VOGONS


Reply 21 of 57, by BSA Starfire

User metadata
Rank Oldbie
Rank
Oldbie

I whole heartedly agree with that, Putas has spent a huge amount of time and effort already establishing the relative performance and abilities of these old cards, you can find pretty much anything you want to know here: https://vintage3d.org/index.php

My thread was just originally about exploring my particular machine with the hardware it already had installed and what it might, or might not manage to achieve, that kinda expanded to the two other 6326 cards I could easily put my hands on. And I'll probably take a look at the Laguna 3D 5464 PCI card I also own simply because I've had it for years and never really done anything with it beyond a test when I first received it.
I'll make a few comments here on the 6326 from my experiences this week anyhow, first up, revision matters, you want a post C3 revision at least, H0 is probably the easiest to find and best, it fixes most of the bugs that spoil the early cards(the revision is marked on the chips, bottom right hand corner) . 8mb does offer some advantages over 4MB on these also, so is preferable.
For what was probably the cheapest 3D card on the market at the time(these apparently retailed for less than ViRGE cards even) the one thing they do offer is a fairly full feature set, decent enough drivers and good compatibility with games of the era, what they don't offer is blazing speed or easy to get working openGL. It will allow you to play games without visual anomalies and at a (what I consider) playable speed that things like the ViRGE, Trio3D, 3D/2X, pre G200 Matrox cards, Trident 3Dimage, pre RAGE PRO Ati and Laguna3D won't(but as always this varies with some games favouring or at least not hating some cards, so on occasion a S3 3D/2X might be better, well maybe!). The visual quality is actually pretty decent and also very consistent,that's something few of the other mentioned cards can manage. But this is a dirt cheap low end accelerator and you can't expect performance from it like a Voodoo or RIVA128(that cost 6 times as much). Windows 2D GUI acceleration I'd describe as adequate, at least on my system, but I doubt it'd seem that special on a much faster machine. DOS stuff works fine, no scrolling bug on Commander Keen for example, it's not especially fast again, but then not the slowest I've tried for DOS either.
So far on the H0 revision cards I haven't found anything that is just plain broken, it all works. The C3 is a different story though and does have a lot of issues, similar to the virge and trident 3Dimage, although not as bad as either. I'll add in here, I'm not a first person shooter fan(they are just not my cup of tea) so the only game of this genre I tried was Unreal Gold(the only FPS I own,actually I don't mind this game too much, not a favourite but I did like the sense of exploration and discovery), no tests have been done by me with quake games or related, it's a case of if I'm not interested in playing it I don't care how it runs.
Well I guess that is about all I have to say on this. It's been a good lockdown project and I've learned a fair bit and perhaps walked into the woods off the beaten path too, something this old hardware will let you do and makes it special and rewarding.
I'm sure your all waiting with baited breath to see the awesome power of the Cirrus Logic Laguna 3D with it's insane Rambus memory so I guess that'll be what I do next.

286 20MHz,1MB RAM,Trident 8900B 1MB, Conner CFA-170A.SB 1350B
386SX 33MHz,ULSI 387,4MB Ram,OAK OTI077 1MB. Seagate ST1144A, MS WSS audio
Amstrad PC 9486i, DX/2 66, 16 MB RAM, Cirrus SVGA,Win 95,SB 16
Cyrix MII 333,128MB,SiS 6326 H0 rev,ESS 1869,Win ME

Reply 22 of 57, by BSA Starfire

User metadata
Rank Oldbie
Rank
Oldbie
laguna.JPG
Filename
laguna.JPG
File size
1.1 MiB
Views
792 views
File license
Fair use/fair dealing exception

