It's no more a MiSTery

https://github.com/mist-devel/mist-board/wiki

Moderators: Mug UK, Zorro 2, spiny, Greenious, Moderator Team

Post Reply
slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

Those GAMEX keys are not universal, they're only working in specific games converted by PPera. And even not in all conversions.

Chris23235
Captain Atari
Captain Atari
Posts: 228
Joined: Thu Aug 07, 2014 6:52 pm

Re: It's no more a MiSTery

Post by Chris23235 »

Just tried it again and it worked, maybe it was this specific conversion I tried first. Many thanks for the help everybody.

slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

Here's a new release:
https://github.com/mist-devel/mist-bina ... 200507.rbf

Fixed two small bugs, and added a double bus speed switch: it can be turned on by selecting STeroids mode, or with the MegaSTe cache control. That means instead of implementing a cache, the whole bus speed is doubled. That has an advantage of not limited by a small cache size, but it works on the whole RAM area. Now in 16MHz mode + cache ON, the CPU can work in true double speed.

DanyPPC
Atari Super Hero
Atari Super Hero
Posts: 856
Joined: Tue Feb 21, 2017 7:02 am

Re: It's no more a MiSTery

Post by DanyPPC »

Good ! :cheers:

User avatar
rondc
Atari maniac
Atari maniac
Posts: 82
Joined: Tue Apr 30, 2019 9:59 pm
Location: Spain
Contact:

Re: It's no more a MiSTery

Post by rondc »

Great ! Thanks very much Gyorgy !

Chris23235
Captain Atari
Captain Atari
Posts: 228
Joined: Thu Aug 07, 2014 6:52 pm

Re: It's no more a MiSTery

Post by Chris23235 »

Great! Many thanks, wow the speed is now really impressive, Robocop 3 runs buttersmooth. I also saw that you fixed the text not showing on the grey screen in Robocop 3. Great work, one of the best cores on the MiST just got a lot better.

User avatar
vebxenon
Atari Super Hero
Atari Super Hero
Posts: 986
Joined: Fri Apr 24, 2015 12:10 pm

Re: It's no more a MiSTery

Post by vebxenon »

Thanks very much! :cheers:
Just a computer and videogame lover :)

- Atari Jr 2600 clone
- Atari 7800 Peritel
- Atari XEGS
- Atari Lynx II
- Atari Jaguar
- MiST Board

User avatar
vebxenon
Atari Super Hero
Atari Super Hero
Posts: 986
Joined: Fri Apr 24, 2015 12:10 pm

Re: It's no more a MiSTery

Post by vebxenon »

*Deleted. Problem solved*
Just a computer and videogame lover :)

- Atari Jr 2600 clone
- Atari 7800 Peritel
- Atari XEGS
- Atari Lynx II
- Atari Jaguar
- MiST Board

slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

jamesrc wrote:
Fri Mar 06, 2020 4:42 am
POV 26.

The Ultimate Demo -- This is the most broken I've seen something on this core in a long while! And it's not that broken. Hit Space to get past the credits and greets and then use the left/right cursors and return to select The Sprite Demo.
I think I could fix this finally. Also the BLT3 in SkidRow. Interestingly the proper fix (at least what I believe is the proper fix) still allow de-syncing in some cases (about 2 times out of 10), and this is ack'ed in the scrolltext, too (that's why I believe the proper fix is found). Maybe that was the last MFP problem (or not, hehe).
Now just why Closure doesn't work in ST mode...

However finally the forum is back!

Chris23235
Captain Atari
Captain Atari
Posts: 228
Joined: Thu Aug 07, 2014 6:52 pm

Re: It's no more a MiSTery

Post by Chris23235 »

That's great news. And its also great news that the forum is back. Nice to see you all again.

slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

With some help from Troed, Closure finally works in ST mode!
https://github.com/mist-devel/mist-bina ... 200621.rbf
I hope this is really true now:
viewtopic.php?p=379700#p379700

User avatar
vebxenon
Atari Super Hero
Atari Super Hero
Posts: 986
Joined: Fri Apr 24, 2015 12:10 pm

Re: It's no more a MiSTery

Post by vebxenon »

slingshot wrote:
Sat Jun 20, 2020 10:18 pm
With some help from Troed, Closure finally works in ST mode!
https://github.com/mist-devel/mist-bina ... 200621.rbf
I hope this is really true now:
viewtopic.php?p=379700#p379700
Outstanding!! :cheers: :cheers: Thanks very much!
Just a computer and videogame lover :)

- Atari Jr 2600 clone
- Atari 7800 Peritel
- Atari XEGS
- Atari Lynx II
- Atari Jaguar
- MiST Board

User avatar
DrOG
Atari Super Hero
Atari Super Hero
Posts: 727
Joined: Sun Jul 31, 2016 8:23 pm
Location: Gyula, Hungary

Re: It's no more a MiSTery

Post by DrOG »

Avesome! :cheers:

DanyPPC
Atari Super Hero
Atari Super Hero
Posts: 856
Joined: Tue Feb 21, 2017 7:02 am

Re: It's no more a MiSTery

Post by DanyPPC »

Yeah !

