Updated ST Core?

[RealLarry]

Pals, I'm so sorry to say that the core I built yesterday is the wrong one. After cloning the repo I copied the relevant parts to /tmp before I switched to mist-experiments. This was done before coffee and I think I'm getting older for the times being
Regardless of this poo I've had my coffe now, pulled the repo again and built a new (and correct) core, including the latest changes from last night.
Repeating the tests now.

[RealLarry]

First results:
- MIDI is still working as normal
- serial communication via usb2rs232 is somewhat working. ST receives characters and text as expected, but when sending it sends every char eight times..so when sending a "hello" the opposite receives "hhhhhhhheeeeeeeelllllllloooooooo". Reproducable against older cores (including my "wrong" core) where serial I/O is working.

[RealLarry]

Next result:
- Ethernet. Can't get this to work, receives "NE reset bit not set. Fatal" with net.st after every reset, even after reseating d-link e100.
The good news are that it is still working with core-181017.

[slingshot]

Committed a change, maybe fixes the serial out. Ethernec is not a surprise, it wasn't modified for the differences of the TG68K and FX68K bus.
Maybe you can try parallel port again, the FPGA sends out the parallel data to the ARM if I set the USB I/O: parallel option in the OSD System menu.

[jamesrc]

Pals, I'm so sorry to say that the core I built yesterday is the wrong one. After cloning the repo I copied the relevant parts to /tmp before I switched to mist-experiments. This was done before coffee and I think I'm getting older for the times being
Regardless of this poo I've had my coffe now, pulled the repo again and built a new (and correct) core, including the latest changes from last night.
Repeating the tests now.

I plan to go home and watch sports for three hours tonight. But after that, when those of you in the UK are all asleep in your beds, I will try the new core.

Thanks RealLarry!

[DanyPPC]

With new core RoboCop 3, Apprentice and Another World work, but some games do not work anymore, like PacMania STE 2013 version.

Tested only these games. Thanks for the improvements on this core.

[jamesrc]

Hi Slingshot!

Okay! I've sat down and have some testing results and a couple of bug reports. This testing was on the version of the core that RealLarry uploaded.

First off, this is fantastic. I can't say "Thank you" enough.

A good place to start is the full colour screen from Wings of Death. Here's how it looked on the old core:



By comparison, here it is with the new core:



This is much much improved, but still not perfect. There are a few incorrectly coloured pixels in various places on the image. The most noticeable area is on the lighter coloured set of rocks at the right hand side of the castle (to the right of the screen about midway down). The lower left of these rocks has a number of miscoloured pixels.

As my actual ST is in a closet right now, here's a reference rendering by way of Hatari. You'll note these off-coloured pixels aren't present.



Next I tried a couple of things that have never worked for me. First was Pompey Pirates Menu Disk 1 Version 2. This has a very timing sensitive intro that opens the borders. On the previous core, the screen would go completely black, a few seconds of music would play and it would crash out to the desktop. I'm pleased to report that the music now plays properly until I select a game by pressing a number. The screen is still black, which I suspect is due to some video timings outside the range that my monitor can display. I'll find out if this behaviour is the same on an era-appropriate television soon enough as I've got a Sony PVM en route to me.

I also tried Tri-Heli 2. A game which many seem to dislike, but which I'm rather fond of. This is one of the few games to open the borders and overscan and... the screen goes black in much the same way as that Pompey Pirates menu. I can still hear the game playing, mind, so again I think it's generating a signal that my current monitor can't deal with. Oh well.

Here's one I wasn't expecting to work at all. The B.I.G demo will now run on a MiST! Unfortunately, some timings are still off. As you can see from the screen shot, the selection bar is mis-aligned and there's a dark grey area at the top. A crazy, flickering dark grey area which I have a video of, but I think the screenshot will suffice:



One of the Psych-O-Screens is similarly broken, and not only renders incorrectly but causes my screen to turn black for several seconds at a time before returning, repeatedly.



As I'm trying all sorts of weird open borders overscan stuff, I decided to dream big and give Obsession a go.

I started with the version on Dbug164 using .ST images and couldn't get the game to start due to an unrelated MiST bug which I'm going to put in bold here in the hopes that you'll take pity on me and fix it.

When it comes time to put Disk B in when loading Obsession, it doesn't wait for you to press fire or anything, but instead automatically detects when the disk has been changed. Whatever mechanism a real ST (or emulator) would use to detect this doesn't fire on the MiST's ST core. You put the second disk in, the detection never happens, and... nothing happens. It sits there waiting for Disk B forever.

Thankfully, this version of Obsession can just be copied in full to a hard drive and no disk swapping is required.

Does it work?



Oh.

My.

God.

Yes, it does. And as far as I can tell from staring at Hatari and Obsession and the MiST for about 30 minutes, it's pixel perfect. It's playable. The MiST can run Obsession!

I feel compelled to offer proof that this really is running on the MiST....



The last thing I tested was Pacmania STE. Like DanyPPC, I found it didn't work. Just froze on a white screen without the game ever completely loading. I retested on the old core and it works just fine.

Two more notes:

Occasionally when I boot this new core, the screen will stay black, or I'll get a mostly black screen with a white bar on the right hand side. Sometimes I know the core is running because I can hear the sounds an ST makes when I press keys. Hitting reset will produce the same result, or a white screen with thin pixel wide black bars. I have to reload the core completely or power cycle the MiST to get things to boot normally. I have never encountered this with the original core.

