appiah4 wrote:

Is yours an OEM card?

It is. It's an IBM OEM card. FRU 22P0704

That's not that bad, actually it is about what I expected. 32MB version should be able to reach around 36FPS with those settings, and probably 40fps at 1024x768x16 which is roughly Voodoo 3 2000 levels of performance - that is a solid in my book. I have a 32MB version of this same card and look forward to testing it in my GX110 with a 1GHz P3 CPU myself.

Considering that, according to the Anandtech test, the G400 is able to hit 51fps with a P3-600 under Win98SE I would expect that the G450, which is supposedly only a tiny hair slower in games, would be able to muster something over 40fps, especially when paired with a faster processor. But... I'm going to run this again under 98se later tonight so I'll find out what's really going on.

To indulge my curiosity, could you test it in Q3 with the Radeon 7000 and post the results as well? 32FPS at 800x600x32 , more than I thought it would do.

Sure. I'm removing the Matrox stuff as I type this.

And this is the Radeon 7000's results on the same machine:

I kinda expected better out of it, maybe closer to 60fps, but then again I do have everything maxed except filtering (it's still bilinear).

Ah man, poor Matrox. Did you determine the fastest Matrox driver by trying them all? Perhaps the G400Max will fare better.

feipoa wrote:

Ah man, poor Matrox. Did you determine the fastest Matrox driver by trying them all? Perhaps the G400Max will fare better.

Honestly, I think it's the 2000/XP driver in general. I'm about to flip this thing to Win98SE in a few. I'm almost 100% positive it'll run a lot faster then.

damn... flipping through my parts and I ran across my 32MB Radeon SDR PCI card (aka 7200). Coulda swore I sold it off a couple of years ago. Less RAM than the 7000, but it's a faster car. That thing should hit 60FPS for sure.

misterjones wrote:

feipoa wrote:

Ah man, poor Matrox. Did you determine the fastest Matrox driver by trying them all? Perhaps the G400Max will fare better.

Honestly, I think it's the 2000/XP driver in general. I'm about to flip this thing to Win98SE in a few. I'm almost 100% positive it'll run a lot faster then.

In my testing I found the 582 driver was fastest (on the G450) in Quake3, I tried it on the G400max and the G450. At 800x600x32 I got 40.9FPS with my G450 AGP, with the 596 driver I got 36.5. I did find the G400max was much faster under 98 though. Especially with the turbogl driver.

Edit: My results were a bit different between the G450 and G400max. The best driver on the max was 584, on the G450 it was 582.

I'm a little confused. Could you clarify which was the best driver version in Quake 3 for the G450 and for G400Max under a) Win98SE, b) WinNT4, c)W2K, and d) XP? Were these driver revisions still the best under Quake 1 and Quake 2?

EDIT: I'm also curious what the driver revision outcome is for the G200, G550, and Parhelia.

Finally, the Radeon 32 SDR aka Radeon 7200 (as it's recognized as by the driver)

^^^ That's more like it!!! I was going to use the 7000 in my retro emulation machine, but this just makes more sense to me. It'll give better frame rates with ePSXe and Project64.

appiah4 wrote:

Even Matrox's site lists the card as having only a 32MB option.

I guess that's just worth as much as Matrox stating on their site that the 5.96.004 driver doesn't support D3D.

The 16 MB versions may be indeed OEM cards, as that's how most Matrox cards were sold anyway.

feipoa wrote:

I'm a little confused. Could you clarify which was the best driver version in Quake 3 for the G450 and for G400Max under a) Win98SE, b) WinNT4, c)W2K, and d) XP? Were these driver revisions still the best under Quake 1 and Quake 2?

EDIT: I'm also curious what the driver revision outcome is for the G200, G550, and Parhelia.

Here is my driver testing for the G400/G450 under windows 98: https://docs.google.com/spreadsheets/d/1OXrT5 … dit?usp=sharing Keep in mind that vsync was stuck on in Unreal and it was in D3d. OpenGL mode just crashes. I think it's because it is the GoG version of Unreal gold. I did use a refresh rate of 75Hz, which is the highest my panel supports.

As for the Parhelia, I didn't test the performance of all the drivers with Quake1 and 2. I did try those games on each driver. GLQuake seems to run fine with most drivers. Quake2 has problems with many of the drivers. The best drivers for Quake2 for the Parhelia are either 105, 106, 107 or 203. 105-107 run fastest while 203 looks the best imo. 203 was the first Parhelia driver that gave the game nice round particles.

I don't have a G550 atm so I can't test it, I do have a G200 and I do plan on testing it. My G200 is an OEM card with lower clocks and SDRAM though.

I also have not done thorough driver testing with NT4. I can confirm it has the same driver bug as 2k/XP in Quake3 though. Maybe they all use the same OpenGL ICD? I have done driver testing with the G450 under XP and Ive done some testing with Win2k. I could link them at some point. The Win2k testing is pretty much the same as XP though, it does use the same drivers.

You should explore Parhelia's fragment AA. I don't think I've seen anyone test that across a bunch of games.

cxm717 wrote:

feipoa wrote:

I'm a little confused. Could you clarify which was the best driver version in Quake 3 for the G450 and for G400Max under a) Win98SE, b) WinNT4, c)W2K, and d) XP? Were these driver revisions still the best under Quake 1 and Quake 2?

EDIT: I'm also curious what the driver revision outcome is for the G200, G550, and Parhelia.