Here's the Cirrus Logic 5464 Laguna 3D! Wow, what a beast! RDRAM for the win! I believe this is a 4MB card? that's what newest HWINFO thinks too anyway .
OK well lets be honest here, I'm really not expecting much of this card even compared to the SiS 6326, I'm expecting an experience more akin to the Trident 3D Image and later ViRGE and TRIO 3D(perhaps I'll have to give both those another look in the future) cards if I'm honest, well, lets see.
Card is installed and working with the ver 2.0 drivers obtained from vintage3D ,created in Jan 2000, same Cyrix MII machine we have been using so far.
Well I guess I badly underestimated this little card, it's running Star Wars episode 1 racer, TBH it looks nowhere near as nice as the SiS 6326 H0 rev does, but it plays, only a few obvious graphical glitches in the energy stream between the engines and the suns on Tatooine, but frankly they are not bad, just showing a weird green texture rather than the correct stuff. The FPS isn't great, but it is actually playable on this poor lowly machine, menu's are all fine and readable. The surprise is that this is better than the C3 revision 6326 with SGRAM & AGP that is built into the motherboard, that was an unplayable mess, this you can play and enjoy.
Daytona USA also plays great, no glitches or anomalies. It's slower than the 6326 Ho but unlike the C3 has no issues with drawing the sky. I mean really it's not running badly at all.
Final Reality gives a score of 1.47(the lowest score we have seen so far), however all looks fine and the audio glitches that occur with the 6326 while still present are more short lived and intermittent unlike those when the 6326 is installed, although they do happen earlier in the test, during the rotating starship scene in this case rather than at end in city scene.
Motoracer demo is a bit of a mess, start of the race stutters and stalls, then when it gets going although frames pick up lovely and fast it does drop a lot of textures, randomly it seems, big white areas on cliffs and buildings,also especially the railings and fences in the later part of the race, certainly not playable.
Need for Speed III also runs, no glitches at all in the menu's and is playable as far as FPS goes too on the same settings that I used on the 6326 cards. There are a couple of graphical oddities, a thin black strip along each side of the road markings not present on the SiS and also a black cube around the cursor in the pause menu but it's perfectly playable and again far faster and nicer than software rendering on this game.
Unreal gold fails to start, I'm pretty sure this is due to too little video RAM being available on the card after reading vintage3D. I'm getting the same error Putas does with 2MB cards. Was a bit much to expect anyway to be honest!
Ziff davis 3D winbench 98 has no problems with the HUD text like the C3 6326 did but does drop a lot of textures, epsecially in the shrine. I assume this is due to the small amount of video memory, the space station scenes however run pretty nicely.
That's about it for now.

286 20MHz,1MB RAM,Trident 8900B 1MB, Conner CFA-170A.SB 1350B
386SX 33MHz,ULSI 387,4MB Ram,OAK OTI077 1MB. Seagate ST1144A, MS WSS audio
Amstrad PC 9486i, DX/2 66, 16 MB RAM, Cirrus SVGA,Win 95,SB 16
Cyrix MII 333,128MB,SiS 6326 H0 rev,ESS 1869,Win ME

Reply 23 of 57, by matze79

User metadata
Rank l33t
Rank
l33t

Grab SiS 305 Card!

The Laguna is very bad.. i also own it, there is almost nothing running as it should. Expect a lot of drawing errors 😁

https://dosreloaded.de - The German Retro DOS PC Community
https://www.retroianer.de - under constructing since ever

Co2 - for a endless Summer

Reply 24 of 57, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t

Laguna 3D is different. Cirrus Logic 5464 was horrible, but Cirrus Logic 5465 is more or less fine as first gen accelerator. I don't see a reason to torture yourself with SIS 6326 though, no matter which revision, it's too slow.

Get up, come on get down with the sickness
Open up your hate, and let it flow into me

Reply 26 of 57, by BSA Starfire

User metadata
Rank Oldbie
Rank
Oldbie
matze79 wrote on 2020-05-17, 08:07:

Grab SiS 305 Card!

The Laguna is very bad.. i also own it, there is almost nothing running as it should. Expect a lot of drawing errors 😁

I do have a SIS 315 64MB AGP card, not sure the difference with 305, but it's pretty decent, as I recall from testing years ago about the same as a geforce 2 mx. If I ever come across a PCI 305 I'll give it ago.

