Sphere478 wrote on 2021-12-22, 03:32:

So when are these going to become a thing?

Something has to curb these insane voodoo prices 🤣…

Running a Glide wrapper on a Geforce FX 5900 or Radeon 9700 doesn't do it for you ? 😉

darry wrote on 2021-12-22, 04:02:

Sphere478 wrote on 2021-12-22, 03:32:

So when are these going to become a thing?

Something has to curb these insane voodoo prices 🤣…

Running a Glide wrapper on a Geforce FX 5900 or Radeon 9700 doesn't do it for you ? 😉

“Chants” we want real hardware “chants”

Lol😂

darry wrote on 2021-12-22, 04:02:

Sphere478 wrote on 2021-12-22, 03:32:

So when are these going to become a thing?

Something has to curb these insane voodoo prices 🤣…

Running a Glide wrapper on a Geforce FX 5900 or Radeon 9700 doesn't do it for you ? 😉

As long as it’s a PCI version

Sphere478 wrote on 2021-12-22, 04:06:

darry wrote on 2021-12-22, 04:02:

Sphere478 wrote on 2021-12-22, 03:32:

So when are these going to become a thing?

Something has to curb these insane voodoo prices 🤣…

Running a Glide wrapper on a Geforce FX 5900 or Radeon 9700 doesn't do it for you ? 😉

“Chants” we want real hardware “chants”

Lol😂

To paraphrase Pink Floyd :

"We don't need no emulation..." 😉

Sphere478 wrote on 2021-12-22, 03:32:

Something has to curb these insane voodoo prices 🤣…

Considering Misters' finally starting to crack the 32-bit console ceiling with Playstation and Saturn cores (being the peak of the already expensive FPGA they use), and AO486 being nowhere near an FPU, this is highly unlikely.

heck all of the OG SST1 voodoo isn't even emulated completely in software yet by anything that's out there (at the time of this writing).

leileilol wrote on 2021-12-22, 06:37:

Sphere478 wrote on 2021-12-22, 03:32:

Something has to curb these insane voodoo prices 🤣…

Considering Misters' finally starting to crack the 32-bit console ceiling with Playstation and Saturn cores (being the peak of the already expensive FPGA they use), and AO486 being nowhere near an FPU, this is highly unlikely.

heck all of the OG SST1 voodoo isn't even emulated completely in software yet by anything that's out there (at the time of this writing).

I think it's an inevitability given the right timespan, but what that timespan is is still up in the air. I personally hope it's more like 5-10 years than 30, especially with scrappers buying up parts at $30+ in order to "salvage" $0.07' worth of gold. Unfortunately, they're now gone, but Near's Ares emulator has a pretty accurate RCP implementation and that's just the work of a handful of people (mainly Near), so it's not a task outside the bounds of a few mortal humans -- just not a priority when paid jobs take up all your days.

Sphere478 wrote on 2021-12-22, 04:06:

darry wrote on 2021-12-22, 04:02:

Sphere478 wrote on 2021-12-22, 03:32:

So when are these going to become a thing?

Something has to curb these insane voodoo prices 🤣…

Running a Glide wrapper on a Geforce FX 5900 or Radeon 9700 doesn't do it for you ? 😉

“Chants” we want real hardware “chants”

Lol😂

I know it isn't the greatest DX8 chip, but I really want an FPGA replica of a Spectre card to plug into my Athlon 750 build. I only learned about them yesterday, and they're already occupying 50% of my brain's cycles.

lepidotós wrote:

I think it's an inevitability

It isn't, because demand for that is really really low. Glide emulation is more than at acceptable level for a decade now. And emulating Voodoo at near 100% level of authentic experience is counterproductive anyway. Now you want FPGA on top of that very niche thing. Yeah, won't happen anytime soon.

lepidotós wrote on 2022-11-14, 04:33:

I think it's an inevitability given the right timespan, but what that timespan is is still up in the air. I personally hope it's […]
Show full quote

leileilol wrote on 2021-12-22, 06:37:

Sphere478 wrote on 2021-12-22, 03:32:

Something has to curb these insane voodoo prices 🤣…

Considering Misters' finally starting to crack the 32-bit console ceiling with Playstation and Saturn cores (being the peak of the already expensive FPGA they use), and AO486 being nowhere near an FPU, this is highly unlikely.

heck all of the OG SST1 voodoo isn't even emulated completely in software yet by anything that's out there (at the time of this writing).

