It's no more a MiSTery

[slingshot]

One demo, BMT3, is a full overscan screen scroll with a picture of a wolf, but (on my TV at least) the sync is lost and tears part way down the screen, but the central scroll line is stable.

I found only this as BMT3:
https://demozoo.org/productions/153521/

[alexh]

https://demozoo.org/productions/267709/

[DanyPPC]

Anyway this core is a force !

[jamesrc]

One demo, BMT3, is a full overscan screen scroll with a picture of a wolf, but (on my TV at least) the sync is lost and tears part way down the screen, but the central scroll line is stable.

I've been meaning to try to reproduce the problem since you posted, but I've been out of town and now have a sick toddler to take care of, so haven't found time.

If I get a spare minute this weekend I'll do it.

[slingshot]

One demo, BMT3, is a full overscan screen scroll with a picture of a wolf, but (on my TV at least) the sync is lost and tears part way down the screen, but the central scroll line is stable.

I've been meaning to try to reproduce the problem since you posted, but I've been out of town and now have a sick toddler to take care of, so haven't found time.

If I get a spare minute this weekend I'll do it.

It's true, the background palette effect is shifted. Probably still some remaining MFP issue.

[Total Eclipse]

Out of interest, did you also find problems with not all keypresses being picked up while you were walking around the menu?

[slingshot]

Out of interest, did you also find problems with not all keypresses being picked up while you were walking around the menu?

It's the same in STeem.

[jamesrc]

It's true, the background palette effect is shifted. Probably still some remaining MFP issue.

Ah! Thanks -- in that case I wouldn't bother trying to reproduce this particular error and will instead continue the PoV slog when I get time.

I think I tested five more disks last weekend.

[slingshot]

I think I tested five more disks last weekend.

Then it's not a bad ratio with only one partially failing screen.

[jamesrc]

Then it's not a bad ratio with only one partially failing screen.

It's a testament to how far you've brought this core that I am now legitimately surprised when anything doesn't work, even things that push the limits. I'm definitely seeing more failures due to things like TOS versions and RAM settings than I am to any limitation in the core.

I'd probably get through these disks at a faster clip, but I'm exercising each screen for long enough to read all scrolltexts. With exceptions for things like the BIG scroller.

I purchased a MiST in the hope that it could be my full time ST. It takes up way less desk space than my STe and quickly allows me to swap configurations. It's reached the point where that dream seems like a reality, and it's largely down to your sterling efforts.

[slingshot]

A small update:
https://github.com/mist-devel/mist-bina ... 191207.rbf

Toggle between normal/STe joystick ports with F11.

[jamesrc]

A small update:
https://github.com/mist-devel/mist-bina ... 191207.rbf

Toggle between normal/STe joystick ports with F11.

Haven't tried it yet, but does this just apply to USB controllers (which would make sense) or does it apply to the DB9 ports as well?

[slingshot]

Haven't tried it yet, but does this just apply to USB controllers (which would make sense) or does it apply to the DB9 ports as well?

The core doesn't know if it's USB or DB9, it just sees joystick 0,1(2,3 - but these are always connected to the parallel port) -> applies to both.

[DanyPPC]

Thanks Slingshot, the problem is solved in Great Giana Sisters STE (2014), not in Gemini Wings disk version.

[jamesrc]

Want to see something odd?

No? Well, I'm going to post it anyway. Here's something that fails differently on the MiSTery core as compared to both Hatari and STeem SSE.

On PoV 16, and it appears to be a Version 3 which I've attached to this post to ensure we're all looking at the same disk, the S.U.C.C.E.S.S. demo.

For reference, here's how it's supposed to look. This is on the MiSTery core using the ST chipset and running TOS 1.4. The results are the same with TOS 1.2 which I also tested.



Works flawlessly!

Now, here's how it looks using the STe chipset and running TOS 1.62.



Looks like the palette changes just aren't happening as the scanline moves down the screen. Interestingly, this effect also seems present as the demo unpacks itself. With the ST chipset you see the coloured lines characteristic of a demo unpacking. With the STe chipset, the screen just pulses different solid colours with no changes mid-frame.

Running under both Hatari and Steem SSE presents the same result on an ST with TOS 1.02/1.4. The demo works.

The demo also fails to work properly when the emulator as configured as an STe with TOS 1.62. However, it fails in a different way. The depack routine looks normal with actual coloured bars, and the demo looks like this with the bottom of the scroller corrupted.



It doesn't look exactly like that. The effect is only present once every three or four frames, creating some very annoying flicker.

