Hybrid AO486 core?

bbond007
Top Contributor
Posts: 531
Joined: Tue May 26, 2020 5:06 am
Has thanked: 87 times
Been thanked: 211 times

Re: Hybrid AO486 core?

Unread post by bbond007 »

thorr wrote: Thu Jan 20, 2022 11:06 pm LOL, well what are you waiting for? ;-) :-D
My DOSBox was compiled on 02/03/2020, so almost 2 years ago...

I'm going try and xcompile dosbox-0.74-3 and see if the results are any different...
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

PCem looks like an excellent resource for creating cores for all these systems. I wish I had the time and knowledge to help. I would love to see a similar lineup of x86 systems in the MiSTer in FPGA. That way when you play a game, it runs at the right speed with the right graphics, native resolutions, sound, etc. and has BIOS setup screens like a real PC.
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

bbond007 wrote: Thu Jan 20, 2022 11:46 pm
thorr wrote: Thu Jan 20, 2022 11:06 pm LOL, well what are you waiting for? ;-) :-D
My DOSBox was compiled on 02/03/2020, so almost 2 years ago...

I'm going try and xcompile dosbox-0.74-3 and see if the results are any different...
Awesome! Please share so we can all use it, if it works. It would be nice to have as a backup and something to compare to when running things in the ao486 core.
User avatar
Reed_Solomon
Posts: 65
Joined: Sun May 24, 2020 7:37 pm
Has thanked: 40 times
Been thanked: 8 times

Re: Hybrid AO486 core?

Unread post by Reed_Solomon »

If FPU tasks could somehow be offloaded to the ARM side for calculations and results passed back to the ao486 core as if it were like plugging in a 386sx math coprocessor chip you could turn off and on from the mister ao486 menu, that would be more than enough for me.
Neocaron
Top Contributor
Posts: 375
Joined: Sun Sep 27, 2020 10:16 am
Has thanked: 209 times
Been thanked: 87 times

Re: Hybrid AO486 core?

Unread post by Neocaron »

So I asked in the Minimig thread, but what are the OC capability of the arm chip on the DE-10? Anyone tried to give it an OC? Is that even possible?

Remastering Classic Game Cinematics: My new Youtube fun, check it out :D
https://www.youtube.com/@neocaron87

bbond007
Top Contributor
Posts: 531
Joined: Tue May 26, 2020 5:06 am
Has thanked: 87 times
Been thanked: 211 times

Re: Hybrid AO486 core?

Unread post by bbond007 »

Neocaron wrote: Fri Jan 21, 2022 12:47 am So I asked in the Minimig thread, but what are the OC capability of the arm chip on the DE-10? Anyone tried to give it an OC? Is that even possible?
Using the free version Quartus you can possibly (through trial and error) make a ao486.RBF that is stable at 100mhz on YOUR device only --> viewtopic.php?f=13&t=3580
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

bbond007 wrote: Fri Jan 21, 2022 1:21 am
Neocaron wrote: Fri Jan 21, 2022 12:47 am So I asked in the Minimig thread, but what are the OC capability of the arm chip on the DE-10? Anyone tried to give it an OC? Is that even possible?
Using the free version Quartus you can possibly (through trial and error) make a ao486.RBF that is stable at 100mhz on YOUR device only --> viewtopic.php?f=13&t=3580
He is asking about the Arm chip, like when you overclock a Raspberry Pi. Personally I would not want to even try it with my MiSTer, but it is an interesting question.
User avatar
Caldor
Top Contributor
Posts: 930
Joined: Sat Jul 25, 2020 11:20 am
Has thanked: 112 times
Been thanked: 111 times

Re: Hybrid AO486 core?

Unread post by Caldor »

cursedverses wrote: Thu Jan 20, 2022 11:41 pm
Caldor wrote: Thu Jan 20, 2022 8:05 pm Bochs was used because it is open source and I am pretty sure PCem is not open source. Also I do not think it is possible to port PCem to Linux / ARM. I guess that might not be an issue if it is to be made into FPGA code anyway though.
I'm not sure about the past, but the previous maintainer posted sources onto github under GPL 2 - https://github.com/sarah-walker-pcem/pcem

Maybe it wasn't picked due to it's dependency on various firmware/BIOS files?
Interesting... yeah, not sure why Bochs was then picked. Maybe Bochs was done in a way that was easier to automatically translate to FPGA logic.
dmckean wrote: Thu Jan 20, 2022 9:08 pm The faults in the core are not going to be solved with the hybrid path. The faults in the core would need to be cleaned up before work even started.
Based on what? If the faults in the core are in the CPU part and that whole part is then no longer done by the core... well... then the problems would most likely be solved.