Many thanks :cheers:

Chris23235
Captain Atari
Captain Atari
Posts: 228
Joined: Thu Aug 07, 2014 6:52 pm

Re: It's no more a MiSTery

Post by Chris23235 »

Many many thanks for your continued work.

ijor
Hardware Guru
Hardware Guru
Posts: 4006
Joined: Sat May 29, 2004 7:52 pm
Contact:

Re: It's no more a MiSTery

Post by ijor »

slingshot wrote:
Sat Jun 20, 2020 10:18 pm
With some help from Troed, Closure finally works in ST mode!
https://github.com/mist-devel/mist-bina ... 200621.rbf
I hope this is really true now:
viewtopic.php?p=379700#p379700
Great job, as usual, slingshot. But I suspect something is wrong. You seem to have a single Shifter model, or I miss something?. That is already not accurate, of course. But what is more important, something is wrong because these demos should not work in both platform with the same Shifter logic. I am not talking about the STE Shifter extra features like hw scrolling or DMA shound that are obviously different. But about the core clock generation, shift and and reload logic. That can't be (or at least for accuracy, shouldn't be) the same under both platforms. So it seems that in a way you are running something that it should not work.
Fx Cast: Atari St cycle accurate fpga core

slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

ijor wrote:
Sun Jun 21, 2020 8:19 pm
But about the core clock generation, shift and and reload logic. That can't be (or at least for accuracy, shouldn't be) the same under both platforms. So it seems that in a way you are running something that it should not work.
As I'm not aware about the core logic differences - I cook from what I have. STe demos are working so far. There's one tiny thing which must be differ, that's "flushing" after a line, since the extra fetch when fine scrolling is active will leave the shifter in an unstable state.
That's the clearing of the Reload signal:
https://github.com/gyurco/gstmcu/blob/m ... ideo.v#L97

ijor
Hardware Guru
Hardware Guru
Posts: 4006
Joined: Sat May 29, 2004 7:52 pm
Contact:

Re: It's no more a MiSTery

Post by ijor »

slingshot wrote:
Sun Jun 21, 2020 9:00 pm
As I'm not aware about the core logic differences - I cook from what I have. STe demos are working so far.
I don't have the details, but I know they are different. Yes, I understand STe demos work, that's precisely the problem. If that's the main goal for you, then that is perfectly ok. But sometimes not working is more accurate than working.

It's in some ways, analog to the wake up issue. Some demos require a specific wake up, other demos require another one. No wake up can run all demos in its original unpatched form (most of them were fixed years later to avoid the wake up dependency).
There's one tiny thing which must be differ, that's "flushing" after a line, since the extra fetch when fine scrolling ...
No, or at least that shouldn't be the only difference. There are difference even among the multiple versions of ST Shifter (there are 4 different parts). And there are some demos that don't run on some ST Shifter versions.
Fx Cast: Atari St cycle accurate fpga core

slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

ijor wrote:
Sun Jun 21, 2020 9:39 pm

No, or at least that shouldn't be the only difference. There are difference even among the multiple versions of ST Shifter (there are 4 different parts). And there are some demos that don't run on some ST Shifter versions.
When I played with the shifter, I've noticed that even the phase of the pixel clock change can make a demo work or broken (for example, introducing the infamous 16px vertical black bars). That thing can be emulated for an extent. Actually I would happily learn about other differences, but I didn't find any internal docs (only your reverse-engineered schematics). I'm not really expecting to have significant differences in the core logic, but probably in the register latching strobes or clock phases.

Edit: about the wake up state - looks like the STe GSTMCU timings is the equal of WS3 of the GLUE+MMU chips.

jamesrc
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 141
Joined: Fri Aug 12, 2005 11:33 pm
Location: Austin, TX, USA
Contact:

Re: It's no more a MiSTery

Post by jamesrc »

slingshot wrote:
Tue Jun 16, 2020 5:09 pm
I think I could fix this finally. Also the BLT3 in SkidRow. Interestingly the proper fix (at least what I believe is the proper fix) still allow de-syncing in some cases (about 2 times out of 10), and this is ack'ed in the scrolltext, too (that's why I believe the proper fix is found). Maybe that was the last MFP problem (or not, hehe).
Now just why Closure doesn't work in ST mode...

