Work on the Minimig core?

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

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

Post Reply
kolla
Captain Atari
Captain Atari
Posts: 349
Joined: Thu Sep 17, 2015 11:39 pm
Contact:

Re: Work on the Minimig core?

Post by kolla »

robinsonb5 wrote: Sat Jan 16, 2021 11:20 pm If your monitor can recognise those modes then you should be able to create them in P96Mode.
I tried, I gave up, my monitor (BenQ of some sort) isn't very helpful, it only recommend me to use 1920x1080 already, and doesn't inform me much about the signals it receives - it is there, but I have to dig three levels of menus with insane button bindings to reach it :)

A related question - could double scan be supported, so that one can set up modes like 640x240 and 640x256?
-- kolla
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

kolla wrote: Tue Jan 26, 2021 4:10 am I tried, I gave up, my monitor (BenQ of some sort) isn't very helpful, it only recommend me to use 1920x1080 already, and doesn't inform me much about the signals it receives - it is there, but I have to dig three levels of menus with insane button bindings to reach it :)
Funny how user-experience designers always seem to be on holiday when monitors are designed.
A related question - could double scan be supported, so that one can set up modes like 640x240 and 640x256?
Yes, I'd like to move the scandoubler out from the internals of Minimig at some point, so it can be used for the RTG system too - then it would be possible to define a 320x200 or 320x240 mode for DOOM!

The other way to do it is to give the RTG system support for row modulos, so it can fetch each row twice, but I'd prefer to use the first method if possible.
DanyPPC
Atari Super Hero
Atari Super Hero
Posts: 998
Joined: Tue Feb 21, 2017 7:02 am

Re: Work on the Minimig core?

Post by DanyPPC »

@robinsonb5:

Have had you time to fix the borderblank problem ?
Thanks.
kolla
Captain Atari
Captain Atari
Posts: 349
Joined: Thu Sep 17, 2015 11:39 pm
Contact:

Re: Work on the Minimig core?

Post by kolla »

DanyPPC wrote: Tue Jan 26, 2021 1:35 pm @robinsonb5:

Have had you time to fix the borderblank problem ?
Thanks.
It seems fixed in the core I built, so I guess “yes” :)

http://amiga.kolla.no/minimig_beta.lha
-- kolla
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

kolla wrote: Tue Jan 26, 2021 1:43 pm It seems fixed in the core I built, so I guess “yes” :)

http://amiga.kolla.no/minimig_beta.lha
Yes, it was fixed fairly soon after being reported - sorry, I didn't realise the fix hadn't made it into a binary release yet.
DanyPPC
Atari Super Hero
Atari Super Hero
Posts: 998
Joined: Tue Feb 21, 2017 7:02 am

Re: Work on the Minimig core?

Post by DanyPPC »

Many thanks
slingshot
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2011
Joined: Mon Aug 06, 2018 3:05 pm

Re: Work on the Minimig core?

Post by slingshot »

It was included in the latest.
kolla
Captain Atari
Captain Atari
Posts: 349
Joined: Thu Sep 17, 2015 11:39 pm
Contact:

Re: Work on the Minimig core?

Post by kolla »

Is it just me, or is system time running waaay too fast with the current cores?
If it is just not me, can something be done to fix it? :)
-- kolla
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

kolla wrote: Tue Feb 16, 2021 11:05 am Is it just me, or is system time running waaay too fast with the current cores?
If it is just not me, can something be done to fix it? :)
It's probably because I'm misusing the AGA video framing registers to create the RTG sync pulses - I suspect the system's using either HSync or VSync pulses to measure real time, and assuming that these are 15KHz or 50/60Hz respectively. (The HSync and VSync signals feed the TOD inputs on the CIAs) AmigaOS must be smart enough to compensate for this when using doublescan screenmodes (right? right?!) so there's probably a way to fix it.

How much faster than "real life" is it running?

Some Amigas used a tick from the power supply instead of the VSync pulse to trigger one of the CIA's TOD counters - so that's a possible solution if it can't be fixed in software.
kolla
Captain Atari
Captain Atari
Posts: 349
Joined: Thu Sep 17, 2015 11:39 pm
Contact:

Re: Work on the Minimig core?

Post by kolla »

robinsonb5 wrote: Tue Feb 16, 2021 12:55 pm How much faster than "real life" is it running?
1 minute in real life is 1 minute and 10 secs on the Minimig core - so quite a lot faster :)
-- kolla
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

kolla wrote: Tue Feb 16, 2021 1:39 pm 1 minute in real life is 1 minute and 10 secs on the Minimig core - so quite a lot faster :)
Thanks, that suggests its using the vblank frequency - I will investigate tonight... :)
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

Plot twist!

I've just tested this on an RTG screenmode with a much-higher-than-50Hz vblank frequency, using a shell window and the date command:

Code: Select all

7.Workbench:> date
Saturday 15-Feb-03 19.01.25
<...exactly one minute later...>
7.Workbench:> date
Saturday 15-Feb-03 19.02.25
(No Real-time clock here, hence the screwy date.)

So which program are you using to measure?
kolla
Captain Atari
Captain Atari
Posts: 349
Joined: Thu Sep 17, 2015 11:39 pm
Contact:

Re: Work on the Minimig core?

Post by kolla »

robinsonb5 wrote: Tue Feb 16, 2021 8:55 pm So which program are you using to measure?
Just C:Date, like you :)

I have done some simple tests now, and it really seems to be related to screenmodes!

