For DOS and Win 3.x users, I suppose I should add this link here: https://oummg.com/manual/adapters.html

That represents a test report table of different old ISA and PCI cards, focused on pre-VESA use in DOS and Win 3.x, though some VESA testing is also included. The tests are conducted from CRT Terminator compatibility perspective, but do offer a lot of information also for VGA use.

For performance, a rudimentary Doom timedemo FPS number is included.

clb wrote on 2025-11-29, 18:26:

For DOS and Win 3.x users, I suppose I should add this link here: https://oummg.com/manual/adapters.html

That represents a test report table of different old ISA and PCI cards, focused on pre-VESA use in DOS and Win 3.x, though some VESA testing is also included. The tests are conducted from CRT Terminator compatibility perspective, but do offer a lot of information also for VGA use.

For performance, a rudimentary Doom timedemo FPS number is included.

This list might be better for this thread since it covers the newer cards-- https://gona.mactar.hu/DOS_TESTS/

msdos9 wrote on 2025-04-28, 16:19:

Guys, I read the entire thread and concluded that
the S3 Virge series cards are the most compatible with dos and win 3.x-9x. Right?

For me Rage Pro / Rage XL is one at top list if we count out some dos games.
Does play DOS games,Win3.x works with media playback,Win9x just shines.

Win9x API
1. DirectDraw 2D API hardware acceleration
2. Direct 3D is 3D API hardware acceleration
3. ATi-CIF API hardware acceleration
4. OpenGL API hardware acceleration

Win9x ATi MultiMedia
5. ATi MultiMediaCenter 6.3 or ATi DVD + add ATi TV Wonder Remote
6. TV Tuner missing? Use AIW Rage RPO or add ATi TV Capture / S-Video Card (Probably connects over AMC)
ATi Multimedia Connector (AMC)

Speed
7. Rage Pro is faster than S3 Virge and Rage XL AGP (4xMemory=64bit) speed is even faster than Rage Pro AGP (First real 2xAGP card).

Other OS
8. ATi Rage Pro has great OS support, not just these three.

Nearly 30 years ago I had an original ATI All-in-Wonder paired with a Voodoo 1 card in my machine. Worked great.

I asked ChatGPT and it said S3 Trio64V2+ has broadest compatibility for 2D DOS games. That surprised me, but it's actually what I used with a 3Dfx Voodoo.

NJRoadfan wrote on 2025-12-01, 21:01:

Nearly 30 years ago I had an original ATI All-in-Wonder paired with a Voodoo 1 card in my machine. Worked great.

I had AIW Rage PRO at some point, but mostly i had regular Rage Pro and it was my main card, dvd worked and xvid movies also.
I had no Voodoo to play with. At that time MotorRacer / NFS3 was one of the best looking games that i played with that ATiRage.

Cirrus Logic PCI cards are a solid choice for a non-3D accelerated option. Decent performance, good compatibility.

marxveix wrote on 2025-12-01, 21:54:

NJRoadfan wrote on 2025-12-01, 21:01:

Nearly 30 years ago I had an original ATI All-in-Wonder paired with a Voodoo 1 card in my machine. Worked great.

I had AIW Rage PRO at some point, but mostly i had regular Rage Pro and it was my main card, dvd worked and xvid movies also.
I had no Voodoo to play with. At that time MotorRacer / NFS3 was one of the best looking games that i played with that ATiRage.

Quite a few replies advocate the Rage128Pro so I'm going to throw a wild card into the hat - the RV100 or Radeon 7000.

Why? Because RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro.

Given the original Radeon was a big fanfare DX7 T&L card, and the Radeon name then sneakily reused on an upgraded Rage128Pro, actual customers of the Radeon 7000 were sorely disappointed.

Among DX7+ T&L seekers the Radeon 7000 is mocked, misunderstood, and discarded - but among DOS/DX5/DX6 seekers maybe it should be reconsidered?

And, among the Radeon 7000 there are three flavours: 64bit SDRAM (slowest), 64bit DDR, 128bit SDRAM (fastest). The RV100 is not capable of supporting 128bit DDR, but there are myths that it can because retailers used 64bit SDRAM as the baseline and then claimed 128bit SDRAM was Double Data Rate.. because 128bit is double 64bit. When marketing took that insight and designed retail boxes to imply 128bit DDR, customers were sorely disappointed!

In its day this card was delivering disappointments on top of disappointments, so maybe it's no wonder nobody mentions it!

But, perhaps, the Radeon 7000 is the single most misunderstood graphics card in the history of personal computers?

Being not a real Radeon, the Radeon 7000 brings to DOS none of the compatibility glitches that the real Radeon brought to DOS.

And, perhaps, the Radeon 7000 it is the fastest 2D/3D DOS video card ever produced?

MattRocks wrote on 2025-12-07, 22:29:

Quite a few replies advocate the Rage128Pro so I'm going to throw a wild card into the hat - the RV100 or Radeon 7000. […]
Show full quote

marxveix wrote on 2025-12-01, 21:54:

NJRoadfan wrote on 2025-12-01, 21:01:

Nearly 30 years ago I had an original ATI All-in-Wonder paired with a Voodoo 1 card in my machine. Worked great.

