Unicorns season: Prince of Persia for the A8!

[+x=usr(1536)]

3 hours ago, rensoup said:

So let's get started...

 

Are you using fujinet for loading ? Any way I can test this in Altirra ?

Yep, using the FujiNet to load the game is a necessity for me - I don't have any usable blank floppies right now, and even if I did I'd only be able to use the SD version as none of my drives are DD.

 

As for Altirra, I'm not aware of any software emulations of the FujiNet that might work with it.  Than again, I'm not using Windows, so am also not 100% aware of what may or may not be available for it.

Quote

Yes, I need to not forget to fix that

No worries ?

Quote

a video recording would be nice 

Let me see what I can do.  Right now, the physical arrangement of the room that I'm testing in makes that...  Challenging.

Edited June 14, 2021 by x=usr(1536)

[Synthpopalooza]

So, technical question:  How did you get the flickering font effect?  Was that a Antic 2/4 interlace, or did it involve PMG overlays?

[rensoup]

1 hour ago, x=usr(1536) said:

Yep, using the FujiNet to load the game is a necessity for me - I don't have any usable blank floppies right now, and even if I did I'd only be able to use the SD version as none of my drives are DD

I don't really have a clue how fujinet works, it's strange that the SD version works though... anybody else uses fujinet ?

Perhaps the fujinet author has some clues ?

 

Quote

1 hour ago, x=usr(1536) said:
Let me see what I can do.  Right now, the physical arrangement of the room that I'm testing in makes that...  Challenging.

if that's too much trouble, a screenshot would do

 

Edited June 14, 2021 by rensoup

[rensoup]

35 minutes ago, Synthpopalooza said:

So, technical question:  How did you get the flickering font effect?  Was that a Antic 2/4 interlace, or did it involve PMG overlays?

it's just modeF, flickering between 2 screens.

[tmp]

i was looking into why atr loads so slow here, i'm seeing tons of read retries, usually it reads 3 sectors fine, issues read cmd for 4th and then while the sector data is still being transmitted, it starts to retry that sector

sometimes i see 6 successful reads and then 7th retries but it's mostly 4th

 

[Mazzspeed]

I believe RespeQt is also showing sector retries. The loading process actually seems slower than SIO 1x:

 

 

[rensoup]

It's unclear how many crashing issues there are during loading... but I may have solved one.

 

600/800XL seem fine but switching to 1200XL in Altirra caused a crash at boot about 50% of the time. (I didn't think I had any kind of randomization switches turned on though ?)

 

My DLI assert was catching DLIs firing off although there shouldn't have been any at that point.

 

I was relocating a Display list, setting the DL pointer, but I forgot to update its loop address so it ended up jumping to some random location...

 

That was an easy fix... which didn't work at all...

 

It was still firing off DLIs ! I was pretty clueless at that point. I would trace the code after I had set my very basic blank DL and a very basic NMI for the VBI !

The culprit seems to be the DL DMA which was disabled, I'm guessing it was still somehow processing some old DL... so I reenabled it before setting the new DL and NMI and it seems ok now.

 

I've got a bunch more configs to try but they may well still crash... ?

 

 

[rensoup]

16 minutes ago, tmp said:

i was looking into why atr loads so slow here, i'm seeing tons of read retries, usually it reads 3 sectors fine, issues read cmd for 4th and then while the sector data is still being transmitted, it starts to retry that sector

sometimes i see 6 successful reads and then 7th retries but it's mostly 4th

 

So you'd say it's at worst 25% slower ?

 

11 minutes ago, Mazzspeed said:

I believe RespeQt is also showing sector retries. The loading process actually seems slower than SIO 1x:

 

I was told that because XBIOS copies bytes slowly it might miss IRQs and issue read retries, I have no clue if that's true...

 

There's the loading anim too which may slow things down but I don't think it should make that much of a difference because the anim is only updated every 16 frames (I did that specifically to try to avoid slowing it down)

 

You could try the same test with the last 2020 release and see if it got slower because of the anim...

 

Another thing you could try is the same test with Stuntcar since it also uses XBIOS

[Mazzspeed]

2 minutes ago, rensoup said:

You could try the same test with the last 2020 release and see if it got slower because of the anim...

