Ziriax 100% (Ultimate Edition) - Page 4

Keir · 27 February 2018, 12:10

I debugged this for FlashFloppy and the bug is in Ziriax track loader.

For efficient sector scanning it skips 300*2 MFM bytes at the track gap. However this assumes the track is at least (544*11 + 300)*2*8 = 100544 bitcells long.

However a DD track is 100000 bitcells +/- perhaps 2%. Usually drives run a bit slow and you end up with a long enough track, but this is certainly not guaranteed.

Works on HxC because it generates long-ish tracks for ADF (101372 bitcells).

I fixed the track loader to skip 0x11A (=141*2) MFM bytes instead (conservatively assumes track is >= 98000 bitcells) and that works fine for me on FlashFloppy.

ross · 27 February 2018, 12:37

Quote:

Originally Posted by kaffer

I debugged this for FlashFloppy and the bug is in Ziriax track loader.

For efficient sector scanning it skips 300*2 MFM bytes at the track gap. However this assumes the track is at least (544*11 + 300)*2*8 = 100544 bitcells long.

However a DD track is 100000 bitcells +/- perhaps 2%. Usually drives run a bit slow and you end up with a long enough track, but this is certainly not guaranteed.

Works on HxC because it generates long-ish tracks for ADF (101372 bitcells).

I fixed the track loader to skip 0x11A (=141*2) MFM bytes instead (conservatively assumes track is >= 98000 bitcells) and that works fine for me on FlashFloppy.

HxC is not wrong (neither Ziriax loader).

Bitcells on Amiga (generated by Paula) is a multiple of CCK.
So in not exactly 2us but (1/7093790)*2*7=1,97356us
In a rotation you have (,200/dur)=101340 cells

Keir · 27 February 2018, 13:07

Quote:

Originally Posted by ross

HxC is not wrong (neither Ziriax loader).

Bitcells on Amiga (generated by Paula) is a multiple of CCK.
So in not exactly 2us but (1/7093790)*2*7=1,97356us
In a rotation you have (,200/dur)=101340 cells

You need to allow for motor variations, eg Teac 235HF allows up to +/- 2%. Even assuming standard length 101,340 bitcells it would therefore be sane to allow a min track length of just over 99,000 bitcells.

Also consider that your ADF may not be written on an Amiga but by eg Catweasel, Kryoflux, Supercard Pro.

EDIT: The insight on Paula bitcell generation is useful though thanks. I will confirm by logic analyser and probably increase my ADF track length in that case. But I certainly still argue that your loader is not sufficiently conservative.

ross · 27 February 2018, 13:16

Quote:

Originally Posted by kaffer

You need to allow for motor variations, eg Teac 235HF allows up to +/- 2%. Even assuming 101,340 bitcells it would therefore be sane to allow a min track gap of just over 99,000 bitcells.

Also consider that your ADF may not be written on an Amiga but by eg Catweasel, Kryoflux, Supercard Pro.

Also consider that ADF is for Amiga so your firmware need to accomplish this.
But you are free to hack the track loader

EDIT: Anyway for next track loader I will be more relaxed

Jeff_HxC2001 · 27 February 2018, 13:40

Quote:

Originally Posted by ross

HxC is not wrong (neither Ziriax loader).

Bitcells on Amiga (generated by Paula) is a multiple of CCK.
So in not exactly 2us but (1/7093790)*2*7=1,97356us
In a rotation you have (,200/dur)=101340 cells

100% Agree ! Of course bitrate on amiga is not strictly 250000bps !

There is several demos and games relying on the true hardware bitrate/track length.

Note : NTSC and PAL Amiga machines have a different bitrate too

Using a fixed 100000 bits track for Amiga is definitively a mistake.

ross · 27 February 2018, 13:58

Quote:

Originally Posted by Jeff_HxC2001

Note : NTSC and PAL Amiga machines have a different bitrate too

Yes, and even more cells on NTSC

ross · 27 February 2018, 14:53

I remembered seeing a thread in which the question was deepened and I finally found it:
http://eab.abime.net/showthread.php?...erry+track+gap

Can be useful.
There is a very interesting suggestion by Toni (like usual) to dynamically calculate (and skip) the gap!
It would be interesting to know what Blueberry did.

