It's no more a MiSTery

[slingshot]

As some already noticed, there were some (heavy) work on a new'ish STe core. It's based on cycle-exact components, like @ijor's FX68k, the GSTMCU (GLUE+MMU combo) was recreated using the original schematics recovered by @czietz, and a whole new ikbd with support for all custom code uploading demos by @MasterOfGizmo.

So the time for the first public release arrived:
https://github.com/mist-devel/mist-bina ... es/mistery

Please note that it was only tested by me and MasterOfGizmo for some extent, so there might be annoying bugs or crashes due to last minute changes (and I was lazy to retest everything again and again). Also don't expect that many STe demos will work if they're depend on a cycle-exact blitter. Try out, report issues and enjoy!
(And don't forget to update the firmware before using it).

[Chris23235]

Great work, many many thanks for this new core. I tried Lethal Excess and it finally boots, it still has graphical glitches, but on the old core the game never managed to even come to the loading screen.

[slingshot]

Great work, many many thanks for this new core. I tried Lethal Excess and it finally boots, it still has graphical glitches, but on the old core the game never managed to even come to the loading screen.

Turn off blitter in the main menu, and it won't have glitches.

[jamesrc]

I might be the weekend before I do, but I cannot wait to try this!

[RealLarry]

Try out, report issues and enjoy!

Many thanks for this great work! Are external devices, like USB2RS232 and USB2IEEE1284, still supported? I could continue my tests then...

[DrOG]

Big thanks to everyone who is involved in this project!

[sebdel]

Why do I have to work today ;_; I can't wait to try it!

slingshot, could you share your plans for this core? Will you tackle some of the "Current issues/limitations" or they're not worth the effort? What about overclocking, is the steroid mode still there even if it defeat the purpose of cycle accuracy?

Either way, thank you so much for this work!

[goran]

Thank You!

[slingshot]

Many thanks for this great work! Are external devices, like USB2RS232 and USB2IEEE1284, still supported? I could continue my tests then...

In theory yes, they're there, but I didn't try.

What about overclocking, is the steroid mode still there even if it defeat the purpose of cycle accuracy?

STeroid mode is 16MHz turned on unconditionally, that's the maximum for the design (with 32 MHz base clock). That won't change. But the missing CPU cache probably will added sooner or later.

[sebdel]

I couldn't wait, I went back home for lunch. Every demo I tried worked This is all I ever wanted out of the MiST. Finally!

It has glitches on some fullscreens in DSOTS and cuddly, where the screen is dark and there are artifacts at the very top or the far right of the screen. I think the dark image could be my monitor's fault (out of range) and the artifacts could have been invisible on a regular RGB monitor. I would have to cross check with an emulator.

If you want to test:
DSOTS: the very first fullscreen with the ULM logo has a white glitch at the top.
Cuddly: the infamous fullscreen with the TCB logo at the top has a wide column full of crap at the far right of the screen.
Both are really dark.

[alexh]

Good work. Thanks for your continued support of the Scene

[slingshot]

If you want to test:
DSOTS: the very first fullscreen with the ULM logo has a white glitch at the top.
Cuddly: the infamous fullscreen with the TCB logo at the top has a wide column full of crap at the far right of the screen.
Both are really dark.

I'm not sure if these far left/right columns are visible on 15kHz devices, maybe just on VGA when the scandoubler is used. E.g. DSOTS first screen is also dark for me on VGA, but looks good on TV. It simply draws too close to the hsync, in the back porch area. The monitor uses this to calibrate the black level.

[slingshot]

This is fullscreen on my TV:

P1010004.JPG

[rondc]

At the moment everything I've been testing works well.
I'll tell you when I do more concrete tests at night.

Anyway, great work !

[jamesrc]

"PAL clock only (32.084 MHz)"

I can't test right now -- this means 60hz mode won't work at the moment?

[slingshot]

Of course NTSC will work, the PAL and NTSC mode switch does not depend on the main clock crystal, just NTSC machines were equipped with a bit different crystal:
32.0424 MHz for the ST
32.2159 MHz for the STe

[jamesrc]

Of course NTSC will work, the PAL and NTSC mode switch does not depend on the main clock crystal, just NTSC machines were equipped with a bit different crystal:
32.0424 MHz for the ST
32.2159 MHz for the STe

I did not know this. Thanks.

[Chris23235]

Great work, many many thanks for this new core. I tried Lethal Excess and it finally boots, it still has graphical glitches, but on the old core the game never managed to even come to the loading screen.

Turn off blitter in the main menu, and it won't have glitches.

I see, that's great. Just curious, in which way did the programmers of Lethal Xcess use the blitter, so that it glitches now? Is it a timing problem?

So far everything I tried, that didn't work on the old core now works. Robocop 3 won't load in STE mode, it only works in ST mode, but I am not sure if the game worked on an original STE at all.

[jamesrc]

Guys (and especially slingshot):

This is absolutely jaw-dropping work. It runs everything I've thrown at it so far.

Here's Pompey Pirates Menu 1 V2 at 15khz on my Sony PVM, a full screen menu that steadfastly refused to work until today.

[jamesrc]

You probably knew I was going to throw obscure stuff at this until something didn't work. But this core is so good that I feel bad reporting problems. :/

1. Sometimes on power on or reboot I get a black screen, or an ST screen with some coloured lines. Repeated cold/warm reboots get me past this, eventually.

2. POV Demo Disk 117 - the menu music is not right. It sounds like there's at least a channel missing. What's there sounds like... a 48K Spectrum trying to play a sample through its beeper chip.

3.The recent City Frames demo. Now, granted, this could be Blitter timing as it's an STe Demo, but hear me out. It runs great for about three minutes from the time the music starts. No complaint. Then it reaches a part of the demo that causes the MiST to output... something my monitor can't sync to. It's much the same effect I'd get running a full screen in 15khz before this was supported. Hitting Space at this point will successfully return me to the desktop. Here's a quick photo:



If you think this is likely Blitter timings then I shall say no more about it.

Finding waaaaaaaaay more stuff than works than doesn't. I'm in awe of this, I really am.

[slingshot]

3.The recent City Frames demo. Now, granted, this could be Blitter timing as it's an STe Demo, but hear me out. It runs great for about three minutes from the time the music starts. No complaint. Then it reaches a part of the demo that causes the MiST to output... something my monitor can't sync to. It's much the same effect I'd get running a full screen in 15khz before this was supported. Hitting Space at this point will successfully return me to the desktop. Here's a quick photo:

This one looks similarly broken to the STrange robots demo, which is a heavy blitter user.

[slingshot]

I see, that's great. Just curious, in which way did the programmers of Lethal Xcess use the blitter, so that it glitches now? Is it a timing problem?

The game opens the top and bottom borders, there must be some cycle-counted code paths there.

[vebxenon]

Thanks very much!

[RealLarry]

Hmmm...the mist-vhi.prg (Viking/SM194) seem to doesn't work anymore. I'm getting a white screen on all normal modes and black/white stripes an STEroids.

[slingshot]

Hmmm...the mist-vhi.prg (Viking/SM194) seem to doesn't work anymore. I'm getting a white screen on all normal modes and black/white stripes an STEroids.

Did you enable 14M RAM?