I had AIW Rage PRO at some point, but mostly i had regular Rage Pro and it was my main card, dvd worked and xvid movies also.
I had no Voodoo to play with. At that time MotorRacer / NFS3 was one of the best looking games that i played with that ATiRage.

Quite a few replies advocate the Rage128Pro so I'm going to throw a wild card into the hat - the RV100 or Radeon 7000.

Why? Because RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro.

Given the original Radeon was a big fanfare DX7 T&L card, and the Radeon name then sneakily reused on an upgraded Rage128Pro, actual customers of the Radeon 7000 were sorely disappointed.

Among DX7+ T&L seekers the Radeon 7000 is mocked, misunderstood, and discarded - but among DOS/DX5/DX6 seekers maybe it should be reconsidered?

And, among the Radeon 7000 there are three flavours: 64bit SDRAM (slowest), 64bit DDR, 128bit SDRAM (fastest). The RV100 is not capable of supporting 128bit DDR, but there are myths that it can because retailers used 64bit SDRAM as the baseline and then claimed 128bit SDRAM was Double Data Rate.. because 128bit is double 64bit. When marketing took that insight and designed retail boxes to imply 128bit DDR, customers were sorely disappointed!

In its day this card was delivering disappointments on top of disappointments, so maybe it's no wonder nobody mentions it!

But, perhaps, the Radeon 7000 is the single most misunderstood graphics card in the history of personal computers?

Being not a real Radeon, the Radeon 7000 brings to DOS none of the compatibility glitches that the real Radeon brought to DOS.

And, perhaps, the Radeon 7000 it is the fastest 2D/3D DOS video card ever produced?

Do you have any sources that show that the RV100 is based on or more functionally similar to the Rage 4 (Rage 128) than the Rage 6/R100? I don't see any evidence of that anywhere online.

RV100 has exactly 1/2 the pixel pipelines , TMUs and ROPs of R100 (1:3:1 vs 2:6:2), and has no vertex shader. Rage 128 has a totally different configuration (2:2:2). R100 and RV100 are made on the same 180nm process, where Rage 4 is built on 250nm. The R100 and RV100 use the same driver packages too, where Rage 128 Pro drivers have never been combined with Radeon drivers as far as I know.

The die shots are all available too and there are too many differences between them to say that one is based on the other.
https://upload.wikimedia.org/wikipedia/common … 59911615%29.jpg
https://upload.wikimedia.org/wikipedia/common … 84036132%29.jpg
https://upload.wikimedia.org/wikipedia/common … 13734476%29.jpg

Also, I think that Nvidia cards (among others) have far fewer problems in DOS compared to ATi Rage or Radeon cards, and DOS performance tends to be higher as well.

A Radeon 7000 is probably a perfectly fine card for a Windows 9x retro machine, but I don't think there's any evidence of it being any better than other cards from the time. Also, there are no Radeon or Rage 128 drivers for Windows 3.x at all as far as I can tell.

Rage128 Pro is one of the fastest cards for dos, someone has tested it, but its the speediest in lower resolutions as i remember.
Rage XL is the latest ATi card for Win 3.1. Rage128/Rage128Pro is out of this os support, Rage128 supports Win95/NT3.5 or 4.0.

marxveix wrote on Yesterday, 01:00:

Rage128 Pro is one of the fastest cards for dos, someone has tested it, but its the speediest in lower resolutions as i remember.

Ah, yeah, the 320x240 version of the test I linked above has the Rage 128 Pro at the top.
https://www.vgamuseum.info/images/vlask/bench/quake320.png

Still, a 205fps vs 199fps lead over cards that seem to be significantly better in other situations isn't a huge win.

... it is still a win though. 😅

Ozzuneoj wrote on Yesterday, 00:31:

Do you have any sources that show that the RV100 is based on or more functionally similar to the Rage 4 (Rage 128) than the Rage […]
Show full quote

MattRocks wrote on 2025-12-07, 22:29:

Quite a few replies advocate the Rage128Pro so I'm going to throw a wild card into the hat - the RV100 or Radeon 7000. […]
Show full quote

marxveix wrote on 2025-12-01, 21:54:

I had AIW Rage PRO at some point, but mostly i had regular Rage Pro and it was my main card, dvd worked and xvid movies also.
I had no Voodoo to play with. At that time MotorRacer / NFS3 was one of the best looking games that i played with that ATiRage.

Quite a few replies advocate the Rage128Pro so I'm going to throw a wild card into the hat - the RV100 or Radeon 7000.

Why? Because RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro.

Given the original Radeon was a big fanfare DX7 T&L card, and the Radeon name then sneakily reused on an upgraded Rage128Pro, actual customers of the Radeon 7000 were sorely disappointed.

Among DX7+ T&L seekers the Radeon 7000 is mocked, misunderstood, and discarded - but among DOS/DX5/DX6 seekers maybe it should be reconsidered?

