MiSTer FPGA Forum

I played a lot with cache28 preview using DOS and Windows 95 yesterday and as long as I use 90 MHz and don't switch to 100 MHz, everything seems to be working perfectly already! I even played Doom in fullscreen and it was smooth! The game Stunts works perfect too, brings back memories. Even Wing Commander is now superfast using 90 MHz, needs slowing down by changing settings, just like on a real PC back in the day.

I found yesterday that midi music in Windows 95 sounds much better and is stereo when I select FM: OPL3 in Core Settings. Music in DOS games also seems to sound slightly better with OPL3 selected. Maybe Placebo but I will keep it on OPL3 for now.

This core is awesome, and cache28 preview especially is already amazing!

Heh, by any chance, do you know if Gravis Ultrasound support would be complicated ? (late DOS titles + demoscene

There is another new boot1.rom in the repo, not sure if it works better than the unofficial one we have been using, my system is still out for repair/replacement, so I can't test.

Edit: looking at the changes made, I can't tell if it would work any differently than the previous official boot1.rom. Seems like just a small cleanup of the code.

I'm not 100% sure about this, but, I think the shared_folder feature has had some support added, in a newer dev commit, so, cache29 might make that feature usable, but, don't quote me on this.

Edit: looks like it will require a new program, called MisterFS.exe to be run, for DOS, at least, and will be included in the releases folder of the dev branch (it's there, now, if someone wants to see if it works...) I really wish I still had my system to try it, myself.

tontonkaloun wrote: ↑Thu Jul 30, 2020 3:09 pm

Cebion wrote: ↑Thu Jul 30, 2020 2:38 pm I seem to have issues mounting disks after Dos 6.22 has loaded. On startup with a bootdisk works just fine.
Can someone verify on Cache 28?

With QEMM?
try disabling smartdrv if it's in your autoexec.bat !!

Well it may be a QEMM Problem.
I did run optimzie again and now it shows me some gibberish item contents.
But no smartdrv in my auotexec.bat

IAmParadox wrote: ↑Thu Jul 30, 2020 3:36 pm I'm not 100% sure about this, but, I think the shared_folder feature has had some support added, in a newer dev commit, so, cache29 might make that feature usable, but, don't quote me on this.

Edit: looks like it will require a new program, called MisterFS.exe to be run, for DOS, at least, and will be included in the releases folder of the dev branch (it's there, now, if someone wants to see if it works...) I really wish I still had my system to try it, myself.

I tried the new boot1.rom, it gives an ASCII Art MiSTer Logo and the message VGABios svn-current 30/07/2020. In Windows 95, it works neither as Tseng ET4000 nor as Tseng ET4000 Win32, so I guess it is still the old VGA Bios.

tontonkaloun wrote: ↑Thu Jul 30, 2020 3:09 pm

Cebion wrote: ↑Thu Jul 30, 2020 2:38 pm I seem to have issues mounting disks after Dos 6.22 has loaded. On startup with a bootdisk works just fine.
Can someone verify on Cache 28?

With QEMM?
try disabling smartdrv if it's in your autoexec.bat !!

The strangest error

Chris23235 wrote: ↑Thu Jul 30, 2020 4:23 pm

IAmParadox wrote: ↑Thu Jul 30, 2020 3:36 pm I'm not 100% sure about this, but, I think the shared_folder feature has had some support added, in a newer dev commit, so, cache29 might make that feature usable, but, don't quote me on this.

Edit: looks like it will require a new program, called MisterFS.exe to be run, for DOS, at least, and will be included in the releases folder of the dev branch (it's there, now, if someone wants to see if it works...) I really wish I still had my system to try it, myself.

I tried the new boot1.rom, it gives an ASCII Art MiSTer Logo and the message VGABios svn-current 30/07/2020. In Windows 95, it works neither as Tseng ET4000 nor as Tseng ET4000 Win32, so I guess it is still the old VGA Bios.

I just noticed they updated the boot1.rom on the dev branch:
https://github.com/MiSTer-devel/ao486_M ... eases/bios

Not sure whether its the one to use with patch28 or not, I will try to CRC check it. I guess they might have made their own bios? Afaik, there are open source ones around. I noticed while searching for the et4000 bios that there was some article going into how exactly that GPU works. Well... or at least tried to figure it out and somewhat reverse engineer it.

IAmParadox wrote: ↑Thu Jul 30, 2020 3:36 pm I'm not 100% sure about this, but, I think the shared_folder feature has had some support added, in a newer dev commit, so, cache29 might make that feature usable, but, don't quote me on this.

