nd22 wrote on 2023-06-01, 10:10:

appiah4 wrote on 2023-06-01, 09:10:

KT333 wins because Universal AGP.

That may be another point of view, very valid when taking into account Voodoo cards!

I'd say VIA and SiS, with universal AGP for 3dfx compatibility.
Both chipsets work fine with Geode NX processors, which is also a big plus. Cool as cucumbers and with software controllable multipliers.

stef80 wrote on 2023-07-10, 07:50:

nd22 wrote on 2023-06-01, 10:10:

appiah4 wrote on 2023-06-01, 09:10:

KT333 wins because Universal AGP.

That may be another point of view, very valid when taking into account Voodoo cards!

I'd say VIA and SiS, with universal AGP for 3dfx compatibility.
Both chipsets work fine with Geode NX processors, which is also a big plus. Cool as cucumbers and with software controllable multipliers.

As I said for 3dfx compatibility this is a very important point to take into consideration. However this is not the purpose of the tests which is to find the highest performing chipset from each generation for socket 462.
It seems I gained some experience as testing on KT600 platform is proceeding far faster than expected! When I first started testing with the first board - Abit AN7 - it took me 2-3 months to complete all tests now I am already half way through after 1 week. Abit KV7 is stable as a rock, has no incompatibilities - first 2 modules of ram that I put in worked - and as a general impression is a pleasure to work with!

I'd still say nForce2 Ultra 400 was the fastest socket A chipset, and with working async AGP/PCI lock.
KT600 used set of fixed ratios for locking, not sure about KT880.

Also fixed multipliers. Half of KT880 motherboards failed to actually lock anything.

You are both correct: nforce2 is the platform of choice when it comes to Athlon XP overclocking.
That being said I already use the best CPU for socket 462 and I do not intend to overclock any of my retro systems as long life and stability is paramount! In fact the only CPU I ever overclocked is Q9650 back in 2009 that is still running today at 3.6 GHz! Even my modern system with 2600K is running at stock speeds!
I am confident that by the end of the week I will finish testing Abit KV7 after just 2 weeks!

The Serpent Rider wrote on 2023-07-06, 19:31:

Majority of Thoroughbreds were unlocked from the box.

Some info:

1Q: Is it true all XP (Thorougbred) processors come unlocked from the factory?
2A: Yes, but there's a catch. Only the higher multipliers (> x12.5) are unlocked, NOT the lower ones. In order to unlock the lower multipliers (< x12.5) you must unlock your processor.
3
4* It has been reported that the latest EPoX 8RDA+ BIOS unlocks ALL multipliers, without the need for an unlock!
5* Epox EP-8K9A series boards will unlock the lower multipliers without having to modify the cpu/ZIF socket (submitted by Need4Speed)

If this is true than how comes I can not change the multiplier in BIOS to any of my CPU's ? The motherboard is AN7 - one of the best for socket 462.
Also how can I change the multiplier if the system refuses to POST if I put them in NV7-133R?

Because Athlon XP 2000 has not arrived yet testing the 266 MHz chipsets will have to wait!
I finished testing VIA KT600 chipset and I am going to present the results.
So now let's get back to the 400 MHz chipsets which represent the final stage for socket 462! Nforce2 and KT880 are already done so I will not put up the pictures again but I will present the situation for each benchmark with nforce2 set as the baseline - 100%: that is if in one test KT880 is faster it will have a score over 100% for example and if KT600 is slower will have a score under 100%.
The motherboard used is Abit KV7. I have both the KV7 and KV7-V; the only difference between them is the audio solution used: KV7 uses the VIA VT1616 codec while KV7-V uses Realtek ALC650. Because it was clear when testing KW7 that uses ALC650 that Realtek codec is of very poor quality it makes no sense to test it again!

Complete system specs:
CPU: Athlon XP 3200
RAM: Corsair 2*1gb
GPU: Leadtek geforce 7800gs
Board: Abit KV7
HDD: WD raptor 150gb
PSU: Enermax 365w
Cooler: Titan
Case Cooler master
Windows: XP SP3
Drivers: chipset drivers VIA 4 in 1 4.56v, graphic drivers forceware 175.19, audio drivers 7.00b
Impressions after using the system for 2 weeks: memory compatibility is excellent; I simply grabbed the first 2 sticks and laid my eyes on and it worked, later I put some Kingmax modules and it also worked; start up, shutdown and general system responsiveness is very good, no difference from AN7 or KW7; compatibility with modern power supplies is very good - tested with Corsair RM650 - however I used a period correct Enermax just to cycle through my old PSU's.

A few pictures with the components and the fully assembled system:

Summary from Everest and CPU-Z:

All tests are done just as usual at 3 resolutions: 1024*768, 1280*1024 and 1600*1200. All tests are repeated 3 times. First testing is done using default settings the following are done using maximum settings allowed by the game/benchmark with the exception of Vertical sync.
First test is Code creatures which utilizes DirectX8:
Resolution 1024-768 default settings without and with sound:

So now for the comparison with KT880 and nforce2 - average number of frames per second is what I compare!
1. Sound OFF
Nforce2: 70.4 fps - 100%
KT880: 64.6 fps - 91.76%
KT600: 66.3 - 94.17%
2. Sound ON:
nforce2: 69.0 fps - 100%
KT880: 58.0 fps - 84.05%
KT600: 65.0 fps - 94.20%
First 2 tests and it seems that VT1616 audio codec is making a difference! Sure, nforce2 with Sound storm is winning both with and without sound tests but the difference between Realtek and VIA is there!

Next is the 1024-768 resolution with maximum settings which in the case of Code creatures is only represented by AA. Without and with sound:

Now the comparison with KT880 and nforce2 - average number of frames per second is what I compare!
1. Sound OFF
Nforce2: 61.4 fps - 100%
KT880: 58.9 fps -95.92 %
KT600: 59.7fps - 97.23%
2. Sound ON:
nforce2: 60.9 fps - 100%
KT880: 55.9 fps - 91.78%
KT600: 59.6 fps - 97.86%
Nforce2 wins again but the margin is really small! VT1616 is indeed making a difference!

We are increasing the resolution to 1280*1024 and we get the following results without and with sound on Abit KV7/VIT KT600:

The comparison to nforce2 and KT880:
Sound OFF:
nforce2: 51.3 fps - 100%
KT880: 49.6 fps - 96.68%
KT600: 50.4 fps - 98.24%
Sound ON
nforce2: 50.8 - 100%
KT880: 47.5 - 93.50%
KT600: 49.9 - 98.22%
nFORCE2 wins this time too but, when compared to KT600, the result is within the margin of error!

Finally we arrive at the daddy of all resolutions during socket 462 era: 1600*1200! Without and with sound:

When compared to previous results, KT600 stands up pretty good!
Sound OFF:
nforce2: 42.8 - 100%
KT880: 42.2 - 98.59%
KT600: 42.5 - 99.29%
Sound ON:
nforce2: 42.6 - 100%
KT880: 41.5 - 97.41%
KT600: 42.6 - 100%
At this resolution the results are all bunched up close because even geforce 7800gs is starting to limit performance in this directx8 benchmark!
In conclusion Code creatures is won by nforce2 and despite the help provided by VT1616 audio codec KT600 can not overthrow NVIDIA!

Code creatures testing is done, we are continuing with Aquamark 3 - a directx9 benchmark.
Resolution 1024*768 default settings: