VOGONS


First post, by cxm717

User metadata
Rank Member
Rank
Member

Hey, so I decided to do a small driver comparison on my Rage128GL card. I used my Aopen AX6BC mobo with 128MB of PC100, a Pentium2@450MHz and a Creative Sound Blaster Live sound card.

I used mostly older games for this test but I am going to go back and test some newer games with these drivers also (probably some games from 1999 and 2000).

I am looking for more drivers. So if you have some that I did not test, post a link or something.

Here are the results:

alldrv128.png
Filename
alldrv128.png
File size
531.42 KiB
Views
816 views
File license
Fair use/fair dealing exception

Reply 2 of 16, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t

To summarize: 6219 is the top dog. Huh, who would have thought that late drivers were even more horrible in some places.

Great job btw, could be used as a reference.

All the other kids, with the pumped up kicks
You'd better run, better run, faster than my blaster

Reply 4 of 16, by appiah4

User metadata
Rank l33t
Rank
l33t

Yeah png screenshots in say 3dmark and, I dont know, maybe Quake 2 would be great for IQ comparison.. 62xx seems to perform but maybe it is buggy and does not render correctly?

Great work as ever though.

Last edited by appiah4 on 2019-03-01, 06:05. Edited 1 time in total.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 6 of 16, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t

With old cards I am suspicious about faster drivers having more visual speed hacks

Rage128 already has "optimised" linear filtering.

All the other kids, with the pumped up kicks
You'd better run, better run, faster than my blaster

Reply 7 of 16, by cxm717

User metadata
Rank Member
Rank
Member
Nemo1985 wrote:

This is very interesting, thank you for sharing the results!
By any chance do you have the links where to to download the drivers?

https://megaupload.nz/a4051bvbb4/drivers_zip

Reply 8 of 16, by cxm717

User metadata
Rank Member
Rank
Member
appiah4 wrote:

Yeah png screenshots in say 3dmark and I dont know Quake 2 would be great for IQ comparispn.. 62xx seema to perform but maybe it ia buggy and does not render correctly?

Great work as ever though.

There are some bugs that I noticed running the benchmarks. Most of them are hard to get in a screenshot though. 6162 and 6219 both had problems in unreal with textures that would get bright sometimes, it happens in both D3D and OGL but it wasn't very bad in D3D. 6249 has an OGL bug that would case the screen to get corrupted when either loading a game or changing the resolution from in the game. 6277 had some flashing textures in drakan. 6103 has corrupted textures in all the OGL games I tried. The drivers after 6111 had weird looking lightmaps in quake. All the drivers have this mouse lag in D3D, disabling vsync made it much worse and added stuttering to a few games. Here are some screenshots I took with each driver: https://megaupload.nz/Ve661dvab9/ss_zip. You can see the unreal bug I was talking about in the 6219 unreal screenshot. Maybe when I get some time I could try to get some video of the bugs with fraps.

Edit: Forgot screen flashing and texture corruption in decent3 in OGL on the 6162 driver in 800x600 and up resolutions. There was also problems with trying to force a zbuffer to either 16 or 24 in OGL. Some drivers would give quake2 or quake3 a different zbuffer and if you tried to force it in the control panel the game would run very slow (less than 1fps).

Reply 9 of 16, by appiah4

User metadata
Rank l33t
Rank
l33t
cxm717 wrote:
appiah4 wrote:

Yeah png screenshots in say 3dmark and I dont know Quake 2 would be great for IQ comparispn.. 62xx seema to perform but maybe it ia buggy and does not render correctly?

Great work as ever though.

There are some bugs that I noticed running the benchmarks. Most of them are hard to get in a screenshot though. 6162 and 6219 both had problems in unreal with textures that would get bright sometimes, it happens in both D3D and OGL but it wasn't very bad in D3D. 6249 has an OGL bug that would case the screen to get corrupted when either loading a game or changing the resolution from in the game. 6277 had some flashing textures in drakan. 6103 has corrupted textures in all the OGL games I tried. The drivers after 6111 had weird looking lightmaps in quake. All the drivers have this mouse lag in D3D, disabling vsync made it much worse and added stuttering to a few games. Here are some screenshots I took with each driver: https://megaupload.nz/Ve661dvab9/ss_zip. You can see the unreal bug I was talking about in the 6219 unreal screenshot. Maybe when I get some time I could try to get some video of the bugs with fraps.

Edit: Forgot screen flashing and texture corruption in decent3 in OGL on the 6162 driver in 800x600 and up resolutions. There was also problems with trying to force a zbuffer to either 16 or 24 in OGL. Some drivers would give quake2 or quake3 a different zbuffer and if you tried to force it in the control panel the game would run very slow (less than 1fps).

First of all, holy shit at my terrible spelling on a phone - I hate touch cap keyboards..

That out of the way, this is no surprise. Even during the Radeon 8500 days initial drivers for architectures were usually very fast but either buggy or overzealously 'optimized' (remember the Radeon 8500 QUACK.EXE scandal?) so as they mature these bugs or 'optimizations' probably get filtered out of the driver stack and are replaced with more accurate rendering. So yeah, even though 6219 is the fastest of the lot it certainly has IQ issues and I would probably install a more mature driver (such as 6249 through 6292) that does not destroy performance (7192 and onwards).