Memory addressing and such is afaik done in the CPU. The problems overall that I have come across... like needing to replace DOS4GW with DOS32A, even when the RAM is set to 16MB or less, is most likely due to issues with the CPU.

So if you have any arguments as to how the issues with the core would not be solved by doing this, then I am interested in knowing, because I cannot follow the logic.
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

bbond007 wrote: Thu Jan 20, 2022 11:46 pm My DOSBox was compiled on 02/03/2020, so almost 2 years ago...

I'm going try and xcompile dosbox-0.74-3 and see if the results are any different...
Did you try this yet? Any chance you can share it?
User avatar
amstan
Posts: 56
Joined: Sat Feb 12, 2022 8:08 am
Location: Mountain View, California
Has thanked: 5 times
Been thanked: 26 times
Contact:

Re: Hybrid AO486 core?

Unread post by amstan »

I installed dosbox 0.74.3 yesterday and tried Warcraft II at Caldor's prompt. It was much slower than the fpga core, like actively lagging.

It's possible a lot of it was due to the gpu blitting being slow on the ARM? I'm not sure. Are there any cpu benchmarks I should test with? I could also install qemu and try running x86 stuff in there (I guess I just need to point it at the same .vhd file as my ao486 core?).
rhester72
Top Contributor
Posts: 1323
Joined: Thu Jun 11, 2020 2:31 am
Has thanked: 15 times
Been thanked: 213 times

Re: Hybrid AO486 core?

Unread post by rhester72 »

re: those championing more pipelines and more sophisticated branch prediction...it's worth noting that at least in the case of AArch64, the ARM group found that ditching pipelines and BP in favor of more cache was a win.
dmckean
Posts: 310
Joined: Sat Jan 16, 2021 7:03 am
Has thanked: 396 times
Been thanked: 95 times

Re: Hybrid AO486 core?

Unread post by dmckean »

rhester72 wrote: Sat Feb 12, 2022 8:25 pm re: those championing more pipelines and more sophisticated branch prediction...it's worth noting that at least in the case of AArch64, the ARM group found that ditching pipelines and BP in favor of more cache was a win.
That doesn't surprise me considering that disabling all the cache in this core makes it run similar to a 5mhz 286.
bbond007
Top Contributor
Posts: 531
Joined: Tue May 26, 2020 5:06 am
Has thanked: 87 times
Been thanked: 211 times

Re: Hybrid AO486 core?

Unread post by bbond007 »

thorr wrote: Thu Feb 03, 2022 1:18 am
bbond007 wrote: Thu Jan 20, 2022 11:46 pm My DOSBox was compiled on 02/03/2020, so almost 2 years ago...

I'm going try and xcompile dosbox-0.74-3 and see if the results are any different...
Did you try this yet? Any chance you can share it?
Looks like I did compile it again on 01.20.2022.

The results are pretty much identical to the pervious (Superscape 3DMARK still 11.6 in 32BPP & 11.3 in 16BPP). Definitely not 486 level performance - More like 386DX.

I suppose I could take the time and do a proper installer like I've done for ScummVM and Basilisk II, however, it is really only useful for running Tandy/pcJr stuff and offers not much else...

DOSBox is not compatible with DOSLFN.COM and also does not support interrupt 0x21, 0x714e & 0x714f API (FindFirst & FindNext) for LFN (long filename), so that is a bummer too...
amstan wrote: Sat Feb 12, 2022 8:14 am It's possible a lot of it was due to the gpu blitting being slow on the ARM? I'm not sure. Are there any cpu benchmarks I should test with? I could also install qemu and try running x86 stuff in there (I guess I just need to point it at the same .vhd file as my ao486 core?).
I found running the framebuffer in 16BPP was actually a little slower than 32BPP.

I would suggest going with this pack:
https://www.philscomputerlab.com/dos-be ... -pack.html
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

bbond007 wrote:
I suppose I could take the time and do a proper installer like I've done for ScummVM and Basilisk II, however, it is really only useful for running Tandy/pcJr stuff and offers not much else...
Thanks. The Tandy sound is the main reason I want it. For the early Sierra games. I wish the Mister supported it. Also it may offer the ability to play more games that are not working in ao486 yet.
held
Posts: 209
Joined: Sun Sep 26, 2021 2:18 pm
Has thanked: 27 times
Been thanked: 32 times

Re: Hybrid AO486 core?

Unread post by held »

bbond007 wrote: Thu Jan 20, 2022 11:46 pm My DOSBox was compiled on 02/03/2020, so almost 2 years ago...

I'm going try and xcompile dosbox-0.74-3 and see if the results are any different...
Why not compile the latest dosbox-SVN ? That's where all the good stuff is :D

The 0.74-3 is almost 3 years old 2019-06-26, 18:53
vanfanel
Posts: 179
Joined: Sun May 24, 2020 6:53 pm
Has thanked: 13 times
Been thanked: 24 times

Re: Hybrid AO486 core?

