interesting aspect of this discussion comes from a
radically different distribution of machines in Europe and the US -- I
know fairly few people over here who bought the 500, 600 or 1200 -- I
don't have sales figures, but I'll bet the A2000, 3000 and 4000 are
much larger proportions of the market over here than there. You folks
perceive a huge A1200 AGA market, and I actually have never seen an
A1200 in the flesh.
In itself, that's an interesting topic for discussion, and something
AT ought to be thinking about -- I don't think people here just are
fond of console-type systems-- in general, they want a big box with
easy expandability, and with the A4000 coming back priced comparably to a
dual-Pentium, 32Mb/2Gb PC, and bare-bones 1200s competing with 486's,
I'm not sure what's going to happen...
: AGA/GFX-Card software is the way forward. If the new (well, rereleased)
: machiens have any hope at all to survive, the need games that at least LOOK
: good compared to the PC stuff - which is hard enough anyway, having seen some
: recent PC stuff... eitehr way, 32-col or EHB just won't cut it anymore...
There is something to what you're saying, but if new software
alienates a large part of the current user base, and AT is hell-bent
on insane prices, I fear that it'll be the last nail in the coffin --
people who have been hanging on will ditch their ECS Amigas and
instead of upgrading, will (cringing) relent and buy PCs.
-- Jim
Janne Siren
: won't implement memory protection except for specially-coded software
: -- but it is a critical step, and I think AT is making a *big* mistake
: by continuing to build new machines which will not support it out of
: the box.
I don't. I rather see Amigas in the shops NOW than new and improved Amigas
in the shops october 1996... AT knows they've got to improve the system
to survive, but Amigas back in shops is just as important to stabilize the
market.
They can't deliver memory protected Amigas now, so this is better than
nothing.
--
Janne Siren
si...@mikrobitti.fi
Tom D Tek 7BvO
>Andrew Robson (a.g.r...@unn.ac.uk) wrote:
>: Sure the majority have OCS/ECS (I have an OCS '500 here collecting dust) but
>: how many of them are in use? How many of those owners have an accelerated
>: mahcine that can handle a decent 'doom' clone? Most OCS/ECS machiens will
>: still be 6...@7.14MHz ...
>
>I don't know anyone with an A2000 or greater who have less than an '030/25.
The Yank is right...
>
>You know, I think an interesting aspect of this discussion comes from a
>radically different distribution of machines in Europe and the US -- I
>know fairly few people over here who bought the 500, 600 or 1200 -- I
>don't have sales figures, but I'll bet the A2000, 3000 and 4000 are
>much larger proportions of the market over here than there. You folks
>perceive a huge A1200 AGA market, and I actually have never seen an
>A1200 in the flesh.
This is true... I have been to Europe twice now... And have found that
OVER THERE, the Amiga is USED MOSTLY as a Game platform...
AND because of the difference in electronics prices, I found that MOST
people did not even have a HARD DRIVE! MUCH LESS an accelorated machine!
The users here in the states tend to be more Power user/Productivity
oriented...
A generalization.. yes.. but accurate..
>In itself, that's an interesting topic for discussion, and something
>AT ought to be thinking about -- I don't think people here just are
>fond of console-type systems-- in general, they want a big box with
>easy expandability, and with the A4000 coming back priced comparably to a
>dual-Pentium, 32Mb/2Gb PC, and bare-bones 1200s competing with 486's,
>I'm not sure what's going to happen...
Yes.. we're all waiting...
>
>
>: AGA/GFX-Card software is the way forward. If the new (well, rereleased)
>: machiens have any hope at all to survive, the need games that at least LOOK
>: good compared to the PC stuff - which is hard enough anyway, having seen some
>: recent PC stuff... eitehr way, 32-col or EHB just won't cut it anymore...
here we go on GAMES again... There'z more to life than PacMan and Mario!
(or Doom as the case may be!)
Tom D Tek 7Bu)
__________________________________________________________________________
/ Thomas W. Weeks | __ Commodore Amiga 500/030 40MHz \
| Authorized C= Technician | /// Hacked Into Black Tower Case, |
| Full Time Elec. Eng. Tech | __ /// 5M RAM, SyQuest 44M, 160M SCSI |
| and Telecom Student | \\\/// IBM 286 Hardware Emulator, |
| Texas A&M University | \XX/ Home Brew Audio A/D, HP48GX |
| **EMAIL** |"Amiga, The Computer for the Creative Mind"TM |
| tww...@venus.tamu.edu |"Macintosh, The Computer for the Rest of Us"TM|
\____________________________|_____________________________________________/
Andrew Robson
> on a slow '020 (admittedly with some help from the blitter) to run --
> The Mac standard these days is pretty much a 33 MHz '040, better than
> even most A4000's and with faster RAM, and 90 MHz PPCs are starting to
> get common.
Yep, but if thats the case, surely we *SHOULD* be using 256 colours? Plus, how
many Amigas are using 'slow '020's' ? Mine isn't. Few of my friends are.
Remember that most [all?] new stuff on the PC and Mac is written in C. Asm
still kicks C's ass on CISC processors, and most decent Amiga stuff is 68K...
> Not denying that it's highly nontrivial-- even MacOS 8 (aka Copland)
> won't implement memory protection except for specially-coded software
> -- but it is a critical step, and I think AT is making a *big* mistake
> by continuing to build new machines which will not support it out of
> the box.
I disagree. Get the boxes in the high-street, get some publicity, get some
sales, try and kickstart (no pun intended :) ) the market again - including the
software - very little new stuff planned compared to say 2 years ago :(
Then, start doing a new Amiga - give it RISC (PPC, HP, whatever - PPC has the
mainstream following, but it's none too fast from what I have seen) port the
OS, dump all the old crap left in there from 1.3 (BCPL! Aghrrrr!) add virtual
memory, mem protection, multi-user stuff... Add PCI, a decent GFX-chipset,
16-bit sound... and forget backwards compat. as far as games go... try and add
some kinda 'emulation box' system for older stuff with 68K emulation - I say
emulation box cause that'll let you mess with stuff in the OS - such as adding
mem protection - without needign kludges to prevent old OS-legal stuff
breaking.
Thats the way forward IMHO (then again, what do I know? I'm a programmer, not a
marketoid...)
> Yes, but think back to the early days of AGA and even with KillAGA,
> what a large percentage of games (which are, after all, what we're
> discussing here) wouldn't run at all.
>
Actually, I found that most of my games worked ... okay, a couple here and
there, but hardly my entire s/w collection.
> : the same as any other big-box Amiga. If thats what you have, get a GFX-card.
>
> But alas, they're not all that well supported...
>
I know :( But if someone starts...
> I don't know anyone with an A2000 or greater who have less than an '030/25.
I don't know anyone with a 2000. (3000's, 4000's, even a 1000...) Oh, I know
someone with an A1500 - that's a C=UK rebadged 2000 supplied without the HD...
>
> You know, I think an interesting aspect of this discussion comes from a
> radically different distribution of machines in Europe and the US -- I
> know fairly few people over here who bought the 500, 600 or 1200 -- I
> don't have sales figures, but I'll bet the A2000, 3000 and 4000 are
> much larger proportions of the market over here than there. You folks
> perceive a huge A1200 AGA market, and I actually have never seen an
> A1200 in the flesh.
Yep, this is very true. Remember that the best-selling Amiga model is teh 1.3
A500. Most of these sales were in europe. They have an ECS agnus, and an OCS
denise, and have to have a motherboard kludge to allow 1MB chip - otherwise its
512K Chip, 512K slow RAM ... The A1200 is quite popular, and there are a fair
few A4000's knocking about in various guises - I even know an 060 owner ...
BTW the A1200 casing sux.
>
> In itself, that's an interesting topic for discussion, and something
> AT ought to be thinking about -- I don't think people here just are
> fond of console-type systems-- in general, they want a big box with
> easy expandability, and with the A4000 coming back priced comparably to a
> dual-Pentium, 32Mb/2Gb PC, and bare-bones 1200s competing with 486's,
> I'm not sure what's going to happen...
>
I wish I could have afforded a big-box when I bought my A1200, but I was
unemployed at the time, and had to save for months to get it...
In the UK at least, the A500 sold well as a games machine to kids who knew the
could get pirated stuff off their mates... they did'nt want a computer, they
just wanted a games-system with good games - and back then, the Amiga was teh
best... Again, when I got the my 500 (still have it, folks...) it was the best
I could afford - I was in skool then... tho it did get an HD, extra mem etc.
(eventually...)
The current models are overpriced, admittedly... (and underspecced.. ) .. nut
thats another discussion ;)
> There is something to what you're saying, but if new software
> alienates a large part of the current user base, and AT is hell-bent
> on insane prices, I fear that it'll be the last nail in the coffin --
> people who have been hanging on will ditch their ECS Amigas and
> instead of upgrading, will (cringing) relent and buy PCs.
>
Its a vicious circle - if we stick with stuff that works on a bog std. 500,
it's gonna impress no-one - they'll take one look and buy a PC (or a play
station, but I don't think we can compete in the games-only arena, with an 030
+ mem costing around the same as a Play station..)
If we go with stuff that requires AGA, we - as you point out - risk alienating
the current user-base, and losing them to the Dark Side[tm] ...
I still think the way would be to do stuff (technical, huh?) that will wirk
with AGA *OR* a gfx-card... this way the biggest risk is losing the A500 1.3
crowd, and frankly, with the Play station out there I think we already lost
'em...
On a totally dif subject, hands up all those who'd like to see a
Daytona/ridge-racer clone for the Amiga? Probably impossible to do properly
(well, without a decent 040 and gfx card..) but I'd love to see it - and it
might stop me throwing ''pund chunks'' at Sega Rally and Daytona....
> -- Jim
Regards,
A.
_____
(o o) .oO What? No .sig?
-ooo--(_)--ooo-
Email a.g.r...@northumbria.ac.uk
Andre Punter
TDT> This is true... I have been to Europe twice now... And have found
TDT> that
TDT> OVER THERE, the Amiga is USED MOSTLY as a Game platform...
You're probably right. But there's nothing wrong with that, in my opinion.
btw, there are a lot of pc's (at home) that are used moslty (or 'almost
only') for games.
TDT> AND because of the difference in electronics prices, I found that MOST
TDT> people did not even have a HARD DRIVE! MUCH LESS an accelorated
TDT> machine!
yeah, that's a pitty.
__
/-/ndre, just another average gameplayer..:)
E-mail: and...@twilight.xs4all.nl
.-------------------------------------------------------------------------.
| -=> The TWiLiGHT Zone BBS <=- -=> +31-(0)-20-6764584 <=- -=> 24 Hrs <=- |
| -=> Modules Heaven! <=- -=> 9600 to 14400 Baud <=- -=> /\MIG/\ Only <=- |
`-------------------------------------------------------------------------'
-- Via Xenolink 1.97, XenolinkUUCP 1.1
john enger
: Andrew Robson <a.g.r...@unn.ac.uk> wrote:
: >> Yes, Marathon on the Mac is in 32768, but it's not relying
: >> on a slow '020 (admittedly with some help from the blitter) to run --
: >> The Mac standard these days is pretty much a 33 MHz '040, better than
: >> even most A4000's and with faster RAM, and 90 MHz PPCs are starting to
: >> get common.
FWIW you can now get a Full '040 33Mhz Accelerator for the 1200 for 399 UKP.
Perhaps this would be a good CPU to stick in an updated 1200 while we all
wait for the RISC machines...
Just think what you could do with 4Mb and a 33/040...
BTW there is something bothering me (sort of) about the price difference
between 1200 accelerators, and 4000 boards. A case in point being the 060
boards - 599 on the 1200, 849 on the 4000. Similar differences occur with
040 boards too...
Finally, as you can now get a 030/50 Mhz, I really believe that this should
become the new standard processor. People like me with 40 Mhz boards will
get by fine... We do need cheaper boards for the big box Amy's though...
Just another 2c (Which makes about 50 bucks by now!)
========================= john....@bbsrc.ac.uk =====================///===
A1200, 40 Mhz '030, 2+4Mb RAM, 80Mb HD, 1960 MSync Monitor. ///
-------------------------------------------------------------------\XX/-----
john enger
: Finally, as you can now get a 030/50 Mhz, I really believe that this should
^ for 199
Paul Chan
>
>> The crippling factor here is the bitplane gfx of the A1200. With no way
>> to upgrade it to a VGA chunky display without using up cpu power I would
>> estimate that to match a fast 486 game on the Amiga may well need a 50Mhz
>> 030 or 25Mhz 040SLC (a new low cost version just out) with a lot of custom
>> re-programming to get the speed back - not just a straight port.
>
>I am sorry but that is crap. The lack of chunky mode is not
>the reason, even if AGA had chunky mode 68020 would not be
>able to render 320x256 screen at decent frame rate.
>
>Now, there are many very efficient chunky2planar routines,
>like 160x128 blittescreen for A1200, and 160x128 screen
>it can render at decent frame rate.
>
>Obviously you are not aware of very fast c2p routines and
>the fact that it is better to render the game screen to
>fast ram and then convert it to bitplanes.
OK. I`ll concede that point, but a faster cpu is still my concern. Tecnhically,
the Amiga would do very well from having a 28Mhz 020 on board. I think that the
engineering people will use a 25Mhz EC030 or EC020 as the primary cpu. The
marketing people would much prefer a 50Mhz 030 or an 040 so they could push
the Amiga properly.
Where do you think the new minimum level cpu should lie?
James E. Reich
: So, if we go AGA, we lose the A500 owners - big deal.
NONONONONONONONONONONO!
What you lose is most of the US market. At this point, most of the
machines over here are accelerated A2000s and A3000s with OCS, ECS or
graphics cards. A500's, A600's and A1200's were never very big over
here-- console machines haven't been popular since the C64, and A4000's
were too late and too expensive to sell much.
: >I still think the way would be to do stuff (technical, huh?) that will wirk
: >with AGA *OR* a gfx-card... this way the biggest risk is losing the A500 1.3
: >crowd, and frankly, with the Play station out there I think we already lost
: >'em...
Right, but then you have the problem of supporting multiple graphics card
standards. Although we constantly beat on the undisciplined PC programmers
for their bloated software, a goodly percentage of the bloat is from support
of multiple graphics and network boards -- proceed with caution.
Still, this has a much better chance of working -- if a lot of stuff
supported gfx cards, A2000 users might be talked into upgrading. Talking
accelerated A2000 users into buying a A4000 is a tough sell given the
high price and uncertain future.
-- Jim
mur...@thelair.zynet.com
and A3000 systems. Let me qualify my statement by saying that the future
software should support either graphics cards or AGA, or both. I don't
think I'm going out on a limb here. It just doesn't make sense, no
matter how much I'd like to run Virtua Fighter on this A500, to cater to
the lower end specs of the Amiga. Let people who either have A2000s or
A3000s with graphics cards, or AGA machines, get new products that have a
specific bent toward graphics. The ECS crowd has been supported pretty
well by public domain, and should be able to do so for some time. It is
simple logic that a marketable, competitive machine can't kludge itself
to death.
Moo.
Muraii/\Lorelei
: --
: "Procrustes in modern dress, the nuclear scientist will prepare the bed
: on which mankind must lie; and if mankind doesn't fit--well, that will be
: just too bad for mankind." Dr. Aldous Huxley, (c) 1946
--
"Procrustes in modern dress, the nuclear scientist will prepare the bed
on which mankind must lie; and if mankind doesn't fit--well, that will be
just too bad for mankind." Dr. Aldous Huxley, (c) 1946
mur...@thelair.zynet.com
: I agree that the RISC Amigas should break completely from the compatiblity
: problem. If you need compatibility then just keep your old Amiga. In addition,
: it`s not as if all the old Amiga software was indispensible. There must be less
: than a dozen developers making quality release software like Final Writer (for
: instance). They can be persuaded to write new versions for new machines.
I have just bought, about three months ago, my first Amiga, an A500. I
must say, I totally agree with you here. I have become a convert to this
type of machine, that lets me do so much without me even realizing that
I'm being able to do so much. The only time I realize just how
impressive this 8-10-year-old machine is, is when I play with a PC.
Sure, the PC generally has a better standard video subsystem, and now
better support, but actually working with anything that isn't totally
maxed out can be a pain.
It is for the respect I've gotten for the Amiga that I hope it's future
isn't hobbled by people with my caliber of machine bitching and moaning
for compatibility. You can't put the drive shaft of a Model-T into a
Vector can you? No. Why? Because the automobile has _evolved_. The
only good thing for the Amiga is evolution, and it's about time, from
what I hear.
I'm going to get an A3000 soon, and maybe, down the line, hopefully, one
of this new breed of Amiga about whose specifications everyone is
arguing. But, if I want to play games I can play now on this A500, or
games I may play on the A3000, I'll keep one or both around.
I want a future for the Amiga. There is none if AT tries to capitulate
to everyone's wishes.
: >
: >If we go with stuff that requires AGA, we - as you point out - risk alienating
: >the current user-base, and losing them to the Dark Side[tm] ...
: So, if we go AGA, we lose the A500 owners - big deal. They either couldn`t
: afford to upgrade (which means they won`t really buy a game) or they wanted to
: stick with the games coming out now (mostly so they can get a pirated copy) -
: and we really don`t need people like them.
If I can't afford a new proto-Amiga, that doesn't mean everyone else
should suffer. Again, I agree.
: >
: >I still think the way would be to do stuff (technical, huh?) that will wirk
: >with AGA *OR* a gfx-card... this way the biggest risk is losing the A500 1.3
: >crowd, and frankly, with the Play station out there I think we already lost
: >'em...
: Right on. Stick with the group that`s already invested MORE than the standard
: spec. They are loyal, and would BUY a game that used their setup. Who cares if
: there are less of them? I would say they are more in touch with the Amiga
: situation and a larger percentage would buy PC style (big) games.
Prezactly. If I can get a graphics card for the A3000, which I hope to
do, it would be nice to be able to play whatever games I want with that
fairly expensive hardware.
There may be a way to have some _degree_ of compatibility, but if it
would be at the cost of blowing the lid off the competition and coming
full boar with a badass machine, then fuck compatibility.
Moo.
Muraii/\Jhargon
Peter McGavin
>a faster cpu is still my concern. Tecnhically,
>the Amiga would do very well from having a 28Mhz 020 on board. I think that the
>engineering people will use a 25Mhz EC030 or EC020 as the primary cpu. The
>marketing people would much prefer a 50Mhz 030 or an 040 so they could push
>the Amiga properly.