Edit: looks like it will require a new program, called MisterFS.exe to be run, for DOS, at least, and will be included in the releases folder of the dev branch (it's there, now, if someone wants to see if it works...) I really wish I still had my system to try it, myself.

I hope that RMA is processed fast.

Cebion wrote: ↑Thu Jul 30, 2020 5:05 pm

tontonkaloun wrote: ↑Thu Jul 30, 2020 3:09 pm

Cebion wrote: ↑Thu Jul 30, 2020 2:38 pm I seem to have issues mounting disks after Dos 6.22 has loaded. On startup with a bootdisk works just fine.
Can someone verify on Cache 28?

With QEMM?
try disabling smartdrv if it's in your autoexec.bat !!

The strangest error
20200730_170343-screen.png

let me see your config.sys and autoexec.bat file

Cebion wrote: ↑Thu Jul 30, 2020 5:05 pm

tontonkaloun wrote: ↑Thu Jul 30, 2020 3:09 pm

Cebion wrote: ↑Thu Jul 30, 2020 2:38 pm I seem to have issues mounting disks after Dos 6.22 has loaded. On startup with a bootdisk works just fine.
Can someone verify on Cache 28?

With QEMM?
try disabling smartdrv if it's in your autoexec.bat !!

The strangest error
20200730_170343-screen.png

Check you have the correct parameters when loading EMM386/QEMM, you are stepping on a region of the upper memory area that is causing that problem.

I think the reoptimimization screwed with the original values. But there is no return

We really should make a nice config.sys and autoexec section so that everyone has a guideline.

Cebion wrote: ↑Thu Jul 30, 2020 6:02 pm I think the reoptimimization screwed with the original values. But there is no return

We really should make a nice config.sys and autoexec section so that everyone has a guideline.

The memory area has moved? Hmmm... that could explain why I cannot get Game Wizard to run I guess. Maybe I should try to see what memmaker suggests... I do not remember if it added the more specific values to the config. I found them on some thread and have been using a certain configuration every since. I have not tried using QEMM with AO486 though, I just use regular EMM386.

Caldor wrote: ↑Thu Jul 30, 2020 6:12 pm

Cebion wrote: ↑Thu Jul 30, 2020 6:02 pm I think the reoptimization screwed with the original values. But there is no return

We really should make a nice config.sys and autoexec section so that everyone has a guideline.

The memory area has moved? Hmmm... that could explain why I cannot get Game Wizard to run I guess. Maybe I should try to see what memmaker suggests... I do not remember if it added the more specific values to the config. I found them on some thread and have been using a certain configuration every since. I have not tried using QEMM with AO486 though, I just use regular EMM386.

Before I ran qemm the first time I've set up a specific region and it worked perfect.
Then I added some stuff like a custom doskey and some game installations and qemm reported that it needed to run optimize again.
After that the floppy issues began. Temporarily I switched back to no ems and the memory region and everything works again.

I'll post my configs later so we can check what qemm did.
Qemm 8.03 btw, I only installed it because Bond wrote in Midilink that it is necessary for softmpu.

What is it SoftMPU does? Is it more than the ability to play midi-files? I can play midi files under DOS using SBPRO software, I also added the SBPRO Midi driver. Not sure how that relates to MPU. Is it to do with using the UART Midi solution? I saw a video someone made that showed switching the midi... instrument around for different midi files I think it was. Pretty cool. Pretty sure that is not possible with this pretty simple midi player the SBPRO driver has.

Yeah, SoftMPU lets you use the UART midi function to send midi commands to either Fluidsynth or Munt (MT-32 emu) running on the Mister's ARM processor. You can also send the Midi signals over network to another computer running a synth on it. If really like this function since the Mister's ARM isn't quite capable enough to run Munt perfectly. BBond007 made a nice Windows program, MiSTer Midilink UDP Listener, that makes it really easy to setup.

SuperBabyHix wrote: ↑Thu Jul 30, 2020 6:40 pm Yeah, SoftMPU lets you use the UART midi function to send midi commands to either Fluidsynth or Munt (MT-32 emu) running on the Mister's ARM processor. You can also send the Midi signals over network to another computer running a synth on it. If really like this function since the Mister's ARM isn't quite capable enough to run Munt perfectly. BBond007 made a nice Windows program, MiSTer Midilink UDP Listener, that makes it really easy to setup.

