Geforce2 MX*

Rick6, this is brilliant!

Does Painkiller have a built-in benchmark?

That 6800 does very well, I thought the 7800GS would be ahead more.

Great Benchmarking Project and very unique as only Asrock had motherboards with Pci-e and AGP on them.

From Benchmarking I've done The 7800gs only pulls significantly ahead of the 6800gt in later titles, like 3dmark06 and Crysis, even though Crysis is practically
unplayable with it, maybe low settings would work.
The 7800gs G70 based has fewer vertex and pixel units than a 6800gt but clocks up a lot higher on the gpu.

Rick, are You sure the HD3850 AGP is 256mb, I thought they were all 512mb except for pci-e models.

It would've been neat if you had the same card in AGP and PCI-e.

I wonder how much quicker my 7900 gs agp (only paid $25 for it) is vs the 7800 gs, this also make me feel good about not having a 3850 or 4670 agp 🤣.
6800 GT is a great bang for the buck while people shovel out more money on the 7800 gs. Would love to see the 850 xt and x1950 pro agp in this list at some point 😀

leileilol wrote:

Geforce2 MX*

Nahh, GeForce 2 with MMX™ technology sounds waaay better! 🤣

No geforce 2 ultra? Curious how that would of ranked.

I always enjoy huge bar graphs like these that show many years of hardware in a single test.

I love how some of the tests show the mighty Geforce 2 GTS as bringing nearly 50% of the performance of an HD 7870. CPU\Engine scaling from older games is quite a mysterious thing... though the 4200+ is certainly a bottleneck for any post 2005 mid to high end graphics card. Would be interesting to see a few results with a significantly faster CPU... or a significantly slower one! Disable a core and downclock that 4200+ as far as it'll go. 😁

GF2 Ultra might be a bit rare now, but GF FX 5900 or Quadro FX 2000 / 3000 are relatively easy to obtain. That OC'd GF3 Ti was surely having a good time until Rick threw DOOM3 at it. 😀

Well done for dedicated the time to do this. I also have a PCIe / AGP board but just haven't had opportunity to pull off a similar test (go big or go home). I think it's probably still worth me doing for a number of reasons e.g. including additional graphics cards and using a faster Intel CPU. Thanks for the inspiration!

OS : Win XP SP3 (with all updates) ?

archsan wrote:

leileilol wrote:

Geforce2 MX*

Nahh, GeForce 2 with MMX™ technology sounds waaay better! 🤣

Haha, damn this classic mistake 😁
It's like that other mistake "DuEl Core CPU" like if there was going to be a deathmatch between them.

Imperious wrote:

Rick, are You sure the HD3850 AGP is 256mb, I thought they were all 512mb except for pci-e models.

You're right, it's a 512MB video card. After so many benchmarks it's easy to make some typos, thank you for pointing them out, it's all corrected now!

Thank you all for the feedback! I would also love to be able to add something from the Geforce 8/9 generation but my geforce 8800GT died, my Geforce 8400GS died during the tests, my 8800GTS 512 was in use on another computer and my 9600GT was with a friend. Something tells me that the Geforce 8xxx or 9xxx would be able to standout in most tests and 3Dmarks here. Maybe i'm wrong.

agent_x007 wrote:

OS : Win XP SP3 (with all updates) ?

SP3 yes, maybe not with all updates. I haven't connected that computer to the internet for a while now.

PhilsComputerLab wrote:

Does Painkiller have a built-in benchmark?

Yes! Type in the console "benchmark c5l2". There are also other benchmarks i believe. It should be either c5l1 or c5l0

I don't want to be a buzz-kill but did you verify with gpu-z that the pci-e video cards are actually running at 16x speed in this system? I say this because a high percentage (practically all) motherboards that offer both agp and pcie like this hybrid one usually run the pcie slot at 8x speed.. usually it's the only way the agp one can function in the same system.

PCIe Myth.

Man, Unreal Tournament is all over the place...DirectX works in mysterious ways on that game 😀

kithylin wrote:

I don't want to be a buzz-kill but did you verify with gpu-z that the pci-e video cards are actually running at 16x speed in this system? I say this because a high percentage (practically all) motherboards that offer both agp and pcie like this hybrid one usually run the pcie slot at 8x speed.. usually it's the only way the agp one can function in the same system.

I didn't check the speed of the pci-e slot but i expect to be the slowest possible since it's really old. Even so, with such old cpu, would newer card benefit from a faster pci-e in older games and benchies like these? Anyway, i'm going to install the ATI HD 4670 pci-e again and take a gpu-z shot just to make sure.

Munx wrote:

Man, Unreal Tournament is all over the place...DirectX works in mysterious ways on that game 😀

You mean UT99 or UT2K4? UT99 was running a opengl renderer. Anyway, i think that engine was capped at a certain point by the cpu, making a mess on the chart with the graphic cards. I mean, the Geforce 3 TI200 when overclocked was actually by nearly 12 fps than the ATI HD 4670 😁

EDIT: oh, and 20 fps faster than the AMD HD 7870...this engine makes no sense when it comes to benchmarking.

Weird how the Radeon HD7870 is almost universally beaten or matched by the 3850 & 4670, except in the final 3DMark. Wonder what's going on there...

Could driver optimisation make that big a difference?

kithylin wrote:

I don't want to be a buzz-kill but did you verify with gpu-z that the pci-e video cards are actually running at 16x speed in this system? I say this because a high percentage (practically all) motherboards that offer both agp and pcie like this hybrid one usually run the pcie slot at 8x speed.. usually it's the only way the agp one can function in the same system.

I don't know if this Asrock board is different, but on my 4core dual vsta (lga775), the PCI-e slot only does x4.
http://www.asrock.com/mb/via/4coredual-vsta/