Check the clocks on your card and have you tested the card in another rig? Sometimes cards will perform oddly especially when paired with an older system.

I'll need to swap out with the P4 board (though I seriously need to get that board's caps replaced soon) and try. Even if the P4 is 1.6GHz Wiliamette core.

Can you post 3DMark configuration? You might be CPU limited, as a FX 5700U is supposed to be paired with a Northwood-B/T-Bred AXP when it came out.

My FX 5500 PCI (not PCI-e), gets almost 6300 points in 3DMark 01 SE (with Xeon E5440 @ 3,4GHz)...
So yes, you probably have a problem.

NightSprinter wrote:

I just want to know if the FX5700 Ultra is supposed to do poorly in 3DMark99-2001? In each one, I seem to be getting well under […]
Show full quote

I just want to know if the FX5700 Ultra is supposed to do poorly in 3DMark99-2001? In each one, I seem to be getting well under 6000 for my score.

Specs are:
CPU - Intel Coppermine Pentium III-800 Slot-1
Motherboard - Intel Seattle-II 440BX

That result is perfectly fine for a 800MHz PIII running on a 440BX board. The CPU and motherboard are pulling you back. A lot. In fact you'd get a better score on that platform with a Geforce 4 Ti 4200.

This happens when you use a fast video card in an older / slower machine.

agent_x007 wrote:

My FX 5500 PCI (not PCI-e), gets almost 6300 points in 3DMark 01 SE (with Xeon E5440 @ 3,4GHz)...
So yes, you probably have a problem.

Yes, the 800MHz p3 is the problem.

3dmark is quite CPU-bound.

@NightSprinter - install that video card on a socket A barton or a pentium 4 and you will get twice that score in 3dmark01.

Yeah, though aside from my Voodoo3, all of my other GeForce cards all use SDR memory instead of DDR (even my GF4 MX440 has SDR memory).

NightSprinter wrote:

Yeah, though aside from my Voodoo3, all of my other GeForce cards all use SDR memory instead of DDR (even my GF4 MX440 has SDR memory).

I'd say leave the 5700 ultra in there for now, and replace it with something a bit older (say a GF3 ti200 or a GF4 Ti 4200) if the opportunity arrives. It's kind of wasted on a 800MHz P3, but there's no harm in using it on that machine if you don't need it anywhere else.

The voodoo 3 would do fine in that machine as the 800MHz P3 can't really push high resolutions (1600x1200) in 1999 games at enjoyable framerates - but I guess it really depends on what you want to run on your machine.

Good luck with your build!

I appreciate all the help folks. One reason I got the 5700 was due to a lack of inexpensive GF3 and GF4 Ti cards at the time. Plus with Expendable Rearmed no longer compatible with my Shield Tablet (Android 6.0 and newer gives an error regarding the Marmalade SDK), with the hacked patch this was the only way I could play that game with EMBM turned on. That, plus the DOS compatibility.

Which driver are you using? I'd use something 56.64 or older. Beyond that and I get freezes with some NV cards on 440BX.

The main problem with 5700 is you can't use driver 45.23 or 44.03 which are often the best drivers for Direct3D 7 and older games. With a 5200/5600/5800/5900 you can use those older drivers. 5700 came out later.

I've used everything from 56.64 to 9x.xx with no difference in performance. Though strangely when doing a benchmark of Quake III, the objects/projectiles/gunflares either lose textures or have glitches. This issue happens on all driver versions used.

I swear, ya gotta be kiddin' me, computer. I swapped out the FX with my old Hercules 3D Prophet II MX (card I bought for my first gaming computer ever, which was a budget machine), bumped core/mem to 200/190, and 3DMark99 is giving me a score of over 7,000. Maybe you guys WERE right on this. Shame GeForce3 cards aren't really cheap anymore.

swaaye wrote:

The main problem with 5700 is you can't use driver 45.23 or 44.03 which are often the best drivers for Direct3D 7 and older games. With a 5200/5600/5800/5900 you can use those older drivers. 5700 came out later.

The 45.23 are also my go-to drivers. Older drivers on certain cards can be even faster, but the 45.23 are a nice balance of speed and everything working.

Also consider or take a look at ATI cards. I ditched a GeForce3 in my current Socket A project because a 9600 XT draws almost 20 W less thanks to a smaller process. It also has more memory and supports higher versions of DirectX I believe.

I have a Radeon 9250, but I am also considering some DOS support too. I can re-bench with it later today if you want, Phil. I don't really intend to run DX8 games (still haven't broken the shrinkwrap on my AquaNox jewelcase), that I can rebuild my AMD system with a Barton 2800+ and the FX for that. This system is meant to run up to the best of the DX7 and OpenGL 1.3 titles, with the V2 as a backup for games like FF7&8.

NightSprinter wrote:

I have a Radeon 9250, but I am also considering some DOS support too. I can re-bench with it later today if you want, Phil. I don't really intend to run DX8 games (still haven't broken the shrinkwrap on my AquaNox jewelcase), that I can rebuild my AMD system with a Barton 2800+ and the FX for that. This system is meant to run up to the best of the DX7 and OpenGL 1.3 titles, with the V2 as a backup for games like FF7&8.

That narrows it down then 😀 I think the MX GeForce cards are a good fit in that case.

I do wonder theoretically, how well a GF3 or GF4 Ti would perform on an 800MHz P3.

NightSprinter wrote:

I do wonder theoretically, how well a GF3 or GF4 Ti would perform on an 800MHz P3.

You could try looking here.
I didn't find a list of all the results, but some GF3 and GF4 results are probably included here.

Sounds like we need more benchmarking threads! I like the the Doom3/Quake3 thread where its divided into the few years that came after its release.

Here are some example numbers using a Radeon 8500 64MB (275/300 core/mem), which is GeForce 3 Ti500 level.

System specs
Asus CUBX
P3-600E
512MB PC133

Asus P4S533
P4 Northwood
2666/133
1024MB DDR-333

3DMark2000
1024x768x16
P3-600, 4240
P4-2.66, 11918

1024x768x32
P3-600, 4222
P4-2.66, 10785

3DMark 2001SE
1024x768x32
P3-600, 4443
P4-2.66, 9961

NightSprinter wrote:

I swear, ya gotta be kiddin' me, computer. I swapped out the FX with my old Hercules 3D Prophet II MX (card I bought for my first gaming computer ever, which was a budget machine), bumped core/mem to 200/190, and 3DMark99 is giving me a score of over 7,000. Maybe you guys WERE right on this. Shame GeForce3 cards aren't really cheap anymore.

Sounds about right 😀

I keep saying that newer video cards in older machines will under-perform and slow down the machine. You should make a new thread with screenshots and show your findings.

I remember seeing someone here post benchmarks between various old GeForce series. I have no idea when that was but it could have been years ago. The FX series might have more driver overhead that impacts it with a CPU like a Pentium 3. A test of various 3dfx, GeForce and Radeon series across old CPUs at lower-ish resolutions would be interesting from a driver overhead perspective. This may already have been done though. 😀 Forums sort of absorb useful data and it can be hard to dig up again.