An earlier release? I'll do it now.

 

EDIT: I just tried FujiNet using the DD ATR, speeds are the same, so I assume the same issue?

Edited June 14, 2021 by Mazzspeed

[Mazzspeed]

@rensoup, the 20200514.atr without the loading animation only had one retry under RespeQt. Loading seemed faster, but I didn't time it.

Edited June 14, 2021 by Mazzspeed

[tmp]

6 minutes ago, rensoup said:

So you'd say it's at worst 25% slower ?

i'd say way more since at least in my case, it often ends with (approx) 0.5s-1s delays due to desync and retrying

[tmp]

yes, 0514 looks flawless, i'm not seeing retries

[Mazzspeed]

Stunt Car Racer only has one retry, and that's when the blue loading screen appears. The animation of the little car to the finish line incurs no retries.

[rensoup]

41 minutes ago, Mazzspeed said:

@rensoup, the 20200514.atr without the loading animation only had one retry under RespeQt. Loading seemed faster, but I didn't time it.

 

30 minutes ago, tmp said:

yes, 0514 looks flawless, i'm not seeing retries

 

ouch... so that new anim is a problem... crazy

[xxl]

3 hours ago, rensoup said:

It was still firing off DLIs ! I was pretty clueless at that point. I would trace the code after I had set my very basic blank DL and a very basic NMI for the VBI !

The culprit seems to be the DL DMA which was disabled, I'm guessing it was still somehow processing some old DL... so I reenabled it before setting the new DL and NMI and it seems ok now.

 

2 hours ago, rensoup said:

2 hours ago, tmp said:

yes, 0514 looks flawless, i'm not seeing retries

 

ouch... so that new anim is a problem... crazy

 

you need to check DLI and VBI, it is very important that the runtime of these procedures during I / O (pooling) was as short as possible, I suspect there may be a problem.

also check the earlier version of xB - in the last one the number of allowed retries has been changed (allows you to load even if the number of erroneous readings is large)

 

---

a good effect can be achieved with simultaneous sector loading and decompression - gaps between sectors will be anyway and the amount of data transferred will increase significantly - this effect can be seen in my last appendices with music on GTIA.

Edited June 14, 2021 by xxl

[xxl]

6 hours ago, rensoup said:

Thanks, I thought I'd disabled it a long time ago...

check it out:

xIRQEN          equ xBIOS+$3e8 ; User IRQ (1 byte)

 

I don't know if you load config but it also overwrites xIRQEN

Edited June 14, 2021 by xxl

[Beeblebrox]

Pulled this from the Homebrew completed/WIP topic post( @ZeroPage Homebrew hope you don't mind(?)), to say thanks again to Rensoup and all the team for the excellent progress since the last release and ongoing work/snagging. Fingers crossed ya nail it.

 

Had no idea so many were involved:

 

@rensoup (Programming), @TIX (Character Art), @miker (Music + SFX Bank 1), @VinsCool (Music), @emkay (SFX Bank 2), @makary (SFX Bank 1), @dmsc (Soundplayer), @xxl (Disk Loading), @fox (Packer), @phaeron (Dev+Testing), Rich Talbot-Watkins (Assembler), @elmer (aplib depacker)

 

@rensoup Also, when I first heard about the project a month+ back I started a RastaConversion of the classic PoP title screen just to see what it would look like. I ended up doing a few, (early ones failed), and then I started this one, a 320*200 format so there was space at the bottom for your credits if desired/technically possible. I'd popped it on the back burner but just resumed it yesterday alongside some other RC conversions. 

 

So far it looks like this (employs all PMGs btw):

 

 

Some errors and the lower blue blocking isn't quite right. (Incidentally the lower screen white line error only appears in Altirra and not my A8 hardware).

 

I may well have a tweak of the source image and RC setting and restart a fresh run. Doing it for myself as much as anything, but if a good version emerges which can be used if possible and it helps, then great.

 

XEX attached. Very much a WIP/experiment as I say. 

