Awesome update with great info! GJ Mate!

Thank you very much!

All the results that you have seen in the screenshots above are with Thoroughbred. However before Thoroughbred another core revision is available for the Athlon XP: Palomino. This core runs at 1.75V and the 2000 model with part number AX2000DMT3C has a TDP of 70W unlike the Thoroughbred 2000 that runs at 1.60V and has a TDP of 60W. It runs hotter and consumes more. The question is: performance is lower, better or the same?
To get that answer I tested the Palomino 2000 in all the 4 motherboards (KT7A – RAID; NV7 – 133R; KG7 – RAID; KR7A – 133R) and got all the results.
Before anyone asks: Palomino 2100 does not run in all the boards because it has a multiplier of 13X so I did not test it.

I will not put up the screenshots with the exception of the system info and the scores from the 3dmark 2000 on the KT7A – RAID which is indicative of the general results obtained in the rest of the tests on every single board:

3d mark 2000:
1024*768
1280*1024
1600*1200

The delta between Palomino and Thoroughbred is <3% in all the tests on all the boards. So from the performance point Palomino = Thoroughbred but the latter has the advantages of running cooler, requires a less complex and less noisy CPU cooler and consuming less power – that translates into longer lifespan for all the components.
I can not recommend any Palomino to be used as long as Thoroughbred is available at the same speed grade and does not require you to spend extra money. Sure, if all you got is a Palomino use it, otherwise buy a Thoroughbred and be done with it.

Now that we got the definitive answer to the question: which is the best platform for the 266MHz Athlons & Durons a question arises: does VIA Kt266A is enough to squeeze all the performance out of them or you need something more powerful? Maybe a newer board with a newer chipset could outperform the Abit KR7A – 133raid.
To put it more simply: does nforce2 - the best chipset for 400 MHz FSB Athlon's - is required to get the best performance out of Athlon XP 2000 or KT266A is enough? Do I need to spend serious cash on an nforce2 board or I can get something cheaper on an older chipset such as KR7A/KT266A?

This question will be answered in the next tests that will compare 3 different platforms: nforce2 ultra 400, VIA KT266A and nforce1. The reason I included nforce1 is to find out how much stronger is nforce2 over nforce1 and to see how much you gain going from the weakest DDR chipset to the strongest. Being the winner in the 266 MHz battle KT266A is automatically included.
The motherboards selected are:
Nforce1/415 – D: Abit NV7 – 133raid
VIA KT266A: Abit KR7A – 133raid
Nforce2 ultra 400: Abit AN7
All 3 systems will get the following common components:
CPU: Athlon XP 2000 part number AXDA2000DUT3C
RAM: 2*1gb DDR set at 2-3-3-6 in the BIOS set at 2-3-3-6 in the BIOS.
GPU: Leadtek geforce 7600 GT
In the case of storage I switched to the WD raptor 150gb on all 462 boards with a SATA connector so this will be used with the AN7.
NForce1 goes with a 160gb WD and KT266A with a Samsung 160gb, both IDE.

I will not put up pictures with the systems again; anyone can check previous pages where everything is described in detail. In case of AN7 I simply replaced the 3200 with the 2000 and that's it. Only the scores of the system with nforce1 + sound storm will appear in the results; for all intents and purposes nForce1 + Creative live is equal to nforce1 + sound storm (within 2%).
Only the screenshots with the results on nforce2 ultra will appear, the scores with the nforce1 and KT266A can be seen on the previous pages.
When comparing nForce1, KT266A and nforce2 the order will always be:
1. nForce1
2. KT266A
3. nForce2
We are going straight to the benchmarks and we start with Code creatures pro.

nd, are you planning on doing something similar with socket 478 platform?

Yes, but only after I finished with socket 462. I got all the Abit boards (including rare ones such as TH7II - raid or IC7 - max3 - there, I could not refrain myself 😀 ) with all the chipsets (minus SIS) and I plan someday to do a big performance test on socket 478. However I do not have any Pentium 4 EE, all I got is Northwood, Prescott 3.4 GHz, Northwood 3.06 Ghz FSB 533 and smaller speed grades.

Nforce1 is awarded a 100% score in order to keep the results comparable with those obtained on 266 MHz chipsets and KT266A and nforce2 scores are calculated based on that. If anyone wishes AMD760 and KT133A scores can be seen in the previous pages and compared to nforce2 ultra.

The results:
1. Code creatures pro 1024*768 default settings
2. Code creatures pro 1024*768 default settings + sound
There, I finally managed to put the correct screenshots! First time I put the results from the KT7A - raid by mistake!

3. Code creatures pro 1024*768 max settings
4. Code creatures pro 1024*768 max settings + sound

5. Code creatures pro 1280*1024 max settings
6. Code creatures pro 1280*1024 max + sound

7. Code creatures pro 1600*1200 max settings
8. Code creatures pro 1600*1200 max settings + sound

Well , nforce2 is not off to a good start! It is better than nforce1 in all the tests but KT266A manages to beat it in the lower resolutions!
1. Code creatures 1024*768 default:
nforce1: 100%
KT266A: 116.32%
nforce2: 106.12%
However in the test 8:
8. Code creatures 1600*1200 max + sound:
nforce1: 100%
KT266A: 110.13%
nforce2: 115.03%
So in test 1: 1024*768 default settings nforce2 results is 6% better than nforce1 but 10% slower than VIA KT266A!!
As the resolution increases so does the lead nforce2 has over nforce1: at 1600*1200 max settings with sound it has a 15% advantage!! Also it manages to beat VIA by 5%.

We continue with Aquamark3, a DirectX9 benchmark:
9.1024*768 default settings
10. 1024*768 max settings + sound

Aquamark 3
11. 1280*1024 max settings
12. 1280*1024 max settings + sound

Aquamark3
13. 1600*1200 max settings
14. 1600*1200 max settings + sound