ao486

[Sorgelig]

github is the place to track the development.

[Retr80s]

github is the place to track the development.

Great! I'll check it out! Thanks

[JimDrew]

I need to figure out why FUSION-PC crashes out. It seems that there is some unsupported opcode.

[ameliarobin]

Most of the Windows 10 users are getting such kind of similar errors like adobe error code df024 that definitely solved by them. These booting issues should be recovered by them to access in an efficient manner.

[PiddlyD]

Will the Cutemouse mouse driver work in Windows 95 DOS mode on ao486?

Right now I've got great results with Windows 95 installed on a .vhd on ao486 - but I need to create autoexec.bat and config.sys to load himem/emm, soundblaster, and mouse support... and it has been a long time...

all the discussion here seems to have focused on DOS 6.x - but it sounds like people were getting DOS 6 configured with DOS and SB support - so it stands to reason that it should work with Windows 95 DOS 7, too, right?

[Hewhoisred]

all the discussion here seems to have focused on DOS 6.x - but it sounds like people were getting DOS 6 configured with DOS and SB support - so it stands to reason that it should work with Windows 95 DOS 7, too, right?

The DOS that comes with Win95, surely it will work fine, but if you want a more full featured DOS experience, I would just install 4DOS on top of MDDOS 6.22 as it adds a ton of nice new features, like autocomplete and easier navigation.

[PiddlyD]

all the discussion here seems to have focused on DOS 6.x - but it sounds like people were getting DOS 6 configured with DOS and SB support - so it stands to reason that it should work with Windows 95 DOS 7, too, right?

The DOS that comes with Win95, surely it will work fine, but if you want a more full featured DOS experience, I would just install 4DOS on top of MDDOS 6.22 as it adds a ton of nice new features, like autocomplete and easier navigation.

Thanks, Hewhoisred - I'll look into it. The nice thing is that you can set up as many .vhds as you have time!

I'm just blown away that this little FPGA box can even emulate a fairly modern DOS/Windows device with decent results. It wasn't a trip down memory lane I was expecting the MiSTer to take me on.

[breiztiger]

sorry but ...

http://www.pouet.net/prod.php?which=79996

is it normal that pc speaker does'nt sound ok ?

[Glaurung]

trying prince of persia, a X286 "old" game

The games seems to load perfectly with music and graphics, but the game controls doenst work.I have tried to custom controls, but still doenst work,¿ its a bug or I'm doing something wrong? Tring with keyboard.

Another games controls are working good with keyboard

Anyway, amazing that fpga can run a x86 core!

[Hewhoisred]

Tring with keyboard.

I'm not sure if this will help, but once the game loads, try "ctrl-K" to turn on keyboard mode.

[Glaurung]

Tring with keyboard.

I'm not sure if this will help, but once the game loads, try "ctrl-K" to turn on keyboard mode.

ok now is perfect!

I didint know ctrl+k, other games doesnt need to press ctrl+k to play with keyboard
thanks

[glaucon1984]

I didint know ctrl+k, other games doesnt need to press ctrl+k to play with keyboard
thanks

This Ctrl+K thing is not a MiSTer feature, it's a Prince of Persia feature, so it's normal you don't see it in other games.

Just in case you have lost the original manual that came with your game:
https://strategywiki.org/wiki/Prince_of_Persia/Controls

[Glaurung]

I didint know ctrl+k, other games doesnt need to press ctrl+k to play with keyboard
thanks

This Ctrl+K thing is not a MiSTer feature, it's a Prince of Persia feature, so it's normal you don't see it in other games.

Just in case you have lost the original manual that came with your game:
https://strategywiki.org/wiki/Prince_of_Persia/Controls

ok thanks.

I think this core is a x286, more than a x486.
Seems that this cpu code is very difficult to improve.

[Sorgelig]

I think this core is a x286, more than a x486.

try to run Win95 and other 32bit apps on i286

[BBond007]

I think this core is a x286, more than a x486.

try to run Win95 and other 32bit apps on i286

I would agree if Glaurung would have said 386 or 386sx. there is no FPU and "486" in the name does set speed expectation a little high

There was really very little difference between a 286 and 386 as far as performance other than clock speed. https://www.vogons.org/viewtopic.php?t=46350

The 386 introduced the 32bit flat memory model and MMU...

I like the core very much as the speed is perfectly adequate for many of the DOS games I grew up with and ran on my fast 20mhz 286.

