After around 60 HDtune and HDtach tests here are the results: in the HDtach screenshot you can see my results are not far from the original one form 2004 - please note that the Silicon Image 3112 controller is connected to the rest of the system over the PCI bus which is limited to 133mb/s - if any other member has the 74gb Raptor I expect his results, maybe he tested under better conditions:

kolderman wrote on 2021-03-09, 06:26:

Eh, supporting v5 more important than all that stuff.

Gimme socket 462 with 3.3v agp and 12v aux atx power and I'll give you my right leg.

KT333 is best chipset as it supports the former and ddr400, not sure if ever put on mobo with aux power though.

MSI KT3-Ultra says Hi.
I thought such boards didn't exist, you can imagine my surprise when I ran across this one.

pixel_workbench wrote on 2021-04-16, 19:31:

kolderman wrote on 2021-03-09, 06:26:

Eh, supporting v5 more important than all that stuff.

Gimme socket 462 with 3.3v agp and 12v aux atx power and I'll give you my right leg.

KT333 is best chipset as it supports the former and ddr400, not sure if ever put on mobo with aux power though.

MSI KT3-Ultra says Hi.
I thought such boards didn't exist, you can imagine my surprise when I ran across this one.

This sounded familiar...and yep I bought one back in 2018 according to my ebay emails. Not even sure where it is or if I have it in a PC. Truly a rolls royce of motherboards.

pixel_workbench wrote on 2021-04-16, 19:31:

kolderman wrote on 2021-03-09, 06:26:

Eh, supporting v5 more important than all that stuff.

Gimme socket 462 with 3.3v agp and 12v aux atx power and I'll give you my right leg.

KT333 is best chipset as it supports the former and ddr400, not sure if ever put on mobo with aux power though.

MSI KT3-Ultra says Hi.
I thought such boards didn't exist, you can imagine my surprise when I ran across this one.

My first AMD Board.

I have a KT3 Ultra2 but despite a recap it's not POSTing. I believe it actually fried a few of my Socket A CPUs too 🙁

Already completed 20 PCMark04 and 05 tests - results are not earth shattering and it seems that Silicon image 3112 chip which uses PCI bus is not a limiting factor - 3 generation Raptor exceeds the 7200 RPM drive used in the system but not by leaps and bounds even under the best circumstances - I am disappointed, I sincerely expected more, it seems that access time is the only big advantage that Raptor brings to the table! I can only imagine that 1 and 2 generation Raptor were even slower!

nd22 wrote on 2021-04-28, 09:00:

Already completed 20 PCMark04 and 05 tests - results are not earth shattering and it seems that Silicon image 3112 chip which uses PCI bus is not a limiting factor - 3 generation Raptor exceeds the 7200 RPM drive used in the system but not by leaps and bounds even under the best circumstances - I am disappointed, I sincerely expected more, it seems that access time is the only big advantage that Raptor brings to the table! I can only imagine that 1 and 2 generation Raptor were even slower!

Linear transfer rate tends to be pretty consistent across most hard drives of a similar age. Access time is the thing that varies the most between quiet slow drives and fast noisy ones.
Access time is usually the more important factor in real usage unless working with huge files.

Here are the PC Mark 2002 & 2004 & 2005 results - I did only the HDD tests and nothing else - the results are very, very close to those done by various review sites back in the day:

I consider that the ultimate socket A platform tests are not finished so I will include new ones as I make them but for now I decided to continue with socket 462 from a different angle: instead of choosing the end of the era I decided that I will go with the beginning of socket 462.
The end goal is to cover the whole socket 462 from a performance point of view using the best combination of hardware and software that was not available back then and find the best platform for each stage of socket 462!
What is a stage? I divided the socket 462 era in 3 stages. The criteria is the FSB supported by each chipset.
Stage 1: FSB 200/266 - The beginning !
Chipsets:
1. Via KT133/133A
2. AMD 760
3. VIA KT 266/266A
4. NVIDIA 415/420
All 4 were available in 2000-2001, all support processors with a maximum FSB of 266, all support AGP 4x which is more than enough even for geforce 7800gs; all support only PATA drives; KT133/133A support SDRAM, the rest support DDRAM only.

Stage 2: FSB 333 - The climb!
Chipsets
1. Via KT333
2. Via KT400/400A
3. nforce2
I am unsure that KT333 actually supports processors with FSB 333 because the 2 boards I have do not support it officially! All 3 were available in 2002-2003, work with AGP 4x/8x and with PATA drives only.