So... which is correct? Perhaps someone with an actual STe up and running right now could confirm?

[slingshot]

It's an example when the shifter got confused, and the various bitplanes are shifted. As the STe checks for enabling the display a bit earlier than ST, it's easier to provoke this "effect" with demos made for ST.
Btw, it would be really interesting to compare it with a real STe. As the Level16 fullscreen showed, the core is more close to the original than the emulators But that was a glue effect, this one happens in the shifter.

[jamesrc]

Btw, it would be really interesting to compare it with a real STe.

Hopefully there's someone following this thread who can quickly verify. I don't have easy access to my STe at the moment.

[SaschaFFM]

This looks so great. Unfortunately I do not own a MiST, but rather a MiSTer. Hope that someone will port this core to MiSTer. Would be such a great addition. Keep up the good work.

[jamesrc]

I seem to recall somewhere that a firmware update was needed for some hard drive stuff?

But there's been an update since then. I'm running the latest firmware, 11/29/19, and there seem to be some regressions with hard drive installable versions of games.

Petari's Hard n Heavy STE version (hard drive installable): "STE fine scroll Hard disk v." (HNHSHD.ZIP) on this page: http://atari.8bitchip.info/TestMe/HNH/HnH.html

It locks up when you press a key to start on the trainer selection page. The floppy version above it works. This version starts correctly on the old ST core.

D-Bug's Jim Power Hard Drive version from here: http://d-bug.mooo.com/patch.php also locks up after the Dbug intro and trainer selection. This works on the old ST core, too.

I'm running a ACSI hard drive image and using the AHDI driver. The exact same configuration works both on the old core and under emulation. I've tried both TOS 2.06 and TOS 1.62.

[RealLarry]

Btw, it would be really interesting to compare it with a real STe.

Hopefully there's someone following this thread who can quickly verify. I don't have easy access to my STe at the moment.

Just checked this out before breakfast. It "just works" on my STe, TOS1.62.
The distortion of the scrolling text is ok as camera wasn't fast enough

[jamesrc]

Just checked this out before breakfast. It "just works" on my STe, TOS1.62.
The distortion of the scrolling text is ok as camera wasn't fast enough

Thank you!! The bottom half of the scroller wasn't flickering every few frames?

If not then Steem SSE, Hatari and the MiSTery core are all wrong here... just in different ways.

[RealLarry]

Thank you!! The bottom half of the scroller wasn't flickering every few frames?

Ah, I see...thought that's a effect or glitch from the demo itself Yes, the bottom half of the scroller is flickering every few frames on the real STe.

[slingshot]

Petari's Hard n Heavy STE version (hard drive installable): "STE fine scroll Hard disk v." (HNHSHD.ZIP) on this page: http://atari.8bitchip.info/TestMe/HNH/HnH.html

It locks up when you press a key to start on the trainer selection page. The floppy version above it works. This version starts correctly on the old ST core.

It might sound unbelievable, but depends on what state you left the mouse.

And now the technical explanation if somebody interested:
As the mouse is a rotary encoder, it constantly turns on-off it's outputs as you move it in a Gray-coded scheme. Now it's possible when you stop it, it's left in a postion where some pins are high, some are low. No problem.
But!
As the mouse port also a joystick port, pins left low are detected as a joystick is held in a position. Now the loader turns on joystick event reporting, the stuck "joystick" will generate an immediate event, which generates an ACIA interrupt - a totally unexpected one. No one resets it, and it masks Timer C interrupts, which are needed - at the end, the whole thing freezes.
I really wonder what state an original Atari mouse leaves its outputs when it's not moving.

[jamesrc]

It might sound unbelievable, but depends on what state you left the mouse.

And now the technical explanation if somebody interested:

Huh. That is interesting. Is it just random chance that it worked for me on the old core? I must have tried on the new core a dozen times with different TOS and RAM combinations. I tried on the old core twice and it worked both times.

It's conceivable the floppy version worked because it boots straight up and didn't require me to move the mouse at all.

I really wonder what state an original Atari mouse leaves its outputs when it's not moving.

If someone can answer this question, will it be possible to fix the issue? Or is it an issue that's inherent in the way that the MiST shows a joystick and a mouse connected to that port at all times?

If it's the latter, perhaps another hotkey like the F11 to "disconnect" the mouse could resolve the problem?

[slingshot]

The old core didn't use an exact ikbd implementation. A disconnect mouse option for buggy programs is not the way to go (why it turns on joystick event monitoring if it doesn't expect a joystick event?)