mwdmeyer wrote:

I really should write you a little php script to make it easier! One day, I've been meaning todo that, never any free time!

At first I wanted to put it all into a giant spreadsheet, but then I wanted individual charts and I couldn't do this with Excel, through I'm sure there is a way.

So I'm doing it the hard way, but it's easy to do. I will see how the charts turn out once the number of cards increases...

I would like to bench my machine (signature) too. I will always bench in 1024x768@16Bit.
Have you got the Benchfiles ?

dr.zeissler wrote:

I would like to bench my machine (signature) too. I will always bench in 1024x768@16Bit.
Have you got the Benchfiles ?

I don't have benchfiles. It's all original game CDs.

Video review of the TNT2 range of cards is up!

Nvidia Riva TNT2 series review and benchmarks

What do I think?

I really like the TNT2 cards. Great image quality, 16 and 32 bit colours support, lots of memory (32 MB), supports high resolution textures. The drivers work fine, didn't have any issues. For overclocking coolbits or PowerStrip work great.

In terms of performance the TNT2 works great for older Windows 98 SE games but struggles once you crank up the resolution or try newer games. The charts have been updated on my website: http://www.philscomputerlab.com/pentium-4-306 … benchmarks.html

I used driver version 29.42

Why no 1024x768@16Bit for 3dmark99max ?
btw. my PII-333/V3-3000-AGP 1027x768@16Bit 2700 3Dmarks.

Default setting used.

3dfx Voodoo3 2000, 3000 and 3500 benchmark results

Finally got around to updating the benchmarks: http://www.philscomputerlab.com/pentium-4-306 … benchmarks.html

Trading blows with the TNT2, in some games Nvidia has the edge, in others 3dfx is ahead.

Something is holding back the TNT2s, they can do significantly better.

I have to agree to Putas with his point.

In Q3A the TNT2 really should be faster, imho.
On the other hand, all 3DMarks and both GLQuake + Quake2 results are in line with my own benchmarks, which I performed few years ago. http://hw-museum.cz/benchmark-2-1.php

havli wrote:

In Q3A the TNT2 really should be faster, imho.
On the other hand, all 3DMarks and both GLQuake + Quake2 results are in line with my own benchmarks, which I performed few years ago. http://hw-museum.cz/benchmark-2-1.php

Q3A is the only game (it's actually the demo), were I set the graphics quality higher than the default. I wanted one rather demanding test in there for faster cards.

Apart from that, it's a SIS chipset, default BIOS settings, no tuning. And all cards are benchmarked on the same system. When / if I ever go through all cards I have planned, I will redo everything on a different system 😀

At least for the V3, the results are also in line with previous benchmarks I've conducted. But I haven't done any TNT2 benchmarks before, so this is a first.

But will be cool to see how things change on other processors.

Havli, BTW great site. I often use it to compare and I admire the work you put in to get all these results. Very impressive.

Well, they can do slightly better, even on weaker system. Expendable is also quite suspicious, maybe the driver version is not so great for the game? It is fine to not do tuning, but perhaps not everything is working right with the TNT2.

Putas wrote:

Well, they can do slightly better, even on weaker system. Expendable is also quite suspicious, maybe the driver version is not so great for the game? It is fine to not do tuning, but perhaps not everything is working right with the TNT2.

You keep saying this, but provide nothing to go by. Facts please. Suspicious how? Newer drivers do not work with the TNT cards, produce a black screen in D3D games. Can you test with the same driver and see what results you're getting? Ideally a similar P4 machine.

Define "slightly" better?

And remember, all cards are run on the same system. So any tweaks would just boost all cards. This benchmarks show relative performance, I would worry about absolute performances as there are too many factors at play.

Differences between systems will be covered in the future, just be patient. With V2 for example I know that a fast Pentium III is faster than most Pentium 4, just has to do with the platform.

I've tested the TNT2 M64 a bunch before and your results look inline with what I've seen for Quake 3. TNT2 Ultra looks okay too.

http://www.vogonswiki.com/index.php/3dfx_Benchmarks (down the bottom).

mwdmeyer wrote:

I've tested the TNT2 M64 a bunch before and your results look inline with what I've seen for Quake 3. TNT2 Ultra looks okay too.

http://www.vogonswiki.com/index.php/3dfx_Benchmarks (down the bottom).

Thank you!

Anyway, question to the group:

What should I benchmark next?

I'd like to do Matrox G400 / MAX next, but also thinking about ATI. I only have Radeon 7000 and 7500 DDR and I think this card is a bit faster than the current cards. The next slower card I have is a Rage 128 Pro. I am not that familiar with the ATI cards, so if someone can give me a breakdown that would be awesome.

I definitely want to get a Radeon 8500 and I have a few 9000 series cards too.

G400 would be good. Would still love to see a Voodoo 4 😁

Geforce SDR/DDR would be nice too.

There are so many cards. 😁

mwdmeyer wrote:

G400 would be good. Would still love to see a Voodoo 4 😁

Geforce SDR/DDR would be nice too.

There are so many cards. 😁

I have all of these 😁

But I want to cover a wide range of cards of similar performance first, then go up / down from there if that makes sense.

I'm very familiar with the G400, but feel like ATI needs some love too 🤣

What can people tell me about the Rage 128 Pro? I have a 32 MB version.

Rage 128 Pro should be TNT2 speed I think, pretty fast, drivers are average I believe.

G400 I think is a bit faster than the TNT2, was a good card.

Voodoo 4 is similar performance to TNT2/Voodoo3 😁

What about a Savage 2000?

Got Savage4 and Savage2000, but drivers were always an issue. I have a Inno3D and got it only going with the drivers from the Vogons driver archive. I really should get the Diamond versions of these.

Rage 128 Pro on TNT2 level? Nice! The card never seems to be used in retro builds around here. It's always 3dfx and Nvidia. That could be interesting to see what's going on there 😁

Yeah not sure why. Rage 128 was used in Dell P4s a lot. I don't think the output quality is as good, but 32bit performance doesn't drop as much as the TNT2. Been a while, I might not be correct about that.