Stage 3: FSB 400 - Reaching the peak!
Chipsets
1. VIA KT600
2. VIA KT880
3. nforce2 ultra 400
All available in 2003-2004; AGP 8X, native SATA or as a third party chip. This is mostly done, but I will add any future tests if requested! The winner of this stage is nforce2 ultra!

To keep things simple I set some rules:
1. CPU: the lowest denominator that is the processor supported by all boards in each stage will be used!
2. RAM: equal quantity in all the boards used; the only problem here is that KT133/133A supports up to a maximum of 1.5gb of ram otherwise the rest can use up to 2gb with no problems!
3. GPU: at each stage the video card will be the same; if possible I will use the geforce 7800gs with all the boards!
4. HDD: I do not have identical drives but I was thinking that I can use the same drive and reuse it after all tests are completed on a specific platform! Maybe some disk imaging software can help!
5. Monitor: Eizo S2100 with 1600*1200 max resolution will be used!
6. All 48 tests already used at stage 3 will remain; any further test will be add based on your recommendation!

Motherboards to be used - all Abit boards from my collection:
Stage 1: FSB 200/266 - The beginning !
1. Via KT133/133A: KT7-raid; KT7A; KT7A-raid
2. AMD 760: KG7-raid
3. VIA KT266/266A: KR7A; KR7A-Raid
4. NVIDIA 415: NV7-133R

Stage 2: FSB 333 - The climb!
1. VIA KT333: KX7-333, AT7-MAX
2. VIA KT400/400A: AT7-MAX2; KD7A
3. nforce2: - unknown, any suggestion appreciated!

Stage 3: FSB 400 - Reaching the peak!
1. VIA KT600: KV7; KV7-V
2. VIA KT880: KW7
3. nforce2 ultra 400: NF7-S 2.0, AN7, NF7-S2; NF7-s2G.

Amazing!!

Today, after 2 years, I come back to continue the topic.
The battle for the best 462 platform is now at stage 2: chipsets supporting processors with 333 MHz FSB; that is Athlon XP Thoroughbred B, Thorton and Barton.
I got all Abit boards for socket 462 and I have managed to perform compatibility testing on all of them. Despite having manuals and internet resources this was a gigantic task: most of the info I found is about 75% correct and solving basic problems took weeks; that plus the life that got in the way made the stage 2 becoming unnecessary long – 2 years long!
It also became apparent that stage 1 must be divided in 2 because processor support varies a lot between different chipsets even if they support 266 MHz FSB; but let’s not anticipates and for now let’s focus on FSB 333 MHz.
3 different chipsets, all from VIA, support 333 MHz FSB:
1. VIA KT333
2. VIA KT400
3. VIA KT400A
First question would be: but KT333 does not support 333 MHz FSB processors! Yes it does, according to Wikipedia and the web there are 3 revision of the KT333 chipset: revision CD which is the first and which one supports only 266 MHz FSB processors; revision CE which support 333 MHz FSB processors and lastly revision CF which is a rebadged KT400 and also supports 333 MHz FSB processors. Revision CF is eliminated because testing a KT400 with AGP 8X disabled would not made for a fair comparison. Revision CD is used in the first boards released at the very beginning when VIA manufactured the first batches. The one that is actually important and the “true” KT333 is the revision CE. How can I know which revision you got? If you take the fan or heat sink and clean with isopropyl alcohol the chipset you will see the revision as the last 2 letters after the part number in plain view.
Second question would be: where is nforce2? Nforce2 is already the winner in the 400 MHz FSB processors category. Nforce2 – the version released in 2002 – is actually the initial version and the nforce2 ultra 400 released in 2003 is a revision of the original one and not a brand new chipset therefore this would not be a fair comparison. Also when testing with the same CPU, RAM, GPU but with 2 different motherboards – Abit NF7 – S 1.2 and NF7 – S 2.0 the scores in 3dmark 2001 are identical!
Third question would be: if nforce2 is not included then why KT400A is? All testing have already been completed on the KT400A board before initial testing on NFS – S 1.2 and 2.0 so I choose not to discard the results but to keep them. Most reviews done by various sites back in the day did consider KT400A a revision of the KT400 so this will be taking into account in the final conclusion. But another thing to consider is that the part number of the KT400A is slightly different from the KT400: VT8377A versus VT8377.

