Indivision AGA MK3 for A1200

Caution: Non registered users only see threads and messages in the currently selected language, which is determined by their browser. Please create an account and log in to see all content by default. This is a limitation of the forum software.


Also users that are not logged in can not create new threads. This is a, unfortunately needed, counter measure against spam. Please create an account and log in to start new threads.

Don't Panic. Please wash hands.
  • We now have a core for everyone to test - this is really only a test-core, as it does not meet all timing requirements. We normally aim at 165MHz output pixel clock, but this core is only good for a maximum of 155MHz output pixel clock. However, one important change is there, and that's the chipset communication fix. Please download the file from our Wiki (test core sfrom September 17th) and install the core using the "flash firmware file" menu item from the drop-down menu.


    Please report back if it works as good for you as it does for us - then it's worth putting more effort into meeting timing up to 165MHz pixel clock.


    No changes to the Audio engine yet - this issue was much more important to us, and we wanted to release something before the weekend, so we hopefully have your results on Monday.

    Well, i just did some test on my 1d1 board. With this board i had black screens with some demo's, games and in workbench (< 64 colors) .when cold.

    I cant reproduce this any more when the system is still cold :).


    Tomorrow i will test longer to see if it stays this way when warmed up. Its now only running 30 minutes (Jezus on Es without any blackscreen!).

  • Just wanted to give some feedback on HDMI audio. My Onkyo AV amplifier won't believe there is audio there and just stays silent. If I plug the HDMI straight into my Samsung TV then there is audio, but the pitch is incorrect (Worms in Worms DC sound like they are pitched down) and the audio seems to stutter like it's coming through a tremolo pedal. I've tried changing the sample rate between 44.1kHz and 48kHz and it just sounds the same.

  • We now have a core for everyone to test - this is really only a test-core, as it does not meet all timing requirements. We normally aim at 165MHz output pixel clock, but this core is only good for a maximum of 155MHz output pixel clock. However, one important change is there, and that's the chipset communication fix. Please download the file from our Wiki (test core sfrom September 17th) and install the core using the "flash firmware file" menu item from the drop-down menu.


    Please report back if it works as good for you as it does for us - then it's worth putting more effort into meeting timing up to 165MHz pixel clock.

    Thanks team will report back at some point this weekend.

  • you can either set to "black borders left and right", or you can stretch the picture to be full-screen.

    I don't see such an option in the setup tool, as far as I can tell the image is always stretched.


    Also, is it possible to switch to a different resolution in live config mode? This was suggested in this forum at some point but not documented in the manual.


    BTW, I set up a 960x540 mode on a 1920x1080 monitor and it looks amazing. A bunch more space for the workbench but not a sea of empty space like when I try to push my A3000's Cybervision card to its limits. 8)


    On the Amiga side this is PAL super-hires interlace with a custom screen size of 960x540. PAL overscan needs to be set large enough for 270/540 lines. Be sure that the Indivision triggers on SHires. On the HDMI side it's either EDID 1920x1080@60Hz 150 MHz pixel clock or a copy of that mode with the pixel clock reduced to 125 MHz so the framerate reduces to 50 Hz.


    Although this is a super-hires mode, it's actually not too slow in 256 colors, as the video chip only needs to read 960 pixels per line from chip RAM rather than 1280. This leaves a reasonable amount of chip RAM bandwidth for the CPU and the blitter.


    I would definitely recommend including such a resolution in the defaults.

  • Although this is a super-hires mode, it's actually not too slow in 256 colors, as the video chip only needs to read 960 pixels per line from chip RAM rather than 1280. This leaves a reasonable amount of chip RAM bandwidth for the CPU and the blitter.


    I would definitely recommend including such a resolution in the defaults.

    Thanks I’ll try that mode on my Dell 2311H which is also 1080p and supports 50Hz.