And, among the Radeon 7000 there are three flavours: 64bit SDRAM (slowest), 64bit DDR, 128bit SDRAM (fastest). The RV100 is not capable of supporting 128bit DDR, but there are myths that it can because retailers used 64bit SDRAM as the baseline and then claimed 128bit SDRAM was Double Data Rate.. because 128bit is double 64bit. When marketing took that insight and designed retail boxes to imply 128bit DDR, customers were sorely disappointed!

In its day this card was delivering disappointments on top of disappointments, so maybe it's no wonder nobody mentions it!

But, perhaps, the Radeon 7000 is the single most misunderstood graphics card in the history of personal computers?

Being not a real Radeon, the Radeon 7000 brings to DOS none of the compatibility glitches that the real Radeon brought to DOS.

And, perhaps, the Radeon 7000 it is the fastest 2D/3D DOS video card ever produced?

Do you have any sources that show that the RV100 is based on or more functionally similar to the Rage 4 (Rage 128) than the Rage 6/R100? I don't see any evidence of that anywhere online.

RV100 has exactly 1/2 the pixel pipelines , TMUs and ROPs of R100 (1:3:1 vs 2:6:2), and has no vertex shader. Rage 128 has a totally different configuration (2:2:2). R100 and RV100 are made on the same 180nm process, where Rage 4 is built on 250nm. The R100 and RV100 use the same driver packages too, where Rage 128 Pro drivers have never been combined with Radeon drivers as far as I know.

The die shots are all available too and there are too many differences between them to say that one is based on the other.
https://upload.wikimedia.org/wikipedia/common … 59911615%29.jpg
https://upload.wikimedia.org/wikipedia/common … 84036132%29.jpg
https://upload.wikimedia.org/wikipedia/common … 13734476%29.jpg

Also, I think that Nvidia cards (among others) have far fewer problems in DOS compared to ATi Rage or Radeon cards, and DOS performance tends to be higher as well.

A Radeon 7000 is probably a perfectly fine card for a Windows 9x retro machine, but I don't think there's any evidence of it being any better than other cards from the time. Also, there are no Radeon or Rage 128 drivers for Windows 3.x at all as far as I can tell.

Oof! That is a good challenge.

I assumed all chips have an ancestor. GeForce256 to GeForce2MX is mostly a die shrink with one descended from the other. In contrast, RV100 is not descended from R100 so let's see..

Hardware engineering: All three die photos look distinct to me - scaling is the same for RV100 and R100 but that's unrelated to architecture. If RV100 were simplified R100, we would see the “ghosts” of R100's removed parts (unused TMU areas, or relocated blocks) but we don't see that. Same for Rage128Pro. They look like three unrelated chips!

Software engineering: Software architecture is not coupled to hardware functions. ATI changed the software architecture when they launched Radeon - that shows a maintenance reset. From a business perspective, it's not viable to maintain two software architectures - whatever the currently maintained software architecture is will be used to carry new code. Radeon drivers contain RV100-specific quirks, suggesting RV100 is a parallel development rather than a cut down R100.

Systems testing: Holistically, the CPU cache load of RV100 is similar to Rage128Pro.

So I'll revise my position. It seems now that RV100 follows Rage128Pro requirements - same market assumptions and same functional behaviours, but on a clean sheet! It now looks like RV100 has no direct ancestor, and that makes RV100 even weirder and harder to understand.

Now we need someone to actually test it. I own a Rage128Pro, Radeon 7000, and Radeon DDR (original). I'll put this on my backlog but it might take me years to get around to reporting on.

MattRocks wrote on Yesterday, 02:10:

Oof! That is a good challenge. […]
Show full quote

Ozzuneoj wrote on Yesterday, 00:31:

Do you have any sources that show that the RV100 is based on or more functionally similar to the Rage 4 (Rage 128) than the Rage […]
Show full quote

MattRocks wrote on 2025-12-07, 22:29:

Quite a few replies advocate the Rage128Pro so I'm going to throw a wild card into the hat - the RV100 or Radeon 7000. […]
Show full quote

Quite a few replies advocate the Rage128Pro so I'm going to throw a wild card into the hat - the RV100 or Radeon 7000.

Why? Because RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro.

Given the original Radeon was a big fanfare DX7 T&L card, and the Radeon name then sneakily reused on an upgraded Rage128Pro, actual customers of the Radeon 7000 were sorely disappointed.

Among DX7+ T&L seekers the Radeon 7000 is mocked, misunderstood, and discarded - but among DOS/DX5/DX6 seekers maybe it should be reconsidered?

And, among the Radeon 7000 there are three flavours: 64bit SDRAM (slowest), 64bit DDR, 128bit SDRAM (fastest). The RV100 is not capable of supporting 128bit DDR, but there are myths that it can because retailers used 64bit SDRAM as the baseline and then claimed 128bit SDRAM was Double Data Rate.. because 128bit is double 64bit. When marketing took that insight and designed retail boxes to imply 128bit DDR, customers were sorely disappointed!

In its day this card was delivering disappointments on top of disappointments, so maybe it's no wonder nobody mentions it!

