Re: [stella] Hardware comparisons

Subject: Re: [stella] Hardware comparisons
From: kurt.woloch@xxxxxxxxx
Date: Wed, 26 Apr 2000 14:15:36 +0200
Glenn Saunders wrote:

>>So... do all the old Amiga games run on your editing workstation?

>Ironically, I don't have many commercial Amiga games, although I played 
>some great shareware games like a truly ass-kicking Galaga redux.  I guess 
>that gives me something to look forward to as a hobby down the road.  Most 
>of the games were written for OS 1.3 which is vastly different from even OS

>2.1, let alone 3.x.  Most of the incompatibilities rest there, which is why

>people came up with software degraders and hardware multi OS ROM boards.

Well, I have to admit that I've got almost no original commercial games for
the Amiga, as well as for the C-64, let alone the PC. But I've got hundreds
of disks filled with copies of them. I was 11 when that home computer boom
began, and when I got the C-64 at 13, and the Amiga 500 at 16, the reasons
for them were also that there were so many other kids at school having those
machines with whom I could swap games and other software. I admit that they
were pirate copies. That's true. But everybody I knew did this at that time.
I know that this isn't an excuse for it today... since now I work as a
progammer myself and know how complicated it is to write such programs...
but what should I do now? All these games aren't on the market anymore... I
think even the PC games disappear from the market again when they become a
few years old. 

>However, on my Amiga 2000 with an 040/28 in it and an Amiga 1200 030/50 I 
>ran Wing Commander just fine.

Ah yes. However, Wing Commander is a simulation game. I think on the Amiga,
there were already enough hardware ressources to make the game programmers
not bother that much about low-level-programming like on the 2600. On the
2600, you have to re-write the registers each scanline. On the Atari XL's,
you can, but don't have to, still it makes for better effects in a game. On
the Amiga, you theoretically could, but virtually no one does it, because
you can get nearly all of those better effects by writing Copper lists,
which is the "official" way to do such things. So... basically, things
already move away from the hardware a lot, though not as much as in PC
games. But it could be that really fast games (that do 60 fps in their
scrolling et al) do have more direct hardware accesses and thus do have
problems on faster Amigas.
This is the same like the PC games. Until the mid-90's, nearly all games
existing ran under DOS. These games probably had a lot more hardware
accesses in them as the now released Windows games. True, you could do
smooth scrolling on a 386 VGA system, but only if you accessed the hardware
directly. Today's systems give you the power to do all this using the
"official" way, so games can be written in C instead of assembler, and you
can access the correct drivers and still get a reasonable speed. This just
wasn't possible on the early 80s' computers.

>I think video DMA on the Atari cuts raw throughput of the CPU down by 
>1/3rd, which is why there is a poke you can do to turn off video DMA when 
>you are running a big numbercrunch operation.  A lot of Atari BBS operators

>would do that.  The Atari 1200XL had that assigned to one of its function
keys.

So... in the end you get the same CPU speed as on the 2600, with the
difference that the custom chips will do much more additional work for you.
This leads me to the suspicion that the Atari's bus is actually SLOWER than
that of the C-64 (1,79 MHz vs 2 MHz), but the CPU, WHEN it gets time to do
its job (which is more likely to occur during border display), due to faster
running, still has more cycles left, since in border times, it can use all
of the 1,79 MHz, while on the C-64 this is not the case.

Also, on the TMS-9918 systems, this runs different. The video chip always
has its own memory and reads from there, not ever halting the CPU. On the
other hand, the CPU has to write to video memory via ports on the TMS-9918,
which is relatively slow, though the chip allows auto-incrementing (the
memory location's increased with every byte written or read). The SMS and
Game Gear, by the way, have video chips derived from the TMS-9918, which are
apparently able to read more memory at a time, but this comes at the cost of
only allowing accesses via the CPU during scanlines where nothing's
displayed. At least I read this in a documentation of the SMS. This could
have made programming more complicated. And though the SMS's chip only has
got one VALID video mode, it can be switched into other INVALID ones.

>It actually cuts throughput enough to hinder playing digital samples, 
>actually, which is why sampling programs (or software voice synths like 
>SAM) would blank the screen.

Yes, the sampling programs on the C-64 do this too... at least most of them.
There are exceptions, however. There was a company called Electronic Speech
Systems doing the speech for some games like Ghostbusters, Slap Shot Hockey,
and Kennedy Approach, and they didn't shut off the screen. In Kennedy
Approach, there's even some (limited) action going on while you hear the
pilots and tower talking to each other. On the other hand, the Amstrad CPC
version of Ghostbusters sounds much worse to this. I wonder if this also was
done by ESS.

With love (and still many things to compare)
Kurt Woloch

--
Archives (includes files) at http://www.biglist.com/lists/stella/archives/
Unsub & more at http://www.biglist.com/lists/stella/

Current Thread