So XP+Version 1.0 is the proven best configuration for this?

The benchmarks here demonstrate about a 10% bump with Version 1, but either XP or W7 seem to work just as well. XP predominates because of all the period-correct systems being tested. I'll let the mavens speak though!

tincup wrote:

The benchmarks here demonstrate about a 10% bump with Version 1, but either XP or W7 seem to work just as well. XP predominates because of all the period-correct systems being texted. I'll let the mavens speak though!

For any system older than socket 1366 XP is much faster. The key seem to be memory bandwidh, if you have "enough" of it you dont take as much of a performance hit when using Windows Vista/7/8/10.

While Im not sure I do not think Doom 3 v1.1 and 1.2 are slower then 1.0 but 1.3 is definitely slower. I have used version 1.0 the last 5 weeks but all of my older results are with version 1.3.

I reinstalled v1 [from v113] and FPS jumped from 131.9 FPS to 147.9 - call it an 11% bump. Nuts the way the patched version is so much slower. And the patch doesn't appear to be critical for single-player mode any way, so why bother.

Anything Goes category
Timedemo [Doom3 v1] = 147.9
C2D E6700 2.66, BioStar P4M890-M7-TE/socket 775, EVGA 7900GTX [92.91], 2gb DDR2/pc4300 [single-channel], SBLive!, XP Pro SP3

That's single-channel ram. I tried a similar VIA 890 based MSI mobo that runs dual-channel but the project was another MSI fiasco. Something wrong with the sata/ide controller and no boot drive could be selected - at least it didn't explode haha.. But I got a refund on the board so it was just a waste of time, not time *and* money. But I'm really liking this E6700/7900GTX dual XP/W98 build.

Now we're cooking with gas! Here's the trash picked 775 system in its final configuration, with a Core 2 Quad and a fresh install of Win7 x64. This machine will probably handle basement HTPC duty.

I'm not entirely sure why, but the Quad seems to be much faster clock for clock than the C2D E6600, at least in Doom3. The E6600 @ 3GHz was only pushing 180.6 fps under Win7 and 228.5 fps under XP. I also noticed that the memory score in Windows Experience Index shot up after I installed the Quad (from 5.5 to 7.3), although that may be because Win7 x64 can actually use all 8 gigabytes of RAM (I was using the 32-bit version of Windows with the other CPUs I tested).

Core 2 Quad Q6700 @ 3.33GHz/1333FSB (8MB L2), GeForce GTX 560 (341.44), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, Win7 64.

Standard Def Steve wrote:

Now we're cooking with gas! Here's the trash picked 775 system in its final configuration, with a Core 2 Quad and a fresh install of Win7 x64. This machine will probably handle basement HTPC duty.

I'm not entirely sure why, but the Quad seems to be much faster clock for clock than the C2D E6600, at least in Doom3. The E6600 @ 3GHz was only pushing 180.6 fps under Win7 and 228.5 fps under XP. I also noticed that the memory score in Windows Experience Index shot up after I installed the Quad (from 5.5 to 7.3), although that may be because Win7 x64 can actually use all 8 gigabytes of RAM (I was using the 32-bit version of Windows with the other CPUs I tested).

Core 2 Quad Q6700 @ 3.33GHz/1333FSB (8MB L2), GeForce GTX 560 (341.44), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, Win7 64.

I think I will have to bench my Q6850@4.0 again as its behind your q6700 😁. I wonder why the patches makes Doom 3 so much slower...

Standard Def Steve wrote:

Now we're cooking with gas! Here's the trash picked 775 system in its final configuration, with a Core 2 Quad and a fresh install of Win7 x64. This machine will probably handle basement HTPC duty.

I'm not entirely sure why, but the Quad seems to be much faster clock for clock than the C2D E6600, at least in Doom3. The E6600 @ 3GHz was only pushing 180.6 fps under Win7 and 228.5 fps under XP. I also noticed that the memory score in Windows Experience Index shot up after I installed the Quad (from 5.5 to 7.3), although that may be because Win7 x64 can actually use all 8 gigabytes of RAM (I was using the 32-bit version of Windows with the other CPUs I tested).

Core 2 Quad Q6700 @ 3.33GHz/1333FSB (8MB L2), GeForce GTX 560 (341.44), 8GB DDR2-800 5-5-5, Asus P5N-E SLI (nForce 650i), X-Fi Ti, Win7 64.

This is pretty close to the FX-8350 @ 4.8/8gb PC2133 I posted (269.4, though Doom3 v1 could push this up 10%). Though I've leaned on AMD for years on a price basis this highlights the pretty serious clock for clock superiority of the intel part.

I was able to wring a little more performance out of the q6700 by tightening up the memory timings. Turns out that 667MHz @ 4-4-4-12-1T is faster than 800MHz @ 5-5-5-12-2T.

Core 2 Quad Q6700 @ 3.33GHz/1333FSB (8MB L2), GeForce GTX 560 (341.44), 8GB DDR2-667 4-4-4-12-1T, Asus P5N-E SLI (nForce 650i), X-Fi Ti, Win7 64.

I did finally test a Pentium 3 system.

The BIOS that let me run AGP 4X overclocked wont overclock as far as the latest official beta BIOS. The difference between AGP 2X and AGP 4X is small so its not a huge issue. Steves system with DDR memory is faster but if only someone could mod the better CUSL2-C BIOS to enable AGP 4X with overclocked settings I think I would get close 😀.