Keir · 27 February 2018, 15:37

Quote:

Originally Posted by Jeff_HxC2001

100% Agree ! Of course bitrate on amiga is not strictly 250000bps !

There is several demos and games relying on the true hardware bitrate/track length.

Note : NTSC and PAL Amiga machines have a different bitrate too

Using a fixed 100000 bits track for Amiga is definitively a mistake.

This precise Amiga timing had passed me by completely.

Now confirmed on the logic analyser and, notwithstanding whether it is sensible for a trackloader to depend on it, it makes sense to lengthen the generated track accordingly for extra compatibility.

Jeff_HxC2001 · 27 February 2018, 16:15

Quote:

Originally Posted by ross

I remembered seeing a thread in which the question was deepened and I finally found it:
http://eab.abime.net/showthread.php?...erry+track+gap

Can be useful.
There is a very interesting suggestion by Toni (like usual) to dynamically calculate (and skip) the gap!
It would be interesting to know what Blueberry did.

And here an quite old discussion on the HxC forum :
http://torlus.com/floppy/forum/viewt...0&p=1837#p1834

Superman · 27 February 2018, 17:08

Now working with flash floppy firmware 0.9.12a

ross · 27 February 2018, 19:11

Quote:

Originally Posted by kaffer

Now confirmed on the logic analyser and, notwithstanding whether it is sensible for a trackloader to depend on it, it makes sense to lengthen the generated track accordingly for extra compatibility.

Good! Thanks Keir.

Quote:

Originally Posted by Jeff_HxC2001

And here an quite old discussion on the HxC forum :
http://torlus.com/floppy/forum/viewt...0&p=1837#p1834

Interesting reading, thanks Jeff.

Quote:

Originally Posted by Superman

Now working with flash floppy firmware 0.9.12a

Amiga1992 · 13 March 2018, 16:41

Will anybody back-port changes or support this ADF version on WHDLoad? Would be really good to have.

ross · 13 March 2018, 17:42

Quote:

Originally Posted by Akira

Will anybody back-port changes or support this ADF version on WHDLoad? Would be really good to have.

Volunteer wanted!

Nah, i'll look at it, but.. this in not an 'original' so only an 'unofficial' version can be made.

Or an 'official' version will arise

For those who are wondering what the problems of the existing whdload version:
- high score's handler is broken
- can crash on level 3 and in some particular situation
- garbage can appear ofter final boss and exit from extro need patching
- invincibility makes you play with an annoying blinking ship and better trainer is needed
- minor fix here and there

Amiga1992 · 13 March 2018, 18:49

I thought that sometimes slaves could be modified to support ADFs when only a non-original is worth it. I think this would be the case, but let the WHDLoad slave makers decide!
I posted a req' on the appropriate forum.

Galahad/FLT · 13 March 2018, 22:20

Quote:

Originally Posted by Akira

I thought that sometimes slaves could be modified to support ADFs when only a non-original is worth it. I think this would be the case, but let the WHDLoad slave makers decide!
I posted a req' on the appropriate forum.

The rule is originals only, but there is an acceptance that some games are incredibly rare, so occasional support for a cracked version here or there is tolerated.

I did the install for Cavemania, if i didnt support the Quartex crack, about 4 people would have been genuinely able to install the game from an original.

As Ziriax is a bit of a special case, it would cause no problems, and the author of the game is happy with what Ross did, so thats the legality covered in any case

Amiga1992 · 14 March 2018, 00:55

Thanks for clarifying Galahad, that was my exact understanding.
Hopefully Stingray can and wants to look into it, as he was involved with the slave.

StingRay · 14 March 2018, 07:34

Quote:

Originally Posted by Akira

Hopefully Stingray can and wants to look into it, as he was involved with the slave.

My work on the slave is not finished, I only updated it to fix the most annoying bugs. I'll definitely add a much better trainer and once done I'll check the complete game for any problems and will fix them. All this will take a little while though as I'm currently busy with other projects.

Amiga1992 · 14 March 2018, 15:05

Thanks for looking into it, you are always busy I know.
Whenever it comes it will be very welcome.

jdryyz · 22 March 2019, 06:57

I have two systems I tested this on: an accelerated (ACA1232) KS 3.1 PAL A1200 and an A2000 with GVP 030 (1M chip, 13M fast, KS3.1.4).

