Breakthrough for the ao486 core announced - Cache

[flynnsbit]

I tried all the ET4000 drivers I could find for Windows 95 and I tried several different DOS drivers as well. Nothing seems to work for me... but in many cases it does seem to run except it just freezes the display. I suspect if I connect the MiSTer to a CRT monitor, it might switch to VESA mode... but in Windows it keeps seeing it as a display driver problem to go into any VESA mode, even 640x480 with 256 collors.

But also, it shows it to only have 256kb RAM when I run Univbe in DOS, but then I try the one that comes with the game Z, it sees the GPU as a Tseng ET6000 with 2MB RAM.

So I think I will just wait and see what gets released, and play around with non-VESA stuff until then.

Ahh... the problem seems to be a new BIOS has to be compiled, the new BIOS for this has not been released yet.

But it seems it might be possible to compile one.
https://github.com/MiSTer-devel/ao486_MiSTer/tree/dev

This explains how to compile a VGA bios... but, guess its not simple to compile a SVGA bios with this guide only explaining how to compile the VGA one. Ahh well.

The actual bios file is in there, in releases directory, but, to get it completely working, a different bios may be required, and, that one hasn't been implemented, at all, afaik, although, there is apparently, a closed source bios, floating around, that works, I might have actually seen a system using this bios that seems to be doing pretty good, with it, but, since it is closed source, I'm pretty sure I wouldn't be able to share it, here. it is et4000 and has a crc of 713CB637 from what I hear.

Checksum treasure hunt over the Internet

I have found... maybe 10 different Tseng 4000ET bioses and have been doing CRC32 checks on them all. I have found one with this checksum here:
http://www.elpod.org/pcem_roms/

The one named: et4000w32.bin

Yup, that's it. Good job finding that in the wild. That thing was an easter egg out there without altavista.com search

[bazza_12]

tried it with win98 as soon as i changed screen resolution, the screen garbled - tried it with windows super vga driver, et4000, et4000w32 and et4000w32.pci driver. then gave up.. also noticed that in dos the tie-fighter screen crashed after the title crawler. although 386pga golf loaded but was slow on the screen redraw.. i think i'll wait for the official release

[IAmParadox]

I tried Death Rally and it keeps crashing, I do get into the game now, the menu and such, but not the game itself... I suspect the problem with 100mhz might also be... less of a problem. Both when it comes to detecting the sound devices and the freezing issues. Just seems strange that the bios affects that... or maybe not, but... should the bios problems with the GPU depend on the speed of the CPU? Guess it might.

The problem is one that happened on real hardware, the speed of the CPU causes the program from detecting the soundcard properly, many games have a patch to fix this issue, and, one solution, at the time, on real hardware, was to use the turbo button to slow the cpu down, until the soundcard is detected, and, then, go back to normal, after. it has absolutely nothing to do with ao486, except to show that it is performing more like a 100Mhz 486, than it was, before. As for my freezing issues, I think I have a faulty de 10, or, maybe there was a change in the hardware that the mister devs aren't aware of. And they will probably never know, since they like to close issues about such things, while being snarky, and, a bit rude.

[Caldor]

I tried Death Rally and it keeps crashing, I do get into the game now, the menu and such, but not the game itself... I suspect the problem with 100mhz might also be... less of a problem. Both when it comes to detecting the sound devices and the freezing issues. Just seems strange that the bios affects that... or maybe not, but... should the bios problems with the GPU depend on the speed of the CPU? Guess it might.

The problem is one that happened on real hardware, the speed of the CPU causes the program from detecting the soundcard properly, many games have a patch to fix this issue, and, one solution, at the time, on real hardware, was to use the turbo button to slow the cpu down, until the soundcard is detected, and, then, go back to normal, after. it has absolutely nothing to do with ao486, except to show that it is performing more like a 100Mhz 486, than it was, before. As for my freezing issues, I think I have a faulty de 10, or, maybe there was a change in the hardware that the mister devs aren't aware of. And they will probably never know, since they like to close issues about such things, while being snarky, and, a bit rude.

