Discussion: Rygar AGA Edition

[mcgeezer]

Originally Posted by ross

Unfortunately it would no longer be distributable as an ADF (but eADF) with all associated problems..

And I need this to work on my Gotek.

Me thinks we will stick with what we gots...

In other news...

I've now spent 3 hours trying to isolate a stupid bug and I just about have found where it is.... it will have to wait until tomorrow.

The joys of writing games in assembler!

[jotd]

I was suggesting this as a joke. Standard ADF is the way. Maybe with extreme compression/executable packing with powerpacker/imploder it could reduce size ?

[mcgeezer]

Originally Posted by jotd

I was suggesting this as a joke. Standard ADF is the way. Maybe with extreme compression/executable packing with powerpacker/imploder it could reduce size ?

Sorry buddy it was lost on me so I can only apologise.

I've done a bit of testing with packers lately and I might be able to get a better compression with 'PackFire' I think it was.... not tried yet though.

[Foul]

try this one : http://neoscientists.org/~bifat/bina...llery/cranker/

Modern packer and multi-plateform...

[mcgeezer]

Originally Posted by Foul

try this one : http://neoscientists.org/~bifat/bina...llery/cranker/

Modern packer and multi-plateform...

I tried cranker already.... propack beat it by a long way.

[Foul]

ok

[ross]

Originally Posted by jotd

I was suggesting this as a joke.

Zut! I was already planning the format

Originally Posted by mcgeezer

I've done a bit of testing with packers lately and I might be able to get a better compression with 'PackFire' I think it was.... not tried yet though.

Surely you can gain in respect to other compressors 'cause it's based on LZMA (7z compression).
But it require two muls/token for the range decoder so sure is slow..
(also you must also exclude Shrinkler which is amazing but absolutely unusable in your case)

Originally Posted by Foul

try this one : http://neoscientists.org/~bifat/bina...llery/cranker/

Modern packer and multi-plateform...

Yes, good but also not so useful for Rygar. It's based on LZO so compression is mediocre (but anyway fast..)


EDIT:
You have to give ARJm7 a chance, it will surprise you

[mcgeezer]

Originally Posted by ross

Surely you can gain in respect to other compressors 'cause it's based on LZMA (7z compression).
But it require two muls/token for the range decoder so sure is slow..
(also you must also exclude Shrinker which is amazing but absolutely unusable in your case)

The speed of depacking does not worry me at all...

I can display a loading progress bar when the game is loading. For the round loading I can load the next round while the sanctuary bonus is being shown using the VBI so loading and depacking will always finish in the required time.

[mcgeezer]

Originally Posted by ross

EDIT:
You have to give ARJm7 a chance, it will surprise you

Seen this mentioned a few times now so will try it!

[Don_Adan]

Originally Posted by ross

Zut! I was already planning the format


Surely you can gain in respect to other compressors 'cause it's based on LZMA (7z compression).
But it require two muls/token for the range decoder so sure is slow..
(also you must also exclude Shrinkler which is amazing but absolutely unusable in your case)


Yes, good but also not so useful for Rygar. It's based on LZO so compression is mediocre (but anyway fast..)


EDIT:
You have to give ARJm7 a chance, it will surprise you

If I remember right after many years. LZMA need two mulu.w only for 68000. Because Rygar is AGA (68020+), only one mulu.l is necessary. Then depacking will be much fastest for 68020+. LZMA need tables too, I dont remember exactly how big, 16KB-32KB. Then is not very good if all memory (2MB of chip ram) is used for game/demo. Anyway, i think that ARJ can do packing job without problem for Rygar AGA.

[ross]

Originally Posted by Don_Adan

If I remember right after many years. LZMA need two mulu.w only for 68000. Because Rygar is AGA (68020+), only one mulu.l is necessary. Then depacking will be much fastest for 68020+. LZMA need tables too, I dont remember exactly how big, 16KB-32KB. Then is not very good if all memory (2MB of chip ram) is used for game/demo. Anyway, i think that ARJ can do packing job without problem for Rygar AGA.

You are right
Actually is a single 11bitsx21bits unsigned multiplication, so the only way on 000 is to spread on two 16x16 mulu
(the result fits in 32 bits, this mean no problem also on 060).

Even ARJ need a quite big buffer for the tables (circa 12 KB), but can be recycled the MFM buffer that is a little more than that.

[mcgeezer]

Could someone point me in the direction of the ARJ packer please? I'm having a little trouble finding it on Aminet.

I'll pack the files tonight and see what the outcome is.

[jotd]

Maybe packer is just available on windows (which isn't a big deal as long as the unpacker is available as 68k code). If it's the same, ARJ was developped on msdos by Robert Jung (hence the name of the format)

[Don_Adan]

Originally Posted by mcgeezer

Could someone point me in the direction of the ARJ packer please? I'm having a little trouble finding it on Aminet.

I'll pack the files tonight and see what the outcome is.

http://leonard.oxg.free.fr/download/...odeVersion.zip

arjbeta.exe this is PC exe, similar like for LZMA.

[Don_Adan]

Originally Posted by ross

You are right
Actually is a single 11bitsx21bits unsigned multiplication, so the only way on 000 is to spread on two 16x16 mulu
(the result fits in 32 bits, this mean no problem also on 060).

Even ARJ need a quite big buffer for the tables (circa 12 KB), but can be recycled the MFM buffer that is a little more than that.

I dont remember than ARJ need big tables for depacking. No space for big tables in 0.5M game, like Turrican 2 or BC Kid. But maybe I remember something wrong.

[ross]

Originally Posted by Don_Adan

I dont remember than ARJ need big tables for depacking. No space for big tables in 0.5M game, like Turrican 2 or BC Kid. But maybe I remember something wrong.

From the original Mr Ni! source:

workspacesize   EQU 11312

And yes, in all LZ-H unpacker the huffman tree can be also not in table form, but then speed is a no-go on bare 68k.
I'm pretty sure that BC.Kid recycled the MFM buffer for the job

[ross]

Originally Posted by mcgeezer

Could someone point me in the direction of the ARJ packer please? I'm having a little trouble finding it on Aminet.

I'll pack the files tonight and see what the outcome is.

You have a PM

[mcgeezer]

For those who are interested I'm very likely going to run a rough live You tube stream tonight of how I go about creating the configuration for rounds 9,10 and possibly 11.

You might find it interesting and I'm happy to answer any questions.

It'll be after my little boy goes to bed (around 9.30 GMT+1 / BST).

Here's my channel on Youtube.

https://www.youtube.com/channel/UChX...jLg-zYZY7cDc7A

Geezer

[mcgeezer]

And here is a recording of the stream I did earlier...

[ Show youtube player ]

[malko]

nice