And I'm going to be presumptuous here and ask if you can fix a bug that has nothing to do with you, as far as I know - it's a long-standing issue: When you choose the "Cold boot" reset option, or press the rightmost button to reset the MiST -- it removes whatever virtual floppy you have in drive A and inserts whatever the default disc stored in your config is. This is frustrating because inserting a disk and resetting the ST is a natural way to boot something. I find I have to quickly bring up the UI after the reset and select the disk I want. Thankfully, once you get into the file browser, the last disk you choose is the one that's preselected, so you can easily switch it in time to boot from it.

In conclusion...

YOU ARE AN AWESOME HUMAN BEING AND I STILL SAY YOU SHOULD HAVE A TIP JAR OR DONATION BOX FOR YOUR WORK ON THIS PROJECT.

[DanyPPC]

Obsession Pinball works on old core too. It's an STE game.

[slingshot]

Pacmania works now, the bug was just with the Microwire device. But still a lot to do in the Shifter area. The problem is my knowledge about the machine is not in the same level as @MasterOfGizmo or @ijor.

[jamesrc]

Pacmania works now, the bug was just with the Microwire device. But still a lot to do in the Shifter area. The problem is my knowledge about the machine is not in the same level as @MasterOfGizmo or @ijor.

Thanks -- this is still great.

[jamesrc]

Obsession Pinball works on old core too. It's an STE game.

Really? Weird. I thought I'd tried it at some point in the past and it hadn't worked.

[Estrayk]

I remember a good game that doesnt work with old core:

LETHAL XCESS

[jamesrc]

I remember a good game that doesnt work with old core:

LETHAL XCESS

I've been meaning to try that on the new core. I will when I get a chance.

My monitor came in, so I spent the morning getting my MiST running with CRT goodness.

[vebxenon]

My MiST needs to be repaired (power supply ports needs to be soldered again) but thanks for these updates!

[Eero Tamminen]

Obsession Pinball works on old core too. It's an STE game.

Really? Weird. I thought I'd tried it at some point in the past and it hadn't worked.

It does need overscan, but the STE right border opening is much easier than ST overscan.

[jamesrc]

Pals, I'm so sorry to say that the core I built yesterday is the wrong one. After cloning the repo I copied the relevant parts to /tmp before I switched to mist-experiments. This was done before coffee and I think I'm getting older for the times being
Regardless of this poo I've had my coffe now, pulled the repo again and built a new (and correct) core, including the latest changes from last night.
Repeating the tests now.

I honestly don't know how much work I'm requesting here -- if it's "not much", would you be averse to pulling the latest changes and building another core?

If it's more than "not much" then I am a patient man and shall wait. Or try to figure out what I need to do myself.

[slingshot]

I would ask you to have some patience please. Maybe I made a mistake to announce this work publicly, it was too early. It's not for public use. I only wanted to know if the external interfaces - which I cannot test - are still working. Better to forget this branch for a while. If someone wants to try it - please build it yourself. Also I ask you to not spread rbf version of this half-work, I'm aware of many bugs in it, it doesn't help at all to report them.

[RealLarry]

...and maybe it was a mistake by me to publish these test cores. But this was done to the best of my knowledge and certain to help testing and improving slingshot's great work.
But I think that referring to the experimental repo wasn't a mistake; look, you've found at least one willing tester by that way

[slingshot]

No problem at all, just I want to avoid people to have false hopes

[vebxenon]

I would ask you to have some patience please. Maybe I made a mistake to announce this work publicly, it was too early. It's not for public use. I only wanted to know if the external interfaces - which I cannot test - are still working. Better to forget this branch for a while. If someone wants to try it - please build it yourself. Also I ask you to not spread rbf version of this half-work, I'm aware of many bugs in it, it doesn't help at all to report them.

We are here for testing and to help all the developers

[jamesrc]

No problem at all, just I want to avoid people to have false hopes

I'm just excited that it's being worked on, and that things that didn't work before are working now.

I'll refrain from reporting issues, but if there's anything else specific you want people to test, I'm happy to give it a try.

[desUBIKado]

I would ask you to have some patience please. Maybe I made a mistake to announce this work publicly, it was too early. It's not for public use. I only wanted to know if the external interfaces - which I cannot test - are still working. Better to forget this branch for a while. If someone wants to try it - please build it yourself. Also I ask you to not spread rbf version of this half-work, I'm aware of many bugs in it, it doesn't help at all to report them.

Mistake? No, only plenty of fun for most of us.

Trying 1st TEST core: https://gaming.youtube.com/watch?v=5D5njYOeh-k (in spanish)

Trying 2nd TEST core: https://gaming.youtube.com/watch?v=FbCmPmUgW1w (in spanish)

[jamesrc]

I note that the B.I.G Demo now runs flawlessly, although digging through the commit history I understand this may be a temporary fix and not the proper fix.

Anyway, sorry if we've caused any undue pressure, Slingshot. Don't worry - I'm not using your experimental core for anything important and I'm not spreading it beyond the confines of this thread.

[slingshot]

I note that the B.I.G Demo now runs flawlessly, although digging through the commit history I understand this may be a temporary fix and not the proper fix.

Anyway, sorry if we've caused any undue pressure, Slingshot. Don't worry - I'm not using your experimental core for anything important and I'm not spreading it beyond the confines of this thread.

Yes, opening top and bottom borders are fully working, but making this in sync with possible left and right border effects are requiring much work. I'm currently building a simulation using the original schematics of the GST MCU (for fun, and to understand the circuits better). And what will come out of these will be implemented in the core...when it'll be ready.

[slingshot]

I did a little update, I hope I found the annoying bug when the screen went black in color modes on VGA. The core is as good as it is now, I won't do much updates, since I want to implement the GST MCU faithfully, which will take a lot of time. Hope it'll be ready this year
But until that, here's the current state.