>
>Where do you think the new minimum level cpu should lie?
I think moving up the 020..030..040..060 ladder will barely keep up
with the competition and is only a temporary solution anyway.
The new minimum level should be RISC as soon as possible. Ideally the
OS should already be ported and brilliant new hardware should already
be designed.
--
Peter McGavin. (p.mc...@irl.cri.nz)
Jyrki Saarinen
> Tecnhically, the Amiga would do very well from having a 28Mhz 020 on board.
> I think that the engineering people will use a 25Mhz EC030 or EC020 as the
> primary cpu. The marketing people would much prefer a 50Mhz 030 or
> an 040 so they could push the Amiga properly.
>
> Where do you think the new minimum level cpu should lie?
Well, EC020/28 would be very cheap for A1200. 030/50 would not
be very expensive either -> low-power 040 would be quite cheap
too.
I dont know.. ;)
Paul Chan
>: Where do you think the new minimum level cpu should lie?
>
>I think that the minimum spec Amiga should be 060 and CV64 - of course
>its not realistic. In reality, at least an 030 should be used, with 4-6Mb
>of RAM, and a 500Mb Harddrive. A full 030 or an EC030 wouldnt make
>much difference (execpt in price??) to the games player, and even to some
>semi-professional users. The next low end Amiga should have two, maybe
>three card slots, easily upgradable CPU, 4 spaces on the motherboard for
>RAM expansion, 4Mb of FAST RAM, an accelerated version of AGA (at least
>so that a game could run in HIRESLACE 256, quickly) and maybe upgraded
>sound.
>
>A system like that wouldnt require too many changes to the current AmigaOS,
>it would only really be a matter of actually getting the hardware together -
>and of course software co`s to actually write programmes for this machine,
>and not the A500..
The irony here and now is that Dr Peter Kittel has been saying that low end
Amigas really don`t need a hard drive as standard as it would increase the base
price.
Now, two thing come to mind:
1. Omigod! Dr Peter is stuffing loads of expensive and cool hardware into the
A1300 and can`t afford to keep a HD in there for his projected (very low) price
point! Maybe he intends to have a CD drive as standard and supply the OS and
plenty of applications software on one CD?
2. Dr Peter has caught C=-itis and has decided that a 4Mb equipped 030 Amiga
(if reports about the A1300 are true) doesn`t actually need a HD at the base
level and users will make do with a DD floppy drive. Even with clever
compression, games would need 2 disk swaps and 2mins+ of loading to fill up
4Mb. Applications lke Wordworth would take longer to load. (Back to the
spectrum, guys)
Anyhow here are the problems I have with the A1300.
1. A1300? Not much of a name, and it`s an odd number! I would prefer A2400.
Actaully, I would prefer that AT gave it a *name* to distinguish from the rest
of the Amiga range which disappears into a haze of 3/4 digit numbers. Phoenix
sounds a lot better than A2400 - of course, that name is taken, so get out your
mythology books and hunt out a totally cool name for the low end Amiga.
2. It had better not be a "long cream log" ie an A1200 with a CD drive bolted
on where the PCMCIA slot used to be. That *would* be sad.
3. The spec has to be right. We can`t go cheap on the CPU. I`d be impressed by
a low end 040SLC in the low end Amiga - I`d probably buy one as it would put
the cpu in my A4000/EC030 to shame! They can still code in 020 for
compatibility, but for 3D games (which are all the rage) the 040 is king. Does
anyone else agree that releasing all the next Amigas with the 040SLC as base
level would make us upgrade? I bet it will! This would be the instant sales
boom that the Amiga would be looking for if we could get an 040 SLC Amiga out,
with a CD drive, for 399-499 UK pounds (599-699 US).
Paul Chan
>Paul Chan <paul> writes:
>
>>Andrew Robson <a.g.r...@unn.ac.uk> wrote:
>
>>that to match a fast 486 game on the Amiga may well need a 50Mhz 030 or 25Mhz
>
>>040SLC (a new low cost version just out)
>
>Incidentally do these things have an integral MMU & FPU ?
>
>Just curious...
Don`t quote me here, but I think the SLC doesn`t have an FPU, and ISN`T a hot
power hog like the old 040 was. It *might* not have an MMU, but that would be
silly on an 040...
Jyrki Saarinen
> I think moving up the 020..030..040..060 ladder will barely keep up
> with the competition and is only a temporary solution anyway.
Agreed.
> The new minimum level should be RISC as soon as possible. Ideally the
> OS should already be ported and brilliant new hardware should already
> be designed.
I think some 3rd party developer could make a PPC card for
A1200 for example and have a 68020 emulator in ROM. Then
later more and more native software would come..
John Hendrikx
>> Yes, Marathon on the Mac is in 32768, but it's not relying on a slow '020
>> (admittedly with some help from the blitter) to run -- The Mac standard
>> these days is pretty much a 33 MHz '040, better than even most A4000's
>> and with faster RAM, and 90 MHz PPCs are starting to get common.
AR> Yep, but if thats the case, surely we *SHOULD* be using 256 colours?
We should be using 256 colors if there was a DECENT upgrade path to get those
256 colors. At the moment there is no decent upgrade path as the 4000's are
down-right expensive (and the 4000/030 doesn't even have MMU/FPU) and the A1200
simply is 'not good enough' for people who used to own a 2000 or 3000. And if
you also consider that most earlier expensions (for 500's and 2000's for
example) can't be used anymore in new machines it starts to look really gloomy.
AR> Plus, how many Amigas are using 'slow '020's' ? Mine isn't. Few of my
AR> friends are. Remember that most [all?] new stuff on the PC and Mac is
AR> written in C. Asm still kicks C's ass on CISC processors, and most
AR> decent Amiga stuff is 68K...
We shouldn't get rid of 256 colors yet, instead we should become flexible and
write games in such a way they can support any number of colours (from 16 to
16M, if necessary using a B&W display). There are just too many heavily
expanded ECS machines out there to simply 'ignore', and they will remain out
there as long as there is no decent upgrade path. Only when a completely new
Amiga arrives which will surpass AGA and the 68060 you will see these ECS users
upgrade again, they won't upgrade to a machine which offers almost nothing, but
comes with a whole lot of other problems, check this out:
I own an expanded A2000 with 68030, 68882 and MMU plus 7MB of RAM (and some
SCSI devices as well). Which computer should I buy?
A1200? Yeah right, where would I put my CD-ROM player, my 2 Harddrives
etcetera... not to mention it doesn't even come with SCSI.
A4000/030? Yeah right, and spend loads of cash just to get 256 colors and then
-lose- my 68882 and MMU? Also, no SCSI which means I'll have to buy a
controller for it as I won't throw away all my SCSI stuff.
A4000/040? Spend even MORE cash just to get a machine which has a memory
interface more than twice as slow as my current machine and generally not all
that much faster? Get real.
A5000/PPC/PCI? Well, now that is getting interesting, it is *much* faster than
my current machine, likely has a decent memory interface this time, and since
it has got PCI slots adding the things I 'miss' is VERY cheap (like a SCSI
controller for $80). Also I could add a cheap PCI graphics card (if one wasn't
installed standard) and get True 24-bit (or maybe 48-bit by that time :-)))
So, for me, and for probably a lot of us power-ECS users out there a new Amiga
is the only decent path to upgrade to, and until that time I would urge people
not to ignore the ECS users but instead be flexible enough to make games run
independant of what the user has got installed.
AR> I disagree. Get the boxes in the high-street, get some publicity, get
AR> some sales, try and kickstart (no pun intended :) ) the market again -
AR> including the software - very little new stuff planned compared to say 2
AR> years ago :( Then, start doing a new Amiga - give it RISC (PPC, HP,
AR> whatever - PPC has the mainstream following, but it's none too fast from
AR> what I have seen) port the OS, dump all the old crap left in there from
PPC not fast enough? Are you kidding? It beats the crap out of a 150MHz
Pentium and PPC hasn't even started yet. Also they are cheap as they are the
best sold RISC chip currently in existence, in other words, if the PPC chip
wouldn't be fast enough alone, then thanks to its low price you could simply
use 2 PPC chips.
AR> 1.3 (BCPL! Aghrrrr!) add virtual memory, mem protection, multi-user
AR> stuff... Add PCI, a decent GFX-chipset, 16-bit sound... and forget
AR> backwards compat. as far as games go... try and add some kinda
AR> 'emulation box' system for older stuff with 68K emulation - I say
AR> emulation box cause that'll let you mess with stuff in the OS - such as
AR> adding mem protection - without needign kludges to prevent old OS-legal
AR> stuff breaking.
Yeah, something like that is what I have in mind too, infact, I think it is the
only way Amiga can ever become mainstream again (PPC and PCI I mean)
AR> Thats the way forward IMHO (then again, what do I know? I'm a
AR> programmer, not a marketoid...)
It is IMO the only way, the 68060 won't cut it and Motorola certainly doesn't
give us a whole lot of confidence with the 68060's (they are over 2 years late,
and they don't seem to care about it at all anymore -- not to mention that it
is already inferior to all the other current processors out there and it hasn't
been released yet, kinda like what we'd get if we released AAA now). So, going
RISC (and I really mean going PPC) seems to be the only chance we've got left.
>> Yes, but think back to the early days of AGA and even with KillAGA, what
>> a large percentage of games (which are, after all, what we're discussing
>> here) wouldn't run at all.
AR> Actually, I found that most of my games worked ... okay, a couple here
AR> and there, but hardly my entire s/w collection.
The problems isn't with AGA, it is with the price of the AGA machines and what
you get for that price. It simply is not worth spending a whole lot of money
for just to get a similair machine you had previously but now with 256
colors... buying a gfx-card is not only cheaper, it is faster too and you can
have upto 4MB of gfx-ram.
AR> Yep, this is very true. Remember that the best-selling Amiga model is
AR> teh 1.3 A500. Most of these sales were in europe. They have an ECS
AR> agnus, and an OCS denise, and have to have a motherboard kludge to allow
But these unexpanded 500's are not out there anymore (well they are but they
aren't being used anymore so they don't count).
AR> 1MB chip - otherwise its 512K Chip, 512K slow RAM ... The A1200 is
AR> quite popular, and there are a fair few A4000's knocking about in
AR> various guises - I even know an 060 owner ...
AR> BTW the A1200 casing sux.
Yep, I think that casing an Amiga as one gigantic keyboard is what is stopping
people from seeing it as a serious platform. No other serious platform I know
off still does this kind of thing, and I don't blaim them.
AR> I wish I could have afforded a big-box when I bought my A1200, but I
AR> was unemployed at the time, and had to save for months to get it...
I think that a A1200 put in a mini-tower kind of case by AT themselves would at
most add say $150 to the end price. You can obtain cheap mini-towers from the
clone market, and creating a small card with say 3-4 expansion slots is not
going to be expensive at all. It would have made the A1200 look a lot more
professional at and that price (ie, a professional looking machine at such a
bargain price) I would imagine it would sell fast. Too late now though.
AR> The current models are overpriced, admittedly... (and underspecced.. )
AR> .. nut thats another discussion ;)
I think you need to adjust that, the current models are underspecced, heavily I
might add.
Grtz John
-----------------------------------------------------------------------
John.H...@grafix.xs4all.nl TextDemo/FastView/Etc... development
-----------------------------------------------------------------------
-- Via Xenolink 1.98b1, XenolinkUUCP 1.1
John Hendrikx
>> Its a vicious circle - if we stick with stuff that works on a bog std.
>> 500, it's gonna impress no-one - they'll take one look and buy a PC (or a
>> play station, but I don't think we can compete in the games-only arena,
>> with an 030 + mem costing around the same as a Play station..)
PC> Bad news, I agree. How many Amiga owners already have expansions? They
PC> are less likely to change platforms, and more likely to buy a game that
PC> ekes out every last MIP in their system - games like this can impress
PC> people. We only need a few PC titles to appear on our system (and work
PC> almost identically) and we might have a market.
Why haven't there appeared any X-Wing or Tie-Fighter clones on Amiga? These
games are a lot -less- processor intensive than a DOOM-clone and not all that
hard to program. In these kind of games the most graphics you are most likely
to see is one or two (gouraud shaded) fighters at medium distance and some
coloured lines indicating enemy and friendly fire at large distances. In other
words, the most time consuming thing would be to render the 3d-star background.
With a C2P routine which only converts parts which have changed this could
easily run at a 2-3 times higher frame-rate than the TextDemo engine runs at
now (that would be 15 FPS on a 3000/25, full-screen 320x240 1x1 -- with the
cockpit 'on' the screen would be reduced to almost half that and run even
faster).
Joshua Galun
:This too is what I'd like to see. Escom seem to be obscessed by Motorola's 68K
:line tho...
Where did you get THAT particular view from? You say they are
obsessed with the 68k line just because they use them in Amigas, which is
the only line Amigas can use at the moment?
--
/// Joshua Galun \\\
/// Business Manager for Creative Edge Software \\\
\\\/// Check out our WWW page at http://outland.cyberwar.com/~zool \\\///
\\\/ The Amiga, JRR Tolkien, and The Simpsons. 'Nuff said. \///
Andrew Robson
> We should be using 256 colors if there was a DECENT upgrade path to get those
> 256 colors. At the moment there is no decent upgrade path as the 4000's are
> down-right expensive (and the 4000/030 doesn't even have MMU/FPU) and the A1200
> simply is 'not good enough' for people who used to own a 2000 or 3000. And if
> you also consider that most earlier expensions (for 500's and 2000's for
> example) can't be used anymore in new machines it starts to look really gloomy.
Decent Upgrade Path? GFX-Card. Unfortunatly, this isn't an option for me
'cause I can't afford to switch to a 'big-box' amiga of the same spec as my
A1200. I'm kinda in a similar situation to you. I don't wanna give up my 50
Mhz '030. (btw It'd have to ba a 3000/3000T..)
> We shouldn't get rid of 256 colors yet, instead we should become flexible and
> write games in such a way they can support any number of colours (from 16 to
> 16M, if necessary using a B&W display). There are just too many heavily
> expanded ECS machines out there to simply 'ignore', and they will remain out
> there as long as there is no decent upgrade path. Only when a completely new
> Amiga arrives which will surpass AGA and the 68060 you will see these ECS users
> upgrade again, they won't upgrade to a machine which offers almost nothing, but
> comes with a whole lot of other problems, check this out:
This makes sense, but I think supporting ECS chipsets is still a dead end. I
would prefer a minimum of AGA or GFX card. If a user has a heavily expanded
machines there is a reasonable chance he/she will own a GFX-card too. If not,
they can alway buy one.
>
> I own an expanded A2000 with 68030, 68882 and MMU plus 7MB of RAM (and some
> SCSI devices as well). Which computer should I buy?
>
Which ever one ya want!
[...]
>
> A5000/PPC/PCI? Well, now that is getting interesting, it is *much* faster than
> my current machine, likely has a decent memory interface this time, and since
> it has got PCI slots adding the things I 'miss' is VERY cheap (like a SCSI
> controller for $80). Also I could add a cheap PCI graphics card (if one wasn't
> installed standard) and get True 24-bit (or maybe 48-bit by that time :-)))
This too is what I'd like to see. Escom seem to be obscessed by Motorola's 68K
line tho...
>
> So, for me, and for probably a lot of us power-ECS users out there a new Amiga
> is the only decent path to upgrade to, and until that time I would urge people
> not to ignore the ECS users but instead be flexible enough to make games run
> independant of what the user has got installed.
I upgraded from an A500/HD/3MB ... The 1200 was a reasonable upgrade for me at
the time. Had I owned a more powerful machine, perhaps I would have stuck with
it... now I own an A1200/030@50Mhz/FPU/MMU/HD... does that make me a
poeer-AGA user? Regardless, without spending huge amounts of cash, I can't a
hugely improved system. I too am dogged by the lack of a decent upgrade path.
>
> PPC not fast enough? Are you kidding? It beats the crap out of a 150MHz
> Pentium and PPC hasn't even started yet. Also they are cheap as they are the
> best sold RISC chip currently in existence, in other words, if the PPC chip
> wouldn't be fast enough alone, then thanks to its low price you could simply
> use 2 PPC chips.
I was referring to its' speed compared to other RISC chips. Now I admit I have
no real experience with RISC chips, beyond playing with SPARC and DEC RISC
workstations... but from what I can remember, PPC - esp the 601 - ain't too hot
compared to other RISC chips - *NOT* compared to the current 680x0 .. well,
68060... I agree that PPC gives much better price/performance Ratio than the
'060 - its around the same speed for less cash, if memory serves....
btw have you looked at the instruction set of the PPC? Hardly reduced...
Oh, using 2 PPC's? I was thinking, maybe ESCOM/AT should lisence the BeBox and
port AmigaOS to it......?
> Yeah, something like that is what I have in mind too, infact, I think it is the
> only way Amiga can ever become mainstream again (PPC and PCI I mean)
Without doubt. Even with that, I suspect it'll end up being ''just another
NT-running box'' probably with an OTT price tag :(
> It is IMO the only way, the 68060 won't cut it and Motorola certainly doesn't
> give us a whole lot of confidence with the 68060's (they are over 2 years late,
> and they don't seem to care about it at all anymore -- not to mention that it
> is already inferior to all the other current processors out there and it hasn't
> been released yet, kinda like what we'd get if we released AAA now). So, going
The 060 is nice. I've used an Amiga with one. But it costs too much and
delivers too little compared to the other systems out there.
> RISC (and I really mean going PPC) seems to be the only chance we've got left.
You think we've got a chance left? Perhaps. Damn slim tho.
> The problems isn't with AGA, it is with the price of the AGA machines and what
> you get for that price. It simply is not worth spending a whole lot of money
> for just to get a similair machine you had previously but now with 256
> colors... buying a gfx-card is not only cheaper, it is faster too and you can
> have upto 4MB of gfx-ram.
The problem is with AGA. That's the current GFX-subsystem in the Amiga. It's
slow as hell and way outtadate[tm]. And not available for older machines.
> But these unexpanded 500's are not out there anymore (well they are but they
> aren't being used anymore so they don't count).
Why not? They're the majority of the ECS/OCS base, and I know of people still
using 'em. Hell, I know people still uisng 8MHz Atari ST's.
> Yep, I think that casing an Amiga as one gigantic keyboard is what is stopping
> people from seeing it as a serious platform. No other serious platform I know
> off still does this kind of thing, and I don't blaim them.