Laguna 3D is different. Cirrus Logic 5464 was horrible, but Cirrus Logic 5465 is more or less fine as first gen accelerator. I don't see a reason to torture yourself with SIS 6326 though, no matter which revision, it's too slow.

I had very low expectations, so this was purely for my own amusement and learning, I have other far more capable retro machines for games, basically this is a lockdown project off the cuff with what I had set up at the time. BUT I still think a few of the games I have tried are perfectly enjoyable on the 6326(H0). Can't comment on the Laguna 5465 as I don't own one to try out.

286 20MHz,1MB RAM,Trident 8900B 1MB, Conner CFA-170A.SB 1350B
386SX 33MHz,ULSI 387,4MB Ram,OAK OTI077 1MB. Seagate ST1144A, MS WSS audio
Amstrad PC 9486i, DX/2 66, 16 MB RAM, Cirrus SVGA,Win 95,SB 16
Cyrix MII 333,128MB,SiS 6326 H0 rev,ESS 1869,Win ME

Reply 29 of 57, by zapbuzz

User metadata
Rank Oldbie
Rank
Oldbie

I had a SiS 2mb pci gpu. I found out its BIOS was updated and a newer bios loadable by TSR in dos on bootup.
What I did was interesting. The actual BIOS TSR was just a bios attached to a terminate stay resident executable and I didn't have anything to re-program the GPU bios. Didn't like the idea of TSR as it used resources. In the end after extensive research I discovered I could remove the BIOS chip and load the updated TSR bios (without the TSR EXE) into the motherboard BIOS and within the system BIOS it booted the card flawlessly.
Many PCI GPU's could do the same thing. Especially useful for EPROM chips why the expense of programmers when the PC's BIOS might have room for a VGA bios as well. (Uness thats an EPROM chip too)
This information requires research into BIOS modding and isn't as easy as it sounds.
All commercial BIOS have locks, encryption , and special versions of programs to build a modified BIOS from a vendors BIOS version.
The newer BIOS however gave a little more frame speed a marginal improvement but the GPU was running cooler without a fan.
Maybe the SiS 6326 BIOS chip is a flash chip I don't know but that'd mean another programmer anyhow and for 1 card why not just use the system BIOS. For example Award modular is modular because the systems I/O is installed into the Flash RAM or EPROM as option ROMs. VGA BIOS can be recorded as an option ROM; in what is called BIOS strings.
I wouldn't load a PCI BIOS into a motherboard who has an AGP slot that would need an AGP card and bios to work.
Also not for motherboard with integrated GPU.(Although you might find a BIOS ROM for that) Some people have had success loading it as being together with the integrated but I wouldn't.
Still, I remembered I wished I had the AGP version of the SiS 6326 😀

Reply 30 of 57, by Takedasun

User metadata
Rank Newbie
Rank
Newbie
Putas wrote on 2020-05-10, 16:48:

I am glad somebody else deep dives into the often shamed 6326, congrats!

"Since the beta driver release code named Java with OpenGL ICD was buried in the depths of the internet"
https://vintage3d.org/sis.php

-------------------------------------------------------------
SiS 6326 Driver Installation Guide Rev. Java (07/02/1999)
-------------------------------------------------------------

*** Driver Installation Procedure for Windows 95/98 ***

1. Find the file "Setup" after decompressing file "326win9X".
2. Run the program "Setup" by double clicking on it in Windows.
3. Follow up indications to proceed the installation procedure.
4. The driver installation will be completed after system rebooting.

*** Release notes ***

1. DirectX supporting
Driver Java supports DirectX6 and DirectX previous verions. However,
there is not any DirectX driver included in SiS6326 Java. User who needs
to run DirectX compatible softwares may need to install DirectX driver
besides SiS6326 driver. Most software packages include DirectX driver
for users. If it is not included, please download from Microsoft web-site.