* PAL modes - system time is mostly correct, and if anything, just a wee little slow
* NTSC and RTG modes - system time is 8-10 secs too fast per minute

I found a fun way to confirm the above... if I use AmiTimeKeeper (an NTP client) and tell it to correct time from NTP every 60 seconds, the correction is for example 9825360 microseconds (close to 10 seconds) when RTG screen is foremost, but only 152587 microseconds (close to 0 seconds) when PAL screen is foremost :)
-- kolla
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

kolla wrote: Wed Feb 17, 2021 11:09 am Just C:Date, like you :)

I have done some simple tests now, and it really seems to be related to screenmodes!
Well that does make sense, since the CIAs' Time-of-Day counters are edge-triggered by video signals. (One CIA counts HSyncs, the other counts VSyncs.)
* PAL modes - system time is mostly correct, and if anything, just a wee little slow
* NTSC and RTG modes - system time is 8-10 secs too fast per minute
OK great - useful information. I must admit I didn't test from MiST last night, I used my devboard instead - but that shouldn't (!) make any difference.
I also only tested with my old WB3.0 installation from the A4K days - I'll test again with 3.1.4 and see if that makes any difference.

[Edit: yes, it does - system time gains time if the VBlank frequency is higher than 50Hz under Kickstart 3.1.4, but not under Kickstart 3.1 - even using the same installation - so it seems to be a ROM thing. Does that match what you see?]
I found a fun way to confirm the above... if I use AmiTimeKeeper (an NTP client) and tell it to correct time from NTP every 60 seconds, the correction is for example 9825360 microseconds (close to 10 seconds) when RTG screen is foremost, but only 152587 microseconds (close to 0 seconds) when PAL screen is foremost :)
Well that just raises more questions - like how are you contacting an NTP server from MiST?!
kolla
Captain Atari
Captain Atari
Posts: 349
Joined: Thu Sep 17, 2015 11:39 pm
Contact:

Re: Work on the Minimig core?

Post by kolla »

IP connection over a ppp link over the serial port to a raspberry pi zero taped ontop of the SD slot, which again has a usb wifi dongle... 115200 baud, 10kB/sec :)

Image
-- kolla
8bitguy1
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 101
Joined: Sat Feb 23, 2013 3:54 am

Re: Work on the Minimig core?

Post by 8bitguy1 »

That looks cool!, I'm hoping the Minimig core gets updated with the 88772 type usb to ethernet like the ST cores have....
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

kolla wrote: Wed Feb 17, 2021 11:56 pm IP connection over a ppp link over the serial port to a raspberry pi zero taped ontop of the SD slot, which again has a usb wifi dongle... 115200 baud, 10kB/sec :)
Nice! Much tidier than the NodeMCU ESP module I have crudely lashed up to my DE10-lite devboard!

I've found a clue regarding the clock problem: Kickstart 3.1.4 believes that Minimig has a power supply tick signal (which it doesn't) - so it's setting the GfxBase->DisplayFlags TODA_SAFE bit. The system thus believes that the CIA-A TOD counter is fixed at the power supply frequency, when in fact it varies with the video mode. Now to figure out why Kickstart 3.1.4 is reaching that erroneous conclusions, while Kickstart 3.1 isn't.
Last edited by robinsonb5 on Fri Feb 19, 2021 11:20 am, edited 1 time in total.
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

[Ignore double-post - was trying to edit!]
kolla
Captain Atari
Captain Atari
Posts: 349
Joined: Thu Sep 17, 2015 11:39 pm
Contact:

Re: Work on the Minimig core?

Post by kolla »

robinsonb5 wrote: Fri Feb 19, 2021 11:18 am
kolla wrote: Wed Feb 17, 2021 11:56 pm IP connection over a ppp link over the serial port to a raspberry pi zero taped ontop of the SD slot, which again has a usb wifi dongle... 115200 baud, 10kB/sec :)
Nice! Much tidier than the NodeMCU ESP module I have crudely lashed up to my DE10-lite devboard!
Mmm, I have a plan to replace the raspberry pi zero with a orange pi zero, which has dedicated serial port, external wifi antenna, and also an RJ45 NIC ready when the Minimig/MiST eventually gets ethernet support ;)
I've found a clue regarding the clock problem: Kickstart 3.1.4 believes that Minimig has a power supply tick signal (which it doesn't) - so it's setting the GfxBase->DisplayFlags TODA_SAFE bit. The system thus believes that the CIA-A TOD counter is fixed at the power supply frequency, when in fact it varies with the video mode. Now to figure out why Kickstart 3.1.4 is reaching that erroneous conclusions, while Kickstart 3.1 isn't.
Aha, yes this makes sense, so it thinks the MiST is A2000? :D
My kickstart is a blend of components from various 3.x releases - do you exactly which "rom module" it is that does this mistake? timer.device? battclock.resource?
-- kolla
robinsonb5
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 146
Joined: Sat May 16, 2015 3:02 pm

Re: Work on the Minimig core?

Post by robinsonb5 »

kolla wrote: Fri Feb 19, 2021 2:06 pm Aha, yes this makes sense, so it thinks the MiST is A2000? :D
My kickstart is a blend of components from various 3.x releases - do you exactly which "rom module" it is that does this mistake? timer.device? battclock.resource?
At the moment I think it's graphics.library! That's the module which is responsible for figuring out whether the TOD clock is driven by a power_tick signal, and it does at the same time as it checks for a genlock.
Post Reply

Return to “MiST”