I had freezing issues yesterday as well... but it was just strange. I did suspect it was because maybe I had used it for a long time. Made me think it would be nice if the OSD had a way to show temperature information. But it just seems better now, I have not yet tried it with Doom though, which is the game where I am sure it was a problem. But I mean, I could even in the setup program go to the OSD, switch it to 100mhz, go back to the setup and have it tell me it could not find the soundcard, then change it to 30mhz and right away it would find it. But that works for me now.

tried it with win98 as soon as i changed screen resolution, the screen garbled - tried it with windows super vga driver, et4000, et4000w32 and et4000w32.pci driver. then gave up.. also noticed that in dos the tie-fighter screen crashed after the title crawler. although 386pga golf loaded but was slow on the screen redraw.. i think i'll wait for the official release

What bios have you been using? Seems to matter quite a bit whether its the right bios. All kinds of strange behavior when not using the right one it seems. In Windows 95 it needs to be the regular ET4000 driver it seems. So I am guessing its the same in Windows 98. I am trying to make a clean disk with no games installed and such with Windows 98 SE that I can share.

I can confirm, this is the bios that is "working" as I mentioned, but, it might not actually be the correct one, or, is named wrong, I switched from the et4000 drivers, in windows, to the et4000w32 drivers, and, it doesn't work, although, it does work with the et4000 drivers.

Same conclusion for me... the /W32 drivers keep failing, but the regular ET4000 drivers seem to work. I found... or read about... a v8 driver, while this one that comes with Windows 95 OSR2 seems to only be v4.0. That might help in some way.

But I am just happy to see Windows 95 now running with 16bit in 800x600. I will try to see if this means I can finish my Windows 98 SE install. Running at 100mhz at all times does not seem to cause any problems for me anymore.

Yup, that's it. Good job finding that in the wild. That thing was an easter egg out there without altavista.com search

Thanks Yeah, I could not help but find it a bit funny doing an easter egg hunt on the Internet. Its amazing how much the search keywords used can mean for the searches made. Adding " " at the right places and sometimes not having too many keywords. It tends to become quite a rabbit hole to dive into when searching for something obscure where you think you already went over everything, but then you find some new keyword that helps you find some other stuff that helps give more clues.

[IAmParadox]

This page does a good job of explaining the issues with speed affecting soundcard detection.

http://sierrahelp.com/Utilities/SoundUt ... iERRA.html

[kathleen]

Just checked with the ET4000W32.BIN and I'm so happy because my favorite program DESKMATE from tandy works like a charm now, while before, with the old bios it didn't work at all. Big thanks already for this !
I tried Win 3.11 with the same ET driver, it works also but only in 16 colors. Geos has also the same graphic issues than before but I guess they are more related to the AO486 core. Anyway, I'm enjoying more and more with this so wanted core. Thanks again for the hard work you've been doing, it is really appreciated.

[Apolonius]

Anyone have tried to run games with MIDI as sound card?, for me doesn't work. Only games that use Roland MT-32 work fine, the games that use Sound Canvas or General Midi not work. I'm using now the Cache24 version.

[Caldor]

This page does a good job of explaining the issues with speed affecting soundcard detection.

http://sierrahelp.com/Utilities/SoundUt ... iERRA.html

It is not that I doubt the problem exists, it just seemed it was now working better... but it was just some games that handles this problem better it seems. Duke3D and Doom still has the problem. Running stuff in Windows helps though. But Windows 98 is still causing problems and freezes. Not sure if that is to do with the 90 vs 100mhz problem, or if its because W98SE would like to have FPU.

I found out something else strange. With this new setup, Duke Nukem 3D runs quite a bit slower. I think I will just focus on finding out what games that require VESA I can run now, while waiting for the public official release.

[flynnsbit]

Windows 98 with the ET driver is unstable above 16 colors, Windows 95 with the ET4000 driver can run all the way up to 1024/768 256 color, though I keep it at 640x480 True color. W95 has been the most stable for me with the new driver. W98 will freak and show rainbow lines and then blue screen pretty often. Stick to W95 or Dos for now or wait for official. We are on the bleeding edge here, riding these dev builds as soon as things are checked in.

[Caldor]

Windows 98 with the ET driver is unstable above 16 colors, Windows 95 with the ET4000 driver can run all the way up to 1024/768 256 color, though I keep it at 640x480 True color. W95 has been the most stable for me with the new driver. W98 will freak and show rainbow lines and then blue screen pretty often. Stick to W95 or Dos for now or wait for official. We are on the bleeding edge here, riding these dev builds as soon as things are checked in.

