wait a few weeks.
and for 2 games I don't see that much use for adding a video card.

Well, there's more than 2 games that actually use this video card tbh. So far, I've only found 5, but I'm sure more exist.

Hi shogun, how are you? I thought you were working on xbox 1 emulation 😀
A pci framework is already part of svn, you should be able to add a class that implements your device
about mmio access, have a look at the 3dfx patch, in particular the changes to memory.cpp and the voodoo page handler class in voodoo_main.
I guess qbix has a point, about nv1 software support, probably another hardware would be a better choice, like powervr.
there should be a console emulator around that already implements povervr emulation and its developer even posted on this forum in the 3dfx thread.
Anyway, probably 3dfx is enough for most of the 3d acceleration needs in dosbox; you might concentrate on emulating other features.

Oh, hey kekko! Heh, wasn't sure if you were the same one from ngemu or what, 🤣. While I'm here, I'd like to say thanks for your hard work on emulating SST-1 (3Dfx Voodoo). I am still working on Xbox1, but honestly, I need a break from that from time to time, otherwise I can't work on it efficiently! So I do keep a few other projects handy for when I need to deviate a bit.

Maybe I should have been a bit more direct, but I was kinda hinting at some sort of way to add support for multiple 3D accelerators and let the user choose which one they want to use. But thinking a bit more about what DOSBox's purpose is, that sort of thing might be more suitable for Bochs or QEMU. Unfortunately for me, I find Bochs and QEMU too confusing to work with as it's source is advanced and complex.

But anyway, I'm not saying qbix doesn't have a valid point (because it very well may not be worth it), but let's just say that it's no longer business... it's personal! 3Dfx may be my passion, but NVIDIA is my specialty (primarily NV3 (Riva128) and NV2A (Xbox)), so I thought I'd do the old school SEGA fans a favour. Adding to the fact that not a single NVIDIA GPU has ever been emulated (except the NV2A, *very* partially), it's driving me nuts that my copy of Panzer Dragoon 1 doesn't work without NV1 so it's more motivation right there as I like challenges that no one has ever done or succeeded at. As for emulating PowerVR, I have absolutely no experience with that GPU as it never interested me in the past (which might change). Not saying that I can't do it, but what games actually utilize PowerVR besides Quake II?

Other than that, I don't see any other ways I could really contribute to DOSBox. I was also hoping that adding NV1 support could also help me learn more about it too.

blueshogun96 wrote:

Not saying that I can't do it, but what games actually utilize PowerVR besides Quake II?

A hell of a lot more than NV1, that's for sure. Almost every UnrealEngine1 game for example... I have a Youtube channel with video captures of PowerVR PCX2 running games it's intended to, and NOT intended to run. See the PowerVR Fun Thread in my sig for..... well PowerVR Fun.

By the way, in a SEGA slant, there's a Virtual On Cybertroopers release that uses PowerVR but I never had this myself.

For PowerVR PCX2 emulation, starting off with nullDC PVR emulation code's a good start, because that's emulating an evolution of the chipset (Midas). You'd have to work your way BACKWARDS.

Heh, guess I haven't been paying enough attention to PowerVR all these years.... Now I'm starting to get curious as to how many games used exclusively, and how complex the hardware might be. I can't think of a single card from the 90s that's more complicated than NV1 and NV3 (Riva128).

I won't say it isn't worth emulating because I like to see everything/anything emulated just for the sake of emulating it. Sorry if I sound really stubborn with this topic, but even though there were only a few games using NV1, certain games can't be run without it (my copy of Panzer Dragoon has no software fallback). While 3Dfx and PowerVR emulation is thriving, NVIDIA is completely neglected 🙁 Therefore I'd still like to at least try. I'll still look into PowerVR just for fun, but I can't promise I'll implement it, assuming I can wrap my head around how it works or if there's enough documentation and other resources needed to emulate it.

blueshogun96 wrote:

Heh, guess I haven't been paying enough attention to PowerVR all these years.... Now I'm starting to get curious as to how many games used exclusively, and how complex the hardware might be. I can't think of a single card from the 90s that's more complicated than NV1 and NV3 (Riva128).