I think it's an inevitability given the right timespan, but what that timespan is is still up in the air. I personally hope it's more like 5-10 years than 30, especially with scrappers buying up parts at $30+ in order to "salvage" $0.07' worth of gold. Unfortunately, they're now gone, but Near's Ares emulator has a pretty accurate RCP implementation and that's just the work of a handful of people (mainly Near), so it's not a task outside the bounds of a few mortal humans -- just not a priority when paid jobs take up all your days.

Sphere478 wrote on 2021-12-22, 04:06:

darry wrote on 2021-12-22, 04:02:

Running a Glide wrapper on a Geforce FX 5900 or Radeon 9700 doesn't do it for you ? 😉

“Chants” we want real hardware “chants”

Lol😂

I know it isn't the greatest DX8 chip, but I really want an FPGA replica of a Spectre card to plug into my Athlon 750 build. I only learned about them yesterday, and they're already occupying 50% of my brain's cycles.

- AFAIU, the hardware (Rampage chip) was never finished, as in fully achieving production-ready design goals.
- The drivers for it were very raw
- There aren't many of those in existence and they are expensive (I've heard of prices around 11K US$ )
- Setting aside the technical limits of current FPGA tech, anybody willing and able to undertake such a project and successfully take it to completion would end up with the functional equivalent of an unfinished, buggy, slow card that never did anything special compared to the chips/cards that actually made it to market and actually worked .

Who would want such a clone/reproduction (as opposed to the real thing, which is collectible if not really all the useful), except maybe some diehard fans (many would probably only care about the "real thing" because of its rarity, not its usefulness) ?

And how much would such a thing and up costing ?

And most importantly, who among those who actually own an original one would be willing to have its main chip essentially destroyed by decapping so its innards could be analyzed in order to implement a theoretical FPGA reproduction ?

I mean, it's nice to dream, but again setting aside technical limits of current FPGA tech, I'm pretty convinced that practically every other 3DFx chip that actually made it to market and probably quite a few competing ones are more likely to get made than something like this.

Just my 2 cents worth of an opinion.

The Serpent Rider wrote on 2022-11-14, 05:00:

lepidotós wrote:

I think it's an inevitability

It isn't, because demand for that is really really low. Glide emulation is more than at acceptable level for a decade now. And emulating Voodoo at near 100% level of authentic experience is counterproductive anyway.

That too .

It's far more likely that a low(ish)-level Voodoo hardware clone that runs as an emulation layer on something like an ARM SOC interfaced to the PCI bus will come to fruition someday (possibly adapting code from DOSBox-X or PCEM). While not being a 100% authentic Voodoo, if it works with original drivers and give a similar visual experience, I think there might be a market for it , if it can be made cheaply enough.

No idea how feasible or sane that would be .

Ultimate solution would obviously be FPGA VSA100. But all things considered, that's just a Glide accelerator with nicer graphics (better filtering, anti-aliasing, 32-bit color, etc), i.e. things which are easily achieved with Glide wrapping/emulation on much more modern and powerful hardware anyway. Oh, the irony!

The Serpent Rider wrote on 2022-11-14, 05:00:

lepidotós wrote:

I think it's an inevitability

It isn't, because demand for that is really really low. Glide emulation is more than at acceptable level for a decade now.

Glide emulation is very helpful, especially for laptops and I use it plenty, but if there weren't demand for Voodoos, they wouldn't sell for $200+ on eBay. AdLib Gold I'd say is something comparable -- while it's easier to clone since it uses all off-the-shelf components, it could probably be emulated just as well. And yet, there's an AdLib Gold run going on (or that's already finished?). I mean, why shouldn't I stay on this Pentium N3540 and just live in PCem or 86box then?

