Should game devs have to support new hardware?

[grond]

Originally Posted by DMWCashy

Grond needs to show his cards now, we are all calling.

I don't play by rules set by an internet mob.

[Tigerskunk]

Originally Posted by grond

I don't play by rules set by an internet mob.

[h0ffman]

It depends on what your intention is. If you're goal is to make a game you want to sell, then you really have an obligation to make sure it works on as much hardware as possible. If you're releasing it for free you are under zero obligation. Do whatever you want. The community might (will) complain that it doesn't work on [insert name of dodgy accelerator card] but its your game and YOUR TIME!.

[malko]

Maybe some lights can be put on the topic for "asm noobs" like me to have a better understanding on the real issue ?
Without being too technical @McGeezer & @Steril707 what is disturbing so much that at the end there is less fun in coding ? Can you elaborate ?
I see normal that a code run faster on a 68020 (or 68060) than on a 68000. Is the code to handle such situation that reluctant that it may discourage from coding ?
Isn't the base instruction set the same for all the CPU family ?
Or is the "handling"* of a zorro controller that difficult to prevent a computer crash ?
You said you gained knowledge from the help of ross. Is that knowledge difficult to reuse ?
As I am learning 68k asm on my spare time and for my pleasure, it's a real question. Thanks for you enlightenment.

*by handling I mean on the Amiga world (no 1000 Chinese copy of the same hardware with compatibilies issues) : "Oh, you're there. Ok but I don't care as I don't need you. And because you are there I will not let you crash my code as I know what addresses you like to visit, you naughty card ".

[Dunny]

The problem here is that some people have "upgraded" their Amigas to use Vampires or 060s and they're now salty that coders 30 years ago didn't build-in support for their follies.

So if the software says "Requires A500 512KB RAM" and it fails to run on your A4000 then you're the one at fault, not the developer.

[Tigerskunk]

Originally Posted by h0ffman

If your goal is to make a game you want to sell, then you really have an obligation to make sure it works on as much hardware as possible.

Why?
You could just state it's for stock A500s or stock A1200s.
Whom do you owe this obligation to?
People with accelerator cards?

Originally Posted by malko

Maybe some lights can be put on the topic for "asm noobs" like me to have a better understanding on the real issue ?
Without being too technical @McGeezer & @Steril707 what is disturbing so much that at the end there is less fun in coding ? Can you elaborate ?

I see normal that a code run faster on a 68020 (or 68060) than on a 68000. Is the code to handle such situation that reluctant that it may discourage from coding ? Or is the "handling"* of a zorro controller that difficult to prevent a computer crash ?

As I am learning 68k asm on my spare time and for my pleasure, it's a real question. Thanks for you enlightenment.

I can only talk from my perspective here, Graeme's might differ from mine.

Off course you can create code in a way that it runs on as many systems as possible, but you will never be able to ensure that anyway.
There are way too many configs around and you cannot test all of them.

Even regularly testing on A500 and A1200 is a chore.
I had a couple of incidences where I haven't been testing on either for a longer time, and when I did again stuff looked and behaved different due to some factors.

This shit takes time.
The thing is, creating a game is already taking up hundreds to thousands of hours.
Doing this a the proverbial bedroom coder, you want to maximize your fun with this.

Ensuring your game runs on different configs is time consuming and NOT FUN.

Most of the stuff I did for Inviyya involving different configs felt like work. And if somethings feels like work and not fun, I wanna get paid for it.

Which won't happen with those few sales that are to be expected.

So, I decided for my next game I will be doing it for having fun again, and not that someone is happy he can use his Frankenstein 68060 setup PCish Amiga the way he thinks and not like I want the game to be played (which is oldschoolish via disk and A500/1000).

Not hating here on anyone for having his own idea about what an Amiga is, but if it's a hobby, it needs to feel like one and be fun for me, and not feel like work.

And QA stuff and doing regression testing on different setups is not my idea of fun.

So, in the future it's A500/Floppy/ADF for me, and if it runs on something else it's amazing, but not my goal (and I guess with Ross's startup code being present in my engine, it will run on most stuff anyway).