Yeah... I was trying some more DOS games... it sure runs great. Bedlam works, Settlers 2, Albion... not all at full speed, but they run.

That Duke3D runs slower now makes me think maybe I should try Daggerfall again or System Shock. But I have decided to focus on Windows 95. I noticed that opening "This Computer" in Windows 98 will always cause that glitch with a lot of colors and lines and such. Very unstable indeed. It did autodetect the GPU though. I have been trying to install Alchohol 120% or Daemon Tools in Windows 95, but without any luck so far. Looking into NoCD versions of games now instead.

[NegSol]

I did a Need for Speed test
https://youtu.be/dguBqVsNpU8
It is already playable - I like it!

[rsn8887]

Amazing work!

I am playing Settlers 2 Gold using the “cache24” preview version of the core from this thread, et4000w32.bin used as boot1.rom and univbe 4.0 freeware vbe driver started under ms-dos. It works great even with full music and sound fx. Scrolling the map is a bit sluggish, but not too bad.

C0B063A7-F084-41C1-810A-AF03C8DBCF3D.jpeg (111.35 KiB) Viewed 9851 times

Windows 95 in 800x600 16bit works great too, using Tsing ET4000 drivers (standard ET4000 drivers, not the ET4000/w32 ones) that come bundled with Windows 95. After Dark 4.0 screensavers also work That brings back memories.

45469EA6-71CA-4A32-8926-3A2F1625C1F1.jpeg (85.85 KiB) Viewed 9851 times

Edit:

Z by Bitmap Brothers works perfectly in svga, too, using univbe 4.0:

171AB149-2557-454F-BFC8-F8230AE00136.jpeg (96.04 KiB) Viewed 9825 times

[bbond007]

ao486_Cache25_061fc53 AVI video test "Big Buck Bunny" --> https://youtu.be/ZLNP7y1BjwU

[IAmParadox]

I just tested Doom and Doom2 with cache25 and they start, with music, and play as good as they have been, without dropping to 30mhz, first. I can't promise the soundcard fix in this beta will fix that issue with all games that wouldn't initialize the soundcard at the higher speed, but, I have a feeling, it does.

I still get freezing with 100mhz, but, like I said, I think it's faulty hardware.

Also, the memory option doesn't seem to work, still get 65MB (max in dos 6.22) when set to 16MB but, I also get 65MB under DOS 7 (win95) instead of the 260MB, so, I'm thinking it's labeled wrong, and should say 64MB not 16MB, or, is set wrong, in code.

[Caldor]

I just tested Doom and Doom2 with cache25 and they start, with music, and play as good as they have been, without dropping to 30mhz, first. I can't promise the soundcard fix in this beta will fix that issue with all games that wouldn't initialize the soundcard at the higher speed, but, I have a feeling, it does.

I still get freezing with 100mhz, but, like I said, I think it's faulty hardware.

Also, the memory option doesn't seem to work, still get 65MB (max in dos 6.22) when set to 16MB but, I also get 65MB under DOS 7 (win95) instead of the 260MB, so, I'm thinking it's labeled wrong, and should say 64MB not 16MB, or, is set wrong, in code.

Yours freezes outside of Windows? I have not had it freeze in games, like Doom and such. It mainly seems to be Windows that has such issues, and even then it does not lock up completely. So yeah... you might need a second one just to make sure

Nice to hear Cache25 is out, I will try it out when I get home.

[IAmParadox]

I just tested Doom and Doom2 with cache25 and they start, with music, and play as good as they have been, without dropping to 30mhz, first. I can't promise the soundcard fix in this beta will fix that issue with all games that wouldn't initialize the soundcard at the higher speed, but, I have a feeling, it does.

I still get freezing with 100mhz, but, like I said, I think it's faulty hardware.

Also, the memory option doesn't seem to work, still get 65MB (max in dos 6.22) when set to 16MB but, I also get 65MB under DOS 7 (win95) instead of the 260MB, so, I'm thinking it's labeled wrong, and should say 64MB not 16MB, or, is set wrong, in code.

Yours freezes outside of Windows? I have not had it freeze in games, like Doom and such. It mainly seems to be Windows that has such issues, and even then it does not lock up completely. So yeah... you might need a second one just to make sure

Nice to hear Cache25 is out, I will try it out when I get home.

