OMG i had no idea!!! (Atari LYNX related)

[fdave]

STUN Runner

mmmmmmm, scaley sprites.

Still need a decent emulated version

Won't MAME run it if you throw a sufficiently amazing PC at it?

I know there were speed issues when the driver was first written, due to it having all manner of nasty Hard Drvin style hardware and DSPs and whatnot, but weren't people joking about how it would run 60fps if in theory you had a 2Ghz PC (which we now have...)?

Dunno...

[skip]

Won't MAME run it if you throw a sufficiently amazing PC at it?

I know there were speed issues when the driver was first written, due to it having all manner of nasty Hard Drvin style hardware and DSPs and whatnot, but weren't people joking about how it would run 60fps if in theory you had a 2Ghz PC (which we now have...)?

Dunno...

I think it runs okay on good modern PCs. I do know the following though:

1. It runs at 5 frames/hour on my 1GHz PowerBook
2. It was shitcanned from one of the Midway Arcade collections for the XBox

I have a feeling it might be included on one of the upcoming ones, but whether it's emulated or recoded I don't know. Don't care either, as long as it's good

[noodles]

By the time they got them it was much, much too late.

One of the hardware designers (name escapes me - might have been Dave Hainie(sp?)) who was involved with the last of the C= work (AAA, Hombre etc) commented that soon after Amiga was bought out, an off-the-shelf PC graphics card could do more, faster, and at a much reduced cost. Even a VirgeDX would have shat all over anything the last Amiga designs could do.

Yup, that was Haynie; I posted a link to some of the documents he made available from the laters days of the Amiga (including some on AAA & Hombre) back on page 2 : http://www.thule.no/haynie/ - well worth a read if you've never seen them

I'm not sure I'd say a VirgeDX could shit all over AA/AAA - it may have been faster (probably much faster) and have higher resolutions but I don't think it was anywhere near as flexible?

And also probably easier to program. Making things nicer for
the developer was clearly not a priority on the PC.
It's the whole combination of Amiga hardware that made it so pleasant
to code for : 68000 assembly + custom chips + built-in stereo sound just
was an awesome combination.

There's a filmed interview of Haynie recalling his times at C= floating out there. Well worth a look. He clearly lays the blame on Mehdi Ali and Irving Gould's other cronies for sabotaging all efforts and having zero vision.
On top of that, Commodore was re-incorporated in the Bahamas in the later years, which was highly suspicious, as that prevented many small
shareholders from attending yearly stockholder meetings.

[corrosive]

And also probably easier to program. Making things nicer for
the developer was clearly not a priority on the PC.
It's the whole combination of Amiga hardware that made it so pleasant
to code for : 68000 assembly + custom chips + built-in stereo sound just
was an awesome combination.

There's a filmed interview of Haynie recalling his times at C= floating out there. Well worth a look. He clearly lays the blame on Mehdi Ali and Irving Gould's other cronies for sabotaging all efforts and having zero vision.
On top of that, Commodore was re-incorporated in the Bahamas in the later years, which was highly suspicious, as that prevented many small
shareholders from attending yearly stockholder meetings.

You mean "The Deathbed Vigil"? Got a copy - think Haynie still does it on his website (http://www.frogpondmedia.com/dbv/).

The Amiga was easy to program, and it helped me a lot at college, simply because before that point I'd only had access to the usual 8-bit micros, and while at college I suddenly had access to a VAX, at home, I'd just got an Amiga - great for learning about a modern operating system, but also great for learning about low-level programming too. Before the Amiga assembly language didn't "click" with me (I could do bits, but not really an entire program), after trying out 68000, it all fell into place

There was something about the designs from that linage (2600, Atari 8-bits, Amiga, Lynx) that just made you go "Wow".

[Dio]

commented that soon after Amiga was bought out, an off-the-shelf PC graphics card could do more, faster, and at a much reduced cost. Even a VirgeDX would have shat all over anything the last Amiga designs could do.

I'm not sure I'd say a VirgeDX could shit all over AA/AAA - it may have been faster (probably much faster) and have higher resolutions but I don't think it was anywhere near as flexible?

And also probably easier to program. Making things nicer for the developer was clearly not a priority on the PC. It's the whole combination of Amiga hardware that made it so pleasant to code for : 68000 assembly + custom chips + built-in stereo sound just was an awesome combination.

Writing large programs in any assembly language is nothing short of torture - you can trust me on this, and I love assembly language. x86 is no worse than 68k - in fact, it can be easier because as it has evolved it became a more orthogonal architecture than 68k is (from the point of view of the integer instruction set, at least).

Turn the Z-buffer off and all those early PC 3D cards absolutely fly. The main disadvantage they would have had over a blitter-based architecture was low small batch performance. There were some flexibility headaches (mostly API-enforced, rather than hardware-enforced - intriguingly enough it's only more recently the hardware itself has become more limiting in terms of it's 2D pixel addressing flexibility).