I agree. Drop the low-end and start out with a mid-range pizza box ot
mini-tower. Perhaps with one PPC chip in and slots for expansion. If you make
AmigaOS multiprocessing, then there is no need to throw your existing CPU away
when you upgrade, just bang in another CPU, and.... (It kinda gets to me there
is an '020 sitting in the box I am typing on that isn't being used... okay, it
wouldn't make a huge difference, but it'd be real nice....)
> I think that a A1200 put in a mini-tower kind of case by AT themselves would at
> most add say $150 to the end price. You can obtain cheap mini-towers from the
> clone market, and creating a small card with say 3-4 expansion slots is not
> going to be expensive at all. It would have made the A1200 look a lot more
> professional at and that price (ie, a professional looking machine at such a
> bargain price) I would imagine it would sell fast. Too late now though.
$150? With slots I assume you mean... I can get a tower case for around 30UKP
- $45 - INCLUDING a decent PSU ...
>
> I think you need to adjust that, the current models are underspecced, heavily I
> might add.
That's what I said. But it'll take time to get machienes out there. Supplying
the A4000T with a GFX-card, an 060 and a reduced price tag would go a long way
to improve matters, til they get a RISC box out there. It's a lot to ask, I
know, but....
BTW as a point of interest, A1200's seem to be selling pretty well in the UK
right now...
>
> Grtz John
>
Regards,
A.
> -----------------------------------------------------------------------
> John.H...@grafix.xs4all.nl TextDemo/FastView/Etc... development
> -----------------------------------------------------------------------
> -- Via Xenolink 1.98b1, XenolinkUUCP 1.1
=============================================================================
James McArthur
: > I think moving up the 020..030..040..060 ladder will barely keep up
: > with the competition and is only a temporary solution anyway.
: Agreed.
: > The new minimum level should be RISC as soon as possible. Ideally the
: > OS should already be ported and brilliant new hardware should already
: > be designed.
: I think some 3rd party developer could make a PPC card for
: A1200 for example and have a 68020 emulator in ROM. Then
: later more and more native software would come..
Certainly. If an 040 board can be made for the 1200, then a PPC is possible.
Jyrki Saarinen
> : I think some 3rd party developer could make a PPC card for
> : A1200 for example and have a 68020 emulator in ROM. Then
> : later more and more native software would come..
>
> Certainly. If an 040 board can be made for the 1200, then a PPC is
> possible.
Now I dont get it. How these two things are related?
PPC board had to contain a 68020-emulator on ROM though.
And AT and 3rd party developers should agree how
PPC things would be made..
Paul Chan
>Paul Chan (paul) wrote:
>
>: The irony here and now is that Dr Peter Kittel has been saying that low end
>: Amigas really don`t need a hard drive as standard as it would increase the base
>: price.
>
>disk. If things keep going the way they're going, it's possible that any hard
>disk inside an Amiga could depreciate 50-60% just sitting on a shelf for a
>few weeks. Better to buy a drive at current prices and get the store to
>install it when you buy the Amiga.
Yes, we know all about depreciation on hard drives, PC manufacturers tend to
price things monthly, and assemble PC`s days before they are sold. Amigas can`t
really be sold on this method unless the manufacturing centre is in Europe/USA
- NOT Philippines (with 6 week shipping delay).
A better, stable way of raising the Amiga standard is to include an IDE CD-ROM
as standard *instead* of a hard drive. Include *all* software in a bundle on
one CD, and ask the user to supply blank floppies as storage/scratch disk. Not
the best situation but perhaps a more stable price, and we might start seeing
more CD games if every Amiga came with a CD drive as standard from now on.
>
>
>: 3. The spec has to be right. We can`t go cheap on the CPU. I`d be impressed by
>: a low end 040SLC in the low end Amiga - I`d probably buy one as it would put
>: the cpu in my A4000/EC030 to shame! They can still code in 020 for
>: compatibility, but for 3D games (which are all the rage) the 040 is king. Does
>: anyone else agree that releasing all the next Amigas with the 040SLC as base
>: level would make us upgrade? I bet it will! This would be the instant sales
>: boom that the Amiga would be looking for if we could get an 040 SLC Amiga out,
>: with a CD drive, for 399-499 UK pounds (599-699 US).
>
>No way! '060 for sure!
The 060 is overpriced and in low supply. The 040 isn`t the ideal choice, nbut
it`s the one between the 030 (too low powered) and the 060 (as I said, too
rare/expensive) :) :) :) :)
Andrew Roehrig
: The irony here and now is that Dr Peter Kittel has been saying that low end
: Amigas really don`t need a hard drive as standard as it would increase the base
: price.
Not sure of Peter's logic, but it actually makes sense to not include a hard
disk. If things keep going the way they're going, it's possible that any hard
disk inside an Amiga could depreciate 50-60% just sitting on a shelf for a
few weeks. Better to buy a drive at current prices and get the store to
install it when you buy the Amiga.
: 3. The spec has to be right. We can`t go cheap on the CPU. I`d be impressed by
: a low end 040SLC in the low end Amiga - I`d probably buy one as it would put
: the cpu in my A4000/EC030 to shame! They can still code in 020 for
: compatibility, but for 3D games (which are all the rage) the 040 is king. Does
: anyone else agree that releasing all the next Amigas with the 040SLC as base
: level would make us upgrade? I bet it will! This would be the instant sales
: boom that the Amiga would be looking for if we could get an 040 SLC Amiga out,
: with a CD drive, for 399-499 UK pounds (599-699 US).
No way! '060 for sure!
ar
Peter McGavin
>> I think moving up the 020..030..040..060 ladder will barely keep up
>> with the competition and is only a temporary solution anyway.
>
>Agreed.
>
>> The new minimum level should be RISC as soon as possible. Ideally the
>> OS should already be ported and brilliant new hardware should already
>> be designed.
>
>A1200 for example and have a 68020 emulator in ROM. Then
>later more and more native software would come..
Any takers? I think emulating the OS would be difficult because of
the custom chips. Porting it would be much more practical IMO, but
would probably still take a small team of good software engineers
about a year. Only AT is in a good position to port the OS because AT
owns the OS source code. Unfortunately, just like Commodore
management, AT doesn't seem to be interested. Maybe someone could get
a license to port the OS from AT.
If Commodore management had reinvested in Amiga development 3..4 years
ago, instead of pocketing the profits and scarpering, I think the
first Amiga RISC consoles would have been out months or even years
ago.
Imagine a RISC Amiga in an A500/A1200 style box running AmigaDOS,
outperforming standard 486/Pentium machines for CPU, graphics and
sound, while costing less, with, say 4MB RAM standard (expandable to
64MB with a couple of 32MB SIMMs, say), hard disk, TCP/IP, universal
multimedia viewer (including WWW browser and mpeg player), a few
bundled awe-inspiring games, maybe a bundled 680x0 emulator, and
optional CDROM. With proper marketing (and maybe even without), such
a system would sell millions today, IMHO.
A higher-end model would have PCI and Zorro3 (Zorro4?) slots and maybe
a higher MHz CPU. Developer tools and docs, C (C++?) compiler,
application builder, hardware docs, etc, would all be available on
CDROM (docs browsed with universal viewer).
Restarting development now is more difficult because years have been
lost and future competition will be fierce (Bebox, playstations,
future PCs, etc) but at least the OS only has to be ported, not
written from scratch.
--
Peter McGavin. (p.mc...@irl.cri.nz)
Paul Chan
>In a message of 02 Oct 95 Paul Chan wrote to All:
>
> >> Its a vicious circle - if we stick with stuff that works on a bog std.
> >> 500, it's gonna impress no-one - they'll take one look and buy a PC (or a
> >> play station, but I don't think we can compete in the games-only arena,
> >> with an 030 + mem costing around the same as a Play station..)
>
> PC> Bad news, I agree. How many Amiga owners already have expansions? They
> PC> are less likely to change platforms, and more likely to buy a game that
> PC> ekes out every last MIP in their system - games like this can impress
> PC> people. We only need a few PC titles to appear on our system (and work
> PC> almost identically) and we might have a market.
>
>Why haven't there appeared any X-Wing or Tie-Fighter clones on Amiga? These
>games are a lot -less- processor intensive than a DOOM-clone and not all that
>hard to program. In these kind of games the most graphics you are most likely
>to see is one or two (gouraud shaded) fighters at medium distance and some
>coloured lines indicating enemy and friendly fire at large distances. In other
>words, the most time consuming thing would be to render the 3d-star background.
> With a C2P routine which only converts parts which have changed this could
>easily run at a 2-3 times higher frame-rate than the TextDemo engine runs at
>now (that would be 15 FPS on a 3000/25, full-screen 320x240 1x1 -- with the
>cockpit 'on' the screen would be reduced to almost half that and run even
>faster).
Basically, the Amiga needs a special programming team, not just some guy
tweaking the C code, to get the best of it`s custom chips. Right now, I suspect
that all the decent programmers are being assigned to PC CDROM and superconsole
development with the flunkeys doing ports. Nobody is bothering with the Amiga
because it`s sales are far too low. And that`s because people don`t tend to buy
the trashy games that have been appearing for the last few months, plus the
exit of many gamers to the bright-lights big-city PC/superconsole. The vicious
circle srikes back. Only AT can drag us out of this mess:
The have 2 options:
1. Make a machine that is very powerful for the same/less as a superconole and
advertise it right.
2. Form a games software house and get some decent games written (probably at
an initial loss). This would be true if the next Amiga on sale had the same
poor prie/performance ratio as the current A1200.
James McArthur
: "Jyrki Saarinen" <jsaa...@kone.fipnet.fi> writes:
: >> I think moving up the 020..030..040..060 ladder will barely keep up
: >> with the competition and is only a temporary solution anyway.
: >
: >Agreed.
: >
: >> The new minimum level should be RISC as soon as possible. Ideally the
: >> OS should already be ported and brilliant new hardware should already
: >> be designed.
: >
: >I think some 3rd party developer could make a PPC card for
: >A1200 for example and have a 68020 emulator in ROM. Then
: >later more and more native software would come..
: Any takers? I think emulating the OS would be difficult because of
: the custom chips. Porting it would be much more practical IMO, but
: would probably still take a small team of good software engineers
: about a year. Only AT is in a good position to port the OS because AT
: owns the OS source code. Unfortunately, just like Commodore
: management, AT doesn't seem to be interested. Maybe someone could get
: a license to port the OS from AT.
Arent the people behind the Cyberstorm 060 trying to get a PPC
chip into that accelerator. An issue of the Australian Commodore
and Amiga Review suggested somthing like this a couple of months
ago, but I've found no other reference to it.
: If Commodore management had reinvested in Amiga development 3..4 years
: ago, instead of pocketing the profits and scarpering, I think the
: first Amiga RISC consoles would have been out months or even years
: ago.
: Imagine a RISC Amiga in an A500/A1200 style box running AmigaDOS,
: outperforming standard 486/Pentium machines for CPU, graphics and
: sound, while costing less, with, say 4MB RAM standard (expandable to
: 64MB with a couple of 32MB SIMMs, say), hard disk, TCP/IP, universal
: multimedia viewer (including WWW browser and mpeg player), a few
: bundled awe-inspiring games, maybe a bundled 680x0 emulator, and
: optional CDROM. With proper marketing (and maybe even without), such
: a system would sell millions today, IMHO.
One word: SOFTWARE
Look at the PC; workers tend to buy what they have in their office so
that they can do work at home. For any new computer to be successful it
is going to have to be compatible with other platforms software.
Companies are either going to have to make Amiga specific versions, or
the software would have to be emulated. All the advantages in the world
wont sell a computer nowadays if it doesnt have any (prospect?) of getting
software.
IMO, AT should get software developers on side first, then start to
develop brand new Amiga systems.
: A higher-end model would have PCI and Zorro3 (Zorro4?) slots and maybe
: a higher MHz CPU. Developer tools and docs, C (C++?) compiler,
: application builder, hardware docs, etc, would all be available on
: CDROM (docs browsed with universal viewer).
I like this idea :-)
: Restarting development now is more difficult because years have been
: lost and future competition will be fierce (Bebox, playstations,
: future PCs, etc) but at least the OS only has to be ported, not
: written from scratch.
Hey, if Apple can make System 7.5 run on the PPC, then AT can make
the AmigaOS run on a PPC too :-)
: --
: Peter McGavin. (p.mc...@irl.cri.nz)
--
___ ___ ______ ______ ______ ______ ___ ___ ______
/ / / // ___ \ / ___ \ / ___ \ /\ / ___ \ / / / // ___ \
/ /__/ // / // / // / // // / // /__/ // / /
\___ // / // / // / // // / / \___ // / /
/ // / // / // / // // / / / // / /
\__/ \______/ \______/ \______/ \/ \______/ \__/ \______/
Peter McGavin
>Imagine a RISC Amiga in an A500/A1200 style box running AmigaDOS,
>outperforming standard 486/Pentium machines for CPU, graphics and
>sound, while costing less, with, say 4MB RAM standard (expandable to
>64MB with a couple of 32MB SIMMs, say), hard disk, TCP/IP, universal
>multimedia viewer (including WWW browser and mpeg player), a few
>bundled awe-inspiring games, maybe a bundled 680x0 emulator, and
>optional CDROM. With proper marketing (and maybe even without), such
>a system would sell millions today, IMHO.
>
>a higher MHz CPU. Developer tools and docs, C (C++?) compiler,
>application builder, hardware docs, etc, would all be available on
>CDROM (docs browsed with universal viewer).
>
>lost and future competition will be fierce (Bebox, playstations,
>future PCs, etc) but at least the OS only has to be ported, not
>written from scratch.
I just want to add that AT don't seem to appreciate the opportunity
they have, now they own the rights to the Amiga's OS. IMHO, the OS
was Commodore's most valuable asset when they sold out.
For developing a new state-of-the-art home computer, the competition
are faced with either:
1) writing a new OS from scratch, which would take maybe 10 years
to reach the maturity of the Amiga's OS (imagine KS1.0 all over
again), or:
2) porting an existing non-proprietary OS, such as Unix, in which
case they are stuck with having to provide large quantities of
extra CPU-power, memory and disk to satisfy the inefficiencies
and bloat of said OS.
On the other hand, AT only have to port the Amiga's OS to first-class
RISC hardware with a relatively small amount of memory and disk for an
(IMHO) hugely successful machine.
Perhaps ESCOM purchased the Amiga's OS rights simply to prevent anyone
else using it against their own line of PCs.
--
Peter McGavin. (p.mc...@irl.cri.nz)
Kjell Irgens
|> G'Day Jyrki Saarinen, in comp.sys.amiga.games you wrote:
|>
|> : > : A1200 for example and have a 68020 emulator in ROM. Then
|> : > : later more and more native software would come..
|> : >
|> : > possible.
|>
|> : Now I dont get it. How these two things are related?
|> : PPC board had to contain a 68020-emulator on ROM though.
|> : And AT and 3rd party developers should agree how
|> : PPC things would be made..
|>
|> possible, it'll run too hot", but now there is an A1200/040 board.
...because Motorola suddenly released new less power consuming 040 chips.
-------- _ -------------------------------------------------- ------------
4k/040 _-^ \ THOR Homepage: http://www.cs.uit.no/~kjelli/thor / Gravity-Force
Amiga \ \ Kjell Irgens <kje...@stud.cs.uit.no> / Race 0: 32.84s
------- \_-^ ----------------------------------------------- ---------------
Peter McGavin
>G'Day Peter McGavin, in comp.sys.amiga.games you wrote:
>: Imagine a RISC Amiga in an A500/A1200 style box running AmigaDOS,
>: outperforming standard 486/Pentium machines for CPU, graphics and
>: sound, while costing less, with, say 4MB RAM standard (expandable to
>: 64MB with a couple of 32MB SIMMs, say), hard disk, TCP/IP, universal
>: multimedia viewer (including WWW browser and mpeg player), a few
>: bundled awe-inspiring games, maybe a bundled 680x0 emulator, and
>: optional CDROM. With proper marketing (and maybe even without), such
>: a system would sell millions today, IMHO.
>
Once the OS is ported, it shouldn't take much effort to port major
existing Amiga applications: editors, paint programs, graphics
utilities, etc. Maybe applications would run out-of-the-box using a
680x0 emulator, otherwise they would have to be recompiled to native
code, perhaps with minor changes. Whatever, a sensible amount of
compatibility between OS versions is very important.
It is unlikely that hardware-banging games would work, but who needs
those when game developers are rushing to support a machine with
better specs than a Pentium PC in every way? :-) Compare with the
BeBox situation, where everything is being written/ported from scratch
(or is it?)
>IMO, AT should get software developers on side first, then start to
>develop brand new Amiga systems.
Good idea. They need to do this by boosting confidence. I want to
see a demonstration that something incredible is in the works.
--
Peter McGavin. (p.mc...@irl.cri.nz)
Student Mark Smit (SP+T H.W)
-----<SNIP>------
>Compare with the BeBox situation, where everything is being written/ported from
>scratch
> (or is it?)
-----<SNIP>------
> Peter McGavin. (p.mc...@irl.cri.nz)
Hmm, lots of talk about Beboxes these days. Have I missed something out or what.
I`ve never heard of the thing before.. What`s it?? Is this going to be my new
girlfriend or what. Does she have some BIIIIG Boobies :) (or, who can give me some
specs...)
Mark
Sam Taylor
I known these kind of discussions are usually pretty pointless ... but
they're also fun! So here's my opinions on what should go in the next
1200-style box.
1. CPU == 040SLC
It has to be the 040SLC and if possible it has to be clocked faster
than 25MHz. The 030 just doesn't cut it anymore, and although there
will probably be greater compatibility headaches with the 040 (games
which don't figure on having a data cache) the 040 is the only
serious option. If AT are going to make a RISC Amiga (yes please...)
they are going to need a reasonable amount of time. Realistically,
that means the new-1200-style machine will need to be on the shelves
for probably at least 18 months. In 18 months time the 040 is going
to look pretty tired, but the 030 would look a complete joke. Also,
by using the 040 now, they can increase the clock speed over time to
give the thing a little more life. Even now an 030 machine would
have to be clocked at 50 MHz to attract any interest at all, which
2. MEMORY == 2MB Chip, 2MB Fast, 1 empty SIMM slot
2MB is fine for Chip ram, but we *must* have at least 2MB of fast ram
on board, with at least one spare SIMM slot on the motherboard. That
may piss off the expansion board manufacturers, but there are other
things they can make. The fast memory architecture also needs to be
*efficient* and should support burst mode.