Also keep in mind the mention of timescales. If the Mac Plus can get new hardware expansions this year, I don't think it's out of the question for a Voodoo clone to arrive at some point, even if it's just using the reclaimed VSA-100s off AliExpress.

darry wrote on 2022-11-14, 05:05:

- AFAIU, the hardware (Rampage chip) was never finished, as in fully achieving production-ready design goals. - The drivers for […]
Show full quote

lepidotós wrote on 2022-11-14, 04:33:

I think it's an inevitability given the right timespan, but what that timespan is is still up in the air. I personally hope it's […]
Show full quote

leileilol wrote on 2021-12-22, 06:37:

Considering Misters' finally starting to crack the 32-bit console ceiling with Playstation and Saturn cores (being the peak of the already expensive FPGA they use), and AO486 being nowhere near an FPU, this is highly unlikely.

heck all of the OG SST1 voodoo isn't even emulated completely in software yet by anything that's out there (at the time of this writing).

I think it's an inevitability given the right timespan, but what that timespan is is still up in the air. I personally hope it's more like 5-10 years than 30, especially with scrappers buying up parts at $30+ in order to "salvage" $0.07' worth of gold. Unfortunately, they're now gone, but Near's Ares emulator has a pretty accurate RCP implementation and that's just the work of a handful of people (mainly Near), so it's not a task outside the bounds of a few mortal humans -- just not a priority when paid jobs take up all your days.

Sphere478 wrote on 2021-12-22, 04:06:

“Chants” we want real hardware “chants”

Lol😂

I know it isn't the greatest DX8 chip, but I really want an FPGA replica of a Spectre card to plug into my Athlon 750 build. I only learned about them yesterday, and they're already occupying 50% of my brain's cycles.

- AFAIU, the hardware (Rampage chip) was never finished, as in fully achieving production-ready design goals.
- The drivers for it were very raw
- There aren't many of those in existence and they are expensive (I've heard of prices around 11K US$ )
- Setting aside the technical limits of current FPGA tech, anybody willing and able to undertake such a project and successfully take it to completion would end up with the functional equivalent of an unfinished, buggy, slow card that never did anything special compared to the chips/cards that actually made it to market and actually worked .

Who would want such a clone/reproduction (as opposed to the real thing, which is collectible if not really all the useful), except maybe some diehard fans (many would probably only care about the "real thing" because of its rarity, not its usefulness) ?

And how much would such a thing and up costing ?

And most importantly, who among those who actually own an original one would be willing to have its main chip essentially destroyed by decapping so its innards could be analyzed in order to implement a theoretical FPGA reproduction ?

I mean, it's nice to dream, but again setting aside technical limits of current FPGA tech, I'm pretty convinced that practically every other 3DFx chip that actually made it to market and probably quite a few competing ones are more likely to get made than something like this.

Just my 2 cents worth of an opinion.

It would cost way too much and do nothing a GeForce3 Ti200 couldn't, but there's a bit of fun to that. Kind of like purposely buying an S3 ViRGE or Diamond Edge3D 2000, or using mechanical hard drives.

darry wrote on 2022-11-14, 05:15:

That too . […]
Show full quote

The Serpent Rider wrote on 2022-11-14, 05:00:

lepidotós wrote:

I think it's an inevitability

It isn't, because demand for that is really really low. Glide emulation is more than at acceptable level for a decade now. And emulating Voodoo at near 100% level of authentic experience is counterproductive anyway.

That too .

It's far more likely that a low(ish)-level Voodoo hardware clone that runs as an emulation layer on something like an ARM SOC interfaced to the PCI bus will come to fruition someday (possibly adapting code from DOSBox-X or PCEM). While not being a 100% authentic Voodoo, if it works with original drivers and give a similar visual experience, I think there might be a market for it , if it can be made cheaply enough.

No idea how feasible or sane that would be .

That would likely be the most sane way to do it, if maybe a little on the boring side. The ARMSID is basically that, but for the MOS SID chips. Connect it to AGP and you could have a one-card solution for both good DirectX/OpenGL and Glide, but seeing as Glide is from what I've read more or less an OpenGL subset (correct me if I'm wrong, I want to know but can't find any information at all about this), I suppose that's somewhat pointless; or, marginally pointed if you're optimistic.

AdLib Gold I'd say is something comparable

It isn't. No reliable way to emulate, at least on period appropriate system. Also it's not FPGA, as you've mentioned.

but if there weren't demand for Voodoos, they wouldn't sell for $200+

Exactly, new batch of Voodoo chips would likely find some audience, but FPGA won't.

lepidotós wrote on 2022-11-14, 05:23:

Glide emulation is very helpful, especially for laptops and I use it plenty, but if there weren't demand for Voodoos, they would […]
Show full quote

The Serpent Rider wrote on 2022-11-14, 05:00:

lepidotós wrote:

I think it's an inevitability

It isn't, because demand for that is really really low. Glide emulation is more than at acceptable level for a decade now.