Unread post by vanfanel »

held wrote: Fri Feb 18, 2022 4:50 pm
Why not compile the latest dosbox-SVN ? That's where all the good stuff is :D

The 0.74-3 is almost 3 years old 2019-06-26, 18:53
Has DOSBox seen any substantial development on the last years? I thought it was an stagnated project because of architectural decisions that make it impossible to improve it much...
jordi
Posts: 282
Joined: Thu Jun 11, 2020 10:11 am
Has thanked: 112 times
Been thanked: 87 times

Re: Hybrid AO486 core?

Unread post by jordi »

vanfanel wrote: Fri Feb 25, 2022 11:23 am
held wrote: Fri Feb 18, 2022 4:50 pm
Why not compile the latest dosbox-SVN ? That's where all the good stuff is :D

The 0.74-3 is almost 3 years old 2019-06-26, 18:53
Has DOSBox seen any substantial development on the last years? I thought it was an stagnated project because of architectural decisions that make it impossible to improve it much...
there are actually attempts on modernizing it
https://github.com/dosbox-staging/dosbox-staging
held
Posts: 209
Joined: Sun Sep 26, 2021 2:18 pm
Has thanked: 27 times
Been thanked: 32 times

Re: Hybrid AO486 core?

Unread post by held »

jordi wrote: Fri Feb 25, 2022 12:10 pm
vanfanel wrote: Fri Feb 25, 2022 11:23 am
held wrote: Fri Feb 18, 2022 4:50 pm
Why not compile the latest dosbox-SVN ? That's where all the good stuff is :D

The 0.74-3 is almost 3 years old 2019-06-26, 18:53
Has DOSBox seen any substantial development on the last years? I thought it was an stagnated project because of architectural decisions that make it impossible to improve it much...
there are actually attempts on modernizing it
https://github.com/dosbox-staging/dosbox-staging
That one is not from the DOSBox developers :D the SVN is here and has all the latest goodies.

DOSBox-staging is a DOSBox fork just like DOSBox-X or DOSBox-ECE
bbond007
Top Contributor
Posts: 531
Joined: Tue May 26, 2020 5:06 am
Has thanked: 87 times
Been thanked: 211 times

Re: Hybrid AO486 core?

Unread post by bbond007 »

held wrote: Fri Feb 25, 2022 2:34 pm DOSBox-staging is a DOSBox fork just like DOSBox-X or DOSBox-ECE
Unfortunately it needs SDL 2.0. I don't believe SDL2.0 supports framebuffer
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

bbond007 wrote: Fri Feb 25, 2022 11:44 pm Unfortunately it needs SDL 2.0. I don't believe SDL2.0 supports framebuffer
So can we just add OpenGL? ;-)
User avatar
Caldor
Top Contributor
Posts: 930
Joined: Sat Jul 25, 2020 11:20 am
Has thanked: 112 times
Been thanked: 111 times

Re: Hybrid AO486 core?

Unread post by Caldor »

vanfanel wrote: Fri Feb 25, 2022 11:23 am
held wrote: Fri Feb 18, 2022 4:50 pm
Why not compile the latest dosbox-SVN ? That's where all the good stuff is :D

The 0.74-3 is almost 3 years old 2019-06-26, 18:53
Has DOSBox seen any substantial development on the last years? I thought it was an stagnated project because of architectural decisions that make it impossible to improve it much...
DOSBox X has some new updates, but its also based on SDL1 and SDL2, not sure if either supports a framebuffer:
https://github.com/joncampbell123/dosbo ... INSTALL.md
bbond007
Top Contributor
Posts: 531
Joined: Tue May 26, 2020 5:06 am
Has thanked: 87 times
Been thanked: 211 times

Re: Hybrid AO486 core?

Unread post by bbond007 »

thorr wrote: Thu Jan 20, 2022 11:51 pm Awesome! Please share so we can all use it, if it works. It would be nice to have as a backup and something to compare to when running things in the ao486 core.
Sorry it took so long, The making a usable installer seems like hard part...
thorr wrote: Sun Feb 13, 2022 3:42 am Thanks. The Tandy sound is the main reason I want it. For the early Sierra games. I wish the Mister supported it. Also it may offer the ability to play more games that are not working in ao486 yet.
Tandy/pcJr is what I wanted it for too, and it works well for that. I have included a Tandy config.

0.74-3 --> https://github.com/bbond007/MiSTer_DOSBox
Malor
Top Contributor
Posts: 860
Joined: Wed Feb 09, 2022 11:50 pm
Has thanked: 64 times
Been thanked: 195 times

Re: Hybrid AO486 core?