It boots fine from the A1200 but just freezes on the A2000 (booted from 1.0fw Gotek or real floppy). WHDLoad version works on this machine. I want to see the excellent Ross trainer though!

Perhaps a KS 3.1.4 problem? I can test more with WinUAE.

dlfrsilver · 22 March 2019, 10:18

Quote:

Originally Posted by Akira

I thought that sometimes slaves could be modified to support ADFs when only a non-original is worth it. I think this would be the case, but let the WHDLoad slave makers decide!
I posted a req' on the appropriate forum.

Please note that i made an extended ADF usable with whdload, same for many other originals not easily available.

they are available on the EAB FTP.

27 February 2018, 12:10	#61
Keir Registered User Join Date: May 2011 Location: Cambridge Posts: 682	I debugged this for FlashFloppy and the bug is in Ziriax track loader. For efficient sector scanning it skips 3002 MFM bytes at the track gap. However this assumes the track is at least (54411 + 300)28 = 100544 bitcells long. However a DD track is 100000 bitcells +/- perhaps 2%. Usually drives run a bit slow and you end up with a long enough track, but this is certainly not guaranteed. Works on HxC because it generates long-ish tracks for ADF (101372 bitcells). I fixed the track loader to skip 0x11A (=1412) MFM bytes instead (conservatively assumes track is >= 98000 bitcells) and that works fine for me on FlashFloppy. Last edited by Keir; 27 February 2018 at 12:22.*

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Ziriax - Great amiga shmup	Shatterhand	Nostalgia & memories	42	18 April 2016 14:43
Ziriax, Zarathrusta and Deliverance	CodyJarrett	project.Amiga Lore	6	07 March 2013 02:50
Gloom 3 ''Ultimate Gloom'' Zombie Edition (CD32)	bagpipes	support.WinUAE	25	22 September 2010 19:53
ZIRIAX at last !!!	dlfrsilver	project.SPS (was CAPS)	6	29 March 2006 23:11
adf to disk and ziriax	tassy	New to Emulation or Amiga scene	4	10 March 2002 22:42

27 February 2018, 14:53	#67
ross Defendit numerus Join Date: Mar 2017 Location: Crossing the Rubicon Age: 53 Posts: 4,468	I remembered seeing a thread in which the question was deepened and I finally found it: http://eab.abime.net/showthread.php?...erry+track+gap Can be useful. There is a very interesting suggestion by Toni (like usual) to dynamically calculate (and skip) the gap! It would be interesting to know what Blueberry did.

27 February 2018, 17:08	#70
Superman Super Member Join Date: Sep 2014 Location: Wakefield Age: 48 Posts: 1,334	Now working with flash floppy firmware 0.9.12a

13 March 2018, 16:41	#72
Amiga1992 Registered User Join Date: May 2001 Location: ? Posts: 19,645	Will anybody back-port changes or support this ADF version on WHDLoad? Would be really good to have.

13 March 2018, 18:49	#74
Amiga1992 Registered User Join Date: May 2001 Location: ? Posts: 19,645	I thought that sometimes slaves could be modified to support ADFs when only a non-original is worth it. I think this would be the case, but let the WHDLoad slave makers decide! I posted a req' on the appropriate forum.

14 March 2018, 00:55	#76
Amiga1992 Registered User Join Date: May 2001 Location: ? Posts: 19,645	Thanks for clarifying Galahad, that was my exact understanding. Hopefully Stingray can and wants to look into it, as he was involved with the slave.

14 March 2018, 15:05	#78
Amiga1992 Registered User Join Date: May 2001 Location: ? Posts: 19,645	Thanks for looking into it, you are always busy I know. Whenever it comes it will be very welcome.

22 March 2019, 06:57	#79
jdryyz Registered User Join Date: Jul 2014 Location: USA Posts: 139	I have two systems I tested this on: an accelerated (ACA1232) KS 3.1 PAL A1200 and an A2000 with GVP 030 (1M chip, 13M fast, KS3.1.4). It boots fine from the A1200 but just freezes on the A2000 (booted from 1.0fw Gotek or real floppy). WHDLoad version works on this machine. I want to see the excellent Ross trainer though! Perhaps a KS 3.1.4 problem? I can test more with WinUAE.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)