1024*768 Ultra

Pentium III 1400-S @1575MHz, CUSL2-C i815, 512MB PC133@150MHz CL2-2-2-7, Geforce 6800GT 92.91 AGP2X, SB Live, XP-SP3.

Pentium III 1400-S @1575MHz, CUSL2-C i815, 512MB PC133@150MHz CL2-2-2-7, Geforce 6800GT 92.91 AGP4X, SB Live, XP-SP3.

Pentium III 1400-S @1680MHz, CUSL2-C i815, 512MB PC133@160MHz CL2-2-2-7, Geforce 6800GT 92.91 AGP2X, SB Live, XP-SP3.

Stumbled on an interesting ECS P4 s775 mATX board [P4M890T-M v2] with dual channel ram support and pretty robust OC options. When I get the chance, hopefully soon, I'll swap motherboards with the Biostar E6700 build and see what I come up with demo wise. I'm interested in comparing results for single vs dual channel ram.

Skyscraper wrote:

I did finally test a Pentium 3 system.

Ha! Nice. 1.68GHz...man I wish my Advance 12T would give me FSB options higher than 150MHz.

Not period correct, but I'm working on a Athlon 64 project, so took the opportunity to benchmark Doom 3 😀

Motherboard: http://www.asrock.com/mb/NVIDIA/K8NF6P-VSTA/
CPU: S754 AMD Athlon 64 3400+
2 GB DDR 400
GeForce 7900 GTX
X-Fi Titanium PCIe

XP SP3, Steam version of Doom 3, latest Nvidia GeForce driver, latest Nvidia nForce 4 driver, latest Creative X-Fi Titanium PCIe driver

just for fun I had a go on my 8400GS PCI 33MHz(g98 with some OC) with a 2.8GHz core 2 quad CPU
1024 ultra did it in 41.8FPS (v1.31, windows 7)

GPUz recorded 440MB of vram being used, GPU usage over 95% all the time and bus interface load around 15-20% (the measurement is not accurate because the card uses a pci to pcie bridge chip, and I think this value is of PCIE x1 as 100%, I've never seeing this usage going higher than 44% and that was with a opencl bus test, gaming is normally at 33% max), CPU usage went around 33%, since 25% is one full core, I guess the game took advantage of having more than one,

edit: since my result was added let me give the full specs

25. 41.8 FPS, SPBHM, Xeon E5420@2.8GHz, Asrock G31M-VS2, 2GB DDR2@750 c5, Geforce 8400GS G98 PCI@680/580, onboard ALC662, Win-7 x86

and let me add
20.0 FPS, SPBHM, Pentium 3 650@866, Asus CUBX-E, 512MB PC133 C3, Geforce 8400GS G98 PCI@680/580, AWE64, Win XP SP3

Back in 2007 when I was using an Opteron 185 as my main system, adding a 2nd 8800GTS barely improved the performance of most games, even at 1600x1200. Fast forward to 2015, throw more processor ponies at the situation, and SLI pretty much doubles the performance, even at ridiculously low resolutions like 1024x768.

i7-4930K (4.5GHz/1.25v), 32GB DDR3-2133 9-11-11-1T (1.55V), Asus Rampage IV Gene, onboard audio, Win7-64

Single 8800GTS 640MB (341.44):
1024x768 Ultra:

8800GTS 640MB SLI
1024x768 Ultra:

Here's your first Phenom result. I just got my AMD Phenom II x4 940 up and running with XP Home SP3. Using 332.21 (the earliest on Nvidia's website) with a 9800 GT. 4GB of DDR2 667. I don't know how to display memory timings, so I took a photo of my BIOS screen. Doom 3 is retail version 1.0.1262.

some more non period correct hardware (I actually don't have anything left from 2004-2005 I think unfortunately)

233.6 FPS Core i5 2310, Pegatron IPMSB-H61, 8GB DDR3 1333, HD 5850, onboard ALC662, Win XP sp3
this i5 performs more or less the same as the CPUs from when the card was launched (like Nehalem with a few hundred MHz more)

on windows 8.1 with updated drivers it was a little slower (like 210 fps)

110.7 FPS, Xeon E5420@2.8GHz, Asrock G31M-VS2, 2GB DDR2@750 c5, HD4670 DDR2, onboard VT1705, Win10 x86
using all default windows drivers...
apart from the OS it's basically a PC from 2008 with midrange-lowend graphics (with the unfortunate DDR2 version of the VGA, less than half the bandwidth from a standard 4670...)

all using 1.31

Nice scores 😀

The list should be (more or less) up to date.

A Willamette P4 is not a good CPU for running Doom 3...

1024*768 Ultra: 31.3 FPS 😁

P4 Willamette 1900@2375MHz, Asus P4B i845, 512MB PC133@166MHz CL3-3-3-7, Geforce 6800GT 92.91, onboard audio. XP-SP3

Skyscraper wrote:

A Willamette P4 is not a good CPU for running Doom 3...

1024*768 Ultra: 31.3 FPS 😁

P4 Willamette 1900@2375MHz, Asus P4B i845, 512MB PC133@166MHz CL3-3-3-7, Geforce 6800GT 92.91, onboard audio. XP-SP3

You know what sucks even more than a Willy P4?
A Northwood Celeron. 25.2 fps at 2.6GHz, and that's with DDR! 🤣

I wonder how poorly a Willamette Celeron would perform. 128k of cache *shudders*