However finally the forum is back!
Ooooh! I'll go back and re-test this. I've been a little distracted since my Spectrum Next arrived and with the forum down. :)

I'm feeling the itch to move forward with testing, now.

Thank you again for everything!

ijor
Hardware Guru
Hardware Guru
Posts: 4006
Joined: Sat May 29, 2004 7:52 pm
Contact:

Re: It's no more a MiSTery

Post by ijor »

slingshot wrote:
Sun Jun 21, 2020 10:00 pm
Actually I would happily learn about other differences, but I didn't find any internal docs (only your reverse-engineered schematics). I'm not really expecting to have significant differences in the core logic, but probably in the register latching strobes or clock phases.
There aren't any docs about that that I know and some parts are pretty rare. I don't expect huge differences either. But as you said, even minor differences can make something break. I wouldn't worry too much about the differences. I am just worried, or at least I thought I should note, that something in the model is not right if it is running what should actually fail.
Edit: about the wake up state - looks like the STe GSTMCU timings is the equal of WS3 of the GLUE+MMU chips.
I meant about wake up states in ST mode because the issue is similar.
Fx Cast: Atari St cycle accurate fpga core

slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

ijor wrote:
Tue Jun 23, 2020 4:06 am
There aren't any docs about that that I know and some parts are pretty rare. I don't expect huge differences either. But as you said, even minor differences can make something break. I wouldn't worry too much about the differences. I am just worried, or at least I thought I should note, that something in the model is not right if it is running what should actually fail.
But do you know what should actually fail and what should not? Because if these "small" detail is not known fully, then I think it's better to make as many things work as possible. The pixel clock phase and when the actual resolution latched are the two factors which affects the shifter behaviour greatly, however would need some original model to compare the FPGA implementation with it.
There are no shifter "tricks" - except the Spectrum 512 palette change, but I wouldn't consider as a trick either. So I see two states only: "working" or "broken" (like de-sync of the planes and the black bars - both can be provoked with the current shifter, too).
Edit: about the wake up state - looks like the STe GSTMCU timings is the equal of WS3 of the GLUE+MMU chips.
I meant about wake up states in ST mode because the issue is similar.
I just mentioned the effective WS in ST mode using the GSTMCU timings looks like WS3 now. Maybe if a similar model would exists for the Shifter, than it could be easily implemented. But it cannot be detected in software, so I fear it's not going to happen.

slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

ijor wrote:
Tue Jun 23, 2020 4:06 am

There aren't any docs about that that I know and some parts are pretty rare. I don't expect huge differences either. But as you said, even minor differences can make something break.
Maybe in some shifter versions, res is latched in the rising edge of CMPCS, in others at the falling edge? That would create a lot of difference (as it does in ST-STe GLUE/GSTMCU).

ijor
Hardware Guru
Hardware Guru
Posts: 4006
Joined: Sat May 29, 2004 7:52 pm
Contact:

Re: It's no more a MiSTery

Post by ijor »

slingshot wrote:
Tue Jun 23, 2020 8:09 am
But do you know what should actually fail and what should not?
E.g., I suspect the 4-bit syncsroll should fail in your model, and something is compensating and making it work. Possibly Closure in ST mode as well.
Because if these "small" detail is not known fully, then I think it's better to make as many things work as possible.
Personally I don't like that approach too much, but that doesn't mean it is necessarily wrong.
There are no shifter "tricks" - except the Spectrum 512 palette change, but I wouldn't consider as a trick either. So I see two states only: "working" or "broken" (like de-sync of the planes and the black bars - both can be provoked with the current shifter, too).
There are at least two shifter based scrolling techniques, that I'd should definitely say they are "tricks". One is the mentioned 4-bit sync scroll. The other is intentionally destabilizing Shifter.
Maybe in some shifter versions, res is latched in the rising edge of CMPCS, in others at the falling edge? That would create a lot of difference (as it does in ST-STe GLUE/GSTMCU).
Don't think so. Shifter can't latch on the CS leading edge. Data is not yet available.
Fx Cast: Atari St cycle accurate fpga core

slingshot
Atari God
Atari God
Posts: 1717
Joined: Mon Aug 06, 2018 3:05 pm

Re: It's no more a MiSTery

Post by slingshot »

ijor wrote:
Tue Jun 23, 2020 12:46 pm

There are at least two shifter based scrolling techniques, that I'd should definitely say they are "tricks". One is the mentioned 4-bit sync scroll. The other is intentionally destabilizing Shifter.
I wasn't aware of the 4-bit sync scroll. I have to do some research how it works (and when should it fail).
Don't think so. Shifter can't latch on the CS leading edge. Data is not yet available.
Just checked this in the sim, true, that cannot happen (CAS goes low at the same time as LOAD).

Post Reply

Return to “MiST”