3. CUSTOM/GRAPHICS CHIPSET == AGA + Chunky mode
We can't expect much change to the AGA, but I think the addition of
a chunky mode is essential. (As an aside, anyone known what became
of the AA+ chipset - I know Commodore dropped it to concentrate on
the AAA, but what stage was it at?) Lots of people have argued that a
chunky display mode is unnecessary because a fast CPU can do c2p while
copying from fast ram to chip ram. While this is certainly true, I
still think a chunky display mode is absolutely essential for a number
of reasons...
First, the time used to do the c2p conversion could be better
used for other computations. I agree that even with a chunky mode it
will be best to render to a fast buffer and then copy to chip ram.
However, the time between chip writes which is used to do the c2p could
be better used for other things. Also, with a chunky mode an 040 could
use MOVE16 to do the copying.
Second, it is a completely trivial modification to the hardware and
it does not necessarily entail large modifications to the OS. Clearly,
making graphics.library aware of chunky screens *would* involve a
considerable amount of work. However, chunky mode is going to be used
almost exclusively for games, of which very few (if any) actually use
graphics.library. Add a tag to OpenScreen, and stop chunky screens
from being public : that's all that's required in the first instance.
Third, programmers shouldn't have to do this stuff! We need to make
programming the Amiga as easy as possible, and it's stupid to make every
programmer re-invent the c2p wheel. Admittedly a standard c2p library
would also a address this, but a native chunky mode will have to come
sooner or later (if there is a later...). By including it now you have
the opportunity to start adapting graphics.library to use chunky screens
as well as planar ones. It would also be a good opportunity to
seriously look at RTG
4. OPERATING SYSTEM == V40 + tweaks + extras from the PD + TCP?
I don't think there is much need to make extensive changes to the OS.
A few performance tweaks and bug fixes would be nice, but I think
that it would be reasonable to plan the next OS release along the
lines Apple used with Sytem 7.5. Basically, they just incorporated
a set of new utilities and tools that everyone had been using anyway.
This approach would be ideal for AmiTech and might that the form of:
o A new set of icons, colours and fonts for the workbench, to give
the OS a new distinctively AmiTech feel ... maybe just using the
Magic icons would be enough
o Import a few essential add-ons from companies and the PD. A few
extra datatypes (GIF, JPEG, maybe CDXL...), a few useful utilities
from the public domain (maybe KingCON, PriMan, PrintMan - with the
authors permission of course), and perhaps licence Digita's
TrueType font libraries (nobody uses Compugraphic fonts these days)
o A lean, efficient TCP/IP stack, optimised for use over SLIP/PPP
set of client applications, but new machines will live and die
by their ability to network. This is one area that the Amiga could
really win over the Playstation/Saturn/3DO
5. STORAGE, EXPANSION AND THE CASE
The other crucial issue is HD/CD drives. Problem is that the little
keyboard computers are still popular, and are a good way to meet the
challenge of the consoles. So, since Escom are good at making cases,
put the motherboard in two different cases:
o A "keyboard" case about the size of the 1200 with room for a
2.5 inch EIDE HD, and with a 500-style side expansion bus. AT
should sell this model without an HD, and let third party people
supply the drives
o A 1000-style "pizza box" case which comes standard with a 3.5 inch
EIDE hard disk, has an internal power supply *and* has room for an
optional CD-ROM. A couple of Zorro III slots would be nice, but
that may not be practical, in which case the same side expansion bus
should be used, and the two buses should match (same height etc)
Amiga owners have been screaming for this for years! It only involves
AmiTech manufacturing two models, gives them much more flexibilty in
marketing and allows them to use cheaper 3.5 inch hard disks. The
expansion bus should replace PCMCIA : an experiment which has failed on
the Amiga (even though I think it will succeed everywhere else) It
would also be a good idea to sell the pizza-box case by itself, so
people can upgrade their keyboard cases over time.
6. OTHER OBVIOUS STUFF == HD Floppy, a Clock, a CD-32 controller?
I understand why the 1200 didn't originally have an HD floppy, but I
still cannot believe that Commodore didn't include a bloody clock!
Nowadays these are both simply essential. I also think including a
CD-32 controller as standard would be a good idea. Although the
machine won't be targetted directly against the consoles it's still
going to have to do it's fair share of game playing, and it should
do this "out of the box". Single button joysticks are too limiting
these days - a multibutton controller is a must
None of this seems hard to me, and it would be a competitive piece of
equipment. A price around US$750 == AUS$900 == 400POUNDS == ??1000DM??
for the keyboard case doesn't sound unreasonable from either the consumer
or producer perspectives. Add maybe 200POUNDS for the pizza-box machine
with a reasonable sized hard disk.
Guess I better go do some work now ... but what do people think of these
ideas?
____ __ ___ ___
/ __) / \ | \/ | Sam.T...@anu.edu.au
\__ \ / () \ | \ / | Distributed High Performance Computing Project
(____/(__/\__)|_|\/|_| Australian National University
Paul Chan
>Dear All -
>
>I known these kind of discussions are usually pretty pointless ... but
>they're also fun! So here's my opinions on what should go in the next
>1200-style box.
I love spec. discussions, especially if AT people are lurking - then maybe
they`ll make an Amiga we will buy!
>
>1. CPU == 040SLC
> It has to be the 040SLC and if possible it has to be clocked faster
> than 25MHz. The 030 just doesn't cut it anymore, and although there
> will probably be greater compatibility headaches with the 040 (games
> which don't figure on having a data cache) the 040 is the only
> serious option. If AT are going to make a RISC Amiga (yes please...)
> they are going to need a reasonable amount of time. Realistically,
> that means the new-1200-style machine will need to be on the shelves
> for probably at least 18 months. In 18 months time the 040 is going
> to look pretty tired, but the 030 would look a complete joke. Also,
> by using the 040 now, they can increase the clock speed over time to
> give the thing a little more life. Even now an 030 machine would
> have to be clocked at 50 MHz to attract any interest at all, which
Yes. Agreed. The 030 @ 50Mhz is the only CPU that would attract me, but would
*still* be dismissed by mainstream PC/Mac mags. The 040, though expensive, is a
marketeers dream. Consider adverts like:
15 times the power of the A1200.
The same CPU as the Mac Performa 630.
Equivalent to 486DX250 or 486DX266
Videographers dream...
etc etc.
Placing an 040 chip into a low end Amiga would quantum leap the state of play
int the Mac yard (possibly long with prices). For an 040 Amiga to work, a box
contining this would have to cost 500 UK pounds or less. This would compete
favourably with the Mac 630.
What AT should do then is:
1. Start development of a few CD 040 games off their own budget.
2. Reinstate a cheap CD license policy (say, 10% of RRP per CD - notice, I`m
not going for a flat fee) for commercial games and applications.
3. Use that CD profit to balance the cost of selling the 040 Amiga at cost.
>
>2. MEMORY == 2MB Chip, 2MB Fast, 1 empty SIMM slot
> 2MB is fine for Chip ram, but we *must* have at least 2MB of fast ram
> on board, with at least one spare SIMM slot on the motherboard. That
> may piss off the expansion board manufacturers, but there are other
> things they can make. The fast memory architecture also needs to be
> *efficient* and should support burst mode.
I would go for 6Mb of RAM, with all of that on the motherboard. Plus an extra
SIMM socket. An A1200 slot should remain for 060 expansion. Motherboard
integration is cheaper, but perhaps putting the 040 onto an A1200 card would be
more cost effective to the purchaser in the long run. They could retain an 020
on the motherboard to make it cheaper and for compatibility.
The new machine should have a Cybergfx compatible SVGA chip with 1Mb video RAM
(expandable to 2Mb). Higher end models should include space for a gfx card
through ZIII slots. IMHO, this feature combined with a trade up offer for
A1200`s, should make A1200 ownerts trade their machines in.
The result then is more "A1300"`s on the street, and AT have lots of spare
A1200 mtherboard with which to service existing A1200 owners. Sales figures of
the new mchine should boost rapidly.
>
>4. OPERATING SYSTEM == V40 + tweaks + extras from the PD + TCP?
> I don't think there is much need to make extensive changes to the OS.
> A few performance tweaks and bug fixes would be nice, but I think
> that it would be reasonable to plan the next OS release along the
> lines Apple used with Sytem 7.5. Basically, they just incorporated
> a set of new utilities and tools that everyone had been using anyway.
> This approach would be ideal for AmiTech and might that the form of:
> o A new set of icons, colours and fonts for the workbench, to give
> the OS a new distinctively AmiTech feel ... maybe just using the
> Magic icons would be enough
> o Import a few essential add-ons from companies and the PD. A few
> extra datatypes (GIF, JPEG, maybe CDXL...), a few useful utilities
> from the public domain (maybe KingCON, PriMan, PrintMan - with the
> authors permission of course), and perhaps licence Digita's
> TrueType font libraries (nobody uses Compugraphic fonts these days)
> o A lean, efficient TCP/IP stack, optimised for use over SLIP/PPP
> set of client applications, but new machines will live and die
> by their ability to network. This is one area that the Amiga could
> really win over the Playstation/Saturn/3DO
This is a case of funding a few PD programmers to bring their stuff up to a WB
release standard. Network games and serial games are much more likely between
friends (not office workers) if the base unit is cheap. IMHO, Internet and
network games are a good lever arm/advertising point for the Amiga.
>
>5. STORAGE, EXPANSION AND THE CASE
> The other crucial issue is HD/CD drives. Problem is that the little
> keyboard computers are still popular, and are a good way to meet the
> challenge of the consoles. So, since Escom are good at making cases,
> put the motherboard in two different cases:
> o A "keyboard" case about the size of the 1200 with room for a
> 2.5 inch EIDE HD, and with a 500-style side expansion bus. AT
> should sell this model without an HD, and let third party people
> supply the drives
ERK! NO! The A1200 forces people to spend more than they should fo expansions:
2.5" drives instead of 3.5", SCSI CD drives when (with a little software and an
internal drive bay) cheaper IDE CD drives would work. External boxes for
everything, when a big-box could overall be better for upgraders...
> o A 1000-style "pizza box" case which comes standard with a 3.5 inch
> EIDE hard disk, has an internal power supply *and* has room for an
> optional CD-ROM. A couple of Zorro III slots would be nice, but
> that may not be practical, in which case the same side expansion bus
> should be used, and the two buses should match (same height etc)
At the low end, if you include a gfx chip to boost the gfx performace, you
don`t really need ZIII slots. There should be a later model further up the
range with these features.
> Amiga owners have been screaming for this for years! It only involves
> AmiTech manufacturing two models, gives them much more flexibilty in
> marketing and allows them to use cheaper 3.5 inch hard disks. The
> expansion bus should replace PCMCIA : an experiment which has failed on
> the Amiga (even though I think it will succeed everywhere else) It
> would also be a good idea to sell the pizza-box case by itself, so
> people can upgrade their keyboard cases over time.
PCMCIA is becoming interesting. I for one could have done with one on my A4000
when I tried to use my Prograb as it locks up my parallel port so I can`t print
from it!
>
>6. OTHER OBVIOUS STUFF == HD Floppy, a Clock, a CD-32 controller?
> I understand why the 1200 didn't originally have an HD floppy, but I
> still cannot believe that Commodore didn't include a bloody clock!
> Nowadays these are both simply essential. I also think including a
> CD-32 controller as standard would be a good idea. Although the
> machine won't be targetted directly against the consoles it's still
> going to have to do it's fair share of game playing, and it should
> do this "out of the box". Single button joysticks are too limiting
> these days - a multibutton controller is a must
Forget the CD32 controller, just license the Playstation joypad ports. You
don`t have to persuade people to make control pads for your machine then -
they`ll use the 10 button one that the PS has.
>
>None of this seems hard to me, and it would be a competitive piece of
>equipment. A price around US$750 == AUS$900 == 400POUNDS == ??1000DM??
>for the keyboard case doesn't sound unreasonable from either the consumer
>or producer perspectives. Add maybe 200POUNDS for the pizza-box machine
>with a reasonable sized hard disk.
>
>Guess I better go do some work now ... but what do people think of these
>ideas?
I`ll post my suggestion later.
MARC EDWARD FORRESTER
>-------- _ -------------------------------------------------- ------------
>4k/040 _-^ \ THOR Homepage: http://www.cs.uit.no/~kjelli/thor / Gravity-Force
>Amiga \ \ Kjell Irgens <kje...@stud.cs.uit.no> / Race 0: 32.84s
>------- \_-^ ----------------------------------------------- ---------------
Oh, don't encourage them.. It's a terrible logo. :>
Jyrki Saarinen
> Any takers? I think emulating the OS would be difficult because of
> the custom chips. Porting it would be much more practical IMO, but
Why? Just run everything off the 68020 emulator.
> Restarting development now is more difficult because years have been
> lost and future competition will be fierce (Bebox, playstations,
> future PCs, etc) but at least the OS only has to be ported, not
> written from scratch.
BeBox looks very nice. It has propably even more problems
than the Amiga has. What do you think about BeBox?
Jyrki Saarinen
> the AAA, but what stage was it at?) Lots of people have argued that
> a chunky display mode is unnecessary because a fast CPU can do c2p
> while copying from fast ram to chip ram. While this is certainly
> true, I still think a chunky display mode is absolutely essential
> for a number of reasons...
Yes it would quite unnecessary if fast ram is present. It not,
maybe then.. Speeding up chip ram would be essential and
chunky could be added, too.
> First, the time used to do the c2p conversion could be better
> used for other computations. I agree that even with a chunky mode
> it will be best to render to a fast buffer and then copy to chip ram.
C2p conversion is done almost freely on fast 030 and 040,
and almost freely on slower CPUs with blitter assisting.
A1200/FAST can do very speedy 2x2 chunky 160x128, chunky
buffer in fast ram and doing pass1 while copying.
160x128 is really enough since you cant expect that a 020/14
would be able to render a 320x256 screen with decent frame rate.
Depends on the games though. But I am talking about modern
tmapping (crap) games.
> However, the time between chip writes which is used to do the c2p
> could be better used for other things.
Hmm. Like what? The picture have been rendered in that stage
of the process already..
john enger
: Basically, the Amiga needs a special programming team, not just some guy
: tweaking the C code, to get the best of it`s custom chips. Right now, I suspect
On the Archimedes, which has a *much* smaller software base than Amy, there
is a company (Krisalis I think?) who port everybody elses games and take
advantage of the machines strengths. For example, the arc version of SC2000
is reputedly the best. Now if this can be done on the arc, it can certainly
be done on Amy. Of course, the arc is only jus gettting stuff like lotus2
and cannon fodder, but it does get them eventually...
Perhaps an enterprising team could do something similar for Amy?
========================= john....@bbsrc.ac.uk =====================///===
A1200, 40 Mhz '030, 2+4Mb RAM, 80Mb HD, 1960 MSync Monitor. ///
-------------------------------------------------------------------\XX/-----
Paul Chan
>Paul Chan (paul) wrote:
>: Basically, the Amiga needs a special programming team, not just some guy
>: tweaking the C code, to get the best of it`s custom chips. Right now, I suspect
>
>On the Archimedes, which has a *much* smaller software base than Amy, there
>is a company (Krisalis I think?) who port everybody elses games and take
>advantage of the machines strengths. For example, the arc version of SC2000
>is reputedly the best. Now if this can be done on the arc, it can certainly
>be done on Amy. Of course, the arc is only jus gettting stuff like lotus2
>and cannon fodder, but it does get them eventually...
>
>Perhaps an enterprising team could do something similar for Amy?
Krisalis have a vitual monopoly in that market, don`t they? And slavering Arc
owners waiting to snap up copies! Yes, I suppose a dedicated conversions team
would be useful to have, but we need original games not 12 month old PC ones,
too.
MARC EDWARD FORRESTER
>Krisalis have a vitual monopoly in that market, don`t they? And slavering Arc
>owners waiting to snap up copies! Yes, I suppose a dedicated conversions team
>would be useful to have, but we need original games not 12 month old PC ones,
>too.
Well, not necessarily. I'd probably buy TIE fighter, for example.
Or Star Control II, or Ultima VII, or Trial by Fire, or..
New games are great, but good games don't age -that- fast.
Juergen Rally Fischer
About new 1200:
>>1. CPU == 040SLC
I don't know what that thingy costs, but you won't be able to keep
in old A1200 prive range => you talk about something different than
a new A1200.
>>2. MEMORY == 2MB Chip, 2MB Fast, 1 empty SIMM slot
I fully agree what you said about this.
>>3. CUSTOM/GRAPHICS CHIPSET == AGA + Chunky mode
Chunky mode added to AGA (AA+): forget it. Too much costs and you
get almost nothing, compared to software solutions.
AGA is not the big breaker of chunky games, you just need a
good cpu. Yes, a big cpu looses a bit (beyond 1 frame) for
copying the typical 320x160 screen. So it'd be good if you
could optionally add...
... Chunky mode on a gfx-card sitting in a Zorro-Slot:
Just use one of those VGA chipsets, nad make the games
support gfx-cards.
Well, Zorro in a keyboard-style-computer, I don't know how to do it...
creativity needed.
>>4. OPERATING SYSTEM == V40 + tweaks + extras from the PD + TCP?
Most important for V40 would be MMU support, but who is to write
it...
>>5. STORAGE, EXPANSION AND THE CASE
A bizza-box case version would be imho a good idea,
but maybe they won't want to risc anything and won't give it a try...
Zorro instead othe A1200-cpu-bus, as I said, you need ideas how
to do it...
>>6. OTHER OBVIOUS STUFF == HD Floppy, a Clock, a CD-32 controller?
Should be no problem do give HD floppy optionally, so DD for those
who aim to the cheapest possible config (important market for Amiga).
CD-32 Controller: sell it separate.
>>equipment. A price around US$750 == AUS$900 == 400POUNDS == ??1000DM??
>>for the keyboard case doesn't sound unreasonable from either the consumer
too much. 1000DM for a keyb-case without Monitor... naah.
Anything above 800DM will IMHO give you a sure flop.
The main thing for the keyb-version is the possibility to do
something like 2x1 DOOMing for beyond 800DM, if this is possible
you got a real chance.
>>or producer perspectives. Add maybe 200POUNDS for the pizza-box machine
>>with a reasonable sized hard disk.
Yes, if the pizza-box version got good concept (slotwise etc)
and not to small components (cpu,mem,HD etc) a price of DM1200 could be
acepted.