There was something about the designs from that linage (2600, Atari 8-bits, Amiga, Lynx) that just made you go "Wow".

They're all very impressive technically. In the end, though, I'd say it was shown that the more sensible decision was to reduce the amount of expensive custom chips and redirect it to cost reductions, more general purpose horsepower, and similar simplicity benefits - certainly in price-sensitive markets like Europe, anyway.

[Flojomojo]

STUN Runner

mmmmmmm, scaley sprites.

Still need a decent emulated version

MAME will run it, but you need a fair amount of juice. It will be featured in Midway Arcade Treasures 3 in a few months, along with SF Rush and Hydro Thunder. At last.

[corrosive]

Writing large programs in any assembly language is nothing short of torture - you can trust me on this, and I love assembly language. x86 is no worse than 68k - in fact, it can be easier because as it has evolved it became a more orthogonal architecture than 68k is (from the point of view of the integer instruction set, at least).

I quite enjoyed writing in assembler, x86 was probably my least favourite though (the others I used were 6809 and 68k); . More orthogonal? With all those special register limitations to remember? I must have missed something in the later x86 processors then . I've done x86 assembler (although admittedly nothing big for a while - this was on a 386 driving a VGA card over ISA), and it didn't feel very orthogonal to me (compared to 68k)(actually, to be fair, I seem to remember that the 32-bit mode instructions did seem to be a bit more orthogonal - unfortunately as soon as you were dealing with less than 32 bits the usual x86 special register problems hit). Did have some nice features though.

They're all very impressive technically. In the end, though, I'd say it was shown that the more sensible decision was to reduce the amount of expensive custom chips and redirect it to cost reductions, more general purpose horsepower, and similar simplicity benefits - certainly in price-sensitive markets like Europe, anyway.

Can't really argue against that, when you're going for a large market, you need the economies of scale. Let someone else produce all the "custom chips" (nVidia, ATI, etc.) - you loose the "custom" bit, but you save rather a lot of money. Unfortunately you also lose the ability to link the hardware and software together closely (although the efficiencies gained by that aren't as important these days, really).

[Dio]

I quite enjoyed writing in assembler, x86 was probably my least favourite though (the others I used were 6809 and 68k); . More orthogonal? With all those special register limitations to remember? I must have missed something in the later x86 processors then . I've done x86 assembler (although admittedly nothing big for a while - this was on a 386 driving a VGA card over ISA), and it didn't feel very orthogonal to me (compared to 68k)(actually, to be fair, I seem to remember that the 32-bit mode instructions did seem to be a bit more orthogonal - unfortunately as soon as you were dealing with less than 32 bits the usual x86 special register problems hit). Did have some nice features though.

Well, I've never programmed outside protected mode, as I only came to PC in the DOS/4GW days, so segment registers are mercifully a closed book to me . AFAICS with the exception of a few minor restrictions on how EBP and ESP can be used in index registers, x86 looks largely orthogonal. And who uses registers smaller than 32 bits nowadays anyway?

With the 64-bit extensions finally giving it a large enough register space to be really useful (16 each of GPR, MMX and SSE), it's a pretty strong ISA I think. Pity it carries so much baggage, though I don't really think it hurts it as a moden architecture.

Unfortunately you also lose the ability to link the hardware and software together closely (although the efficiencies gained by that aren't as important these days, really).

Linking the two is a bad thing - it either traps you into the limitations of an outmoded architecture or prevents upgrades. While Moore's Law still holds, anything that uses a direct access model has a strict lifetime and it's less than ten years. As you say, now the overheads are small, there is rarely significant advantage to doing it.

[corrosive]

Well, I've never programmed outside protected mode, as I only came to PC in the DOS/4GW days, so segment registers are mercifully a closed book to me . AFAICS with the exception of a few minor restrictions on how EBP and ESP can be used in index registers, x86 looks largely orthogonal. And who uses registers smaller than 32 bits nowadays anyway?

Unfortunately, the only x86 dependent job I had was just /before/ protected mode really got everywhere. I'd done a bit before, but this was my first time on a 386 - worst thing was you couldn't be 100% sure that the 32-bit registers were safe in real mode. Some of the memory managers corrupted them IIRC. Still used them, mind (To be fair, I think /most/ of the memory managers were safe by then)

With the 64-bit extensions finally giving it a large enough register space to be really useful (16 each of GPR, MMX and SSE), it's a pretty strong ISA I think. Pity it carries so much baggage, though I don't really think it hurts it as a moden architecture.

Must take a look at x86-64, been unusually slack on this and not read up on it yet.

Linking the two is a bad thing - it either traps you into the limitations of an outmoded architecture or prevents upgrades. While Moore's Law still holds, anything that uses a direct access model has a strict lifetime and it's less than ten years. As you say, now the overheads are small, there is rarely significant advantage to doing it.

Yes, can't argue too much that for modern desktops you'd not want to tie the two together.