Glide emulation is very helpful, especially for laptops and I use it plenty, but if there weren't demand for Voodoos, they wouldn't sell for $200+ on eBay. AdLib Gold I'd say is something comparable -- while it's easier to clone since it uses all off-the-shelf components, it could probably be emulated just as well. And yet, there's an AdLib Gold run going on (or that's already finished?). I mean, why shouldn't I stay on this Pentium N3540 and just live in PCem or 86box then?

Also keep in mind the mention of timescales. If the Mac Plus can get new hardware expansions this year, I don't think it's out of the question for a Voodoo clone to arrive at some point, even if it's just using the reclaimed VSA-100s off AliExpress.

darry wrote on 2022-11-14, 05:05:

- AFAIU, the hardware (Rampage chip) was never finished, as in fully achieving production-ready design goals. - The drivers for […]
Show full quote

lepidotós wrote on 2022-11-14, 04:33:

I think it's an inevitability given the right timespan, but what that timespan is is still up in the air. I personally hope it's more like 5-10 years than 30, especially with scrappers buying up parts at $30+ in order to "salvage" $0.07' worth of gold. Unfortunately, they're now gone, but Near's Ares emulator has a pretty accurate RCP implementation and that's just the work of a handful of people (mainly Near), so it's not a task outside the bounds of a few mortal humans -- just not a priority when paid jobs take up all your days.

I know it isn't the greatest DX8 chip, but I really want an FPGA replica of a Spectre card to plug into my Athlon 750 build. I only learned about them yesterday, and they're already occupying 50% of my brain's cycles.

- AFAIU, the hardware (Rampage chip) was never finished, as in fully achieving production-ready design goals.
- The drivers for it were very raw
- There aren't many of those in existence and they are expensive (I've heard of prices around 11K US$ )
- Setting aside the technical limits of current FPGA tech, anybody willing and able to undertake such a project and successfully take it to completion would end up with the functional equivalent of an unfinished, buggy, slow card that never did anything special compared to the chips/cards that actually made it to market and actually worked .

Who would want such a clone/reproduction (as opposed to the real thing, which is collectible if not really all the useful), except maybe some diehard fans (many would probably only care about the "real thing" because of its rarity, not its usefulness) ?

And how much would such a thing and up costing ?

And most importantly, who among those who actually own an original one would be willing to have its main chip essentially destroyed by decapping so its innards could be analyzed in order to implement a theoretical FPGA reproduction ?

I mean, it's nice to dream, but again setting aside technical limits of current FPGA tech, I'm pretty convinced that practically every other 3DFx chip that actually made it to market and probably quite a few competing ones are more likely to get made than something like this.

Just my 2 cents worth of an opinion.

It would cost way too much and do nothing a GeForce3 Ti200 couldn't, but there's a bit of fun to that. Kind of like purposely buying an S3 ViRGE or Diamond Edge3D 2000, or using mechanical hard drives.

darry wrote on 2022-11-14, 05:15:

That too . […]
Show full quote

The Serpent Rider wrote on 2022-11-14, 05:00:

It isn't, because demand for that is really really low. Glide emulation is more than at acceptable level for a decade now. And emulating Voodoo at near 100% level of authentic experience is counterproductive anyway.

That too .

It's far more likely that a low(ish)-level Voodoo hardware clone that runs as an emulation layer on something like an ARM SOC interfaced to the PCI bus will come to fruition someday (possibly adapting code from DOSBox-X or PCEM). While not being a 100% authentic Voodoo, if it works with original drivers and give a similar visual experience, I think there might be a market for it , if it can be made cheaply enough.

No idea how feasible or sane that would be .

That would likely be the most sane way to do it, if maybe a little on the boring side. The ARMSID is basically that, but for the MOS SID chips. Connect it to AGP and you could have a one-card solution for both good DirectX/OpenGL and Glide, but seeing as Glide is from what I've read more or less an OpenGL subset (correct me if I'm wrong, I want to know but can't find any information at all about this), I suppose that's somewhat pointless; or, marginally pointed if you're optimistic.