  • We now have a core for everyone to test - this is really only a test-core, as it does not meet all timing requirements. We normally aim at 165MHz output pixel clock, but this core is only good for a maximum of 155MHz output pixel clock. However, one important change is there, and that's the chipset communication fix. Please download the file from our Wiki (test core sfrom September 17th) and install the core using the "flash firmware file" menu item from the drop-down menu.


    Please report back if it works as good for you as it does for us - then it's worth putting more effort into meeting timing up to 165MHz pixel clock.


    No changes to the Audio engine yet - this issue was much more important to us, and we wanted to release something before the weekend, so we hopefully have your results on Monday.

    On my 1d1 board it solved the issues, ive been running the amiga for over 2 hours and none black screens any more :) everything runs smooth now.

  • I've had a few times where the screen turned black for a second or two, but not like before the update where there were a few times it kept going black every two seconds in some modes.


    However, I've had a lot of trouble with the MK3 seeing the difference between SHires vs not SHires. I had that before the update but it seems to have gotten worse. A reboot resolves this issue most or all of the time, though.

  • Jens,


    Yesterday I tried to recreate the issue again from cold on the 24th Aug firmware before flashing the Sept firmware, just to get a pre-flash check. For some reason the issue wasn't occurring which I thought was annoying from a consistency point of view. Anyway I launched the IndivisionAGAMK3 app and on a few occasions the app failed to recognise the presence of the card, even though I was getting a picture and appeared to be working - just not recognised, even when starting up without startup-sequence. Eventually it was recognised.


    Continued to test with the Enigma (my test demo) and still it didn't produce the issues I've seen previously. Anyway decided to reset the MK3 back to defaults then flash with Sept. firmware as a test. Result was the same, no change - the demo was working fine. Also noticed large white blocks flashes in workbench while moving through the menus or dragging windows around, but only when running in colour modes above 32, i.e. 256 colours (standard PAL mode). I wasn't getting this when I first had the MK3. I didn't see any such flashing via the RGB port connected to my Philips CM8833-II.



    Today I decided to flash back to Aug firmware and test the demo once more from cold. This flashing process was successful, the firmware version was showing as 20200824 from the Live Edit mode. Running Enigma again I did see the flicking issues for Enigma demo as I've done over the last few weeks, so I thought great, I'll flash immediately to Sept firmware and test again to see if the new firmware has fixed this.


    I launched the IndivisionAGAMK3 app and selected the Sept firmware, and.... no screen. Via the RGB port I see IndivisionAGAMK3 app had crashed during the flashing procedure! Unable to continue I soft rebooted, launched IndivisionAGAMK3 and it reports the MK3 is not detected. If I click continue, it crashes with error #80000005


    Multiple power cycles of increasingly longer periods yields no change, the IndivisionAGAMK3 app does not recognise the MK3 and there is no picture via the card. RGB is fine.


    I'm using KS 3.1.4 and have an ACA1233n-new 26MHz installed.

    Using v1.5 of the IndivisionAGAMK3 app


    What next please?




  • Just wanted to add that having tested a MK2cr, it works perfectly, detected every time including no issues with flashing white while in PAL @256 colours.


    Both on the same A1200 1D4 and with original PSU and one of your PSU's purchased at the same time as the MK3.

  • We now have a core for everyone to test - this is really only a test-core, as it does not meet all timing requirements. We normally aim at 165MHz output pixel clock, but this core is only good for a maximum of 155MHz output pixel clock. However, one important change is there, and that's the chipset communication fix. Please download the file from our Wiki (test core sfrom September 17th) and install the core using the "flash firmware file" menu item from the drop-down menu.


    Please report back if it works as good for you as it does for us - then it's worth putting more effort into meeting timing up to 165MHz pixel clock.

    With the September 17th test core, and a better PSU (CA-PSU), I'm seeing a big improvement in picture stability while running my aca1233n at 55MHz. In standard PAL Hires with 16 colours I didn't get a single instance of screen blanking in around 2-3 hours that I noticed. The JesusOnEs and Enigma demos ran fine too. Just to make sure it wasn't the CS-PSU making the all difference, I temporarily switched back to the August firmware and immediately got full screen flashing/flickering whenever there was Workbench redraw.