 

 

princeofpersiascreen2incomplete8.xex

[Jacques]

@Beeblebrox

Thank you, looks very nice for starters, would love to see a classic title screen in the game.

That's screenshoted with scanlines? How would it look without them?

[Beeblebrox]

6 minutes ago, Jacques said:

@Beeblebrox

Thank you, looks very nice for starters, would love to see a classic title screen in the game.

That's screenshoted with scanlines? How would it look without them?

@Jacques  Load up the XEX in Altirra if you like. If I turn scanlines effect off (with my current Altirra settings still left in place as I ideally don't wanna change them), it looks like this:

 

Edited June 14, 2021 by Beeblebrox
typos

[Jacques]

7 minutes ago, Beeblebrox said:

Load up the XEX in Altirra if you like.

Will do on real machine this evening. I like it without scanlines even more. Thanks!

[Peri Noid]

9 hours ago, rensoup said:

?

how many times before this happens ? anyone else seen this ?

I tried to repeat it and let the computer to run for over 1 hour. The same machine, the same config. But it dodn't happen again. Maybe it was a kind of coincidence. It's been running now for over 70 minutes, I'll let it continue.

[drpeter]

47 minutes ago, Beeblebrox said:

Incidentally the lower screen white line error only appears in Altirra and not my A8 hardware

Interesting. By rights that line error SHOULD appear as it does in Altirra (it's due to the long-recognised RastaConverter player-repositioning bug).

 

Please can you can confirm that it doesn't appear on your hardware with the exact xex that you posted, and if so, confirm the exact A8 hardware you are running it on, including any mods.

 

These screenshots show the line in the screen kernel that should in theory generate the error, and a 'fixed' version (xex attached) with the bug eliminated...

drpeter_princeofpersiascreen2incomplete8_fixed.xex

Edited June 14, 2021 by drpeter

[Beeblebrox]

@drpeter  Ah you are right.  I was thinking of the previous save (ver7). The ver8 I posted (which you have fixed - thanks btw) - does indeed show the white line on my 65XE 320k A8! My bad.

 

FYI here are the (unfixed) Ver8 images side by side, (Altirra on the right, my 65XE 350k A8 on the left). And yeah, I am aware of the jail bars, etc etc. Working on getting my video quality up, (as in the other threads on the topic I've been posting on and getting help with).

 

[TheNameOfTheGame]

1 hour ago, Beeblebrox said:

@rensoup Also, when I first heard about the project a month+ back I started a RastaConversion of the classic PoP title screen just to see what it would look like. I ended up doing a few, (early ones failed), and then I started this one, a 320*200 format so there was space at the bottom for your credits if desired/technically possible. I'd popped it on the back burner but just resumed it yesterday alongside some other RC conversions. 

 

So far it looks like this (employs all PMGs btw):

 

 

 

Very nice title screen.  I hope something like this makes it into the final release!

[Jacques]

6 hours ago, rensoup said:

There's the loading anim too which may slow things down but I don't think it should make that much of a difference because the anim is only updated every 16 frames (I did that specifically to try to avoid slowing it down)

Wouldn't switching of xB into using OS SIO procedures instead of own ones (if possible) allow for using not only 1xSIO but anything the drives and OS allow (turbo, including 3xSIO)?

Maybe it's sensible to ask if these tiny animations really worth such limitation of 1xSIO? In the end the intro would only gain fluency with shorter loading breaks, that cause the intro tu suffer a lot in perception.

 

On the other hand, as for intro itself, after very long initial loading, the music starts and "Broderbund presents" screen appears, followed by "a game by Jordan Mechner" screen. Then there's a sudden interruption in music, and it takes few good seconds of loading to show the title screen with mouse. It really breaks perception and whole fluency, that after so long initial loading, just after two screens there's even more loading. And then the next loading is for the actual intro-animation.

Since the game requires 130XE anyway, isn't it possible to load whole intro stuff once and for all (of course until we enter the actual level to play), keep it's fluency, don't interrupt music with another loading, etc.? Especially if it stays with 1xSIO and nothing more.

I don't mean to put down the effort, just the feedback how it would feel better for me (personally, I may be wrong).

Edited June 14, 2021 by Jacques