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 11 of 16, by cxm717

User metadata
Rank Member
Rank
Member

I also ran some benchmarks on this setup with the Rage128 pro but with a P3 @ 800/133 and 256MB of PC133 CL2.

all128prodrv.png
Filename
all128prodrv.png
File size
287.84 KiB
Views
554 views
File license
Fair use/fair dealing exception

Edit: For these tests I did use the full version of forsaken which has more options for running higher resolutions and in 32bit colour. It's a different demo though, so you can't compare it to the other results. Also the Rage128 pro was at 130/130 for these tests where as the ones with the P2/450 were at 125/143.

Reply 12 of 16, by Takedasun

User metadata
Rank Newbie
Rank
Newbie
cxm717 wrote:

I am looking for more drivers. So if you have some that I did not test, post a link or something.

https://yadi.sk/d/vfgML2Vm3TR4mw/x.xx/Win9x

Rage128-4.11.6060-win9x.zip
Rage128-4.11.6060-win9x_601t01.zip
Rage128-4.11.6068-win9x_610a3z.zip
Rage128-4.11.6071-win9x_610a4e.zip
Rage128-4.11.6103-win9x_611cd05.zip
Rage128-4.11.6105-win9x_6111a1k.zip
Rage128-4.11.6114-win9x.zip
Rage128-4.11.6216-win9x.zip
Rage128-4.11.6220-win9x.zip
Rage128-4.11.6222-win9x_630cd09c_3dnow_sse.zip
Rage128-4.11.6249-win9x_631cd25e.zip
Rage128-4.11.6263-win9x.zip
Rage128-4.11.6664-win9x_612cdh21.zip
Rage128-4.11.6702-win9x_620cdh29e.zip
Rage128-4.11.6713-win9x_630cdh34e.zip
Rage128-4.11.6713-win9x_630cdh37e.zip
Rage128-4.11.6745-win9x_631cdh40e.zip
Rage128-4.11.6745-win9x_631cdh41.zip
Rage128-4.12.6269-win9x.zip
Rage128-4.12.6277-win9x.zip
Rage128-4.12.6281-win9x_633cd18.zip
Rage128-4.12.6281-win9x_633cd21.zip
Rage128-4.12.6282-win9x_6338cd20.zip
Rage128-4.12.6292-win9x.zip
Rage128-4.12.6292-win9x_634cd08.zip
Rage128-4.12.7063-win9x.zip
Rage128-4.13.7078-win9x_640cd19.zip
Rage128-4.13.7078-win9x_whql.zip
Rage128-4.13.7094-win9x_651-010306b.zip
Rage128-4.13.7110-win9x_654-010509b.zip
Rage128-4.13.7140-win9x_656-010620b.zip
Rage128-4.13.7192-win9x_last_official.zip
Rage128-4.13.7202-win9x.zip
Rage128-4.13.8006-win9x_beta.zip

Last edited by Takedasun on 2019-12-19, 11:56. Edited 2 times in total.

Reply 14 of 16, by cxm717

User metadata
Rank Member
Rank
Member
Takedasun wrote:
cxm717 wrote:

I am looking for more drivers. So if you have some that I did not test, post a link or something.

https://yadi.sk/d/vfgML2Vm3TR4mw/x.xx/Win9x

Wow, nice. Thanks. I don't think I'll test all of these (at least for now) just because it would take me forever. I will add some of the really early ones and a few to fill in the gaps between versions.

Btw, any idea on the differences between Rage128-4.12.6281-win9x_633cd21.zip and Rage128-4.12.6281-win9x_633cd18.zip? The ATi driver versions are a bit confusing

Reply 15 of 16, by Takedasun

User metadata
Rank Newbie
Rank
Newbie
cxm717 wrote:
Hey, so I decided to do a small driver comparison on my Rage128GL card. I used my Aopen AX6BC mobo with 128MB of PC100, a Pentiu […]
Show full quote

Hey, so I decided to do a small driver comparison on my Rage128GL card. I used my Aopen AX6BC mobo with 128MB of PC100, a Pentium2@450MHz and a Creative Sound Blaster Live sound card.

I used mostly older games for this test but I am going to go back and test some newer games with these drivers also (probably some games from 1999 and 2000).

I am looking for more drivers. So if you have some that I did not test, post a link or something.

Here are the results:

alldrv128.png
cxm717 wrote:
Nemo1985 wrote:

This is very interesting, thank you for sharing the results!
By any chance do you have the links where to to download the drivers?

https://megaupload.nz/a4051bvbb4/drivers_zip

Wrong versions drivers.

r128w9x-6111a1k - 4.11.6105
r128_612cdh21_6162 - 4.11.6664
r128_630cdh34e_6219 - 4.11.6713

Reply 16 of 16, by cxm717

User metadata
Rank Member
Rank
Member

Where did you find the version info for those drivers? The numbers I used I found in the INF files. Many of these drivers I got from different places (like driverguide), some didn't come with any description or a readme. Each INF does have a date, I could list them by date.

Edit: Found Rage tweak shows driver versions. I'll update the graphs with the correct versions when I have time.