Unread post by Malor »

Not really Mister-related, but my favorite version of DOSBox is DOSBox ECE, aka Enhanced Community Edition. It's still pretty much regular DOSBox, but with some extra patches. The ones I really enjoy are the built-in MUNT and Fluidsynth, so that it works very similarly to having an MT32-Pi. You just have to provide the ROMs, and it handles everything else.
User avatar
Caldor
Top Contributor
Posts: 930
Joined: Sat Jul 25, 2020 11:20 am
Has thanked: 112 times
Been thanked: 111 times

Re: Hybrid AO486 core?

Unread post by Caldor »

bbond007 wrote: Sat Mar 05, 2022 4:27 am
thorr wrote: Thu Jan 20, 2022 11:51 pm Awesome! Please share so we can all use it, if it works. It would be nice to have as a backup and something to compare to when running things in the ao486 core.
Sorry it took so long, The making a usable installer seems like hard part...
thorr wrote: Sun Feb 13, 2022 3:42 am Thanks. The Tandy sound is the main reason I want it. For the early Sierra games. I wish the Mister supported it. Also it may offer the ability to play more games that are not working in ao486 yet.
Tandy/pcJr is what I wanted it for too, and it works well for that. I have included a Tandy config.

0.74-3 --> https://github.com/bbond007/MiSTer_DOSBox
Yes, I tried this yesterday, pretty easy to install and such with this nice install script. The performance is not great, but it works. I guess this supports audio CDs I guess? But its performance is not great so I guess games with audio CDs are probably not relevant for it.
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

bbond007 wrote: Sat Mar 05, 2022 4:27 am Tandy/pcJr is what I wanted it for too, and it works well for that. I have included a Tandy config.

0.74-3 --> https://github.com/bbond007/MiSTer_DOSBox
Thanks!!! I am greatly looking forward to trying this out.
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

bbond007 wrote: Sat Mar 05, 2022 4:27 am Tandy/pcJr is what I wanted it for too, and it works well for that. I have included a Tandy config.

0.74-3 --> https://github.com/bbond007/MiSTer_DOSBox
Thanks very much for sharing this! I got a chance to try this briefly tonight. I got it working, and it is very exciting to have the Tandy sound working. It will be a whole new way for me to experience these old games. The only issue I am having is the joystick seems to be mapped to keyboard keys. If I move the joystick up in DOS, it recalls my last dos command for example. When I try to use it in a game it doesn't work like it should. Is there a way to fix this?

Also, I think the performance is just perfect if not a bit fast for the era of games I want to use with it.

Thanks again.
bbond007
Top Contributor
Posts: 531
Joined: Tue May 26, 2020 5:06 am
Has thanked: 87 times
Been thanked: 211 times

Re: Hybrid AO486 core?

Unread post by bbond007 »

thorr wrote: Wed Mar 09, 2022 5:04 am If I move the joystick up in DOS, it recalls my last dos command for example. When I try to use it in a game it doesn't work like it should. Is there a way to fix this?
Maybe try some different settings here --> https://github.com/bbond007/MiSTer_DOSB ... .conf#L186

Your dosbox-0.74-3_tandy.conf file should have been installed in /media/fat/DOSBox/Configs/
thorr wrote: Wed Mar 09, 2022 5:04 am Also, I think the performance is just perfect if not a bit fast for the era of games I want to use with it.
You can fine tune the performance here. I have the Tandy profile set to "auto" and the regular one set to run full speed. --> https://github.com/bbond007/MiSTer_DOSB ... y.conf#L67
thorr
Top Contributor
Posts: 1311
Joined: Mon Jul 06, 2020 9:37 pm
Has thanked: 634 times
Been thanked: 308 times

Re: Hybrid AO486 core?

Unread post by thorr »

Thanks! I will do some testing when I get a chance and report back.
Coolbho3k
Posts: 60
Joined: Sat Feb 26, 2022 6:36 pm
Has thanked: 7 times
Been thanked: 61 times

Re: Hybrid AO486 core?

Unread post by Coolbho3k »

With my DE10 Nano CPU overclock hack, using @bbond's DOSBox binaries (default profile) and the benchmark files found here: https://www.philscomputerlab.com/dos-be ... -pack.html

On the very first benchmark (3DBench 1.0) I get 16.1 FPS on 1.2 GHz and 10.2 FPS on stock (800 MHz). I think the increase is pretty linearly 50% as expected from the corresponding increase in CPU frequency.

Stock:
stock.jpg
stock.jpg (314.13 KiB) Viewed 5572 times
50% overclock:
overclock.jpg
overclock.jpg (289.49 KiB) Viewed 5572 times
Post Reply