[Mathesar]

I voted "They should support accelerators in all their prods" by accident but I actually meant to vote "They should only target stock Amigas, let the WHD team fix the gltiches".

Like I wrote elsewhere, fast Amiga's were still relevant back in the day. I know, because I had and still have one. And I love it very much.
Nowadays however, for the Amiga as a retro platform, only A500 and A1200 is relevant.
I can imagine however that people want some extra RAM and a harddisk for WHDload. As a bonus, with some FAST ram, demanding Amiga games will run a bit faster too (Lemming's NUKE button, Dune II, Tinyus and the upcoming Dread for example) but that is about it.
No need for 060's and certainly no need for Vampires imho....

[grond]

Originally Posted by Dunny

The problem here is that some people have "upgraded" their Amigas to use Vampires or 060s and they're now salty that coders 30 years ago didn't build-in support for their follies.

If I ever was salty about bad coding practices, it was in the early 90s when the abundance of bad code sabotaged any technical progress of the Amiga beyond the 68000/OCS base configuration. Too many users treated the Amiga as a games console (and probably wondered why it came with a keyboard) and too many coders didn't care about understanding its inner workings.

If you want to test how your code works on anything else than the OCS base configuration, simply run it in UAE with CPU to fastest and do not tick "immediate blitter". This will uncover most faults caused by asynchronicity of the CPU and the custom chips.

[Tigerskunk]

Originally Posted by Dunny

The problem here is that some people have "upgraded" their Amigas to use Vampires or 060s and they're now salty that coders 30 years ago didn't build-in support for their follies.

That said, I found the Vampire 1200 to be very compatible. My game worked without any problems on it even before Ross had a look at my startup code.
Same with 68060 CPUs. Someone had a run with it on those, and also on 68030s. Worked everywhere.

But then you got some people telling you they get weird glitches with the ACA500 plugged in. I plugged my own ACA500+ in, and had no problems.
Once more, how to handle bullshit like this? Without spending weeks over weeks or even years just trying to get your stuff running with every hardware existing.

You just cannot make everybody happy on the Amiga.
In the end it's all about finding your own happiness and fun creating something.

[dreadnought]

Originally Posted by Steril707

You just cannot make everybody happy on the Amiga.
In the end it's all about finding your own happiness and fun creating something.

/thread

[Galahad/FLT]

Originally Posted by grond

If I ever was salty about bad coding practices, it was in the early 90s when the abundance of bad code sabotaged any technical progress of the Amiga beyond the 68000/OCS base configuration. Too many users treated the Amiga as a games console (and probably wondered why it came with a keyboard) and too many coders didn't care about understanding its inner workings.

If you want to test how your code works on anything else than the OCS base configuration, simply run it in UAE with CPU to fastest and do not tick "immediate blitter". This will uncover most faults caused by asynchronicity of the CPU and the custom chips.

You say "bad code", but a lot of it was simply due to how development teams worked when they had to convert several formats and the rather tedious influence of the Atari ST.

I think that had the ST not been such an influence, the urge to make bespoke and better Amiga titles would have occurred two years earlier.

What that also meant was that developers would have been hungry to move onto more than the A500 sooner because those extra two years would have been skills in hitting the technical ceiling on the A500 sooner.

Yes there was bad code, but there was also the publishers who were simply "that will do".

With most programmers on Amiga, and having bug fixed a lot of shit back in the day, most of them had a clear progression where the code was less buggy and got better.

However, the nature of the Amiga with its reliance on hardware, doesnt help much when it comes to 68040/60, because now programmers need to avoid the hardware and simply have the CPU do most stuff, the direct opposite of how programmers were shown to get the best out of the Amiga.

[h0ffman]

Originally Posted by Steril707

Why?
You could just state it's for stock A500s or stock A1200s.
Whom do you owe this obligation to?
People with accelerator cards?

Yeah you could but seems a bit counter productive to your sales if you're going to cut out the large chunk of people who have upgraded their Amigas.

[chiark]

Originally Posted by Galahad/FLT

However, the nature of the Amiga with its reliance on hardware, doesnt help much when it comes to 68040/60, because now programmers need to avoid the hardware and simply have the CPU do most stuff, the direct opposite of how programmers were shown to get the best out of the Amiga.