I won't say it isn't worth emulating because I like to see everything/anything emulated just for the sake of emulating it. Sorry if I sound really stubborn with this topic, but even though there were only a few games using NV1, certain games can't be run without it (my copy of Panzer Dragoon has no software fallback). While 3Dfx and PowerVR emulation is thriving, NVIDIA is completely neglected 🙁 Therefore I'd still like to at least try. I'll still look into PowerVR just for fun, but I can't promise I'll implement it, assuming I can wrap my head around how it works or if there's enough documentation and other resources needed to emulate it.

nVidia isn't being neglected. Apart from NV1, nVidia never had any proprietary API's to emulate. After NV1 they made DX and OpenGL compliant cards so if your VM can emulate those, then it can emulate any nVidia card made from NV3 onwards. Rendition, Videologic, and 3dfx all had their own proprietary 3D standards which is why they need emulation. So few games were ever made for NV1 that the value of emulating them is dubious at best as most people won't be able to get copies of those games, not legal ones anyway.

sliderider wrote:

the value of emulating [NV1] is dubious at best

For you, perhaps, but definitely not for blueshogun96 who wants to scratch his own itch.

Feel free to write an OpenGL pass through or a D3D wrapper, it will be far more useful than criticizing other people's initiatives.

Zorbid wrote:

sliderider wrote:

the value of emulating [NV1] is dubious at best

For you, perhaps, but definitely not for blueshogun96 who wants to scratch his own itch.

Feel free to write an OpenGL pass through or a D3D wrapper, it will be far more useful than criticizing other people's initiatives.

I honestly don't see myself being able to complete a wrapper any time soon (like I'd prefer to do) since I'm not very skilled in reverse engineering nor do I have the time to learn at this point. It may be trivial for the rest of you, but not for me. For some of you, emulation is about what you can play or use, for others, emulation is about the thrill of getting something working and the preservation of a relic. I choose the latter, and that won't change anytime soon. Although kekko may be right. I should probably keep up with my higher priority emulation projects and do this in my spare time rather than get too fixated on it.

Zorbid wrote:

sliderider wrote:

the value of emulating [NV1] is dubious at best

For you, perhaps, but definitely not for blueshogun96 who wants to scratch his own itch.

Feel free to write an OpenGL pass through or a D3D wrapper, it will be far more useful than criticizing other people's initiatives.

I'm not criticizing anything. I'm just pointing out the fact that there is little value in an NV1 wrapper as there was only ever a handful of games that worked with it and not very many copies of those games are available anymore. Using Glide as an example to make my point, a Glide wrapper only has value because there are a lot of games that were written to work with it. If there had only been a few games that worked with Glide, then a Glide wrapper would also be of little value and not worth the time it's going to take to write and debug it. The value of an emulator or API wrapper increases or decreases with the number of games that had support for that API. There was a grand total of 4 games that were playable on an NV1 card and there's maybe a few dozen people who still own legal copies of those games and they all own NV1 cards anyway, so it's not going to generate enough interest to be worth the trouble of doing. The only kind of interest it might generate is among warez kiddies who would then download the games from "abandonware" sites when they can't get legal copies and that's not the kind of interest we need in this hobby.

blueshogun96 wrote:

Therefore I'd still like to at least try. I'll still look into PowerVR just for fun, but I can't promise I'll implement it, assuming I can wrap my head around how it works or if there's enough documentation and other resources needed to emulate it.

For starters, on the Matrox M3D cd there's a SGL.lib, SGL.h, and some other example source code that interfaces with SGL. Images of the CD should be in the PowerVR fun thread

There's a newer SGL.h found in the Deus Ex SDK (WOW! CRAP PAST THE RADAR)

Since this topic is about PCI emulation, did you check out the svn sources? If you need help getting a device added just post here.

sliderider wrote:

Zorbid wrote:

sliderider wrote:

the value of emulating [NV1] is dubious at best

For you, perhaps, but definitely not for blueshogun96 who wants to scratch his own itch.

Feel free to write an OpenGL pass through or a D3D wrapper, it will be far more useful than criticizing other people's initiatives.