But, perhaps, the Radeon 7000 is the single most misunderstood graphics card in the history of personal computers?

Being not a real Radeon, the Radeon 7000 brings to DOS none of the compatibility glitches that the real Radeon brought to DOS.

And, perhaps, the Radeon 7000 it is the fastest 2D/3D DOS video card ever produced?

Do you have any sources that show that the RV100 is based on or more functionally similar to the Rage 4 (Rage 128) than the Rage 6/R100? I don't see any evidence of that anywhere online.

RV100 has exactly 1/2 the pixel pipelines , TMUs and ROPs of R100 (1:3:1 vs 2:6:2), and has no vertex shader. Rage 128 has a totally different configuration (2:2:2). R100 and RV100 are made on the same 180nm process, where Rage 4 is built on 250nm. The R100 and RV100 use the same driver packages too, where Rage 128 Pro drivers have never been combined with Radeon drivers as far as I know.

The die shots are all available too and there are too many differences between them to say that one is based on the other.
https://upload.wikimedia.org/wikipedia/common … 59911615%29.jpg
https://upload.wikimedia.org/wikipedia/common … 84036132%29.jpg
https://upload.wikimedia.org/wikipedia/common … 13734476%29.jpg

Also, I think that Nvidia cards (among others) have far fewer problems in DOS compared to ATi Rage or Radeon cards, and DOS performance tends to be higher as well.

A Radeon 7000 is probably a perfectly fine card for a Windows 9x retro machine, but I don't think there's any evidence of it being any better than other cards from the time. Also, there are no Radeon or Rage 128 drivers for Windows 3.x at all as far as I can tell.

Oof! That is a good challenge.

I assumed all chips have an ancestor. GeForce256 to GeForce2MX is mostly a die shrink with one descended from the other. In contrast, RV100 is not descended from R100..

Hardware engineering: All three die photos look distinct to me - scaling is the same for RV100 and R100 but that's unrelated to architecture. If RV100 were simplified R100, we would see the “ghosts” of R100's removed parts (unused TMU areas, or relocated blocks) but we don't see that. Same for Rage128Pro. They look like three unrelated chips!

Software engineering: Software architecture is not coupled to hardware functions. ATI changed the software architecture when they launched Radeon - that shows a maintenance reset. From a business perspective, it's not viable to maintain two software architectures - whatever the currently maintained software architecture is will be used to carry new code. Radeon drivers contain RV100-specific quirks, suggesting RV100 is a parallel development rather than a cut down R100.

Systems testing: Holistically, the CPU cache load of RV100 is similar to Rage128Pro.

So I'll revise my position. It seems now that RV100 follows Rage128Pro requirements - same market assumptions and same functional behaviours, but on a clean sheet! It now looks like RV100 has no direct ancestor, and that makes RV100 even weirder and harder to understand.

Now we need someone to actually test it. I own a Rage128Pro, Radeon 7000, and Radeon DDR (original). I'll put this on my backlog but it might take me years to get around to reporting on.

Okay, I was just wondering if I was missing something because your post made it sound like everyone was unaware of some fundamental truth that you had little known information on. And I love to learn things like that! But... These are all very matter-of-fact statements and I was curious what convinced you of these things:

MattRocks wrote on 2025-12-07, 22:29:

...RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro. . […]
Show full quote

...RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro.
...
Among DX7+ T&L seekers the Radeon 7000 is mocked, misunderstood, and discarded - but among DOS/DX5/DX6 seekers maybe it should be reconsidered?
....
But, perhaps, the Radeon 7000 is the single most misunderstood graphics card in the history of personal computers?

Being not a real Radeon, the Radeon 7000 brings to DOS none of the compatibility glitches that the real Radeon brought to DOS.

And, perhaps, the Radeon 7000 it is the fastest 2D/3D DOS video card ever produced?

Not trying to be argumentative, but it seems most logical that the RV100 was just cut down from R100 exactly as everything online says it was. Unless there is some technical documentation or clear compatibility difference. I feel like someone would have reported by now if the RV100 was really good at DOS vs the R100, but I haven't seen anything like that. In fact I've seen posts talking about the usual ATi graphical issues in DOS with Rage 128, RV100, R100 and others.

Ozzuneoj wrote on Yesterday, 01:33:

Ah, yeah, the 320x240 version of the test I linked above has the Rage 128 Pro at the top. https://www.vgamuseum.info/images/vla […]
Show full quote

marxveix wrote on Yesterday, 01:00:

Rage128 Pro is one of the fastest cards for dos, someone has tested it, but its the speediest in lower resolutions as i remember.

Ah, yeah, the 320x240 version of the test I linked above has the Rage 128 Pro at the top.
https://www.vgamuseum.info/images/vlask/bench/quake320.png

Still, a 205fps vs 199fps lead over cards that seem to be significantly better in other situations isn't a huge win.

... it is still a win though. 😅

Software rendering benchmark tests how fast the CPU can stream data to VRAM - how well the GPU gets out of the way and lets the CPU pump pixels.