Yes, I have had it freeze outside of windows, and, in fact, in a few other cores, lxde (official Linux desktop environment for mister) will occasionally freeze while loading, and restart the system, once it starts successfully, I haven't had too many issues with lxde, and, even the system, itself has had slowdowns, and minor freezing, which is why I think it's hardware, or something, not just ao486. In fact, ao486 will freeze, for me, before even booting the HDD, when set to 100mhz, sometimes.

Edit: I also have to mention, some of this freezing happened before the cache cores were even around, but, since ao486 in the repo doesn't have the speed settings, I can't go back and test most of this in the official core.

[IAmParadox]

Can anyone get the 10x uart to work, in win95? it causes all kinds of bad stuff, for me, maybe, just unfinished, but, I'm wondering if it works for anyone.

[Caldor]

Can anyone get the 10x uart to work, in win95? it causes all kinds of bad stuff, for me, maybe, just unfinished, but, I'm wondering if it works for anyone.

I was trying to install some stuff in Windows 95 yesterday, but began getting errors like the ones you get in Windows 98 with Cache24. Not sure if that might be better in Cache25, but overall, I think Windows should be expected to be unstable. It might also just be that I had to close Win95 unsafely a few times.

I have not yet tried uart at all though, I have only begun to figure out what it even is. One feature is to use it for file access, and another to use it for some midi stuff?

[IAmParadox]

Can anyone get the 10x uart to work, in win95? it causes all kinds of bad stuff, for me, maybe, just unfinished, but, I'm wondering if it works for anyone.

I was trying to install some stuff in Windows 95 yesterday, but began getting errors like the ones you get in Windows 98 with Cache24. Not sure if that might be better in Cache25, but overall, I think Windows should be expected to be unstable. It might also just be that I had to close Win95 unsafely a few times.

I have not yet tried uart at all though, I have only begun to figure out what it even is. One feature is to use it for file access, and another to use it for some midi stuff?

I use it for internet access, and, I was hoping it would make it faster, but, all it seems to do, for me, is cause programs not to load (or, take a very long time to load) and the ppp connection stops responding, and, disconnects, and won't reconnect.

I'm hoping, someone implements some kind of usb access, that would allow flash drives, and, maybe, wifi dongles, or, usb ethernet adapters to be used... I know they don't want to waste space on the fpga for an ethernet card, but, maybe usb would be a way around it. I also know that usb support wasn't really standard for windows, until Win98, but, usb support was added to win95, using an installable usb stack...

[Caldor]

Yeah, I had hoped UART could be used to access files on the SD card or other storage media maybe. It seems he made a sharedfolders feature for the Minimig Amiga core. But my plan for now is to ignore Windows I think, because it just seems too unstable... well, unless Cache25 made it more stable, I might try it once more.

[IAmParadox]

Yeah, I had hoped UART could be used to access files on the SD card or other storage media maybe. It seems he made a sharedfolders feature for the Minimig Amiga core. But my plan for now is to ignore Windows I think, because it just seems too unstable... well, unless Cache25 made it more stable, I might try it once more.

That shared_folder option is also for ao486, but, I haven't been able to figure out how to make it work.

[IAmParadox]

I'm sending my MiSTer setup in to have it looked at, by Mister Addons, who I bought it from, so, I will be unable to test anything else for a bit, and, I think I'll just crawl up in a corner and go through withdrawals, until I get it back...


Just kidding, I'll find other things to occupy my time, but, I sure will miss it. At least I have my new ED64 to mess around with.

[Caldor]

I'm sending my MiSTer setup in to have it looked at, by Mister Addons, who I bought it from, so, I will be unable to test anything else for a bit, and, I think I'll just crawl up in a corner and go through withdrawals, until I get it back...

Just kidding, I'll find other things to occupy my time, but, I sure will miss it. At least I have my new ED64 to mess around with.

ED64 looks nice. I got the Sega MD from Analogue. FPGA based. Waiting for them to release their new Pocket model. But yeah, I would not like being without the MiSTer right now. You got it as a set or maybe you could buy an extra DE-10 Nano and use that while you send in the other DE-10 Nano? Its not that expensive in itself and it sure seems like it might be nice to have an extra for some other project. I have thought about putting one in some special casing, like a Commmodore 64 or Amiga casing like I did with a Raspberry Pi.

[Coffea]

ao486_Cache25_061fc53 has a memory selector bug.