- - - //- - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
// A1201 HD260 fisc...@Informatik.TU-Muenchen.DE
\X/ ^--fastmem=2xSpeed (Juergen "Rally" Fischer) =:)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
James McArthur
: Paul Chan (paul) wrote:
: : >Paul Chan (paul) wrote:
: : >: Basically, the Amiga needs a special programming team, not just some guy
: : >: tweaking the C code, to get the best of it`s custom chips. Right now, I suspect
: : >
: : >On the Archimedes, which has a *much* smaller software base than Amy, there
: : >is a company (Krisalis I think?) who port everybody elses games and take
: : >advantage of the machines strengths. For example, the arc version of SC2000
: : >is reputedly the best. Now if this can be done on the arc, it can certainly
: : >be done on Amy. Of course, the arc is only jus gettting stuff like lotus2
: : >and cannon fodder, but it does get them eventually...
: : >
: : >Perhaps an enterprising team could do something similar for Amy?
: : Krisalis have a vitual monopoly in that market, don`t they? And slavering Arc
: : owners waiting to snap up copies! Yes, I suppose a dedicated conversions team
: : would be useful to have, but we need original games not 12 month old PC ones,
: : too.
: Oh I agree, we definately need original stuff. But it would be nice to have
: a dedicated conversions team *as well*
Whats stopping a group of Amiga owners from starting up a little programming
group such as this and converting a couple of games?
: ========================= john....@bbsrc.ac.uk =====================///===
: A1200, 40 Mhz '030, 2+4Mb RAM, 80Mb HD, 1960 MSync Monitor. ///
: -------------------------------------------------------------------\XX/-----
--
___ ___ ______ ______ ______ ______ ___ ___ ______
/ /__/ // ___ \ / _____\_/ _____\_/\_/ ___ \/ /__/ // ___ \
\____ // / . // / jam...@it.ntu.edu.au / /\____ // / . /
Paul Chan
>
>: Oh I agree, we definately need original stuff. But it would be nice to have
>: a dedicated conversions team *as well*
>
>Whats stopping a group of Amiga owners from starting up a little programming
>group such as this and converting a couple of games?
Probably the fact that the originators would want a license fee if someone
wanted to convert, say, Wing Commander 3 to the Amiga.
MARC EDWARD FORRESTER
>>Whats stopping a group of Amiga owners from starting up a little programming
>>group such as this and converting a couple of games?
>
>Probably the fact that the originators would want a license fee if someone
>wanted to convert, say, Wing Commander 3 to the Amiga.
Well how do Krysalis do it? Sales commission?
Neil Brewitt
In article <8141989...@hchworth.demon.co.uk> Paul Chan <paul> writes:
> jam...@morinda.it.ntu.edu.au (James McArthur) wrote:
>
> >Whats stopping a group of Amiga owners from starting up a little programming
> >group such as this and converting a couple of games?
>
> Probably the fact that the originators would want a license fee if someone
> wanted to convert, say, Wing Commander 3 to the Amiga.
But to the owner of the WC3 copyright, actually making the license
affordable to an amiga conversion team => sales => more money from an
existing title for no work.
Business is about coming to a compromise about making as much money as you
can for as little effort.
Neil.
--
**** ne...@melkfri.demon.co.uk **** Neil Brewitt
*||||||||||||||||||||||||||||||||* IRC: saggy
** Musician ***** Manchester,UK **
COLMAN CP
>Date: Sat, 07 Oct 95 15:12:38 GMT+1
>Subject: Amiga Spec discussion
>From: john.h...@grafix.xs4all.nl (John Hendrikx)
>Why haven't there appeared any X-Wing or Tie-Fighter clones on Amiga? These
>games are a lot -less- processor intensive than a DOOM-clone and not all that
>hard to program. In these kind of games the most graphics you are most likely
>to see is one or two (gouraud shaded) fighters at medium distance and some
>coloured lines indicating enemy and friendly fire at large distances. In
>other
>words, the most time consuming thing would be to render the 3d-star
>background.
> With a C2P routine which only converts parts which have changed this could
>easily run at a 2-3 times higher frame-rate than the TextDemo engine runs at
>now (that would be 15 FPS on a 3000/25, full-screen 320x240 1x1 -- with the
>cockpit 'on' the screen would be reduced to almost half that and run even
>faster).
>Grtz John
>-----------------------------------------------------------------------
> John.H...@grafix.xs4all.nl TextDemo/FastView/Etc... development
>-----------------------------------------------------------------------
>-- Via Xenolink 1.98b1, XenolinkUUCP 1.1
You don't need to do c2p either - direct to planar goraud shading is
quite quick. Especially with lokk up tables for low end machines
(e.g. Pygmy Projects G-Force demo)
Chris Colman (Findus)
Sam Taylor
support a "chunky" display mode "Jyrki Saarinen" <jsaa...@kone.fipnet.fi>
wrote:
> C2p conversion is done almost freely on fast 030 and 040,
> and almost freely on slower CPUs with blitter assisting.
Well it isn't FREE - the CPU is still doing work. What you mean is that
a quick 030/040 can do the conversion almost as fast as it can copy from
fastmem to chipmem. That still isn't free : it's just one less bottleneck!
> > However, the time between chip writes which is used to do the c2p
> > could be better used for other things.
>
> Hmm. Like what? The picture have been rendered in that stage
> of the process already..
Well rendering is only one part of any game : there are plenty of other
useful things we could be doing while waiting for the next write to chipmem
to take place. The arguement that we don't need a chunky mode seems to be
based on the notion that the CPU would be essentially idle while waiting to
write a fastmem buffer into chipmem. But this is silly! There are tonnes
of things the CPU could be doing in the time we copy our fast buffer into
chipmem. You could even double buffer the fastmem buffer ... render into
one while copying the other to chip.
There are other convincing arguements for having a chunky mode. It's such
a trivial thing, and if AT can't do this then they've got no chance of ever
building a RISC Amiga!
<DREAMING>
Further, it will be essential if Amigas are to ever have a native
24-bit mode. The easiest thing to do to make up for the fact that there
are only eight plane pointers in the bitmap structure, would be to use
chunky planes rather than bit planes. Have a single plane for 256 colour
displays, three planes for 24-bit with one for each of the RGB components.
Would also be easy to have multiple playfields, each with an independent
chunky plane. I'm not suggesting that the 1300 should be able to do this
but if it supported a single chunky plane display, it would make it
easier to integrate this more sophisticated hardware in subsequent models
In fact, now that I think of it, have a chunky display mode might also
be the key to finding a cure for cancer, and would certainly bring
everlasting world peace ..... !
</DREAMING>
Seriously though, there are no good reasons for NOT have a chunky display
mode. Nobody could claim that the Amiga would be worse off for supporting
both chunky and planar displays! For the first new machine AT won't be able
to make major modifications to the display hardware. However, with a chunky
mode, and perhaps a slightly tweaked blitter, the basic AGA is still an
acceptable display system : not earth shattering but not too bad. With a
bit of fast ram and an 040SLC that would be a machine that would sell well
enough for AT to develop a RISC Amiga.
----------------------------------------------------------------------------
Sam.T...@anu.edu.au
Distributed High Performance Computing Project
Australian National University
----------------------------------------------------------------------------
Jyrki Saarinen
> > C2p conversion is done almost freely on fast 030 and 040,
> > and almost freely on slower CPUs with blitter assisting.
>
> Well it isn't FREE - the CPU is still doing work. What you mean is that
> a quick 030/040 can do the conversion almost as fast as it can copy from
> fastmem to chipmem. That still isn't free : it's just one less
> bottleneck!
Well. The choices:
1) rendering to chip ram if we had chunky mode --> slow
2) rendering to fast ram, chunky2planar while copying --> fast
> There are other convincing arguements for having a chunky mode. It's
> such a trivial thing, and if AT can't do this then they've got no chance
> of ever building a RISC Amiga!
Now you are way off. Hacking AGA further is non-trivial and
slow.
Making a PPC based Amiga with PCI bus and off-the-self parts
is very fast.
> Seriously though, there are no good reasons for NOT have a chunky
> display mode. Nobody could claim that the Amiga would be worse off for
> supporting both chunky and planar displays!
I did not say the next Amigas should not have chunky mode, did I?
I just said chunky mode in AGA machines would not help very much.
It would only help chip-only A1200s, and not much, a little.
> to make major modifications to the display hardware. However, with a
> chunky mode, and perhaps a slightly tweaked blitter, the basic AGA is still an
> acceptable display system : not earth shattering but not too bad. With
> a bit of fast ram and an 040SLC that would be a machine that would sell
> well enough for AT to develop a RISC Amiga.
Agreed, but developing a PowerPC Amiga is propably easier and
faster process (off-the-self parts, motherboard support logic,
L2 caches, SCSI, PCI, video, sound) than hacking AGA
further.
Niki Hemmings
Jyrki Saarinen <jsaa...@kone.fipnet.fi> wrote:
>> Seriously though, there are no good reasons for NOT have a chunky
>> display mode. Nobody could claim that the Amiga would be worse off for
>> supporting both chunky and planar displays!
>
>I did not say the next Amigas should not have chunky mode, did I?
>I just said chunky mode in AGA machines would not help very much.
>It would only help chip-only A1200s, and not much, a little.
No, I've got to dissagree here. Ok, for all us coders who know all about
chunky conversion methods, rendering to fastmem, blitterscreen et al, a
chunky mode wouldn't make much difference. But, there are an awful lot of
average coders, and moreover, perfectly competent coders who just wouldn't
know of such methods. The lack of a chunky mode has the effect of
immediately putting people off this sort of thing, and it's only when you
look deeper into the problem that you realise it isn't such a big deal.
Just consider this: would ID have said Doom was impossible on the Amiga
if there was a chunky mode available?
Nik.
Juergen Rally Fischer
: I respones to a suggestion by me, that a future Amiga system should
: > C2p conversion is done almost freely on fast 030 and 040,
: > and almost freely on slower CPUs with blitter assisting.
: Well it isn't FREE - the CPU is still doing work. What you mean is that
: a quick 030/040 can do the conversion almost as fast as it can copy from
: fastmem to chipmem. That still isn't free : it's just one less bottleneck!
: > > However, the time between chip writes which is used to do the c2p
: > > could be better used for other things.
: >
: > Hmm. Like what? The picture have been rendered in that stage
: > of the process already..
: Well rendering is only one part of any game : there are plenty of other
: useful things we could be doing while waiting for the next write to chipmem
: to take place. The arguement that we don't need a chunky mode seems to be
: based on the notion that the CPU would be essentially idle while waiting to
: write a fastmem buffer into chipmem. But this is silly! There are tonnes
: of things the CPU could be doing in the time we copy our fast buffer into
: chipmem. You could even double buffer the fastmem buffer ... render into
: one while copying the other to chip.
I see where you go and I see why you call modern c2p methods "not free" ;)
you mean a copy from fast buffer to chipmem in inner loop of the mapper
for example ? well... :)
: There are other convincing arguements for having a chunky mode. It's such
: a trivial thing, and if AT can't do this then they've got no chance of ever
: building a RISC Amiga!
1st thing we got to cry for is fastmem. 2nd is faster cpu. 3rd is improved
gfx system. So, lets start for #1 and #2 first, it's more likely to come
true.
: <DREAMING>
: Further, it will be essential if Amigas are to ever have a native
: 24-bit mode. The easiest thing to do to make up for the fact that there
: are only eight plane pointers in the bitmap structure, would be to use
: chunky planes rather than bit planes. Have a single plane for 256 colour
: displays, three planes for 24-bit with one for each of the RGB components.
huh ? why that ? why separating the rgb bytes ?
: Would also be easy to have multiple playfields, each with an independent
: chunky plane. I'm not suggesting that the 1300 should be able to do this
: but if it supported a single chunky plane display, it would make it
: easier to integrate this more sophisticated hardware in subsequent models
: In fact, now that I think of it, have a chunky display mode might also
: be the key to finding a cure for cancer, and would certainly bring
: everlasting world peace ..... !
The 1300 should rather have faster chipmem access. writebuffer or whatever.
it's more likely to come true...
: </DREAMING>
: Seriously though, there are no good reasons for NOT have a chunky display
: mode. Nobody could claim that the Amiga would be worse off for supporting
except money market crap. They won't give us AAA...
: both chunky and planar displays! For the first new machine AT won't be able
: to make major modifications to the display hardware. However, with a chunky
: mode, and perhaps a slightly tweaked blitter, the basic AGA is still an
: acceptable display system : not earth shattering but not too bad. With a
: bit of fast ram and an 040SLC that would be a machine that would sell well
1st fastmem 2nd cpu. Then chunky. Chunky without improved chipbandwidth is
no use for typical chunky games.
Crying for chunky gives a totally wrong picture.
On 2x2 blitterscreen my A1201 is faster than it could be on a
SuperVGA chunky mode. Just true.
What the A1200 needs is 1st... 2nd...
I hope you see what I mean...
: enough for AT to develop a RISC Amiga.
: ----------------------------------------------------------------------------
: Sam.T...@anu.edu.au
: Distributed High Performance Computing Project
: Australian National University
: ----------------------------------------------------------------------------
Juergen Rally Fischer
: >Grtz John
if the number of polygons isn't too large.
: (e.g. Pygmy Projects G-Force demo)
so why won't G-Force run on my especially low end machine ? ;)
coz you need hi-end mem for the tables ;)
: Chris Colman (Findus)
Sam Taylor
> > Well it isn't FREE - the CPU is still doing work. What you mean is that
> > a quick 030/040 can do the conversion almost as fast as it can copy from
> > fastmem to chipmem. That still isn't free : it's just one less
> > bottleneck!
>
>
> 1) rendering to chip ram if we had chunky mode =A0 -->=A0slow
> 2) rendering to fast ram, chunky2planar while copying --> fast
I agree that #1 is slow, but that isn't what I was suggesting. I suggest
a third alternative:
3) render to fast chunky buffer, copy without c2p to chip buffer - maybe
using something like MOVE16. In the time between chip writes (ie the
time you do c2p in #2) do other work, like some kind of final filtering
or processing of the image, or something else in fast ram --> fastest!
>> There are other convincing arguements for having a chunky mode. It's
>> such a trivial thing, and if AT can't do this then they've got no chance
>>=A0of ever building a RISC Amiga!
>
> Now you are way off. Hacking AGA further is non-trivial and
> slow.
>
> Making a PPC based Amiga with PCI bus and off-the-self parts
> is very fast.
> ...some stuff cut...
> Agreed, but developing a PowerPC Amiga is propably easier and
> faster process (off-the-self parts, motherboard support logic,
> L2 caches, SCSI, PCI, video, sound) than hacking AGA
> further.
Sorry man I can't agree. Building the hardware for a PPC based Amiga would
be relatively simple, but porting the AmigaOS is going to be a nightmare!
When they do move the OS, I hope they will make some serious modifications
(resource tracking, memory protection) but even if they don't it's going to
be a hell of a job. With luck they might be able to licence Apples 680x0
emulator, but even so can you imagine the problems they will have?!? No,
hacking AGA is a picnic compared to that (although I agree its still not
easy)
Anyway, a PPC Amiga will have to have display hardware of it's own, and if
you want any kind of compatibility it will have to look like AGA, or have
an AGA emulation mode. Sooner or later AmigaTech will have to get intimate
with the internals of the AGA : I say start now and keep it simple. The
benefits justify the effort. I mean if they want to encourage some serious
developers back to the Amiga it has to be able to do a few basics, and
these days chunky display hardware is an essential. Remember, this is as
much a political issue as a technical issue. No commercial developer would
accept a statement like "no we don't have chunky hardware, but some guys on
the net reckon this isn't a problem" ...
Jyrki Saarinen
> 3) render to fast chunky buffer, copy without c2p to chip buffer - maybe
> using something like MOVE16. In the time between chip writes (ie the
> time you do c2p in #2) do other work, like some kind of final
It is just MOVE16 is A) a 040 instruction and B) it cant be
used to chip ram.
> be relatively simple, but porting the AmigaOS is going to be a
> nightmare!
Why? AmigaOS is mostly C.
Juergen Rally Fischer
: "Jyrki Saarinen" <jsaa...@kone.fipnet.fi> wrote:
: > > Well it isn't FREE - the CPU is still doing work. What you mean is that
: > > a quick 030/040 can do the conversion almost as fast as it can copy from
: > > fastmem to chipmem. That still isn't free : it's just one less
: > > bottleneck!
: >
: > Well. The choices:
: >
: > 1) rendering to chip ram if we had chunky mode =A0 -->=A0slow
: > 2) rendering to fast ram, chunky2planar while copying --> fast
: I agree that #1 is slow, but that isn't what I was suggesting. I suggest
: a third alternative:
: 3) render to fast chunky buffer, copy without c2p to chip buffer - maybe
: using something like MOVE16. In the time between chip writes (ie the
: time you do c2p in #2) do other work, like some kind of final filtering
: or processing of the image, or something else in fast ram --> fastest!
true. can't think of anything but buffer clearing for stuff like doom, but
true.
: >> There are other convincing arguements for having a chunky mode. It's
: >> such a trivial thing, and if AT can't do this then they've got no chance
: >>=A0of ever building a RISC Amiga!
: >
: > Now you are way off. Hacking AGA further is non-trivial and
: > slow.
not only slow and expensive but not nessesary (read on to see why :)
A1300:
nothing is easier than using some vga chipsets. or implementing on the
A1300 what even hobby manufacturers manage to do (rgb-port donlge).
The question is if it wouldn't be better to get a better cpu for
the money the additional vga/etc stuff would cost (the answer is yes btw ;)
: >
: > Making a PPC based Amiga with PCI bus and off-the-self parts
: > is very fast.
: > ...some stuff cut...
: > Agreed, but developing a PowerPC Amiga is propably easier and
: > faster process (off-the-self parts, motherboard support logic,
: > L2 caches, SCSI, PCI, video, sound) than hacking AGA
: > further.
: Sorry man I can't agree. Building the hardware for a PPC based Amiga would
: be relatively simple, but porting the AmigaOS is going to be a nightmare!
the AMIGA PPC (read all before flaming please)
Just make kind of A1200 board conectable to PPC (PCI ?) slot. a CD32 is
very cheap compared to a PPC, and the platine would be less expensive!
There's no easier way for Amiga going PPC!
COMPATIBILITY CASE (demos, nasty games):
PPC cpu will run 680x0 emulation and drive AGA chips. 680x0 emu will
be beyond PPC-speed but not much different to usual 020.
presence of AGA/CIAs etc. will make OS-port most easy at low cost!