Here is my driver testing for the G400/G450 under windows 98: https://docs.google.com/spreadsheets/d/1OXrT5 … dit?usp=sharing

Wow, that's a lot of work. The numbers aren't all over the place like I picturing. Seems like 552 might be best for OpenGL and 623 for Direct3D? Although, the difference between 552 and 623 in Direct3D is less than the difference with OpenGL, so 552 seems to be the winner. I'll have to add this mess to my infinite list of things to look into.

feipoa wrote:

cxm717 wrote:

feipoa wrote:

I'm a little confused. Could you clarify which was the best driver version in Quake 3 for the G450 and for G400Max under a) Win98SE, b) WinNT4, c)W2K, and d) XP? Were these driver revisions still the best under Quake 1 and Quake 2?

EDIT: I'm also curious what the driver revision outcome is for the G200, G550, and Parhelia.

Here is my driver testing for the G400/G450 under windows 98: https://docs.google.com/spreadsheets/d/1OXrT5 … dit?usp=sharing

Wow, that's a lot of work. The numbers aren't all over the place like I picturing. Seems like 552 might be best for OpenGL and 623 for Direct3D? Although, the difference between 552 and 623 in Direct3D is less than the difference with OpenGL, so 552 seems to be the winner. I'll have to add this mess to my infinite list of things to look into.

I believe 552 has TurboGL so it's OpenGL driver is rather prone to bugs and stuff?

appiah4 wrote:

feipoa wrote:

cxm717 wrote:

Here is my driver testing for the G400/G450 under windows 98: https://docs.google.com/spreadsheets/d/1OXrT5 … dit?usp=sharing

Wow, that's a lot of work. The numbers aren't all over the place like I picturing. Seems like 552 might be best for OpenGL and 623 for Direct3D? Although, the difference between 552 and 623 in Direct3D is less than the difference with OpenGL, so 552 seems to be the winner. I'll have to add this mess to my infinite list of things to look into.

I believe 552 has TurboGL so it's OpenGL driver is rather prone to bugs and stuff?

Yeah, the OpenGL ICD in 552 is pretty buggy with Quake3 and GLQuake.

I got Vsync disabled in Unreal btw and I updated the spreadsheet. I just had to set UseVsync to True 😕

I also did some testing in XP on a P4 in GLQuake and Quake2 https://docs.google.com/spreadsheets/d/1nbEPP … dit?usp=sharing

One interesting thing Id like to point out is the G450 overclocks pretty well. Stock clocks are 120/150, mine was fine up to 155/193

Edit: I had the drivers mixed up for Quake2 on the Parhelia. Driver version 203 has a bug that gives Quake2 huge particles. It was driver version 113 that fixes that

swaaye wrote:

You should explore Parhelia's fragment AA. I don't think I've seen anyone test that across a bunch of games.

I did try it in a few games. It looks really good. I haven't done any performance testing with it though, I should do that.

appiah4 wrote:

derSammler wrote:

misterjones wrote:

As it shows here as well as in the card properties screenshot, D3D is supported and enabled by the 5.96.004 driver.

Are you sure about the version of the driver?

There is a WHQL certified driver with the same version that has D3D support.

Just for the record: since I was using driver version 5.92.006 with my G450, which is the last to support D3D according to many pages on the web, I decided to give the newer version 5.96.004 a try. So I tested the WHQL one first - and no, 5.96.004 WHQL does not support D3D either! From all the various 5.96.004 driver packages, only the non-WHQL one with the file name "xp2k_596_004.exe" supports D3D. That driver has some bugs compared to the older 5.92.006, however. E.g. it's not possible to assign a color profile to a display, which works fine with 5.92.006. Also, despite having dxdiag pass all D3D tests, all software I tried so far tell me that they can not find a D3D capable device.

I am playing around with a AMD750 Slot A setup at the moment and busted out the G400 Max. It's stable at AGP2x on there which is nice. I tried the TurboGL with Sin, Quake 2 and Quake 3. It runs quite well and looks good. I couldn't figure out how to enable Vsync though. Games were configured for Vsync but it was still disabled. Even forcing OpenGL Vsync with the Matrox Technical Support Utility did nothing. The full OpenGL ICD in some of the final driver releases does respond to this though.

I noticed something interesting while trying to enable Vsync. The MTSU says the GL ICD is using block transfers by default, but to even have the option of Vsync you must switch it to page flipping. So that gives me the impression that Matrox was avoiding Vsync support in order to maximize their GL performance.

That wouldn't surprise me. I recall running into this issue with some cards, but forget which cards. I had assumed that VSYNC was stuck enabled, but perhaps it was in fact disabled to increase benchmark scores.

Which Irongate board did you do your testing on? When I setup my slot A system with the AMD 750 chipset, the G400 Max forced into 1x mode and I couldn't get it in 2x mode. Graphics card advice needed for Gateway 2000 Kadoka system w/1 GHz slot A Athlon

feipoa wrote on 2020-08-10, 21:07:

Which Irongate board did you do your testing on? When I setup my slot A system with the AMD 750 chipset, the G400 Max forced into 1x mode and I couldn't get it in 2x mode. Graphics card advice needed for Gateway 2000 Kadoka system w/1 GHz slot A Athlon

I have a FIC SD11 v1.7. It has the older chipset stepping without working Super Bypass so that's not available. I am using the AMD AGP Miniport 4.80 driver. Powerdesk 5.52 and 6.83. The G400 defaults to AGP 2x with sidebanding with all this.

You might want to try a older/newer version of the MTSTU to try to force 2x mode.
ftp://ftp.matrox.com/pub/mga/utils/tech_sup/2002/

I also have a GeForce 2 Ultra (modified Quadro 2 Pro) working at forced 2X mode. Sidebanding can not be enabled though or it freezes in games.