That's what makes the Amiga "special" to me, and using brute force with a CPU doesn't strike me as elegant - even though getting the best from 040+ will undoubtedly rely on optimising your code to the nth degree.

Is there 'reference' safe startup code anywhere? (Galahad, I think you may well have contributed a lot to such a thing )

[DMWCashy]

Originally Posted by grond

I don't play by rules set by an internet mob.

So are you not going to show us your work now, saying you would ?

If you no longer what to show us any of your work that is fine, but please say so.

As for rules, will you abide by the rules of this forum and thread?

[DMWCashy]

Originally Posted by grond

If you want to test how your code works on anything else than the OCS base configuration, simply run it in UAE with CPU to fastest and do not tick "immediate blitter". This will uncover most faults caused by asynchronicity of the CPU and the custom chips.

Questions if you develop for a day job, as anyone that does would know real world environments are not so black and white. Testing can take exponentially more time that developing the application, so adding an unknown number of configurations to the mix, is just crazy talk.

[Tigerskunk]

Originally Posted by h0ffman

Yeah you could but seems a bit counter productive to your sales if you're going to cut out the large chunk of people who have upgraded their Amigas.

I think I will be able to talk about this with more evidence after my game is in the stores and I see who buys what, but in the end you really really should not create something for the Amiga because of "sales".
This "market", if you could even call it one, is much too small. We are talking about a few hundred sales in all.

Let's say you make 5 to 7 bucks after deducting digital distributor costs, boxed set printing, cover illustration, floppies, manual printing, patch printing, poster printing, etc.
So we are talking about less than a typical coder's months salary here, even before tax deductions.


And now you think it makes sense to spend another 100 hours to make sure your game works on every config under the sun?
I don't say I want to cut out people who have upgraded their Hardware. I just think it doesn't make sense to accomodate for every upgrade or medium.

Have a look into the Wiz sales thread here:
http://eab.abime.net/showthread.php?...&highlight=wiz

The CD32 version of my game was around 20 hours of conversion work and testing on different setups (tf330 etc) already.
And that's for expected twenty to fifty sold units?

Like I said, I wanna do this stuff for fun in the future. And "sales" is really not an argument against that.


add/edit: once again, if this stuff is fun for you, all power to you. It's just not fun for me, so given the choice of leaving Amiga dev completely (which I was thinking about) or just keep doing stuff I love and don't care about the rest, I take the second option and just do my games for the A500/1000.

add/edit2: also, how about the idea of leading those who are interested in gaming on the Amiga in using a stock A500 or A1200 for that.

[h0ffman]

Fair point.

[DanielAllsopp]

Originally Posted by mcgeezer

I'll tell you what bonny lad...

That phrase makes me feel right at home! Get in.

[grond]

Originally Posted by Steril707

But then you got some people telling you they get weird glitches with the ACA500 plugged in. I plugged my own ACA500+ in, and had no problems.
Once more, how to handle bullshit like this?

That would be a case of shrugging your shoulders and moving on in my view. If you are very enthusiastic about it, you could inform the manufacturer (which is still available in this case) and see what he thinks about it. I'm pretty sure he will just as well shrug his shoulders and not care to find out what's the problem.

[grond]

Originally Posted by Galahad/FLT

You say "bad code", but a lot of it was simply due to how development teams worked when they had to convert several formats and the rather tedious influence of the Atari ST.

Yes, I am aware that the situation back then was different. But this discussion is about how coders should write code today. We now have more than plain 68000s out there and we have had them for decades. By now some bad habits of old should have died out.

However, the nature of the Amiga with its reliance on hardware, doesnt help much when it comes to 68040/60, because now programmers need to avoid the hardware and simply have the CPU do most stuff, the direct opposite of how programmers were shown to get the best out of the Amiga.

Well, you can have the 040/060 idle 95% of the time while the unchanged OCS/ECS/AGA hardware still does its job like it does on a base configuration. But writing code such that it will break on 040/060 just because you don't care about anything else than base configs seems very lame to me. There may be some cases were you just can't put in some extra code for keeping things in synchonicity on fast CPUs but they are much rarer than some seem to think.