PPC CASE:
with help of gfx-drivers PPC gfx card will display workbench just like
gfx-emulation does now. OS-soft will work. OS-games (mega IN!) will work.
parts of OS could be PPC compiled for speed. parts of games could use PPC
code.
power + amiga.
maybe some hobby hardware freak could do a A1200-platine-PPC connector
if AT is to silly ;) I guess lot's of people having spend money for PPC
and knowing qualities of Amiga would spend money for the card which should
be less than current CD32 prices.
: When they do move the OS, I hope they will make some serious modifications
: (resource tracking, memory protection) but even if they don't it's going to
: be a hell of a job. With luck they might be able to licence Apples 680x0
: emulator, but even so can you imagine the problems they will have?!? No,
: hacking AGA is a picnic compared to that (although I agree its still not
: easy)
all no problemo with AGA-PPC-card.
: Anyway, a PPC Amiga will have to have display hardware of it's own, and if
yes, the gfx-cards ppcs usually have... or better...
real3d with mapping chips comes in my mind.
: you want any kind of compatibility it will have to look like AGA, or have
: an AGA emulation mode. Sooner or later AmigaTech will have to get intimate
the cheapest aga emulation is the AGA-PPC card, including easier OS-port.
: with the internals of the AGA : I say start now and keep it simple. The
: benefits justify the effort. I mean if they want to encourage some serious
: developers back to the Amiga it has to be able to do a few basics, and
: these days chunky display hardware is an essential. Remember, this is as
: much a political issue as a technical issue. No commercial developer would
: accept a statement like "no we don't have chunky hardware, but some guys on
: the net reckon this isn't a problem" ...
PPCs got chunky mode, what's the prob...
chunky really isn't any more the prob concerning mapping games on Amiga,
to programmers it's entirely clear that the real problems are
1) market crap (i.e to little software)
2) market crap (i.e std Amiga got little cpu)
- - - //- - - - - - -- - - - - - - - - - - - - - - - - - - - - - -
// A1201 HD260 fisc...@Informatik.TU-Muenchen.DE
\X/ ^--fastmem=2xSpeed (Juergen "Rally" Fischer) =:)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
: ----------------------------------------------------------------------------
Sam Taylor
> > 3) render to fast chunky buffer, copy without c2p to chip buffer - maybe
> > using something like MOVE16. In the time between chip writes (ie the
> > time you do c2p in #2) do other work, like some kind of final
> used to chip ram.
Yep, after dusting off my 680x0 reference manual I realised that MOVE16
wouldn't help at all (what I had actually been thinking of was MOVEM).
Anyway, none of this changes my fundamental point, which is that the CPU
could be doing other stuff while it was waiting to to write into chip ram.
> > be relatively simple, but porting the AmigaOS is going to be a
> > nightmare!
>
> Why? AmigaOS is mostly C.
Well there are still significant chunks of 680x0 ... and anyway it's not
just a case of re-compile and we're off. AmigaTech are talking about using
the PowerPC and that means moving to a 64-bit model when you want to run on
a PPC620. Not an easy thing to turn a 32-bit OS into a 64-bit OS, especially
since they were also talking about including resource tracking and memory
protection .... No, they're going to need some time to get that all stable...
Sam Taylor
> Just make kind of A1200 board conectable to PPC (PCI ?) slot. a CD32 is
> very cheap compared to a PPC, and the platine would be less expensive!
> There's no easier way for Amiga going PPC!
> with help of gfx-drivers PPC gfx card will display workbench just like
> gfx-emulation does now. OS-soft will work. OS-games (mega IN!) will work.
> if AT is to silly ;) I guess lot's of people having spend money for PPC
> and knowing qualities of Amiga would spend money for the card which should
> be less than current CD32 prices.
> PPCs got chunky mode, what's the prob...
a chip, not a complete machine, just a simple CPU. Apple and IBM are working
toward a Common Hardware Reference Platform (CHRP) which is based on the
PowerPC and the PCI bus, but at the moment that is still vapourware (like the
PowerAmiga...)
What AT have announced is that they will be using the PPC chips as the CPUs
in their next generation of machines. They haven't mentioned what bus they
will use for these machines. Although I guess they could use some form of
Zorro bus, they would be mad not to adopt the industry standard, which is
PCI. What display hardware architecture they adopt is a crucial question.
Use a standard VGA board and you gain a chunky mode, but loose everything
that made the Amiga unique. Use AGA and you will be forced to make some
changes to bring it up to date.
I am suggesting that AmigaTech maintain the Amigas unique hardware as the
native display architecture for the PowerAmiga range. However, this requires
that some changes be made to the AGA to support basic features such as a
chunky display mode. These changes should be made NOW for the forthcomming
Amiga 1300 (or whatever it gets called) rather than later. The PowerAmigas
are going to have enough compatibility problems as it is, this new machine
should be used to ease the transition process as much as possible. Keeping
the AGA does not prohibit the use of an off-the-shelf graphics co-processor
or DSP chip. Clearly it would be a massive task to turn the poor old
blitter and copper into an efficient t-mapped polygon rendering pipeline.
Bottom line for the 1300 should be this:
1. 040SLC - to give the product a reasonable shelf life and because
this is what the Apple 680x0 ROMs emulate
2. Revised AGA - to update the graphics hardware to support industry
basics, and to resolve compatibility issues now rather than later
3. Modest tweaks to the OS along with a comprehensive Game Developers
Kit like that produced by Microsoft for Windows95
Bottom line for the PowerAmiga should be:
1. PPC604 with the Apple emulator
2. Same basic AGA chipset as the 1300
3. Off-the-shelf graphics co-processor in addition to the old blitter
and copper, to handle tmapping etc.
4. Big rewrite of the OS to include memory protection, networking etc
-SAM-
Juergen Rally Fischer
|> fisc...@Informatik.TU-Muenchen.DE (Juergen "Rally" Fischer) wrote:
|> Juergen I think you may be confused as to what a PowerPC actually is : its
|> a chip, not a complete machine, just a simple CPU. Apple and IBM are working
yep, I know. But isn't the Term PPC also used for the Apple implementation,
which surely got a gfx card ? :) I assumed a plug-in card for PPC-slots (PCI ?).
|> Use a standard VGA board and you gain a chunky mode, but loose everything
|> that made the Amiga unique. Use AGA and you will be forced to make some
|> changes to bring it up to date.
Use AGA+VGA and you're fine.
maybe some amiga-coder-fellows will finally let the chunky.library come
true and chunky-software will run perfectly from A500-7 to PPC/PCI!
IMHO you must keep AGA (costs nothing) and just add a gfx card.
AmigaOS makes it all possible.
|>
|> I am suggesting that AmigaTech maintain the Amigas unique hardware as the
|> native display architecture for the PowerAmiga range. However, this requires
|> that some changes be made to the AGA to support basic features such as a
|> chunky display mode. These changes should be made NOW for the forthcomming
NO! current software techniques allow to do chunky fx with realtime c2p,
c2p while copying on 040, blitterscreen on A1200. It's the bandwidth,
not the gfx-system that is slow for higher cpus.
We just need the good old chipset and additional gfx-card in
higher spec Amigas. It's all easy, no AGA update required!
|> Amiga 1300 (or whatever it gets called) rather than later. The PowerAmigas
|> are going to have enough compatibility problems as it is, this new machine
PowerAmiga containing AGA for compatibility and running 68k emulation
won't give any troubles.
|> should be used to ease the transition process as much as possible. Keeping
|> the AGA does not prohibit the use of an off-the-shelf graphics co-processor
|> or DSP chip. Clearly it would be a massive task to turn the poor old
|> blitter and copper into an efficient t-mapped polygon rendering pipeline.
just add a DSP and let AGA the way it is. cheap easy AND powerful way.
|>
|> Bottom line for the 1300 should be this:
|> 1. 040SLC - to give the product a reasonable shelf life and because
|> this is what the Apple 680x0 ROMs emulate
well, also provide a EC030 version and let user choose. with right
software a 030 can do a lot!
|> 2. Revised AGA - to update the graphics hardware to support industry
|> basics, and to resolve compatibility issues now rather than later
this would make only delays and help nothing.
|> 3. Modest tweaks to the OS along with a comprehensive Game Developers
|> Kit like that produced by Microsoft for Windows95
uhm careful :) Yes, game developer kit needed. hope chunky.lib comes true,
maybe a game.lib . But please not like Mickeysoft did it :) PC coders
said it was no use ;)
|>
|> Bottom line for the PowerAmiga should be:
|> 1. PPC604 with the Apple emulator
|> 2. Same basic AGA chipset as the 1300
yes! AGA costs nothing compared to a PPC chip but gives you Amiga
compatibility, and easier (faster) port of next OS revision. Gfx
done by modern gfx-card, no problem.
|> 3. Off-the-shelf graphics co-processor in addition to the old blitter
|> and copper, to handle tmapping etc.
well, a card called Real3d is AFAIK soon available for PCI.
|> 4. Big rewrite of the OS to include memory protection, networking etc
yes! IF this is done then IMHO Amiga OS is the most powerful one,
and really the ONLY one for realtime-multimedia aplications.
|>
|> -SAM-
|>
|> ----------------------------------------------------------------------------
|> Sam.T...@anu.edu.au
|> Distributed High Performance Computing Project
|> Australian National University
|> ----------------------------------------------------------------------------
|>
-----------------------------------------------------------------------------
fisc...@Informatik.TU-Muenchen.DE (Juergen "Rally" Fischer)
=:)
-----------------------------------------------------------------------------
Peter McGavin
>Yep, after dusting off my 680x0 reference manual I realised that MOVE16
>wouldn't help at all (what I had actually been thinking of was MOVEM).
>Anyway, none of this changes my fundamental point, which is that the CPU
>could be doing other stuff while it was waiting to to write into chip ram.
A catch is that fastmem reads cause the CPU to wait until chip-writes
in progress have finished. (OK I haven't tested this myself but
others have said so.) So really, the CPU can only do register stuff
for "free" between chip writes, apart from fastmem reads scheduled
immediately before chipmem writes.
So it's hard to think what useful things could be done between chipmem
writes other than c2p in registers.
--
Peter McGavin. (p.mc...@irl.cri.nz)
Jyrki Saarinen
> A catch is that fastmem reads cause the CPU to wait until chip-writes
> in progress have finished. (OK I haven't tested this myself but
> others have said so.) So really, the CPU can only do register stuff
> for "free" between chip writes, apart from fastmem reads scheduled
> immediately before chipmem writes.
>
> So it's hard to think what useful things could be done between chipmem
> writes other than c2p in registers.
I agree completely. So the original poster still has to prove
WHAT other activity could be done between chip writes than
chunky2planar. I cant think anything..
Do you agree that a chunky mode in AGA would not help
almost at all? A A1200 with 14MHz CPU is not able
to render a 320x256 tmapped screen at decent rate
anyway, so .. 2x2 blitter-c2p really is the
answer.
-- _
a Stellar programmer _ //
"Amiga - back for the future" \X/
Sam Taylor
> A catch is that fastmem reads cause the CPU to wait until chip-writes
> in progress have finished. (OK I haven't tested this myself but
> others have said so.) So really, the CPU can only do register stuff
> for "free" between chip writes, apart from fastmem reads scheduled
> immediately before chipmem writes.
If that is the case then, it changes things a bit. It depends then how the
CPU interacts with it's cache : can it do fastmem/cache writes while a chip
write is pending? I ought to check my reference manuals to see what they
have to say. I have to agree that if you can only do register stuff, then
you may as well just use the "free" time for c2p.
In resonse to the above post Jyrki Saarinen wrote:
> I agree completely. So the original poster still has to prove
> WHAT other activity could be done between chip writes than
> chunky2planar. I cant think anything..
Well if you can only do register stuff then I agree you may as well do c2p.
If you can still access fastram then there are heaps of things you could
do - buffer clearing, start rendering the next frame, you name it...
> Do you agree that a chunky mode in AGA would not help
> almost at all? A A1200 with 14MHz CPU is not able
> to render a 320x256 tmapped screen at decent rate
> anyway, so .. 2x2 blitter-c2p really is the
> answer.
Hang on - for a stock 1200 a chunky mode would help heaps, it's the higher
spec processors where it starts to be less of an issue! You keep dodging
my main point : we have to have a chunky mode sooner or later - make it now
'cause there will be too much to do when the PowerPC machines are developed.
Anyway, this is all academic. Since this thread started AmigaTech have
announced that the new machine will be called the 1200+ and will be based
on either a 40MHz 030 (EC I presume) or on Coldfire (don't know the clock).
If they choose the crappy old 40MHz 030 then I think there is a real case
for having a chunky mode : C2P is still a bottleneck. I haven't had a
chance to look over the specs of the coldfire chip in detail, but I seem to
recall that it sat somewhere between the 030 and the 040 in performance
terms, in which case C2P may still be a bottleneck. Despite this they are
not going to change the AGA. Guess that means they agree with you ;-)
Seriously though I think its a mistake ... still it won't be the first
mistake the Amiga has had to live with...
Peter McGavin
>Do you agree that a chunky mode in AGA would not help
>almost at all? A A1200 with 14MHz CPU is not able
>to render a 320x256 tmapped screen at decent rate
>anyway, so .. 2x2 blitter-c2p really is the
>answer.
An AGA chunky mode would help by simplifying programming and saving
some memory (no chunky buffer). Performance-wise it won't help much
against good c2p algorithms unless the chipmem speed problem is
solved. CPU speed and memory speed are much more important
bottlenecks to eliminate than chunky/planar, IMO.
--
Peter McGavin. (p.mc...@irl.cri.nz)
Jyrki Saarinen
> Well if you can only do register stuff then I agree you may as well do
> c2p. If you can still access fastram then there are heaps of things you
> could do - buffer clearing, start rendering the next frame, you name it...
Well FAST-ram can not be accessed while chip writes.
Everything I have tested indicates this..
> Hang on - for a stock 1200 a chunky mode would help heaps, it's the
Not likely - chip only A1200 is not able to render a 320x256
tmapped screen with a decent frame rate, nowhere near it.
On a chip-only machine blitter-c2p can be done freely as
CPU renders the next image, while blitter-c2p is slow,
a chip-only A1200 cant really render the next image faster
than blitter can convert it.
> real case for having a chunky mode : C2P is still a bottleneck.
Oh god. Have you actually programmed some c2p/tmapping stuff?
I have.. c2p really is not a bottleneck at all. The rendering
time is, and that is really slower if it is done to CHIP ram
instead of FAST ram.
> terms, in which case C2P may still be a bottleneck. Despite this they
> are not going to change the AGA.
Chaning AGA is propably slow, expensive and difficult.
Please explain how c2p is a bottleneck even on faster
CPUs? I just really dont get it..
For example, a tmapping loop:
line move.w a0,d0
move.b d1,d0
move.l d0,a2
add.l a1,a0
addx.l d2,d1
move.b (a2),(a3)+ ;write pixel
Now, if the chunky buffer which a3 points is in chip,
this will be much slower than drawing to fast.
The difference between chip and fast rendering
may be even bigger than c2p time.
Jyrki Saarinen
> An AGA chunky mode would help by simplifying programming and saving
> some memory (no chunky buffer). Performance-wise it won't help much
> against good c2p algorithms unless the chipmem speed problem is
> solved. CPU speed and memory speed are much more important
> bottlenecks to eliminate than chunky/planar, IMO.
Yep. I was concerned only about performace issues..
How about a FAQ about these game/demo programming
matters? OS friendly, chunky2planar, texture mapping?
I could write the texture mapping part..
Thomas Karlsen
> Is all of this really possible for a machine that will be released in
> March? My impression of AT's plans for the 1300 were that it would be a
> 1200 with 030 processor and CD ROM drive. It would be great if a machine
> with your spec could be made in just 4 months but it would also make
> buying a 1200 now a little pointless.
>
>
> Will.
>
>
In the latest issue of Amiga Magazin (German) Peter T. of AT says that
the new CD will contain MPEG as standard... which will probably mean that
the new Amiga will have 030 40Mhz or Coldfire 40Mhz with CD, Fastram an
MPEG function...
Will kick alot!
Aki Laukkanen
: Yep. I was concerned only about performace issues..
: How about a FAQ about these game/demo programming
: matters? OS friendly, chunky2planar, texture mapping?
: I could write the texture mapping part..
Seconded that :) I may aswell write something down.
--
Daeron
Stephen Smith
>> A catch is that fastmem reads cause the CPU to wait until chip-writes
>> in progress have finished. (OK I haven't tested this myself but
>> others have said so.) So really, the CPU can only do register stuff
>> for "free" between chip writes, apart from fastmem reads scheduled
>> immediately before chipmem writes.
>>
>> So it's hard to think what useful things could be done between chipmem
>> writes other than c2p in registers.
>I agree completely. So the original poster still has to prove
>WHAT other activity could be done between chip writes than
>chunky2planar. I cant think anything..
>Do you agree that a chunky mode in AGA would not help
>almost at all? A A1200 with 14MHz CPU is not able
>to render a 320x256 tmapped screen at decent rate
>anyway, so .. 2x2 blitter-c2p really is the
>answer.
According to my tests, an A1200 14Mhz with FastMem can render
chunky pixels to ChipMem around 17% faster than rendering to
fast memory and then copying. This is with a chunky adapter
on the video port. The copy from fast to chip is longword aligned
and could be considered a 'zero time' c2p routine.
An A1200 with 50Mhz 030 does fast2chip 38% faster than direct to
chip memory. The results vary from machine to machine, but if AGA
had a chunky mode, then time wasted on writing the most efficient
c2p routine wouldn't have happened.
If everyone had a 40Mhz 040 in their machine, then things might
be different, but they don't. A chunky mode in AGA would have
benefited in some ways, but I would have liked the option
of both planar and chunky modes.