Now the motherboards available and what I used:
KT333: Abit KX7 – 333; KX7 – 333R, AT7 – MAX. I used AT7 because it has the highest performance out of the all 3 boards in the initial testing – please note that this is the first AT7 – MAX and not AT7 – MAX 2!
AT7 – MAX and the IT7 – MAX are the first motherboards without legacy connectors with the exception of the floppy disk connector! This means no serial and parallel ports which is not a problem because they are not needed anyway. However the AT7 – MAX also lacks PS/2 ports! This is a critical port, badly needed in retro systems! Until I got this board I had no idea how important is to have keyboard PS/2 connector present. The board does have a USB 2.0 controller from VIA, VT6202, however any keyboard connected to it can not be used when the system initializes to access the BIOS. It has to be connected to the USB 1.1 ports and finding a keyboard that actually works was a nightmare! After 3 months and dozens of keyboards tested, including period correct ones from the early 2000’s, different brands ranging from Logitech to Genius, A4tech and no names ones I almost gave up! I tried a PS/2 to USB adapter to no avail, tried the second one – still no luck! There was not a single keyboard that I could use to set up the BIOS because it was not recognized on system start up. Finally I found one that would work and let me enter the BIOS: Microsoft arc! Let’s right, a wireless keyboard from 2010 made to be used in a modern system worked out of the box on a motherboard made in 2002! It was also the only keyboard that works on IT7 – MAX. Why this keyboard works and how is a mystery unsolved, but it does 😀! Just to be clear: I also tried a Microsoft wired keyboard that I bought 10 – 15 years ago and still it did not worked!
That simple step took me a lot of time just to find a working solution before the next hurdle: the processor! The highest performing processor with 333 MHz FSB is the Athlon XP 3000 running at 2167 MHz. However it would not work on the AT7 – MAX, despite working in the other boards. I tried everything I could think of to no avail; it simply locked up at start up with no signs of life, not even a beep. It the end I choose the next best one: Athlon XP 2800. This one worked like a charm, it was recognized immediately by all the boards including the AT7 – MAX which displayed correctly the processor name and not unknown processor on system start up. I came to the conclusion that for some reason processor with multiplier of 13 and over are not supported either by the board or by the chipset; it only worked in the KT400A and MAX 2 boards!
The rest of the components work without a hitch: Corsair 2*1gb RAM, Leadtek Geforce 7800gs, Maxtor 120gb hard drive, Enermax 465w PSU that is one of the best retro friendly power supply that I have seen.

Here are some pictures with the system up and running:
Abit AT7 with VIA KT333 chipset

The BIOS settings used; please note that enhance for benchmark setting is disabled by default and has to be manually enabled in order to have better results in benchmarks.

The legacy CPU-Z and Everest 5.50 screenshots:

.. and system summary:

KT400: Abit KD7 – G, AT7 – MAX 2. I used the AT7 – MAX 2 because it is faster in preliminary testing. Now this board does have both PS/2 connectors so I had no problems here! Better still, it has a silicon image controller that, while still connected over the PCI bus, it is a blessing: I could use a 74 GB Raptor that vastly increases system responsiveness and shortens the boot times! Used different Corsair 1 GB sticks, same Leadtek 7800gs and same Enermax PSU. Again I stumbled upon a problem with the CPU: Athlon XP 3000 would not increase any score, in fact it would decrease some of them – 3d mark 2000&2001 – when compared against the 2800! I retested many times, reinstalled XP, used a period correct geforce4 TI4600 but the problem remains even with that card: 3000 was not any better than 2800! In the end I tested all the boards with the 2800.
Here are some pictures with the system up and running:
Abit AT7 – MAX 2 with VIA KT400 chipset

The BIOS settings used; please note that enhance for benchmark setting is enabled by default; so are enhance DRAM performance and enhance AGP performance. Each one provide a very small performance boost, but when they all add up the results will improve!

The system summary as reported by Everest:

KT400A: Abit KD7A is the only model produced by Abit with this particular revision of the KT400. It was the only motherboard that took the 3000 and actually got very slightly better scores in 3d mark! Another point to take into consideration is that all Abit boards with KT333 chipset do not have the auxiliary power connector so a strong 5V rail PSU is required! All KT400/A boards do have the connector so they can be used with a modern PSU. I used some Kingmax memory I had around, a WD 120 GB PATA HDD and the rest stayed the same.
Abit KD7A with VIA KT400A chipset