In MiSTer menus it presents 16MB or 256MB, 16MB results in 64Mb in Win98, 256 results in a screen full of garbage just after boot.

[Caldor]

ao486_Cache25_061fc53 has a memory selector bug.

In MiSTer menus it presents 16MB or 256MB, 16MB results in 64Mb in Win98, 256 results in a screen full of garbage just after boot.

Yes...Cache24 had the same problem I think, but it sounds like in Cache25 we have a way to get around it by setting it to 16mb then. In the post with the Cache25 link he mentions some fixes done with it. Some drive D bug and some other stuff, he does not list everything. People have been asking for a way to reduce the amount of available memory. Especially Alladin seems to have trouble working with more than a certain amount of memory and then you need to make sure there is less memory available to get the game to run, and there is software that can help do that. EatEMS I think they mentioned in another thread. But it makes sense to be able to set an amount of RAM when its possible to set a CPU speed.

Some really good options we are getting for this core. Pretty incredible, since a PC must be about as complex as a Playstation or Nintendo 64. I guess the reason we can get this far is that PCs might not be as closed as the consoles.

[Chris23235]

Some really good options we are getting for this core. Pretty incredible, since a PC must be about as complex as a Playstation or Nintendo 64. I guess the reason we can get this far is that PCs might not be as closed as the consoles.

Not quite as complex as the PSX or N64, because we are talking about pre-3D-Hardware here, the PSX and N64 had a CPU and a GPU, while the SVGA chip in the PC has a much smaller transistor count then the 3D Hardware in the consoles. At least the N64 CPU was quite powerful back then, maybe not as fast as a Pentium, but my guess is at lease on par with the 486.

Open hardware in the sense as open source where neither the PC nor the consoles.

[IAmParadox]

...
In the post with the Cache25 link he mentions some fixes done with it. Some drive D bug and some other stuff, he does not list everything.
...

https://github.com/MiSTer-devel/ao486_M ... ommits/dev

Also, it's EatXMS.

[Caldor]

Some really good options we are getting for this core. Pretty incredible, since a PC must be about as complex as a Playstation or Nintendo 64. I guess the reason we can get this far is that PCs might not be as closed as the consoles.

Not quite as complex as the PSX or N64, because we are talking about pre-3D-Hardware here, the PSX and N64 had a CPU and a GPU, while the SVGA chip in the PC has a much smaller transistor count then the 3D Hardware in the consoles. At least the N64 CPU was quite powerful back then, maybe not as fast as a Pentium, but my guess is at lease on par with the 486.

Open hardware in the sense as open source where neither the PC nor the consoles.

Yeah, I was not arguing it was open source, just that we know more about these architectures, because they were more used, we have several different CPUs using different versions of pretty much the same technology, so it still seems like there would be more to work with when it comes to reverse engineering it. The AO486 core started out as just an AO186 core a few years ago. It has come a long way since then, but its nice to be able to run software and then just keep adding features that enables more software to run.

But I do agree that the 3D GPU part of the N64 and PS1 must make them more difficult to create FPGA cores for.

...
In the post with the Cache25 link he mentions some fixes done with it. Some drive D bug and some other stuff, he does not list everything.
...

https://github.com/MiSTer-devel/ao486_M ... ommits/dev

Also, it's EatXMS.

Wow... that sounds like that is going to make a big difference 4 times bigger L1 cache. I guess it also means he enabled L1 cache again at some point.
I remembered it as EatEMS, thanks for correcting me.

I made a new video, its pretty long, like the last video, but I added a bunch of time stamps this time. I also did so with the first video. I tested a bunch of VESA games. Its so nice how much the core can now handle this well.
https://www.youtube.com/watch?v=eRB4mPVf8NA

[Caldor]

Seems I am the first to download Cache26

I should be on my way to bed but... just need to give it a quick run.

Update: Have not noticed any differences so far.

Game Wizard does not seem to be working since the VESA update... or might even have been these cache updates overall? Its still not working in this one. The performance seems to be the same in the games I have tried so far and Windows 95. Windows 95 still gets the crash with many colors when I try to install DirectX 8. I have not noticed any changes in the menus.

I have not noticed anything running worse either though.

[DecoySports]

ao486_Cache25_061fc53 AVI video test "Big Buck Bunny" --> https://youtu.be/ZLNP7y1BjwU

Where does one obtain that incredible wallpaper?