Ozzuneoj wrote on Yesterday, 02:18:

Okay, I was just wondering if I was missing something because your post made it sound like everyone was unaware of some fundamen […]
Show full quote

MattRocks wrote on Yesterday, 02:10:

Oof! That is a good challenge. […]
Show full quote

Ozzuneoj wrote on Yesterday, 00:31:

Do you have any sources that show that the RV100 is based on or more functionally similar to the Rage 4 (Rage 128) than the Rage […]
Show full quote

Do you have any sources that show that the RV100 is based on or more functionally similar to the Rage 4 (Rage 128) than the Rage 6/R100? I don't see any evidence of that anywhere online.

RV100 has exactly 1/2 the pixel pipelines , TMUs and ROPs of R100 (1:3:1 vs 2:6:2), and has no vertex shader. Rage 128 has a totally different configuration (2:2:2). R100 and RV100 are made on the same 180nm process, where Rage 4 is built on 250nm. The R100 and RV100 use the same driver packages too, where Rage 128 Pro drivers have never been combined with Radeon drivers as far as I know.

The die shots are all available too and there are too many differences between them to say that one is based on the other.
https://upload.wikimedia.org/wikipedia/common … 59911615%29.jpg
https://upload.wikimedia.org/wikipedia/common … 84036132%29.jpg
https://upload.wikimedia.org/wikipedia/common … 13734476%29.jpg

Also, I think that Nvidia cards (among others) have far fewer problems in DOS compared to ATi Rage or Radeon cards, and DOS performance tends to be higher as well.

A Radeon 7000 is probably a perfectly fine card for a Windows 9x retro machine, but I don't think there's any evidence of it being any better than other cards from the time. Also, there are no Radeon or Rage 128 drivers for Windows 3.x at all as far as I can tell.

Oof! That is a good challenge.

I assumed all chips have an ancestor. GeForce256 to GeForce2MX is mostly a die shrink with one descended from the other. In contrast, RV100 is not descended from R100..

Hardware engineering: All three die photos look distinct to me - scaling is the same for RV100 and R100 but that's unrelated to architecture. If RV100 were simplified R100, we would see the “ghosts” of R100's removed parts (unused TMU areas, or relocated blocks) but we don't see that. Same for Rage128Pro. They look like three unrelated chips!

Software engineering: Software architecture is not coupled to hardware functions. ATI changed the software architecture when they launched Radeon - that shows a maintenance reset. From a business perspective, it's not viable to maintain two software architectures - whatever the currently maintained software architecture is will be used to carry new code. Radeon drivers contain RV100-specific quirks, suggesting RV100 is a parallel development rather than a cut down R100.

Systems testing: Holistically, the CPU cache load of RV100 is similar to Rage128Pro.

So I'll revise my position. It seems now that RV100 follows Rage128Pro requirements - same market assumptions and same functional behaviours, but on a clean sheet! It now looks like RV100 has no direct ancestor, and that makes RV100 even weirder and harder to understand.

Now we need someone to actually test it. I own a Rage128Pro, Radeon 7000, and Radeon DDR (original). I'll put this on my backlog but it might take me years to get around to reporting on.

Okay, I was just wondering if I was missing something because your post made it sound like everyone was unaware of some fundamental truth that you had little known information on. And I love to learn things like that! But... These are all very matter-of-fact statements and I was curious what convinced you of these things:

MattRocks wrote on 2025-12-07, 22:29:

...RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro. . […]
Show full quote

...RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro.
...
Among DX7+ T&L seekers the Radeon 7000 is mocked, misunderstood, and discarded - but among DOS/DX5/DX6 seekers maybe it should be reconsidered?
....
But, perhaps, the Radeon 7000 is the single most misunderstood graphics card in the history of personal computers?

Being not a real Radeon, the Radeon 7000 brings to DOS none of the compatibility glitches that the real Radeon brought to DOS.

And, perhaps, the Radeon 7000 it is the fastest 2D/3D DOS video card ever produced?

Not trying to be argumentative, but it seems most logical that the RV100 was just cut down from R100 exactly as everything online says it was. Unless there is some technical documentation or clear compatibility difference. I feel like someone would have reported by now if the RV100 was really good at DOS vs the R100, but I haven't seen anything like that. In fact I've seen posts talking about the usual ATi graphical issues in DOS with Rage 128, RV100, R100 and others.

I haven't seen the RV100 tested. It generally gets ignored, inheriting the same reputation as Rage128Pro or R100 without actually being plugged in.

I genuinely thought RV100 was the last push for Rage4 architecture because both Radeon 7000 and Rage128Pro are clearly DX6 cards - and it was inconceivable to me that anyone would develop two competing DX6 architectures! But, after being challenged to find proof, it looks like ATI did in fact develop two competing architectures. Why? The OEM market (DELL, Compaq, etc.) didn't want expensive R100 and..

Claims that RV100 is a cut down R100 don't convince me because the chips have too little in common, and the differences are pointlessly different. And, nobody seems to have ever done a proper review.