S3 ViRGE and Diamond Edge3D 2000 are actually useful and usable through their respective APIs and bring something unique to the table . Virge chips are also good at DOS 2D (VGA/SVGA) gaming .

The Spectre gets 9FPS at 800x600 in Unreal Tournament ( https://hothardware.com/news/3dfx-rampage-gpu … -20-years-later ) and doesn't do anything unique on the Glide front (previous 3Dfx chips have that covered) and other contemporary Nvidia and ATI products are better in Direct3D and likely OpenGL .

So what is special about the Spectre other than its rarity ?

Let me put this another way. If there were plenty of Spectre cards available at VSA100 based card prices, would you still really want one (EDIT: to actual use, not solely as a collectible) or would you prefer to have an actually fully functional VSA100 based card ?

The Serpent Rider wrote on 2022-11-14, 05:31:

AdLib Gold I'd say is something comparable

It isn't. No reliable way to emulate, at least on period appropriate system. Also it's not FPGA, as you've mentioned.

That and there is actually that supports the Adlib Gold in particular. The same can't be said about Spectre, unless Spectre happens to bring something extra to the table on the Glide front compared to its predecessors, which I doubt .

The Serpent Rider wrote on 2022-11-14, 05:31:

AdLib Gold I'd say is something comparable

It isn't. No reliable way to emulate, at least on period appropriate system. Also it's not FPGA, as you've mentioned.

Is Glide emulation fast enough on, say, a GeForce2 GTS that it compares to a V4? I suppose it could be. I got fairly decent performance (60+ fps, 320x200) out of nGlide on a Mobility Radeon R6. I've heard that the image isn't quite as nice, mentioning colors not being quite as vibrant with a wrapper instead of a real card, but I can't speak to that as someone that doesn't own one. Mainly because even broken (but not totally dead) V3s are often listed for $70 or more these days.

The Serpent Rider wrote on 2022-11-14, 05:31:

but if there weren't demand for Voodoos, they wouldn't sell for $200+

Exactly, new batch of Voodoo chips would likely find some audience, but FPGA won't.

I mean, a new batch of Voodoo chips would probably have to be made (er, flashed) on FPGAs, at least for the first run to get out to testers considering I don't think that Big Green is going to give up any vhdls they may or may not still have, and I haven't heard of them being leaked. Spinning up a chip fab before you've gotten any feedback on accuracy isn't really the most cost effective idea, and while FPGAs aren't exactly cheap, asking TSMC or GloFo to revive their 250 node to make a bunch of Voodoo3s before you've even done any testing to be 100% sure it's identical enough to satisfy the die-hard purists like myself doesn't sound particularly economical in comparison.

darry wrote on 2022-11-14, 05:48:

S3 ViRGE and Diamond Edge3D 2000 are actually useful and usable through their respective APIs and bring something unique to the […]
Show full quote

lepidotós wrote on 2022-11-14, 05:23:

Glide emulation is very helpful, especially for laptops and I use it plenty, but if there weren't demand for Voodoos, they would […]
Show full quote

The Serpent Rider wrote on 2022-11-14, 05:00:

It isn't, because demand for that is really really low. Glide emulation is more than at acceptable level for a decade now.

Glide emulation is very helpful, especially for laptops and I use it plenty, but if there weren't demand for Voodoos, they wouldn't sell for $200+ on eBay. AdLib Gold I'd say is something comparable -- while it's easier to clone since it uses all off-the-shelf components, it could probably be emulated just as well. And yet, there's an AdLib Gold run going on (or that's already finished?). I mean, why shouldn't I stay on this Pentium N3540 and just live in PCem or 86box then?

Also keep in mind the mention of timescales. If the Mac Plus can get new hardware expansions this year, I don't think it's out of the question for a Voodoo clone to arrive at some point, even if it's just using the reclaimed VSA-100s off AliExpress.

darry wrote on 2022-11-14, 05:05:

- AFAIU, the hardware (Rampage chip) was never finished, as in fully achieving production-ready design goals. - The drivers for […]
Show full quote

- AFAIU, the hardware (Rampage chip) was never finished, as in fully achieving production-ready design goals.
- The drivers for it were very raw
- There aren't many of those in existence and they are expensive (I've heard of prices around 11K US$ )
- Setting aside the technical limits of current FPGA tech, anybody willing and able to undertake such a project and successfully take it to completion would end up with the functional equivalent of an unfinished, buggy, slow card that never did anything special compared to the chips/cards that actually made it to market and actually worked .