>-- _
>a Stellar programmer _ //
>"Amiga - back for the future" \X/
--
//Stephen J.Smith ,-_|\ st...@ph4227b.jcu.edu.au
// Physics Department / *\ Amiga 3000/040/25 10meg
\\ // James Cook University \_,-._/ Amiga 1200/030/50 10meg
\\// Townsville QLD v - SetBuffers/Nebula/UropaII
Juergen Rally Fischer
In article <PETERM.95N...@tui.maths.irl.cri.nz>,
|> A catch is that fastmem reads cause the CPU to wait until chip-writes
|> in progress have finished. (OK I haven't tested this myself but
|> others have said so.) So really, the CPU can only do register stuff
|> for "free" between chip writes, apart from fastmem reads scheduled
|> immediately before chipmem writes.
|>
|> So it's hard to think what useful things could be done between chipmem
|> writes other than c2p in registers.
wow indeed an argument. only weird things like "calculate sine in registers"
could be done ;)
The only argument pro real chunky mode vs realtime c2p is you could do
"direct render" stuff, for example rotzoomer, where you could preload
4 chunky bytes in registers and then directly store in chipmem.
no advantages in doing doom, descent, etc.
|> --
|> Peter McGavin. (p.mc...@irl.cri.nz)
Juergen Rally Fischer
|> > So it's hard to think what useful things could be done between chipmem
|> > writes other than c2p in registers.
|>
|> WHAT other activity could be done between chip writes than
|> chunky2planar. I cant think anything..
|>
|> Do you agree that a chunky mode in AGA would not help
|> almost at all? A A1200 with 14MHz CPU is not able
|> to render a 320x256 tmapped screen at decent rate
|> anyway, so .. 2x2 blitter-c2p really is the
|> answer.
for the case "020-14 on 2x2": blitterscreen is faster than local bus vga!
proof:
copy a longword to chipmem : 12 cycles (about)
copy a longword to fastmem (local bus 0 waitstate): 6 cycles (about)
bytes to copy on blitterscreen: 160x128
bytes to copy on vga : 320x256 (!!)
cycles AGA: 160*128/4*12: 61440
cycles VGA: 320*256/4*6 : 122880
really really interesting, eh ?
BTW:!!! doom needs so much cpu rendering that you could do full blitter
c2p without ghost-look on 020-14.
CONCLUSION: doing DOOM or DESCENT on a 020-14+fastmem, AGA gives you faster
framerate than the fastes PCI-SVGA would do!
think about this bevore crying A1200 can't do no doom because
it got AGA! (if it makes sense cpu wise is not the topic here BTW).
|>
|> -- _
|> a Stellar programmer _ //
|> "Amiga - back for the future" \X/
Juergen Rally Fischer
: fisc...@informatik.tu-muenchen.de (Juergen "Rally" Fischer) writes:
: So, what's your point? What's funny about that?
about what ? you didn't quote.
: :)
: >The only argument pro real chunky mode vs realtime c2p is you could do
: >"direct render" stuff, for example rotzoomer, where you could preload
: >4 chunky bytes in registers and then directly store in chipmem.
: >no advantages in doing doom, descent, etc.
: IMHO work perfectly suited for chunky modes instead of c2p work is
: stuff like ham conversion, colourspace conversion, dithering,
: 1x1->2x2 interpolation... Can be done without extra memory access,
: and it's a joy coding true colour routines...
chunky modes instead of c2p ? only on non-amiga. ham ? only on amiga.
"stuff like ham conversion" "for chunky modes". huh ? you got AAA ? ;)
: >------------------------------------------------------------------------
: > fisc...@Informatik.TU-Muenchen.DE (Juergen "Rally" Fischer) =:)
: /======================================================================\
: | Mans Engman (Sladdpost: mo...@lysator.liu.se) |
: | Student of Computer Science in Linkoping, Sweden. |
: | Assembly programmer since the age of eleven. |
: \======================================================================/
Mans Engman
So, what's your point? What's funny about that?
:)
>The only argument pro real chunky mode vs realtime c2p is you could do
>"direct render" stuff, for example rotzoomer, where you could preload
>4 chunky bytes in registers and then directly store in chipmem.
>no advantages in doing doom, descent, etc.
IMHO work perfectly suited for chunky modes instead of c2p work is
stuff like ham conversion, colourspace conversion, dithering,
1x1->2x2 interpolation... Can be done without extra memory access,
and it's a joy coding true colour routines...
>------------------------------------------------------------------------
Juergen Rally Fischer
: >> A catch is that fastmem reads cause the CPU to wait until chip-writes
: >> in progress have finished. (OK I haven't tested this myself but
: >> others have said so.) So really, the CPU can only do register stuff
: >> for "free" between chip writes, apart from fastmem reads scheduled
: >> immediately before chipmem writes.
: >>
: >> So it's hard to think what useful things could be done between chipmem
: >> writes other than c2p in registers.
: >I agree completely. So the original poster still has to prove
: >WHAT other activity could be done between chip writes than
: >chunky2planar. I cant think anything..
: >Do you agree that a chunky mode in AGA would not help
: >almost at all? A A1200 with 14MHz CPU is not able
: >to render a 320x256 tmapped screen at decent rate
: >anyway, so .. 2x2 blitter-c2p really is the
: >answer.
: According to my tests, an A1200 14Mhz with FastMem can render
: chunky pixels to ChipMem around 17% faster than rendering to
: fast memory and then copying. This is with a chunky adapter
well what do you call "render to chipmem" ? writing in bytes ?
then it's slower (4*8=32 chip vs 4*4+8=24 fast).
doing it in register, then save ? ok, maybe this is faster, but what
fx are you talking about ? I guess something like oldish flatshaded
polygons.
On a A1200 with fastmem, doom is never 17% faster when using a
chunky adapter. just some tiny little % (handling interrupts and
a bit more dma when copying (no, not the blitter dma, but 8 planes
vs 4 on dongle) ).
Only on A1200 chiponly the better dma conditions can make advantages
in this range (20% or so).
BUT: instead of buying a dongle which speeds up games 20% (lets not
think about if there exist some) I'd better buy fastmem which
speeds up ANY software on the A1200 by factor 2 averagely.
IMHO it's really freaky that some people try to provide a real
chunky-mode, I don't have a clue about how to build it, but sorry,
IMHO this way won't work.
a chunky library supporting all known methods is the way.
if the PPC board for A1200 comes true, really fancy stuff will
be possible on good old AGA.
btw a chunky library would be the best way of software really using
the dongles, assuming lot use the library, you also can make seldom
chunky methods be supported by lot software.
: on the video port. The copy from fast to chip is longword aligned
: and could be considered a 'zero time' c2p routine.
mhm what about the scrambling I heard of.
: An A1200 with 50Mhz 030 does fast2chip 38% faster than direct to
: chip memory. The results vary from machine to machine, but if AGA
your numbers make sense. 4mbs/sec on 020, almost 6+mb/sec on 030.
your direct render method is about 5mb/sec (on which machine ?)
you must be doing flat shaded polygons, what else runs 5mb/sec on 020-14 ?
please, tmapped games are the problem, and I still say:
if there is a problem, it's not AGA, it's only cpu speed.
it's only the dudes having no fastmem for example.
BUY AT LEAST FASTMEM. don't know the prices, a 030 board should be
only some little $ more. think about it.
: had a chunky mode, then time wasted on writing the most efficient
: c2p routine wouldn't have happened.
now this is very true. but the routines are ready. on PC it's same
with getting 32bit, lots of getting-into-protected-mode overhead.
there are lots of pc games. what does this tell us ?
Amiga is not the problem. market crap is it. I hope for PowerAmiga.
: If everyone had a 40Mhz 040 in their machine, then things might
: be different, but they don't. A chunky mode in AGA would have
If everyone had JUST FASTMEM (damn!), things would be different!
or just a damn cheap EC020-28. things would be very different!
of course a 040/40 is more, but let's stay realistic.
I hope for poweramiga times.
: benefited in some ways, but I would have liked the option
: of both planar and chunky modes.
: >-- _
: >a Stellar programmer _ //
: >"Amiga - back for the future" \X/
: --
: //Stephen J.Smith ,-_|\ st...@ph4227b.jcu.edu.au
: // Physics Department / *\ Amiga 3000/040/25 10meg
: \\ // James Cook University \_,-._/ Amiga 1200/030/50 10meg
: \\// Townsville QLD v - SetBuffers/Nebula/UropaII
-----------------------------------------------------------------------------
fisc...@Informatik.TU-Muenchen.DE (Juergen "Rally" Fischer)
=:)
-----------------------------------------------------------------------------
Mans Engman
>Mans Engman (c92m...@ida.liu.se) wrote:
>: fisc...@informatik.tu-muenchen.de (Juergen "Rally" Fischer) writes:
>:> wow, indeed an argument! only weird things like <------
>:> "calculate-sine-in-registers" could be done ;) <------
>: So, what's your point? What's funny about that?
>about what ? you didn't quote.
>: :)
Yes I did, see above! (is this what's called post-replying? ;)
Something screwed up, perhaps my brain, but I prefer to blame
it on stupid terminal emulation, cursor keys etc fooling me
to cut away the wrong lines...blah blah blah....
>: >no advantages in doing doom, descent, etc.
>: IMHO work perfectly suited for chunky modes instead of c2p work is
>: stuff like ham conversion, colourspace conversion, dithering,
>: 1x1->2x2 interpolation... Can be done without extra memory access,
>: and it's a joy coding true colour routines...
>chunky modes instead of c2p ? only on non-amiga. ham ? only on amiga.
>"stuff like ham conversion" "for chunky modes". huh ? you got AAA ? ;)
Yes, of course! :)
No, seriously, I thought this discussion whether about if a (hypothetical)
chunky mode on Amiga's would improve doom/descent/whatever-clones
despite slow chip access?
Recycling:
>: /======================================================================\
Mans Engman
>No, seriously, I thought this discussion whether about if a (hypothetical)
^^^^^^^^^^^^^^^^
Of course, I meant to write: "was about whether"
Brain malfunction, I'd better reboot myself ;-/
Jyrki Saarinen
> According to my tests, an A1200 14Mhz with FastMem can render
> chunky pixels to ChipMem around 17% faster than rendering to
> fast memory and then copying. This is with a chunky adapter
> and could be considered a 'zero time' c2p routine.
That depends on the rendered image and the number of overwritten
pixels etc.
In highly optimized texture mapping loop writing to fast
ram really helps, since chipmem writes stagger fastmem reads.
Stephen Smith
>> According to my tests, an A1200 14Mhz with FastMem can render
>> chunky pixels to ChipMem around 17% faster than rendering to
>> fast memory and then copying. This is with a chunky adapter
>> on the video port. The copy from fast to chip is longword aligned
>> and could be considered a 'zero time' c2p routine.
>That depends on the rendered image and the number of overwritten
>pixels etc.
>In highly optimized texture mapping loop writing to fast
>ram really helps, since chipmem writes stagger fastmem reads.
Using fast mem usually helps most operations, but in some cases
writing directly to chip, as mentioned above, is actually faster
overall.
The whole point of my post was to point out that having a chunky
mode in AGA is not a waste. You've assumed that if AGA had a chunky
mode that programmers would write pixels to chip. You could still
render to fast mem and then copy straight to chip. This would be
the equivalent of the fastest possible c2p routine ever made. (i.e.
no conversion time)
You need to look at the various speeds of machines and vary the
type of routine. In the case of an AGA with chunky mode, you would
render directly to chip for a standard A1200 with fastmem, but
use a fastmem buffer for higher spec'd machines. There's a cross
over point, that you're not aware of.
There are other benfits of a chunky mode that you may not have
considered, such as line drawing speed. Not all chunky applications
are texture mapping only.
Regards,
Stephen.
>-- _
>a Stellar programmer _ //
>"Amiga - back for the future" \X/
Juergen Rally Fischer
In article <c92manen.817901290@news>, c92m...@ida.liu.se (Mans Engman) writes:
|> Lines: 39
|>
|> fisc...@Informatik.TU-Muenchen.DE (Juergen "Rally" Fischer) writes:
|>
|> >Mans Engman (c92m...@ida.liu.se) wrote:
|> >: fisc...@informatik.tu-muenchen.de (Juergen "Rally" Fischer) writes:
|>
|> >:> wow, indeed an argument! only weird things like <------
|> >:> "calculate-sine-in-registers" could be done ;) <------
|>
|> >: So, what's your point? What's funny about that?
|> >about what ? you didn't quote.
|> >: :)
|>
|> Yes I did, see above! (is this what's called post-replying? ;)
mhm, that's the data on my newsserver:
-------------------posting of Mans Engman
<PETERM.95N...@tui.maths.irl.cri.nz> <49kou4$l...@sunsystem5.informatik.tu-muenchen.de>
Date: Fri, 1 Dec 95 13:58:06 1995
Lines: 25
fisc...@informatik.tu-muenchen.de (Juergen "Rally" Fischer) writes:
So, what's your point? What's funny about that?
:)
-------------------end
=;)
|> >chunky modes instead of c2p ? only on non-amiga. ham ? only on amiga.
|> >"stuff like ham conversion" "for chunky modes". huh ? you got AAA ? ;)
|>
|> Yes, of course! :)
|> No, seriously, I thought this discussion whether about if a (hypothetical)
|> chunky mode on Amiga's would improve doom/descent/whatever-clones
|> despite slow chip access?
on std A1200 a tiny little bit [about .5 of a frame (10ms) imho on fullsreen
2x2. now think how much frames doom needs on std A1200. BTW on a real chunky
mode still one blitter-pass needed to get 2x2/2x1. real 160 mode could help.
but not worth implementing as soon Amigas got higher cpus].
on std A1200 with grafiti even a bit more (better plane-dma conditions).
But buying fastmem instead will give you about factor 2 speed...
on A1200 with fastmem or higher NOT.
Zoltan Hunt
we have for the next year or so), I think that games like Breathless
sure show that with a fast processor, lots of things are still possible,
which is good now that it seems like the A1200/30 will be basically a
A1200 with SIMM (but no extra memory?) and a 40Mhz '030 (Coldfire is
apparently out.)
As for a future chip set (AAA is gone too, according to AmigaReport), I
would love to see (graphic wise) a cunky mode 8 or 9 bit with a CLUT and
then a 16 and 24 bit mode with no palette (AAA had a 16 bit mode like
this that used 5-bit guns, now add one with 8-bit guns), as well, I
really think we could do with some MPEG decoder hardware that could be
used to store compressed images and audio as well as play movies.
Jyrki Saarinen
> Using fast mem usually helps most operations, but in some cases
> writing directly to chip, as mentioned above, is actually faster
> overall.
This case usually happen ONLY when the rendering is like
some demo-effect where there screen is painted left-to-right,
top-to-bottom, at it can be done longwords at a time.
> render directly to chip for a standard A1200 with fastmem, but
> use a fastmem buffer for higher spec'd machines. There's a cross
> over point, that you're not aware of.
Depends very much on the rendering, too. Try rendering tmapped
polygon pixels by pixel to chip/fast-mem and see the difference.
The loop is:
move.w a0,d0
move.b d1,d0
move.l d0,a2
add.l a1,a0
addx.l d2,d1
move.b (a2),(a3)+ ;a2=ptr to texel, a3=ptr to cbuf (chip)
> There are other benfits of a chunky mode that you may not have
> considered, such as line drawing speed. Not all chunky applications
> are texture mapping only.
Yep. Well what this was originally about, was like the lack
of chunky mode or AGA is not reason for no tmapped games,
it is the lack of power in 14MHz 020. OTOH it CAN render
a 160x128 at a decent frame rate.
Stephen Smith
>Stephen Smith (ph...@manta.jcu.edu.au) wrote:
>: In <3823...@kone.fipnet.fi> "Jyrki Saarinen" <jsaa...@kone.fipnet.fi> writes:
>: According to my tests, an A1200 14Mhz with FastMem can render
>: chunky pixels to ChipMem around 17% faster than rendering to
>: fast memory and then copying. This is with a chunky adapter
>well what do you call "render to chipmem" ? writing in bytes ?
>then it's slower (4*8=32 chip vs 4*4+8=24 fast).
No. Writing individual bytes to chip memory vs fast memory is slower,
but in a practical sense, i.e. rendering a screen full of data, it
is faster (A1200/14 +fastmem) to render chunky pixels to chip memory.
Don't take my word for it, try it for yourself. You don't need a
chunky adapter to test it out.
>doing it in register, then save ? ok, maybe this is faster, but what
>fx are you talking about ? I guess something like oldish flatshaded
>polygons.
The test was for pixel throughput, not actual texture mapping, but since
the tests were done always with fast memory, then texture routines will
scale in speed similarly.
>On a A1200 with fastmem, doom is never 17% faster when using a
>chunky adapter. just some tiny little % (handling interrupts and
>a bit more dma when copying (no, not the blitter dma, but 8 planes
>vs 4 on dongle) ).
I didn't say doom would be 17%, but it would be more than a 'tiny little
%' The mode which is the most useful is 160 which has *NO* bus
contention on either ECS or AGA.
>Only on A1200 chiponly the better dma conditions can make advantages
>in this range (20% or so).
>BUT: instead of buying a dongle which speeds up games 20% (lets not
>think about if there exist some) I'd better buy fastmem which
>speeds up ANY software on the A1200 by factor 2 averagely.
>IMHO it's really freaky that some people try to provide a real
>chunky-mode, I don't have a clue about how to build it, but sorry,
>IMHO this way won't work.
Well, I disagree, since I designed the AGX module. The chunky adapters
give ECS machines 256 col chunky mode, that is nothing to be ignored.
As for speed up, well it depends on the system as I mentioned before.
Also, ECS gets an 18bit color palette. These are side-benefits and not
the whole reason the devices were made.
>a chunky library supporting all known methods is the way.
>if the PPC board for A1200 comes true, really fancy stuff will
>be possible on good old AGA.
>btw a chunky library would be the best way of software really using
>the dongles, assuming lot use the library, you also can make seldom
>chunky methods be supported by lot software.
The AGX device will come with a library, but only for opening/closing
chunky screens using the device, and some other basic functions, such as
fast line drawing.
>: on the video port. The copy from fast to chip is longword aligned
>: and could be considered a 'zero time' c2p routine.
>mhm what about the scrambling I heard of.
Any competent programmer can deal with it. It's easy.
>: An A1200 with 50Mhz 030 does fast2chip 38% faster than direct to
>: chip memory. The results vary from machine to machine, but if AGA
>your numbers make sense. 4mbs/sec on 020, almost 6+mb/sec on 030.
>your direct render method is about 5mb/sec (on which machine ?)
>you must be doing flat shaded polygons, what else runs 5mb/sec on 020-14 ?
>please, tmapped games are the problem, and I still say:
>if there is a problem, it's not AGA, it's only cpu speed.
>it's only the dudes having no fastmem for example.
>BUY AT LEAST FASTMEM. don't know the prices, a 030 board should be
>only some little $ more. think about it.
All the tests I did used fast mem. I don't expect anything less.
The AGX module costs around 1/8th of the price of a DKB mongoose card
(50Mhz 030) with 4Meg fastmem, think about it.