MattRocks wrote on Yesterday, 02:37:

I haven't seen the RV100 tested. It generally gets ignored, inheriting the same reputation as Rage128Pro or R100 without actuall […]
Show full quote

Ozzuneoj wrote on Yesterday, 02:18:

Okay, I was just wondering if I was missing something because your post made it sound like everyone was unaware of some fundamen […]
Show full quote

MattRocks wrote on Yesterday, 02:10:

Oof! That is a good challenge. […]
Show full quote

Oof! That is a good challenge.

I assumed all chips have an ancestor. GeForce256 to GeForce2MX is mostly a die shrink with one descended from the other. In contrast, RV100 is not descended from R100..

Hardware engineering: All three die photos look distinct to me - scaling is the same for RV100 and R100 but that's unrelated to architecture. If RV100 were simplified R100, we would see the “ghosts” of R100's removed parts (unused TMU areas, or relocated blocks) but we don't see that. Same for Rage128Pro. They look like three unrelated chips!

Software engineering: Software architecture is not coupled to hardware functions. ATI changed the software architecture when they launched Radeon - that shows a maintenance reset. From a business perspective, it's not viable to maintain two software architectures - whatever the currently maintained software architecture is will be used to carry new code. Radeon drivers contain RV100-specific quirks, suggesting RV100 is a parallel development rather than a cut down R100.

Systems testing: Holistically, the CPU cache load of RV100 is similar to Rage128Pro.

So I'll revise my position. It seems now that RV100 follows Rage128Pro requirements - same market assumptions and same functional behaviours, but on a clean sheet! It now looks like RV100 has no direct ancestor, and that makes RV100 even weirder and harder to understand.

Now we need someone to actually test it. I own a Rage128Pro, Radeon 7000, and Radeon DDR (original). I'll put this on my backlog but it might take me years to get around to reporting on.

Okay, I was just wondering if I was missing something because your post made it sound like everyone was unaware of some fundamental truth that you had little known information on. And I love to learn things like that! But... These are all very matter-of-fact statements and I was curious what convinced you of these things:

MattRocks wrote on 2025-12-07, 22:29:

...RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro. . […]
Show full quote

...RV100 is not a cut down R100. In fact the Rv100 is not descended from the R100 at all. It's descended from the Rage128Pro.
...
Among DX7+ T&L seekers the Radeon 7000 is mocked, misunderstood, and discarded - but among DOS/DX5/DX6 seekers maybe it should be reconsidered?
....
But, perhaps, the Radeon 7000 is the single most misunderstood graphics card in the history of personal computers?

Being not a real Radeon, the Radeon 7000 brings to DOS none of the compatibility glitches that the real Radeon brought to DOS.

And, perhaps, the Radeon 7000 it is the fastest 2D/3D DOS video card ever produced?

Not trying to be argumentative, but it seems most logical that the RV100 was just cut down from R100 exactly as everything online says it was. Unless there is some technical documentation or clear compatibility difference. I feel like someone would have reported by now if the RV100 was really good at DOS vs the R100, but I haven't seen anything like that. In fact I've seen posts talking about the usual ATi graphical issues in DOS with Rage 128, RV100, R100 and others.

I haven't seen the RV100 tested. It generally gets ignored, inheriting the same reputation as Rage128Pro or R100 without actually being plugged in.

I genuinely thought RV100 was the last push for Rage4 architecture because both Radeon 7000 and Rage128Pro are clearly DX6 cards - and it was inconceivable to me that anyone would develop two competing DX6 architectures! But, after being challenged to find proof, it looks like ATI did in fact develop two competing architectures. Why? I guess the OEM market (DELL, Compaq, etc.) didn't want a die shrunk Rage128Pro.

Claims that RV100 is a cut down R100 don't convince me because the chips have too little in common, and the differences are pointlessly different. And, nobody seems to have ever done a proper review.

It did some tests on the rv100 with a K6-3 400, P3-800 and an Athlon 2500+ recently.
Re: K6-III Socket 7 AGP vs PCI Tester The RV100 cards I had were not thrilling. I wouldn’t recommend them for DOS or Win31. There were some compatibility issues under DOS, didn’t have Win3.1 drivers, and they were all discount boards with at best average video quality. If you are going for the discount boards, I’d say the MX 4000 or an FX 5200 gives better bang for the retro $$ under Windows 95.

From my own personal experience, the ATI mach 32 with the fast ramdac was my card of choice from 1992 to 1995. Fast and had 8514/a hardware support for out of the box compatibility. Also had the best picture quality on the market at the time and you could save your refresh rate settings to the card.

MattRocks wrote on Yesterday, 02:36:

Software rendering benchmark tests how fast the CPU can stream data to VRAM - how well the GPU gets out of the way and lets the CPU pump pixels.

Right, and at 640x480 in Quake (software mode) it seems like Nvidia, 3dfx cards, S3, SIS and others got out of the way much better than any ATi cards did.

MattRocks wrote on Yesterday, 02:37:

I haven't seen the RV100 tested. It generally gets ignored, inheriting the same reputation as Rage128Pro or R100 without actually being plugged in.