Also the speed of ao486 video RAM is much-much faster than even the best ISA bus VGA card - more like VLB or PCI

[troydm]

Anyone having issues when uploading big (2-3mb files) using PPP connection over FTP using dos_ftpd.zip package inside MS-DOS 6.22? Like when file upload is almost completed it stucks and then connection reset happens on FTP client and inside ao486 the file is somehow corrupted? This doesn't happens with small files but almost all big zip files that I try to upload cause this issue. I have to reupload files multiple times before it successfully uploads files. I have no issues uploading files over FTP in Amiga core using wu-ftpd daemon so I'm making assumption that the problem is either with the core itself or with dos_ftpd.zip package

[Sorgelig]

yeah, i have the same issue. Probably ftpd is buggy. Need to find a better ftp daemon for ao486
May be someone can write?

[Glaurung]

http://www.brutman.com/mTCP/

[Sorgelig]

http://www.brutman.com/mTCP/

and what is the experience with it besides just throwing link here?

[glaucon1984]

http://www.brutman.com/mTCP/

and what is the experience with it besides just throwing link here?

I personally use it on my real DOS machines with success, but not sure how it would work with ao486 as it needs a Packet Driver and I know nothing about the networking side of ao486.

If ao486 can use PPP over the serial port maybe here there is some useful info about drivers for this:
http://www.dendarii.co.uk/FAQs/network.html

There is a guide to set it up from PhillsComputerLab here:
https://www.youtube.com/watch?v=CgATAo9WVHs

[Newsdee]

I must be doing something wrong, I'm using Win10 but I can't open the VHD under Windows.
Here's what I do:

1) Create a vhd and attach it in Disk Manager
2) Then I format it to FAT from Windows 10 (i.e. FAT16)
3) Transfer the file to MiSTer, mount it from ao486 and reboot
4) It's not recognized by DOS, i.e. I can't d:\ into it
5) If I reinstall DOS it will reformat and wipe out the data
6) If I trasnfer back the (5) VHD file into Windows, it prompts me to reformat it after mounting it

What am I doing wrong? Does the format in 2) have to use specific settings?

[Sorgelig]

Probably MBR/non-MBR format conflicts.
ao486 requires MBR, so it's like raw dump from the whole HDD with possible several partitions inside.

[Newsdee]

Any docs on how can I create a VHD with a MBR? I suppose it would have to be DOS compatible and not Win10?

[Sorgelig]

I partition and format in ao486. Then can use some mounting tools like ImDisk to copy the files to VHD.

[calvinmorrow]

I wanted to start a thread to discuss the ao486 core and my attempts at looking for bottlenecks and opportunities to speed it up. Sorgelig and Alfikpl (the original core author) have done an amazing job bringing this core to life, although it has some known speed issues that may represent some opportunities to improve it a little.

For background, the ao486 core running at 90 MHz has performance roughly equivalent to a 386 SX 16 MHz. There was a GitHub Issue that Sorgelig opened to discuss performance at https://github.com/alfikpl/ao486/issues/4 .

I've spent a fair amount of time trying to look into some operation-level benchmarking which has been difficult due to software and technical limitations, however I've managed to come up with some initial results which may warrant additional discussion.

Code: Select all

100000000 NOP Operations in 3791 Milliseconds
100000000 MEMREAD Operations in 29560 Milliseconds
100000000 MEMWRITE Operations in 7802 Milliseconds
100000000 ADD Operations in 5494 Milliseconds
100000000 SUBTRACT Operations in 5494 Milliseconds
100000000 MULTIPLY Operations in 5494 Milliseconds

Disclaimer: I'm not a full-time programmer and I haven't used C/C++ for quite some time. This is also the first time I've tried using assembly, so its taken a bit of trial and error to get it right.

The C code I'm testing with is using DJGPP GCC under Windows 95 (due to DPMI and other general issues in vanilla DOS) calling volatile asm() in loops. gcc compilation was done with -O2 -funroll-loops -march=i486, -S was called to view the generated assembly. NOP is simply a loop of assembly NOP, Memory Read is a MOV %(ebx), %eax, Memory Write is a MOV %eax, %(ebx), ADD and SUB are subtract constant from register, and MULTIPLY is a LEAL.

The result that stands out the most is of course the Memory Read. If my math is right, its taking somewhere around 26 cycles to perform.

I'll jump into the Verilog next to see if there's anything I can uncover that might explain the discrepancy, though honestly I only started learning Verilog last month so I've got my work cut out for me.