>: had a chunky mode, then time wasted on writing the most efficient
>: c2p routine wouldn't have happened.
>now this is very true. but the routines are ready. on PC it's same
>with getting 32bit, lots of getting-into-protected-mode overhead.
>there are lots of pc games. what does this tell us ?
>Amiga is not the problem. market crap is it. I hope for PowerAmiga.
>: If everyone had a 40Mhz 040 in their machine, then things might
>: be different, but they don't. A chunky mode in AGA would have
>If everyone had JUST FASTMEM (damn!), things would be different!
>or just a damn cheap EC020-28. things would be very different!
>of course a 040/40 is more, but let's stay realistic.
>I hope for poweramiga times.
Realistically, the poweramiga is a while away yet. In the meantime, lets
deal with what we have. Even if poweramigas were available tomorrow, it
will be sometime before software comes out for it. I remember what
happened when AGA first came out, software companies were
hesitant because they weren't going to make huge sales first off.
Juergen Rally Fischer
: fisc...@Informatik.TU-Muenchen.DE (Juergen "Rally" Fischer) writes:
: >Stephen Smith (ph...@manta.jcu.edu.au) wrote:
: >: In <3823...@kone.fipnet.fi> "Jyrki Saarinen" <jsaa...@kone.fipnet.fi> writes:
: >well what do you call "render to chipmem" ? writing in bytes ?
: >then it's slower (4*8=32 chip vs 4*4+8=24 fast).
: No. Writing individual bytes to chip memory vs fast memory is slower,
: but in a practical sense, i.e. rendering a screen full of data, it
WHAT are you rendering ? tmaping polygons needs some overhead if you
want to handle long writes to longword adresses. If you got many polygons,
the fact that the edges are also at odd adresses will make chip version
stall.
: The test was for pixel throughput, not actual texture mapping, but since
: the tests were done always with fast memory, then texture routines will
: scale in speed similarly.
a "one sqare area render" type effect, right ? like a rotzomer.
yup, then write to chip directly.
: >BUT: instead of buying a dongle which speeds up games 20% (lets not
: >think about if there exist some) I'd better buy fastmem which
: >speeds up ANY software on the A1200 by factor 2 averagely.
: >IMHO it's really freaky that some people try to provide a real
: >chunky-mode, I don't have a clue about how to build it, but sorry,
: >IMHO this way won't work.
: Well, I disagree, since I designed the AGX module. The chunky adapters
: give ECS machines 256 col chunky mode, that is nothing to be ignored.
yup yup.
: As for speed up, well it depends on the system as I mentioned before.
speed up only on chip-only amiga which better get at least fastmem
anyway.
: Also, ECS gets an 18bit color palette. These are side-benefits and not
: the whole reason the devices were made.
: >btw a chunky library would be the best way of software really using
: >the dongles, assuming lot use the library, you also can make seldom
: >chunky methods be supported by lot software.
: The AGX device will come with a library, but only for opening/closing
: chunky screens using the device, and some other basic functions, such as
: fast line drawing.
hardwarehack info for chunky.lib programmers to implement in their lib
would be nessesary. how is the dongle activated ? over a fastmem-adress ?
so not reachable via copper ? how to drag such screens ?
: >: on the video port. The copy from fast to chip is longword aligned
: >: and could be considered a 'zero time' c2p routine.
: >mhm what about the scrambling I heard of.
: Any competent programmer can deal with it. It's easy.
yes, but this makes it a mode you'd better not render diectly to!
for example scrambled mapping needs more cycles. So for doom you'd
better do a blitter pass instead of scrambled rendering ;) ;)
: >I hope for poweramiga times.
: Realistically, the poweramiga is a while away yet. In the meantime, lets
: deal with what we have. Even if poweramigas were available tomorrow, it
: will be sometime before software comes out for it. I remember what
: happened when AGA first came out, software companies were
: hesitant because they weren't going to make huge sales first off.
clever coded software can be adapted to run at PPC speed on PPC
with little effort. time critical routines in PPC, rest in emu.
or better: just recompile.
: >: --
Juergen Rally Fischer
: In <3823...@kone.fipnet.fi> "Jyrki Saarinen" <jsaa...@kone.fipnet.fi> writes:
: >In highly optimized texture mapping loop writing to fast
: >ram really helps, since chipmem writes stagger fastmem reads.
: Using fast mem usually helps most operations, but in some cases
: writing directly to chip, as mentioned above, is actually faster
: overall.
but not in texture mapping loops. the need to write 32bit wise to chip
to not loose out would give trouble at the edges.
: You need to look at the various speeds of machines and vary the
: type of routine. In the case of an AGA with chunky mode, you would
: render directly to chip for a standard A1200 with fastmem, but
: use a fastmem buffer for higher spec'd machines. There's a cross
: over point, that you're not aware of.
and in the case of real AGA without chunky, on A1200 with fastmem
you also can directly render chunky into chipmem, or render to
fastmem and then copy. rest is done by blitter.
: There are other benfits of a chunky mode that you may not have
: considered, such as line drawing speed. Not all chunky applications
true.
: are texture mapping only.
yup, all my "need no chunky"-claims go for doing doom, descent and comparable.
The games everyone wants to have.
: Regards,
: Stephen.
: >-- _
: >a Stellar programmer _ //
: >"Amiga - back for the future" \X/
: --
: //Stephen J.Smith ,-_|\ st...@ph4227b.jcu.edu.au
: // Physics Department / *\ Amiga 3000/040/25 10meg
: \\ // James Cook University \_,-._/ Amiga 1200/030/50 10meg
: \\// Townsville QLD v - SetBuffers/Nebula/UropaII
- - - //- - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
// A1201 HD260 fisc...@Informatik.TU-Muenchen.DE
\X/ ^--fastmem=2xSpeed (Juergen "Rally" Fischer) =:)
Stephen Smith
>Stephen Smith (ph...@jcu.edu.au) wrote:
>: fisc...@Informatik.TU-Muenchen.DE (Juergen "Rally" Fischer) writes:
>: The test was for pixel throughput, not actual texture mapping, but since
>: the tests were done always with fast memory, then texture routines will
>: scale in speed similarly.
>a "one sqare area render" type effect, right ? like a rotzomer.
>yup, then write to chip directly.
A texture mapping loop isn't going to make a difference. I've tried
it.
>: >BUT: instead of buying a dongle which speeds up games 20% (lets not
>: >think about if there exist some) I'd better buy fastmem which
>: >speeds up ANY software on the A1200 by factor 2 averagely.
>: >IMHO it's really freaky that some people try to provide a real
>: >chunky-mode, I don't have a clue about how to build it, but sorry,
>: >IMHO this way won't work.
>: Well, I disagree, since I designed the AGX module. The chunky adapters
>: give ECS machines 256 col chunky mode, that is nothing to be ignored.
>yup yup.
>: As for speed up, well it depends on the system as I mentioned before.
>speed up only on chip-only amiga which better get at least fastmem
>anyway.
No, that's not true. Fast memory makes a difference, but only on an
accelerated A1200 with fastmem would you write to fastmem instead
of chip.
I agree that everyone should get fast memory but having it doesn't
suddenly make a chunky mode obselete....
>: The AGX device will come with a library, but only for opening/closing
>: chunky screens using the device, and some other basic functions, such as
>: fast line drawing.
>hardwarehack info for chunky.lib programmers to implement in their lib
>would be nessesary. how is the dongle activated ? over a fastmem-adress ?
>so not reachable via copper ? how to drag such screens ?
I have prepared a document about how the device works. I will place it
on our web site (http://ph4227b.jcu.edu.au) in the next few days.
>: >: on the video port. The copy from fast to chip is longword aligned
>: >: and could be considered a 'zero time' c2p routine.
>: >mhm what about the scrambling I heard of.
>: Any competent programmer can deal with it. It's easy.
>yes, but this makes it a mode you'd better not render diectly to!
>for example scrambled mapping needs more cycles. So for doom you'd
>better do a blitter pass instead of scrambled rendering ;) ;)
Only the modex style modes are scrambled, 160 mode can be true
linear chunky! For a stock A1200 with or without fastmem, you'd write
directly to chipmem because it's faster. For accelerated A1200 with
fastmem, then write to fastmem and do longword copies. The point is,
in any case there is no c2p or blitter...
I'm talking about rendering a 160x256 or 320x256 texture mapped screen
here.
>: >I hope for poweramiga times.
>: Realistically, the poweramiga is a while away yet. In the meantime, lets
>: deal with what we have. Even if poweramigas were available tomorrow, it
>: will be sometime before software comes out for it. I remember what
>: happened when AGA first came out, software companies were
>: hesitant because they weren't going to make huge sales first off.
>clever coded software can be adapted to run at PPC speed on PPC
>with little effort. time critical routines in PPC, rest in emu.
>or better: just recompile.
So Team17 will be giving away the source code to AB3DSE with every
copy then... ;)
>: >: --
Jyrki Saarinen
> No. Writing individual bytes to chip memory vs fast memory is slower,
> but in a practical sense, i.e. rendering a screen full of data, it
> is faster (A1200/14 +fastmem) to render chunky pixels to chip memory.
> Don't take my word for it, try it for yourself. You don't need a
> chunky adapter to test it out.
Try yourself doing tmapped polygons at a longword at the time.
To a 4 aligned memory address.. Becomes slow.
In very SIMPLE demo effects like zoom & rotate hardware
chunky mode would be useful since rendering could be
done in longwords. But shifting that data register
is quite slow on 020/030 anyway.. slower than writing
individual bytes to fast ram chunky buffer.
Stephen Smith
>Stephen Smith (ph...@manta.jcu.edu.au) wrote:
>: In <3823...@kone.fipnet.fi> "Jyrki Saarinen" <jsaa...@kone.fipnet.fi> writes:
>: >In highly optimized texture mapping loop writing to fast
>: >ram really helps, since chipmem writes stagger fastmem reads.
>and in the case of real AGA without chunky, on A1200 with fastmem
>you also can directly render chunky into chipmem, or render to
>fastmem and then copy. rest is done by blitter.
I'd rather have AGA with a chunky mode than look at a screen
rendered through a flyscreen.
>: There are other benfits of a chunky mode that you may not have
>: considered, such as line drawing speed. Not all chunky applications
>true.
>: are texture mapping only.
>yup, all my "need no chunky"-claims go for doing doom, descent and comparable.
>The games everyone wants to have.
In your humble opinion... ;)
>: Regards,
>: Stephen.
>: >-- _
>: >a Stellar programmer _ //
>: >"Amiga - back for the future" \X/
>: --
>: //Stephen J.Smith ,-_|\ st...@ph4227b.jcu.edu.au
>: // Physics Department / *\ Amiga 3000/040/25 10meg
>: \\ // James Cook University \_,-._/ Amiga 1200/030/50 10meg
>: \\// Townsville QLD v - SetBuffers/Nebula/UropaII
>- - - //- - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> // A1201 HD260 fisc...@Informatik.TU-Muenchen.DE
> \X/ ^--fastmem=2xSpeed (Juergen "Rally" Fischer) =:)
>- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Juergen Rally Fischer
In article <1995Dec11.1...@marlin.jcu.edu.au>, ph...@manta.jcu.edu.au (Stephen Smith) writes:
|> Lines: 47
|>
|> I'd rather have AGA with a chunky mode than look at a screen
|> rendered through a flyscreen.
??? what is "rendering though a flyscreen" ? new fx method ? :)
|>
|>
|> >: Regards,
|> >: Stephen.
|>
|>
-----------------------------------------------------------------------------
// fisc...@Informatik.TU-Muenchen.DE (Juergen "Rally" Fischer)
//
\X/ AMIGA: The real multimedia platform.
-----------------------------------------------------------------------------
Stephen Smith
>> No. Writing individual bytes to chip memory vs fast memory is slower,
>> but in a practical sense, i.e. rendering a screen full of data, it
>> is faster (A1200/14 +fastmem) to render chunky pixels to chip memory.
>> Don't take my word for it, try it for yourself. You don't need a
>> chunky adapter to test it out.
>Try yourself doing tmapped polygons at a longword at the time.
>To a 4 aligned memory address.. Becomes slow.
>In very SIMPLE demo effects like zoom & rotate hardware
>chunky mode would be useful since rendering could be
>done in longwords. But shifting that data register
>is quite slow on 020/030 anyway.. slower than writing
>individual bytes to fast ram chunky buffer.
Look, I agree with you that writing bytes to fast memory is
quicker than chip memory, but on *some* systems such as an A1200
(020 14Mhz) and fastmem, it is quicker to render a doomish screen
to a chunky chipmem than any fastmem/c2p/copy combination. As
you boost the processor power it goes the other way, but the whole
point is that a chunky mode on the Amiga would benefit doomish
games for some systems, low end AGA included.
>-- _
>a Stellar programmer _ //
>"Amiga - back for the future" \X/
Regards,
Stephen.
--
//Stephen J.Smith ,-_|\ st...@ph4227b.jcu.edu.au
// Physics Department / *\ Amiga 3000/040/25 14meg
\\ // James Cook University \_,-._/ Amiga 1200/030/50 10meg
\\// Townsville QLD v - SetBuffers/Nebula/UropaII
--
//Stephen J.Smith ,-_|\ st...@ph4227b.jcu.edu.au
// Physics Department / *\ Amiga 3000/040/25 14meg
Jyrki Saarinen
> Look, I agree with you that writing bytes to fast memory is
> quicker than chip memory, but on *some* systems such as an A1200
> (020 14Mhz) and fastmem, it is quicker to render a doomish screen
> to a chunky chipmem than any fastmem/c2p/copy combination. As
Well, even in DOOM-case the framerate difference is very
small.
> you boost the processor power it goes the other way, but the whole
> point is that a chunky mode on the Amiga would benefit doomish
> games for some systems, low end AGA included.
DOOM, yes. Back2front tmapped polygons, no. Perhaps
on a 3D BSP-tree based rendered with front2back rendering
and some scanline algorithms used a chunky mode, in AGA,
would be useful.
Why are we actually speculating this..? ;)
Juergen Rally Fischer
: "Jyrki Saarinen" <jsaa...@kone.fipnet.fi> writes:
: >> is faster (A1200/14 +fastmem) to render chunky pixels to chip memory.
: >Try yourself doing tmapped polygons at a longword at the time.
: >To a 4 aligned memory address.. Becomes slow.
: >In very SIMPLE demo effects like zoom & rotate hardware
: >chunky mode would be useful since rendering could be
: >done in longwords. But shifting that data register
: >is quite slow on 020/030 anyway.. slower than writing
: >individual bytes to fast ram chunky buffer.
: Look, I agree with you that writing bytes to fast memory is
: quicker than chip memory, but on *some* systems such as an A1200
: (020 14Mhz) and fastmem, it is quicker to render a doomish screen
: to a chunky chipmem than any fastmem/c2p/copy combination. As
ok, let's assume this (must be a free-cycle thingy), but then the
byte render to chipmem blitterscreen will be SAME speed!
blitterscreen got chunky byte buffer.
(I assume the rendering needs longer than blitter conversion,
which is true on 020-14 doing doom).
conclusion: chunky would be no speedup doing doom.
: you boost the processor power it goes the other way, but the whole
: point is that a chunky mode on the Amiga would benefit doomish
: games for some systems, low end AGA included.
for doom, you need CPU, CPU and CPU! that's what new A1200 needs.
insisting in a system with chunky mode will give you nothing but
higher price and more delay. not faster doom.
: >-- _
: >a Stellar programmer _ //
: >"Amiga - back for the future" \X/
: Regards,
: Stephen.
: --
: //Stephen J.Smith ,-_|\ st...@ph4227b.jcu.edu.au
: // Physics Department / *\ Amiga 3000/040/25 14meg
: \\ // James Cook University \_,-._/ Amiga 1200/030/50 10meg
: \\// Townsville QLD v - SetBuffers/Nebula/UropaII
: --
Stephen Smith
>Stephen Smith (ph...@jcu.edu.au) wrote:
>: "Jyrki Saarinen" <jsaa...@kone.fipnet.fi> writes:
>ok, let's assume this (must be a free-cycle thingy), but then the
>byte render to chipmem blitterscreen will be SAME speed!
>blitterscreen got chunky byte buffer.
>(I assume the rendering needs longer than blitter conversion,
>which is true on 020-14 doing doom).
Are you seriously trying to tell me that a blitterscreen
would be better than a true chunky mode? Do you think people
would prefer a blitterscreen to a 'normal' display?
>conclusion: chunky would be no speedup doing doom.
In your opinion.
>: you boost the processor power it goes the other way, but the whole
>: point is that a chunky mode on the Amiga would benefit doomish
>: games for some systems, low end AGA included.
>for doom, you need CPU, CPU and CPU! that's what new A1200 needs.
Of course, I agree with that. I'm talking about current systems.
>insisting in a system with chunky mode will give you nothing but
>higher price and more delay. not faster doom.
A chunky mode adapter for the Amiga would be cheaper than an
accelerator, and faster doom.
>: >-- _
>: >a Stellar programmer _ //
>: >"Amiga - back for the future" \X/
>: Regards,
>: Stephen.
>: --
>: //Stephen J.Smith ,-_|\ st...@ph4227b.jcu.edu.au
>: // Physics Department / *\ Amiga 3000/040/25 14meg
>: \\ // James Cook University \_,-._/ Amiga 1200/030/50 10meg
>: \\// Townsville QLD v - SetBuffers/Nebula/UropaII
>: --
Aki Laukkanen
: >blitterscreen got chunky byte buffer.
: >(I assume the rendering needs longer than blitter conversion,
: >which is true on 020-14 doing doom).
: Are you seriously trying to tell me that a blitterscreen
: would be better than a true chunky mode? Do you think people
: would prefer a blitterscreen to a 'normal' display?
No, he is saying it's of no advantage. Because of AGA bandwidth, you'd still
have to render to FAST and then copy the chunky buffer to chip ram. Btw. this
would take virtually the same amount of time than with blitterscreen.
Blitterscreen also offers a full screen 2x2 mode and if AGA had a true
chunky mode, you'd still have to double pixels horizantally either with CPU
(bad - twice the much to write to chip) or blitter. Why not use 1x1 modes then?
Reason is simple, today's base systems are too slow so go buy a faster cpu.
: >conclusion: chunky would be no speedup doing doom.
: In your opinion.
A fact.
: A chunky mode adapter for the Amiga would be cheaper than an
: accelerator, and faster doom.
No, it wouldn't. Try graffiti, on a a1200/fast and higher it's of no use.
--
Daeron