I genuinely thought RV100 was the last push for Rage4 architecture because both Radeon 7000 and Rage128Pro are clearly DX6 cards - and it was inconceivable to me that anyone would develop two competing DX6 architectures! But, after being challenged to find proof, it looks like ATI did in fact develop two competing architectures. Why? The OEM market (DELL, Compaq, etc.) didn't want expensive R100 and..

Claims that RV100 is a cut down R100 don't convince me because the chips have too little in common, and the differences are pointlessly different. And, nobody seems to have ever done a proper review.

Fair enough, I was just curious if there was something empirical behind the claims you made before.

Since the Rage 5 (Flipper) GPU for the Gamecube was being developed along side the similarly complex (both with T&L) Rage 6\R100, that doesn't seem to leave room for RV100 to have been developed separately alongside them all with no Rage code name and no hint at it's existence in the rumor mill or tech news of the time. If cutting out the T&L and half the ROPs\TMUs from the R100 and rearranging the die made the RV100 significantly cheaper to produce (notice how much smaller the die is compared to R100), that alone would be enough to justify it's existence.

Anyway, I'm not sure exactly how someone would go about investigating this with a full review. According to this compatibility chart, the Rage 128 and R100 have issues in most of the same situations anyway. I guess if someone wanted to test an RV100 in some of the situations where an Rage 128 works and an R100 has issues, that could suffice? In the end, even if compatibility in those situations were more like the Rage 128, that wouldn't really prove that the RV100 was a secret Rage 4.5 project that had stayed behind closed doors and was kept secret for 25 years... it would just prove that some differences between R100 and RV100 (which was released later...) result in those particular problems being fixed.

You have me curious now though! I probably have at least one Radeon VE\7000 around here somewhere, so I might give this a try some time. 🙂

Ozzuneoj wrote on Yesterday, 01:33:

Ah, yeah, the 320x240 version of the test I linked above has the Rage 128 Pro at the top. https://www.vgamuseum.info/images/vla […]
Show full quote

Ah, yeah, the 320x240 version of the test I linked above has the Rage 128 Pro at the top.
https://www.vgamuseum.info/images/vlask/bench/quake320.png

Still, a 205fps vs 199fps lead over cards that seem to be significantly better in other situations isn't a huge win.

... it is still a win though. 😅

I love both these great cards, Rage 128 Pro and Rage Pro (with its XL AGP brother). I also want to test Rage Pro AGP vs PCI @ DVD at some point,
because only Rage Pro AGP was marketed for DVD playback as i have seen. Rage XL AGP is faster and better, but Rage Pro got blocky, minecraft
retro feeling to it, Rage XL has all parts smoothened or filetered, Rage Pro most, but not all. Both cards are underrated, drivers could be better. 😀

douglar wrote on Yesterday, 03:21:

It did some tests on the rv100 with a K6-3 400, P3-800 and an Athlon 2500+ recently. Re: K6-III Socket 7 AGP vs PCI Tester The […]
Show full quote

MattRocks wrote on Yesterday, 02:37:

I haven't seen the RV100 tested. It generally gets ignored, inheriting the same reputation as Rage128Pro or R100 without actuall […]
Show full quote

Ozzuneoj wrote on Yesterday, 02:18:

Okay, I was just wondering if I was missing something because your post made it sound like everyone was unaware of some fundamental truth that you had little known information on. And I love to learn things like that! But... These are all very matter-of-fact statements and I was curious what convinced you of these things:

Not trying to be argumentative, but it seems most logical that the RV100 was just cut down from R100 exactly as everything online says it was. Unless there is some technical documentation or clear compatibility difference. I feel like someone would have reported by now if the RV100 was really good at DOS vs the R100, but I haven't seen anything like that. In fact I've seen posts talking about the usual ATi graphical issues in DOS with Rage 128, RV100, R100 and others.

I haven't seen the RV100 tested. It generally gets ignored, inheriting the same reputation as Rage128Pro or R100 without actually being plugged in.

I genuinely thought RV100 was the last push for Rage4 architecture because both Radeon 7000 and Rage128Pro are clearly DX6 cards - and it was inconceivable to me that anyone would develop two competing DX6 architectures! But, after being challenged to find proof, it looks like ATI did in fact develop two competing architectures. Why? I guess the OEM market (DELL, Compaq, etc.) didn't want a die shrunk Rage128Pro.

Claims that RV100 is a cut down R100 don't convince me because the chips have too little in common, and the differences are pointlessly different. And, nobody seems to have ever done a proper review.

It did some tests on the rv100 with a K6-3 400, P3-800 and an Athlon 2500+ recently.
Re: K6-III Socket 7 AGP vs PCI Tester The RV100 cards I had were not thrilling. I wouldn’t recommend them for DOS or Win31. There were some compatibility issues under DOS, didn’t have Win3.1 drivers, and they were all discount boards with at best average video quality. If you are going for the discount boards, I’d say the MX 4000 or an FX 5200 gives better bang for the retro $$ under Windows 95.

From my own personal experience, the ATI mach 32 with the fast ramdac was my card of choice from 1992 to 1995. Fast and had 8514/a hardware support for out of the box compatibility. Also had the best picture quality on the market at the time and you could save your refresh rate settings to the card.