2. Language supporting
Driver Java supports English and Chinese. Other language supporting
is provided by the patch file "Language". After decompressing
the file, a readme file inside was prepared to guide you to setup the
multi-language supporting.

3. Game performance
Driver Java has been fine tuned for enhancing game performance.
Some 3D games such as MotoRacer2, Need for Speed3 that use heavy
textures may need this high performance driver to run. It is suggested
to apply this version for game players.

4. OpenGL supporting
Driver Java supports ICD OpenGL. OpenGL games such as Quake2 can run
with it. Please apply "Default OpenGL" or "Standard OpenGL" if a game
provides options for OpenGL setting.

gl_6326.png
Filename
gl_6326.png
File size
160.76 KiB
Views
634 views
File license
Public domain

Attachments

Reply 31 of 57, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t
Putas wrote on 2020-05-21, 10:39:

No, 6326AGP should do better than 5465 more often than not.

Laguna 3D is significantly faster without bilinear filtering. Frame rate is boosted by 50-60%, which is not achievable on SIS 6326AGP. Heck, it scores around 37 fps in Forsaken 640x480, which is a wet dream for any 6326 card, and even Voodoo Rush is slower. So CL-5465 is quite decent accelerator without filtering.

Get up, come on get down with the sickness
Open up your hate, and let it flow into me

Reply 32 of 57, by BSA Starfire

User metadata
Rank Oldbie
Rank
Oldbie

Wow!

I guess I'm a little slow on what's been going on, but just watched vlaskcz's excellent video on youtube covering the 6326 as part of his series of early 3D cards.

https://www.youtube.com/watch?v=OB9VVTJ48gE&t=924s

It's really neat that another driver has been rediscovered recently for the card from Aopen that supports OpenGL . Judging by the video it seems to work very well. I'll have to dig out the old Cyrix box and give her a try with these rediscovered drivers.

http://www.vogonsdrivers.com/getfile.php?file … 788&menustate=0

Watching vlaskcz video apart from a few complete fails, expendable, aliens vs predator, and some missing fog in a few other titles, it seems that this really was the best budget card , it ran the lot!

I will update the thread when I have tested this "new" driver and perhaps some new games, then we can see how it would have been on a real cheap ass computer of the time!

286 20MHz,1MB RAM,Trident 8900B 1MB, Conner CFA-170A.SB 1350B
386SX 33MHz,ULSI 387,4MB Ram,OAK OTI077 1MB. Seagate ST1144A, MS WSS audio
Amstrad PC 9486i, DX/2 66, 16 MB RAM, Cirrus SVGA,Win 95,SB 16
Cyrix MII 333,128MB,SiS 6326 H0 rev,ESS 1869,Win ME

Reply 33 of 57, by HangarAte2nds!