That does sound pretty great Many of the cores the MiSTer runs are pretty great for creating music, maybe especially the Amiga and DOS computers... I think Atari also, might even have better midi capabilities than the Amiga? I never played around with it much, but have seen people connect keyboards to computers and both have the computer use the keyboard and the keyboard use the computer, to create and play music. Even the C64 could be connected to a keyboard if I remember correctly. There is a video that got shared on Youtube where they go over all the hardware you could use with the C64. Microphones, printers, some simple scanners, modems and so on. Its amazing how much could be done with computers even back then.

Caldor wrote: ↑Thu Jul 30, 2020 7:10 pm

SuperBabyHix wrote: ↑Thu Jul 30, 2020 6:40 pm Yeah, SoftMPU lets you use the UART midi function to send midi commands to either Fluidsynth or Munt (MT-32 emu) running on the Mister's ARM processor. You can also send the Midi signals over network to another computer running a synth on it. If really like this function since the Mister's ARM isn't quite capable enough to run Munt perfectly. BBond007 made a nice Windows program, MiSTer Midilink UDP Listener, that makes it really easy to setup.

That does sound pretty great Many of the cores the MiSTer runs are pretty great for creating music, maybe especially the Amiga and DOS computers... I think Atari also, might even have better midi capabilities than the Amiga? I never played around with it much, but have seen people connect keyboards to computers and both have the computer use the keyboard and the keyboard use the computer, to create and play music. Even the C64 could be connected to a keyboard if I remember correctly. There is a video that got shared on Youtube where they go over all the hardware you could use with the C64. Microphones, printers, some simple scanners, modems and so on. Its amazing how much could be done with computers even back then.

And that's the reason the mister is connected to these bad boys at the moment

what brand are those speakers?

I've been trying to use Oracle VM to test MisterFS, it is actually etherDFS, but, I haven't had much success. I installed MSDOS network drivers, and set up an etherdfs server on one of my linux machines, but, I can't seem to properly access the network, yet, from the VM.

Caldor wrote: ↑Thu Jul 30, 2020 5:11 pmI just noticed they updated the boot1.rom on the dev branch:
https://github.com/MiSTer-devel/ao486_M ... eases/bios

Not sure whether its the one to use with patch28 or not, I will try to CRC check it. I guess they might have made their own bios? Afaik, there are open source ones around. I noticed while searching for the et4000 bios that there was some article going into how exactly that GPU works. Well... or at least tried to figure it out and somewhat reverse engineer it.

I have been looking at this boot1.rom on the dev branch. When opening it it opens with:

IBM Plex86/Bochs VGABios current-svn 30 Jul 2020

(C) 2002-2020 the LGPL VGABios developers Team

So it might be new, but does not seem to be a VESA compatible bios.

In the bios file with the CRC that was said to be the one the VESA stuff is developed for it says:

* Copyright(c)1988 Tseng Laboratories, Inc. 04/28/95 V2.10N

Text found in other ET4000 bioses:

This is not a product of IBM (IBM is a trademark of International Business Machines Corp.)ë[ * Copyright(c)1988 Tseng Laboratories, Inc. 04/15/92 V8.06X

This is not a product of IBM (IBM is a trademark of International Business Machines Corp.)ë[ * Copyright(c)1988 Tseng Laboratories, Inc. 04/07/93 V8.01X

This is not a product of IBM (IBM is a trademark of International Business Machines Corp.)ëY * Copyright(c)1988 Tseng Laboratories, Inc. 04/28/95 V8.00N

Oh, and text found in the bios on the master branch:

IBM Plex86/Bochs VGABios current-cvs 30 Mar 2014

(C) 2008 the LGPL VGABios developers Team

This VGA/VBE Bios is released under the GNU LGPL


Please visit :

. http://bochs.sourceforge.net

. http://www.nongnu.org/vgabios

I will try this newest one on the dev branch to see if it allows for VESA or not. I am thinking there are license / copyright problems with using the ET4000 bios in the eventual release version, and that being the reason they seem to be switching away from Tseng? If they are... still not sure what to make of this.

Testing the Cache28,

Bios from DEV branch (https://github.com/MiSTer-devel/ao486_MiSTer/tree/dev)
Dos 6.22 and Windows OSR 2.5 C
DOS Univbe driver (Universal VESA video BIOS Extension 1.2)

You can fully install Windows 95, but after some time the HD stops working and cannot boot again (I/O Error). You can use the 800x600 resolution but not change the number of colors with the default W95 video driver for ET4000.

The DOS game performance is really impressive, games like Mortal Kombat 2, Darkforce, Warcraft 2, DOOM, etc. work at FULL speed.

