So is 7900 GTX the ultimate Doom 3 era faithful card?

I would say no. GF 7 series cards came out in June 2005, about a year after Doom 3.

According to Skyscraper (first post of this thread):

When it comes to Video Cards, PCI-E was the new hot thing and the top cards were the ATI X800 XT PE and the Geforce 6800 Ultra. The ATI X850 series only adds PCI-E interface to the X800 series so we can include those aswell. SLI and Crossfire were not available to the wide public in August 2004 so no dual action in the period correct list. (late Nov for SLI, early 2005 for Crossfire if I remember correctly)

silikone wrote:

So is 7900 GTX the ultimate Doom 3 era faithful card?

There is a less strict period correct list aswell. The less strict list is dominated by the 7900 GTX so far but that could be because there are lots of results posted with the 7900 GTX and none with the ATI X1950 XTX. I think these two cards should be very close to each other performance wise.

Skyscraper wrote:

I think these two cards should be very close to each other performance wise.

Old benchmarks speak for themselves, Radeon unsurprisingly falls behind in OpenGL when AA is off.

Pentium D 830
Because I used the stock HSF from a C2D E6600, I couldn't overclock any of my Pentium D CPUs. Heck, I had to undervolt this processor just to get it to finish the synthetic CPU tests without throttling. It's insane how hot these things get. Even when it's undervolted, the Pentium D runs hotter than a C2D E6600 overclocked to 3GHz! 😲

At 3GHz/800FSB, the Pentium D managed to beat my P4 520 at 3.73/1066. Doom III's dual core support works quite well!

Windows 7:
Pentium D 830 @ 3GHz/800FSB, GeForce GTX 560 (341.44), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, Win7 x86

Windows XP:
Pentium D 830 @ 3GHz/800FSB, GeForce GTX 560 (341.44), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, XP SP3

I may have problems with mathematics, but isn't 146.4 > 138.8 ?

Anyways, I ran the benchmark on my nearly finished socket 939 box with windows 7 and catalyst 10.2, and it's ... not good 😁
I'll install XP instead I guess, windows 7 was just to validate the hardware was functionnal before I can find my XP CD and do all the cable management and all.

Here it is :
Asrock 939 Dual Sata II, Athlon FX-55 SSE (0.09), Radeon X1950pro (AGP !), 2GB RAM DDR400, all at stock speed : 85.1 FPS.

Not quite period correct, but really close. I could go for a X850 XTPE AGP also, but I like my 1950 IceQ turbo.

What could be causing Windows XP to have such a large performance advantage?

silikone wrote:

What could be causing Windows XP to have such a large performance advantage?

It could be the drivers (7.10 --> 10.2).
It could be the AGP.

I'll have to compare both solutions and we'll see.
I could even get a 1950pro PCIe for a more complete comparison.
I should install XP this week.

edit : also note that very few drivers are available for vista and none for 7 for my motherboard (agp, sata...).
I'm running totally on windows 7 integrated drivers for everything. That could be another reason. Kernel times are a bit high.

sunaiac wrote:

It could be the drivers (7.10 --> 10.2). It could be the AGP. […]
Show full quote

silikone wrote:

What could be causing Windows XP to have such a large performance advantage?

It could be the drivers (7.10 --> 10.2).
It could be the AGP.

I'll have to compare both solutions and we'll see.
I could even get a 1950pro PCIe for a more complete comparison.
I should install XP this week.

edit : also note that very few drivers are available for vista and none for 7 for my motherboard (agp, sata...).
I'm running totally on windows 7 integrated drivers for everything. That could be another reason. Kernel times are a bit high.

The integradted drivers in Windows 7 are often pretty good, thats why they got chosen I guess 😀 While I do think some of the performance issues in Windows 7 are driver related I think its just because the drivers are Windows 6.x drivers, not because they are too new or too old. Doom 3 seem to perform more or less the same regardless of driver version but XP is always faster than Windows 6.x.

Im pretty sure the PCI-E X1950 Pro will score the same as the AGP version, perhaps a little bit worse as the PCI-E port on the Asrock board only runs in 4x mode if I remember right.

101.5 FPS, Skyscraper, A64 FX55 2.6GHz, A8N32-SLI Deluxe Nforce4. 2x1GB DDR475 CL2.5-3-3-5, Radeon X850XTPE 7.10, onboard audio, XP-SP3

85.1 FPS, Sunaiac, A64 FX55 2.6GHz, 939Dual Sata II, 2x1GB DDR400 CL2.5-3-3-6, Radeon X1950pro AGP 10.2, Audigy 2, 7 32bits

That's 20FPS, I really think that's too large a difference to explain with just the OS difference.
Do you have the opportunity to try 7 on your FX55 ?

edit :
http://www.pcstats.com/articleview.cfm?articl … id=1860&page=12 well 5 of those FPS can come from the chipset itself 😀
I'll have to check if I have funky EAX thingies activated too.
I'm really too lazy to install XP, plus my SSD won't like it 🙁

sunaiac wrote:

101.5 FPS, Skyscraper, A64 FX55 2.6GHz, A8N32-SLI Deluxe Nforce4. 2x1GB DDR475 CL2.5-3-3-5, Radeon X850XTPE 7.10, onboard audio, […]
Show full quote

101.5 FPS, Skyscraper, A64 FX55 2.6GHz, A8N32-SLI Deluxe Nforce4. 2x1GB DDR475 CL2.5-3-3-5, Radeon X850XTPE 7.10, onboard audio, XP-SP3

85.1 FPS, Sunaiac, A64 FX55 2.6GHz, 939Dual Sata II, 2x1GB DDR400 CL2.5-3-3-6, Radeon X1950pro AGP 10.2, Audigy 2, 7 32bits

That's 20FPS, I really think that's too large a difference to explain with just the OS difference.
Do you have the opportunity to try 7 on your FX55 ?

Im to lazy to install windows 7 on any of my K8 systems but to me your numbers look as expected.

Here is one of my systems, Win 7 64 vs Win XP 32, the memory settings are not exactly the same but very close. The GPU overclock gave next to nothing.

With new systems that have 25+ gigabyte per second memory bandwidth the difference is alot less but with all older systems even socket 775 the difference is huge.

Some more Pentium D goodness--this time with a PD-935. This one has twice as much cache as the 830 (4MB L2), is clocked 200MHz higher, and runs cool enough that I can operate it at stock voltage with the C2D's stock heatsink. No overclocking though. 😢

Actually, I'm kind of surprised at the results. These numbers aren't half bad. It still sucks at 3DMark 01/05/06, though.

Windows 7:
Pentium D 935 @ 3.2GHz/800FSB (4MB L2), GeForce GTX 560 (341.44), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, Win7 x86

Windows XP
Pentium D 935 @ 3.2GHz/800FSB (4MB L2), GeForce GTX 560 (275.33), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, XP SP3

I have got a system running a Pentium D 945 and love that cpu. It games nicely and is more than enough horsepower for the average user. Plus it does not get hotter than like 45C with my $20 HSF 😁

Looks like Doom3 is also sensitive to cache size. Here's the 1MB Pentium E2220, tested at 2.4/800 and 3.2/1066. Despite being based on Core 2, at stock speed it is only a few frames faster than the Pentium D 935, and is quite a bit slower than the K8-based Opteron 185, clock for clock. When overclocked, it easily leaves the P-D in the dust, but still can't catch up with the Opteron.

Windows 7:
Pentium E2220 @ 2.4GHz/800FSB (1MB L2), GeForce GTX 560 (341.44), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, Win7 x86

Pentium E2220 @ 3.2GHz/1066FSB (1MB L2), GeForce GTX 560 (341.44), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, Win7 x86

Windows XP:
Pentium E2220 @ 2.4GHz/800FSB (1MB L2), GeForce GTX 560 (275.33), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, XP-SP3

Pentium E2220 @ 3.2GHz/1066FSB (1MB L2), GeForce GTX 560 (275.33), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, XP-SP3

Well, I sourced an MSI mATX S775 mobo that has both W98 + XP drivers *and* supports dual-channel memory and more granular OC tweaking so I'll be able to compare demo times with my current single-channel BioStar P4M890-M7-TE setup used so far. The only other MSI board I ever had pretty much exploded on it's first boot up... We shall see...

My timedemo using kind of period correct hardware...

Configuration is:

- Pentium D 940 3.4GHz 4MB cache
- i865 Foxconn with AGP and DDR1
- X1950 PRO 256MB AGP by Sapphire
- X-Fi Xtreme Audio
- 2x512MB DDR400 kingston
- mushkin 650-XP psu

- XP Sp3 (tinyXP)
- Catalyst 10.02

First one windowed, 1280x1024 - ultra - using timedemo timedemo1.demo usecache

Second one in fullscreen (1280x1024) using timedemo timedemo1.demo without the usecache argument

Been contemplating on running it on my ROG G751JY but the results would be silly 😀

Well I guess Windowed 1280x1024 ultra is good enough for the list, I think its a little bit slower than 1024*768 ultra fullscreen.

I'll rerun @ 1024x768.

New top period correct "stock settings" score.

This system only beats the FX55 system because of the older Doom 3 version used. Once I manage to test the FX55 with the older version I expect it to regain the top position.

2x Opteron 248 2.2GHz Sledgehammer, MSI K8T Master2 FAR, 2GB DDR157 CL2-3-3-7, 6800Ultra 92.91, onboard audio, XP-SP3

1024*768 Ultra: 106.0 FPS

Ever wonder how well Doom 3 runs on a Titan X? I certainly do. Here's an FX5200.

Athlon 64 3700+ @ 2.64GHz, GeForce FX 5200 128MB/64-bit (81.98), 2GB DDR-400 CL2, Gigabyte K8NSC-939, onboard audio, XP SP3

640x480 Medium: 19.0 fps
1024x768 Ultra: 6.5 fps
1280x1024 Ultra: 4.5 fps