User metadata
Rank Newbie
Rank
Newbie

I just got an AGP 6326 today. It is a late production model and appears to have 16MB onboard. There can be issues with these cards in DOS apparently but a later driver release fixed most of the issues. I also suspect that when I get to testing it, it will perform well. If they addressed the issues in DOS with later drivers, it stands to reason the later BIOS may just utilize the drivers better also.
I just ordered a Rage 128 Pro 16MB AGP and I have a TNT2 16MB AGP somewhere if I can find it. I am planning to start benchmarking old GPUs and I want to come up with a protocol that will cover 2D and 3D acceleration and basically give a coherent scale that will cover things from the beginning of 2D acceleration in the 1980s and run through 2007 or so when there was a paradigm shift in the way GPUs interacted with CPUs as GPUs started receiving their own native floating point capabilities. In the old days an 8:1 ratio of system RAM to VRAM was appropriate. But today, GPUs have more raw processing power than the CPUs which aid them and a ratio of 2:1 or so is desirable.
I have the ability to pair AGP cards with a Core 2 Duo (albeit running on one core) and DDR2 memory which will make for an excellent test rig for AGP, PCI and and PCIe cards of the vintage persuasion in DOS, 95, 98, 2000, XP and Vista as there will be no chance of bottlenecking any of the cards I am testing. In fact, being based on the Pentium Pro architecture, and lacking later, problematic instruction sets, I cannot think of a better CPU than a Core 2 for empirical testing of such a broad range of video capabilities. I want to get a database going, like Passmark but for potatoes.
I would not hold out much hope for the 6326 because one article I read said it was 60% of an ATi Rage (not 128) Pro and one third of an NVidia Riva 128. The Rage 128 Pro is considered similar to the Riva 128.
Problematically, the shady computing industry released many cards with similar names but rather different performance. People complain about modern components but it is much better. Even in the late 2000s, this obvious scammery with similar names was still going on. I have 2 7800GT cards and one 7800GTX. You would think the GTX would be more powerful but no. There were times a GTS card might be rather good or really crap. They tell you the GTX 1650 is based on the Turing architecture. But you have to take a deep dive before you realize they lopped off the RTX and AI cores from a defective RTX 20 series. Great cards for the price but you are literally buying their garbage. That is something relatively new and is a product of small process manufacturing.

Last edited by HangarAte2nds! on 2021-09-25, 08:10. Edited 1 time in total.

Reply 34 of 57, by zyga64

User metadata
Rank Member
Rank
Member
HangarAte2nds! wrote on 2021-09-22, 05:29:

I would not hold out much hope for the 6326 because one article I read said it was 60% of an ATi Rage (not 128) Pro and one third of an NVidia Riva 128. The Rage 128 Pro is considered similar to the Riva 128.

I'd rather say Rage 128 Pro is similiar (in terms of performance) to Nvidia Riva TNT or TNT2 M64.
Riva 128 can be at best compared to the Rage II Pro (and it will win of course). It's the same generation as Intel 740, Matrox G200, S3 Savage 3D and Permedia 2 (maybe).

1) VLSI SCAMP /286@20 /4MB /CL-GD5422 /CMI8330
2) i420EX /486DX33 /16MB /TGUI9440 /YMF718+GUS
4) i440BX /P!!!750 /256MB /MX440 /SBLive!+Vibra16s
5) i865G /E5800 /2GB /Ti4200 /YMF724

Reply 35 of 57, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t

Rage 128 Pro - TNT2/TNT2 Pro
Rage 128 - TNT2 M64 or overclocked TNT.

Get up, come on get down with the sickness
Open up your hate, and let it flow into me

Reply 36 of 57, by BSA Starfire

User metadata
Rank Oldbie
Rank
Oldbie
HangarAte2nds! wrote on 2021-09-22, 05:29:
I just got an AGP 6326 today (which is a development of the S3 Virge - for the guy who thought this was different, lol). It is a […]
Show full quote