    Admittedly I'm being fairly conservative in my choice of screen mode, but the test core seems to have fixed a lot of issues I was experiencing.

  • I have put 17th test core and tested it. Previously I experienced two problems:

    1. Blackouts on PAL and NTSC screen modes when moving windows which was caused by false detection of SHires modes

    2. Ocassional bumps of the picture on HIGFX mode with showing OSD (also with moving or opening windows around). This one had no blackouts, just a quick bump. I'm still not sure what is the change that causes it.


    For the second (HIGFX) problem I can see an improvement. I did experience it again even with new core but that now happens really rarely. But first problem (false detection of SHires modes on PAL and NTSC) are still present. Subjectivelly it is the same as before. But I have found a workaround (which is fine for me). I have disabled SHires modes for PAL and NTSC so false detection no longer triger change of resolution and therefore also blackouts. Now only OSD is triggered on every move of the window. And I have solve this problem by setting OSD timeout to zero, so no OSD is shown for PAL and NTSC modes. This works for me perfectly since I do not need SHires and in general, I don't need information from OSD. To sumarrize, problem still exists but I'm happy with workaround settings I have made since for what I need, it works OK.


  • Similar issue here after a flash gone wrong (no picture, Indivision not detected by the tool, crashing if continuing anyway). Using the "flashtool" program as described in the wiki worked here. :thumbup:

  • About the incorrect SHires detection: I had a lot of trouble with that yesterday (with the new firmware installed) when I was making new screen mode mappings that use SHires so I kept going in and out of SHires while using the setup tool. But those problems were fixed with a reboot whenever they occurred. I don't think I had a problem when I wasn't using the setup tool.


    I now have my Workbench in 960x570 PAL SHires lace but also use regular PAL from time to time and I haven't seen the issue today.

  • For those where it still happens with the new core: Please try activating the PCLK pull-up resistor. This really sounds like a heavily shifting clock with temperature, which makes it really hard for the hardware to keep track of where things happen within a cycle. The pull-up resistor will help the rising edge of the clock, and that may pull back the glitches "over the edge" where they don't happen any more.

  • Update on testing

    TL;DR: Still getting issues with Sept firmware as I did with Aug - unless I disable ACA1233-n CPU which resolves all issues.


    Method:

    1. Used flashtool via CLI to reset the Indi3 and restored default operation, thanks for the tip @Leone
    2. Restored defaults on Indi3 v1.5 app, created a single new custom VGA mode (1440x1080@50Hz) for PAL. Interestingly getting new behaviour when making slight tweaks to clock values, I get full screen distortion even with known good values, but discovered this can be cured but going back to CLI and running "flashtool info" - screen goes black and comes back normal.
    3. Ran Enigma demo and noted the dropped frames and jumping in the usual places. Note other programs, demos etc. have their issues unless the 030 is disabled.
    4. In workbench, I'm no longer getting flashes in the screen while going through pull down menus in 256 colour mode (with 030 still enabled).... Not sure the re-flash cured that, or the "flashtool info" kick...
    5. I have "PCLK pull-up" selected in the Advanced options.
    6. Everything else at defaults.
  • This sounds like you have overwritten the core. Please run the "flash firmware from file" option again, as everything you're describing points to the old firmware being active.

    Sorry should have been more specific. For step 1 I used "flashtool flash indi_aga_mk3_prod.rbf"


    By "Restoring defaults" I meant to restore the Indi screen modes and VGA settings to defaults. I don't believe this performs a firmware flash?

    Besides which, the OSD shows the date of the new Sep firmware is active.

  • I looked through your postings in this forum, but I can't find any info about the Lisa chip that you have in your computer. What vendor is it? HP, Commodore or NCR?