But I am getting a lot of problems running any SVGA games like Heroes of Might and Magic 1 & 2, Simcity2000 or other like NFS, Wipeout or Quake. Quite similar memory errors.

¿Could the memory map/addresses have change?.

This is my EMM386 config:

EMM386.EXE RAM 32768 FRAME=E000 D=256 X=A000-C7FF I=C800-EFFF

I have been testing the different bioses as well. The dev branch one is still a VGA bios. It does not support VESA as far as I can tell.

so its still this bios they are using for the dev branch and its VESA support I think:
http://www.elpod.org/pcem_roms/et4000w32.bin

It needs to be renamed to boot1.rom to work.

You got Warcraft 2 running? Pretty sure it crashed for me so far. The error it has, if its solved, might also make several other games able to run.

Cebion wrote: ↑Thu Jul 30, 2020 6:26 pm
I'll post my configs later so we can check what qemm did.
Qemm 8.03 btw, I only installed it because Bond wrote in Midilink that it is necessary for softmpu.

EMM386 also works as well. Some people reported problems (like hung notes) when using it with SoftMPU, but that has not been my experience...

Now I use a DOS 6.22 boot menu with QEMM 9.1 and EMM386 configs...

luishg wrote: ↑Thu Jul 30, 2020 9:43 pm
¿Could the memory map/addresses have change?.

This is my EMM386 config:

EMM386.EXE RAM 32768 FRAME=E000 D=256 X=A000-C7FF I=C800-EFFF

I think C800-CFFF is in use now...

Cache29 is up. I also just tested Settlers 2 with Cache28 using the new bios0.rom from the dev branch of the git, and this bios:
http://www.elpod.org/pcem_roms/et4000w32.bin - renamed to bios1.rom.

Caldor wrote: ↑Thu Jul 30, 2020 10:28 pm Cache29 is up. I also just tested Settlers 2 with Cache28 using the new bios0.rom from the dev branch of the git, and this bios:
http://www.elpod.org/pcem_roms/et4000w32.bin - renamed to bios1.rom.

I included the bios0.rom from the DEV branch and the ET4000 Video VGA BIOS linked here viewtopic.php?f=13&t=71

Also renaming it to "bios1.rom" will not work - must be called "boot1.rom"

Cache29 is working for me at 100mhz where as 27 did not. W95 runs like a dream!

bbond007 wrote: ↑Thu Jul 30, 2020 10:55 pm

Caldor wrote: ↑Thu Jul 30, 2020 10:28 pm Cache29 is up. I also just tested Settlers 2 with Cache28 using the new bios0.rom from the dev branch of the git, and this bios:
http://www.elpod.org/pcem_roms/et4000w32.bin - renamed to bios1.rom.

I included the bios0.rom from the DEV branch and the ET4000 Video VGA BIOS linked here viewtopic.php?f=13&t=71

Also renaming it to "bios1.rom" will not work - must be called "boot1.rom"

Yeah, I keep calling it bios instead of boot for some reason. With the boot1.rom, I tried the one linked in the driver thread, and the one I linked. I do not see much difference.

During boot the bios I linked shows a quick:
ET4000/W32
2 MB display memory

With the one in the driver thread, it instead only show the "2 MB display memory" line.

I like that you included all the files in the new release, thanks for that. I have begun adding the files to the older releases, so that I include the rom files that work for each release.

I also noticed that if I use the boot1.rom from the dev branch, and I am guessing the release branch one as well, I can run all the cores. The current release one, the cache20 one and all the new ones.

I think maybe both these bioses are wrong actually. I think the core only has 512kb video memory as the benchmark tool says. That could explain why Warcraft II fails to run with the VESA bios but not without it.

Btw. I have tested Cache29 and now the memory limiter works, it is now reduced to 16mb RAM instead of 64mb RAM.

I have been holding off but heavily thinking about posting this due to all of the current forward progress on this core and not wanting to derail it, but since someone brought up SoftMPU, I figured now is as good as a time as any. I posted in another thread about HardMPU. The HardMPU is essentially an opensource arduino that can go into an ISA slot and pretend to be a real MPU401. Here are links to information about it:
https://github.com/ab0tj/HardMPU
http://www.vcfed.org/forum/showthread.p ... c4dbaa0d14

I am really hoping that someone can look at the HardMPU code to learn about it and implement the MPU401 in hardware on the Mister as part of the ao486 core instead of needing to use SoftMPU. I have a Roland Sound Canvas and a Roland UM-One mk2 and would love to be playing Dos games with it (and Munt).