I just got an AGP 6326 today (which is a development of the S3 Virge - for the guy who thought this was different, 🤣). It is a late production model and appears to have 16MB onboard. There can be issues with these cards in DOS apparently but a later driver release fixed most of the issues. I also suspect that when I get to testing it, it will perform well. If they addressed the issues in DOS with later drivers, it stands to reason the later BIOS may just utilize the drivers better also.
I just ordered a Rage 128 Pro 16MB AGP and I have a TNT2 16MB AGP somewhere if I can find it. I am planning to start benchmarking old GPUs and I want to come up with a protocol that will cover 2D and 3D acceleration and basically give a coherent scale that will cover things from the beginning of 2D acceleration in the 1980s and run through 2007 or so when there was a paradigm shift in the way GPUs interacted with CPUs as GPUs started receiving their own native floating point capabilities. In the old days an 8:1 ratio of system RAM to VRAM was appropriate. But today, GPUs have more raw processing power than the CPUs which aid them and a ratio of 2:1 or so is desirable.
I have the ability to pair AGP cards with a Core 2 Duo (albeit running on one core) and DDR2 memory which will make for an excellent test rig for AGP, PCI and and PCIe cards of the vintage persuasion in DOS, 95, 98, 2000, XP and Vista as there will be no chance of bottlenecking any of the cards I am testing. In fact, being based on the Pentium Pro architecture, and lacking later, problematic instruction sets, I cannot think of a better CPU than a Core 2 for empirical testing of such a broad range of video capabilities. I want to get a database going, like Passmark but for potatoes.
I would not hold out much hope for the 6326 because one article I read said it was 60% of an ATi Rage (not 128) Pro and one third of an NVidia Riva 128. The Rage 128 Pro is considered similar to the Riva 128.
Problematically, the shady computing industry released many cards with similar names but rather different performance. People complain about modern components but it is much better. Even in the late 2000s, this obvious scammery with similar names was still going on. I have 2 7800GT cards and one 7800GTX. You would think the GTX would be more powerful but no. There were times a GTS card might be rather good or really crap. They tell you the GTX 1650 is based on the Turing architecture. But you have to take a deep dive before you realize they lopped off the RTX and AI cores from a defective RTX 20 series. Great cards for the price but you are literally buying their garbage. That is something relatively new and is a product of small process manufacturing.

The SiS 6326 is NOT a development of the S3 ViRGE, two different companies, 2 completely different designs. Are you sure what you have is a 6326? I've never seen a 16 MB version, but the later SiS 305 did come in 16MB version.

With a Core2 motherboard and AGP cards(I assume your talking about one of the Asrock boards), you'll only be able to tun 1.5 volt AGP cards and not the 3.3 volt cards like RIVA 128, Voodoo, SIS 6326, S3 Trio 3D etc...

286 20MHz,1MB RAM,Trident 8900B 1MB, Conner CFA-170A.SB 1350B
386SX 33MHz,ULSI 387,4MB Ram,OAK OTI077 1MB. Seagate ST1144A, MS WSS audio
Amstrad PC 9486i, DX/2 66, 16 MB RAM, Cirrus SVGA,Win 95,SB 16
Cyrix MII 333,128MB,SiS 6326 H0 rev,ESS 1869,Win ME

Reply 37 of 57, by Jasin Natael

User metadata
Rank Member
Rank
Member
HangarAte2nds! wrote on 2021-09-22, 05:29:
I just got an AGP 6326 today (which is a development of the S3 Virge - for the guy who thought this was different, lol). It is a […]
Show full quote

I just got an AGP 6326 today (which is a development of the S3 Virge - for the guy who thought this was different, 🤣). It is a late production model and appears to have 16MB onboard. There can be issues with these cards in DOS apparently but a later driver release fixed most of the issues. I also suspect that when I get to testing it, it will perform well. If they addressed the issues in DOS with later drivers, it stands to reason the later BIOS may just utilize the drivers better also.
I just ordered a Rage 128 Pro 16MB AGP and I have a TNT2 16MB AGP somewhere if I can find it. I am planning to start benchmarking old GPUs and I want to come up with a protocol that will cover 2D and 3D acceleration and basically give a coherent scale that will cover things from the beginning of 2D acceleration in the 1980s and run through 2007 or so when there was a paradigm shift in the way GPUs interacted with CPUs as GPUs started receiving their own native floating point capabilities. In the old days an 8:1 ratio of system RAM to VRAM was appropriate. But today, GPUs have more raw processing power than the CPUs which aid them and a ratio of 2:1 or so is desirable.
I have the ability to pair AGP cards with a Core 2 Duo (albeit running on one core) and DDR2 memory which will make for an excellent test rig for AGP, PCI and and PCIe cards of the vintage persuasion in DOS, 95, 98, 2000, XP and Vista as there will be no chance of bottlenecking any of the cards I am testing. In fact, being based on the Pentium Pro architecture, and lacking later, problematic instruction sets, I cannot think of a better CPU than a Core 2 for empirical testing of such a broad range of video capabilities. I want to get a database going, like Passmark but for potatoes.
I would not hold out much hope for the 6326 because one article I read said it was 60% of an ATi Rage (not 128) Pro and one third of an NVidia Riva 128. The Rage 128 Pro is considered similar to the Riva 128.
Problematically, the shady computing industry released many cards with similar names but rather different performance. People complain about modern components but it is much better. Even in the late 2000s, this obvious scammery with similar names was still going on. I have 2 7800GT cards and one 7800GTX. You would think the GTX would be more powerful but no. There were times a GTS card might be rather good or really crap. They tell you the GTX 1650 is based on the Turing architecture. But you have to take a deep dive before you realize they lopped off the RTX and AI cores from a defective RTX 20 series. Great cards for the price but you are literally buying their garbage. That is something relatively new and is a product of small process manufacturing.

