Rink-a-Dink Redux

[tomcat666]

Seeing as Psygore made a whdload of the original, it would be nice to see the new version running in whdload :

http://www.pouet.net/prod.php?which=61182

Very much improved effects

[StingRay]

Will be released in the next days, I actually planned to make a WHDLoad patch for this demo right from the moment it was shown at Revision (which was an awesome party!).

[tomcat666]

Nice Sting!

[DDNI]

Nice! Thanks.

It won the OldSkool demo too.
http://pouet.net/results.php?which=1550&when=13&f=4

[StingRay]

Sure did. Which was clear right from the beginning (for me at least), one of the best releases at Revision for me.

[DDNI]

Wish I had been there. Always trying to get back to country of my birth, but never quite find the time

[StingRay]

Next year then. You won't regret it.

[hooverphonique]

@stingray.. you might want to hold that off for a few days while paradroid finishes the final version, and then make a patch for that :-)

[StingRay]

I know that he's working on a final but the patch will support both versions and will be updated when (or if) necessary.

[Paradroid]

final will be few days away at least, I'm trying to relax a bit at the moment -_-.zZ

Stingray, if you happen to find why the damn thing crashes on an A4000 (as the small lemon logo flies across the screen) please let me know as it'll save me some trouble. I did have a quick look while I was at the party, but the problem wasn't what I thought it was. I'm betting it's something dumb that is fluke working on the other amigas - I wrote a lot of code while half asleep, so bound to be a few problems lurking in there.

Also, I rewrote the plasmas near the beginning for the A1200 to avoid my blitter naughtiness and leave the caches on, but they run like a lame donkey on the A4000. Does the A4000 have issues when running code in chip ram?

[tomcat666]

I think Retroplay already fixed it for AGA ... check this thread (last post)

http://eab.abime.net/showthread.php?...highlight=rink

It worked fine for me on A1200

[Paradroid]

Originally Posted by Paradroid

I'm betting it's something dumb that is fluke working on the other amigas - I wrote a lot of code while half asleep, so bound to be a few problems lurking in there.

oh dear, just found it...



...I'm trashing code, but with normal floppy reading speeds the code in question will be trashed long after it's executed. Spotted within 2mins of looking at the source now I'm rested, but in my post party deadline state I was stumped for over 2 hours before giving up... Stupid thing is that the line of code that is doing the trashing isn't even needed anymore as I changed my base library to handle the situation it's trying work around, lol

[StingRay]

Oh yes, I know that feeling too well. Good thing you found the bug though! Also, thanks for the "hidden messages are fun" attitude, the first one already made me smile, let's see if there are more.

[Retroplay]

Originally Posted by tomcat666

I think Retroplay already fixed it for AGA ... check this thread (last post)

http://eab.abime.net/showthread.php?...highlight=rink

It worked fine for me on A1200

No I didn't fix anything (I'm no coder), I just downloaded the AGA fixed version from scene.org and zoned it.
http://tinyurl.com/cl83rmq

[Paradroid]

Originally Posted by Retroplay

I just downloaded the AGA fixed version from scene.org and zoned it.

Ah, makes sense now... The "fix" mentioned in the filename isn't anything to do with AGA fixes. The night before the demo had to be submitted I was continuing work on it at the hotel, but instead of sitting at a table I decided to sit on the nice comfy bed... so sooooo comfy.... I didn't set an alarm and woke up, laptop on top of me, late in the morning without time to finish it, so I submitted a broken build. The organisers gave me some time to finish it after the dealine, and that's why the filename says "fixed_party_version"

[hooverphonique]

Originally Posted by Paradroid

Also, I rewrote the plasmas near the beginning for the A1200 to avoid my blitter naughtiness and leave the caches on, but they run like a lame donkey on the A4000. Does the A4000 have issues when running code in chip ram?

I'm not entirely sure, but chipram could be non-cacheable on the A3640 (A4000 040 cpu board), which would give terrible speeds on code in chipram, even for smallish loops..

[Galahad/FLT]

Originally Posted by Paradroid

final will be few days away at least, I'm trying to relax a bit at the moment -_-.zZ

Stingray, if you happen to find why the damn thing crashes on an A4000 (as the small lemon logo flies across the screen) please let me know as it'll save me some trouble. I did have a quick look while I was at the party, but the problem wasn't what I thought it was. I'm betting it's something dumb that is fluke working on the other amigas - I wrote a lot of code while half asleep, so bound to be a few problems lurking in there.

Also, I rewrote the plasmas near the beginning for the A1200 to avoid my blitter naughtiness and leave the caches on, but they run like a lame donkey on the A4000. Does the A4000 have issues when running code in chip ram?

Some a4000's were released with a knackered buster chip where some high words of addresses in the registers are too big for the memory range accessed causing non critical bus errors. Chuck in an error handler to see if that's the cause

[StingRay]

A4000 Chip RAM is VERY slow, avoid if you can!

[hooverphonique]

Originally Posted by Galahad/FLT

Some a4000's were released with a knackered buster chip where some high words of addresses in the registers are too big for the memory range accessed causing non critical bus errors. Chuck in an error handler to see if that's the cause

if you're using an "illegal" address, I guess it's ok to get a bus error, don't you? I mean why would it be considered wrong (due to a "knackered" buster) to get bus errors in this case? (If I misunderstood what this issue is, please elaborate ;-)

maybe this tool helps in such cases? http://aminet.net/package/dev/debug/berm

[Paradroid]

it's the none caching that's the problem. I've just tried disabling the cache on the A1200 and it started running the plasmas at a similar foul speed. I'm now rewriting the whole effect again as all I'd done for the A1200 was take the A500 code, switch the unrolled main loop out for a dbf loop (so it would cache) and stuck in a blitwait. Instead I'll have the CPU write out just the base colours once and have the copper trigger some blits to duplicate the colours to the other rows.

Fingers crossed that'll do the job as I want to go back to sleep

PS. Is my assumption that the A4000 can have just chip ram in it correct, or did they all come with some form of fast ram?