I'm not criticizing anything. I'm just pointing out the fact that there is little value in an NV1 wrapper as there was only ever a handful of games that worked with it and not very many copies of those games are available anymore. Using Glide as an example to make my point, a Glide wrapper only has value because there are a lot of games that were written to work with it. If there had only been a few games that worked with Glide, then a Glide wrapper would also be of little value and not worth the time it's going to take to write and debug it. The value of an emulator or API wrapper increases or decreases with the number of games that had support for that API. There was a grand total of 4 games that were playable on an NV1 card and there's maybe a few dozen people who still own legal copies of those games and they all own NV1 cards anyway, so it's not going to generate enough interest to be worth the trouble of doing.

For starters, there were more than 4 games that use it, if you count the games that were written for DOS and not created by Sega. Second, it's pretty hard to assume that anyone who has an NV1 based game actually has an NV1 to begin with. Like I said, I have NV1 compatible games, but can't find an NV1 to save my life! I'd like to finally be able to play my copy of Panzer Dragoon on my PC, and having bilinear filtered textures and maybe some better texture perspective correction is just a bonus.

Btw, not trying to be rude or anything, but did you not read the last thing I said?? It would fully explain why I still want to pursue this even if no one else really cares.

"For some of you, emulation is about what you can play or use, for others, emulation is about the thrill of getting something working and the preservation of a relic."

I emulate hardware that interests me, NOT everyone else, hence the reason I've poured so much time into Xbox1.

The only kind of interest it might generate is among warez kiddies who would then download the games from "abandonware" sites when they can't get legal copies and that's not the kind of interest we need in this hobby.

This is a valid point/hypothesis here. I'm against warez 100%, but the games aren't that easy to find on pirate websites. I had an easier time finding legal copies of these games...

wd wrote:

Since this topic is about PCI emulation, did you check out the svn sources? If you need help getting a device added just post here.

Sorry for the late reply, but yes. I downloaded the latest SVN sources (last week, iirc) and started adding PCI device registers. I took a look at kekko's code for adding the SST-1 support, but the PCI emulation code is very different. So I'm not sure if I've been doing it right so far. My biggest question at the moment is why Windows says that the PCI bus is disabled even when the support is there. Was bit 31 not set in the PCI host or something?

I haven't done a whole lot on it yet because I'm also working on another emulator and stuff like that, but for now, the main thing is to get the device recognized. Unlike most video cards, the NV1 has more than one function, but it might be ok to only implement one, but I'm not sure. Fortunately for me, the MPU-401 device is already emulated, and that's what NV1 uses for audio. Either way, I wouldn't worry about that.

I'll post more information as it surfaces...

Don't mind sliderider, he is known for his special take on things. Arguing with him will only derail your thread and get nowhere 😉

Are the NV1 games NV1 only or were there special versions of the games tailored to the NV1 (similar to other games that were bundled with video cards back in the days - I know I had a Mechwarrior (1or2) special version bundled with my Matrox Mystique back then. Would only work with a Mystique)?

My biggest question at the moment is why Windows says that the PCI bus is disabled even when the support is there. Was bit 31 not set in the PCI host or something?

You won't need the host and bus devices.

Dominus wrote:

Don't mind sliderider, he is known for his special take on things. Arguing with him will only derail your thread and get nowhere 😉

Are the NV1 games NV1 only or were there special versions of the games tailored to the NV1 (similar to other games that were bundled with video cards back in the days - I know I had a Mechwarrior (1or2) special version bundled with my Matrox Mystique back then. Would only work with a Mystique)?

Nah, I'm not mad at him, but I do get tired of people telling me to emulate whats "useful" to them. And quite honestly, the whole risk of attracting warez monkeys comes with any emulator ever written.

Since I only have 2 NV1 compatible games to test atm, so far, only one of them absolutely requires the NV1 to operate, and that would be Panzer Dragoon; Battle Arena Toshinden doesn't require it. I assumed that this game would have a software renderer as a fall back, but I was wrong. So honestly, I won't assume that the other games also have software fallbacks too.

Don't listen to the naysayers. I, for one, am looking forward to the ability to play Panzer Dragoon with the best image quality.

NV1 was one of those chips that didn't do bilinear filtering so the games look a lot like CPU rendering but with some extras.

Games
3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)

Most are OEM releases and really hard to find.

thanks for the list 😀