There is no relation whatsoever between a S3 Virge and a SiS6326.......or any other SiS card for that matter. Also I've never heard of a 16MB SiS6326? Any pics of this card?

Reply 38 of 57, by Gmlb256

User metadata
Rank Oldbie
Rank
Oldbie
HangarAte2nds! wrote on 2021-09-22, 05:29:
I just got an AGP 6326 today (which is a development of the S3 Virge - for the guy who thought this was different, lol). It is a […]
Show full quote

I just got an AGP 6326 today (which is a development of the S3 Virge - for the guy who thought this was different, 🤣). It is a late production model and appears to have 16MB onboard. There can be issues with these cards in DOS apparently but a later driver release fixed most of the issues. I also suspect that when I get to testing it, it will perform well. If they addressed the issues in DOS with later drivers, it stands to reason the later BIOS may just utilize the drivers better also.
I just ordered a Rage 128 Pro 16MB AGP and I have a TNT2 16MB AGP somewhere if I can find it. I am planning to start benchmarking old GPUs and I want to come up with a protocol that will cover 2D and 3D acceleration and basically give a coherent scale that will cover things from the beginning of 2D acceleration in the 1980s and run through 2007 or so when there was a paradigm shift in the way GPUs interacted with CPUs as GPUs started receiving their own native floating point capabilities. In the old days an 8:1 ratio of system RAM to VRAM was appropriate. But today, GPUs have more raw processing power than the CPUs which aid them and a ratio of 2:1 or so is desirable.
I have the ability to pair AGP cards with a Core 2 Duo (albeit running on one core) and DDR2 memory which will make for an excellent test rig for AGP, PCI and and PCIe cards of the vintage persuasion in DOS, 95, 98, 2000, XP and Vista as there will be no chance of bottlenecking any of the cards I am testing. In fact, being based on the Pentium Pro architecture, and lacking later, problematic instruction sets, I cannot think of a better CPU than a Core 2 for empirical testing of such a broad range of video capabilities. I want to get a database going, like Passmark but for potatoes.
I would not hold out much hope for the 6326 because one article I read said it was 60% of an ATi Rage (not 128) Pro and one third of an NVidia Riva 128. The Rage 128 Pro is considered similar to the Riva 128.
Problematically, the shady computing industry released many cards with similar names but rather different performance. People complain about modern components but it is much better. Even in the late 2000s, this obvious scammery with similar names was still going on. I have 2 7800GT cards and one 7800GTX. You would think the GTX would be more powerful but no. There were times a GTS card might be rather good or really crap. They tell you the GTX 1650 is based on the Turing architecture. But you have to take a deep dive before you realize they lopped off the RTX and AI cores from a defective RTX 20 series. Great cards for the price but you are literally buying their garbage. That is something relatively new and is a product of small process manufacturing.

The only real development based on the S3 ViRGE were the Trio3D cards, I don't know where are you getting this nonsense.

Also the GTX 1650 is indeed based on the Turing architecture but without the raytracing and tensor cores, nothing defective there. This is just market segmentation and low-end stuff were usually aimed for casual or less serious gamers.