Useful, but why test the DX6 cards under DX9c conditions? Depending on what the drivers do, I think that might be a handicap for some cards; FX5200 is notorious for being a DX8 chip sold as DX9 compliant due to driver software emulation.

In terms of capabilities the RV100 VPU aligns closely with TNT2 and Rage128Pro - definitely no T&L/DX7 stuff on any of them. Your test compared 128bit TNT2 and 128bit Rage128Pro against 64bit 7000.

I only did an anecdotal "test" with PC-BSD (an old operating system). I just booted them and shut down - so more a check than a test. The Radeon DDR/9550/9600/9700 all displayed beautifully; while the Rage128Pro/7000 both display the same software glitches around hardware mouse pointer, which I am led to believe is a known Rage128 driver issue (why the Rage128 driver decided it should drive the 7000 is not known - I just assumed the software knows what it's doing). I didn't test further than that as I was just sorting my cards, but it shows something worth further investigation. My 7000 is the 128bit variant, but VRAM behaviours would not be stressed on loading a desktop.

MattRocks wrote on Yesterday, 11:05:

Useful, but why test the DX6 cards under DX9c conditions? Depending on what the drivers do, I think that might be a handicap for […]
Show full quote

douglar wrote on Yesterday, 03:21:

It did some tests on the rv100 with a K6-3 400, P3-800 and an Athlon 2500+ recently. Re: K6-III Socket 7 AGP vs PCI Tester The […]
Show full quote

MattRocks wrote on Yesterday, 02:37:

I haven't seen the RV100 tested. It generally gets ignored, inheriting the same reputation as Rage128Pro or R100 without actually being plugged in.

I genuinely thought RV100 was the last push for Rage4 architecture because both Radeon 7000 and Rage128Pro are clearly DX6 cards - and it was inconceivable to me that anyone would develop two competing DX6 architectures! But, after being challenged to find proof, it looks like ATI did in fact develop two competing architectures. Why? I guess the OEM market (DELL, Compaq, etc.) didn't want a die shrunk Rage128Pro.

Claims that RV100 is a cut down R100 don't convince me because the chips have too little in common, and the differences are pointlessly different. And, nobody seems to have ever done a proper review.

It did some tests on the rv100 with a K6-3 400, P3-800 and an Athlon 2500+ recently.
Re: K6-III Socket 7 AGP vs PCI Tester The RV100 cards I had were not thrilling. I wouldn’t recommend them for DOS or Win31. There were some compatibility issues under DOS, didn’t have Win3.1 drivers, and they were all discount boards with at best average video quality. If you are going for the discount boards, I’d say the MX 4000 or an FX 5200 gives better bang for the retro $$ under Windows 95.

From my own personal experience, the ATI mach 32 with the fast ramdac was my card of choice from 1992 to 1995. Fast and had 8514/a hardware support for out of the box compatibility. Also had the best picture quality on the market at the time and you could save your refresh rate settings to the card.

Useful, but why test the DX6 cards under DX9c conditions? Depending on what the drivers do, I think that might be a handicap for some cards; FX5200 is notorious for being a DX8 chip sold as DX9 compliant due to driver software emulation.

In terms of capabilities the RV100 VPU aligns closely with TNT2 and Rage128Pro - definitely no T&L/DX7 stuff on any of them. Your test compared 128bit TNT2 and 128bit Rage128Pro against 64bit 7000.

I only did an anecdotal "test" with PC-BSD (an old operating system). I just booted them and shut down - so more a check than a test. The Radeon DDR/9550/9600/9700 all displayed beautifully; while the Rage128Pro/7000 both display the same software glitches around hardware mouse pointer, which I am led to believe is a known Rage128 driver issue (why the Rage128 driver decided it should drive the 7000 is not known - I just assumed the software knows what it's doing). I didn't test further than that as I was just sorting my cards, but it shows something worth further investigation. My 7000 is the 128bit variant, but VRAM behaviours would not be stressed on loading a desktop.

I don't think I tested the cards under DirectX 9 conditions just because I installed DirectX 9c, right? I thought it contains all the older DirectX modes and uses them when that's all the card can do.

Radeon 9550/9600/9700 are R3xx family GPUs and when I was testing DOS stuff, I noticed that the R3xx family didn't have many of the older glitches found in the Rage, R100 and R200 families in my experience.
But the Radeon DDR was still R100 and when I was testing, the R100 family it still had a lot of the issues found in the Rage 128, like some of the 640x480 modes displayed incorrectly, etc.

I just tested a lot of cards that I got on the cheap, at first to see if PCI vs AGP mattered in the super socket 7 era. The answer was it didn't matter that much on Super 7 motherboards other than a 10-15% drop off in DOS pixel flinging. But then I expanded the tests to AGP 4x and AGP 8x and by the time you get to AGP 8x motherboards, not only was there a difference in PCI vs AGP performance, but the PCI performance was actually worse on a 2005 motherboard than it was in the Super 7 motherboards.