Who would want such a clone/reproduction (as opposed to the real thing, which is collectible if not really all the useful), except maybe some diehard fans (many would probably only care about the "real thing" because of its rarity, not its usefulness) ?

And how much would such a thing and up costing ?

And most importantly, who among those who actually own an original one would be willing to have its main chip essentially destroyed by decapping so its innards could be analyzed in order to implement a theoretical FPGA reproduction ?

I mean, it's nice to dream, but again setting aside technical limits of current FPGA tech, I'm pretty convinced that practically every other 3DFx chip that actually made it to market and probably quite a few competing ones are more likely to get made than something like this.

Just my 2 cents worth of an opinion.

It would cost way too much and do nothing a GeForce3 Ti200 couldn't, but there's a bit of fun to that. Kind of like purposely buying an S3 ViRGE or Diamond Edge3D 2000, or using mechanical hard drives.

darry wrote on 2022-11-14, 05:15:

That too .

It's far more likely that a low(ish)-level Voodoo hardware clone that runs as an emulation layer on something like an ARM SOC interfaced to the PCI bus will come to fruition someday (possibly adapting code from DOSBox-X or PCEM). While not being a 100% authentic Voodoo, if it works with original drivers and give a similar visual experience, I think there might be a market for it , if it can be made cheaply enough.

No idea how feasible or sane that would be .

That would likely be the most sane way to do it, if maybe a little on the boring side. The ARMSID is basically that, but for the MOS SID chips. Connect it to AGP and you could have a one-card solution for both good DirectX/OpenGL and Glide, but seeing as Glide is from what I've read more or less an OpenGL subset (correct me if I'm wrong, I want to know but can't find any information at all about this), I suppose that's somewhat pointless; or, marginally pointed if you're optimistic.

S3 ViRGE and Diamond Edge3D 2000 are actually useful and usable through their respective APIs and bring something unique to the table . Virge chips are also good at DOS 2D (VGA/SVGA) gaming .

The Spectre gets 9FPS at 800x600 in Unreal Tournament ( https://hothardware.com/news/3dfx-rampage-gpu … -20-years-later ) and doesn't do anything unique on the Glide front (previous 3Dfx chips have that covered) and other contemporary Nvidia and ATI products are better in Direct3D and likely OpenGL .

So what is special about the Spectre other than its rarity ?

Let me put this another way. If there were plenty of Spectre cards available at VSA100 based card prices, would you still really want one (EDIT: to actual use, not solely as a collectible) or would you prefer to have an actually fully functional VSA100 based card ?

Yes in a hypothetical world where functional drivers get hashed out for it, mainly out of base, uncivilized tribalism on faster systems (e.g. an 1100MHz+ Athlon or earlier Athlon XP) if one of the cards ever does die. I just have broad positive feelings towards 3dfx and a general apathy towards NVidia (and broad antipathy towards present-day NVidia). VSA100 as far as I've heard basically already looks like a DX6 card with a wrapper and doesn't perform much better than V3 with less compatibility, so a V2 or V3 clone would be the better pick for having some reason to exist over VSA100, and a Spectre clone with half-decent drivers would be able to show what a really next-gen 3dfx card would be like, and kind of be a ghost of a fanfiction world where they're still around in some form to the present day offering something that isn't a 450w, $1,000 card that has 8K 960fps as its target. I wouldn't want one to be cut up if it wasn't already totally dead, though. It's kind of like the people fixing up Mac OS X 10.6 10a96 to work on PowerPC Macs.

As you (and I) have said, the drivers weren't on the fire for very long, so while it's about as concrete as we'll ever get for official performance with official drivers, I don't put all of my faith in them, looking at the specs I'd say it would have likely been somewhere around the performance of a Ti200 if it released.

I bought a 64DD about a year ago now with the stimulus money because I liked its theoretical functionality that adds on to a console I already love and wish it was less rare, I don't think rarity is what draws me towards Spectre. I'm certainly not celebrating the scrapping trend going around that makes parts even rarer.