Some progress is being made 😉

After many failures and countless hours of experimenting I was finally able to run this ECS P6S5AT with SiS 635T chipset at 166 MHz FSB with Tualatin 512kb 😀

Now the FASTEST Socket 370 motherboard is EVEN FASTER 😁

At this point I’m gonna show you how it performs with PIII-S 1133 MHz, Tualatin core, 512kb (SL6BX).
The rest of the specs are as usual:
- Win98SE mainly
- GeForce 4 Ti 4600, GeForce 5 FX 5950 Ultra, and some GeForce 6800GT(Windows XP)!
- CPU clock = 1577 MHz!!! (13 % more than the original SL6BY 1400 MHz)
- DDR RAM Timings only 2,5-2-2-6 with 1GB GoodRam stick.

Here are the results from Everest MemoryTest:

When we compare those results with the original PIII 1400 MHz at 133 FSB we can see that the latency dropped from 90,6 to 79,4. That gives us 14 % gain. Read, Write and Copy are around 25 % better 😀

In SiSoft Sandra we can see 24 %) gain from the previous results.

In the rest of the tests we can see around 13-19 & gain in performance:

Let's check out the gaming performance 😉

Quake III
GeForce 4 ti 4600:
- from 197,2 FPS to 213,2 FPS (8% gain) - high res.
- from 227,2 FPS to 264,7 FPS (17% gain) - low res.

5950 Ultra gave me 16 % more FPSes In low res, and 12 % more in hihg res.

Serious Sam (OpenGL):

GeForce Ti 4600 gives 6% gain (60,2 to 64,1 FPS) in high res,
GeForce 5 FX 5950 ultra gives us phoenomenal 22 % gain in performance (from 64,8 FPS to 79,2 FPS!)

Here is a nice score that I got on GeForce 6800 GT:

It's still a work In progress project so I can’t share all the details now. There is lots to be done with DDR timings. I have to lower them to cl 2-2-2-5 and the results should be much better 😉

Here are my stupid nicknames for the following configurations so that it is easier to refer to given frequencies.

Pentium III-S 1,4 (SL6BY) @ 133 FSB = 1400 MHz - TualKing

Pentium III-S 1,26 (SL6BW) @ 166 FSB = 1411 MHz - TualSlave

Pentium III-S 1,4 (SL6BY) @ 150 FSB = 1575 MHz - TualQueen

Pentium III-S 1,13 (SL6BX) @ 166 FSB = 1577 MHz - TualWarrior

Pentium III-S 1,4 (SL6BY) @ 166 FSB = 1743 MHz - TualGOD!

In most cases those speed are reachable but there is one that I would love to achieve:
Pentium III-S 1,13 (SL6BW) @ 200 FSB = 1750 MHz - TualBeast!

In the next post I will try to compare those results with the AMD AthlonXP equivalent 😉

Great results, congratulations!

Do you have any news about the modified bios, any progress has been made?

Nemo1985 wrote on 2021-09-17, 08:13:

Great results, congratulations!

Do you have any news about the modified bios, any progress has been made?

Unfortunatelly, BIOS is not ready yet... and it will be VERY hard to do ANYTHING about it without the DataSheet for this mobo, or SiS 635T specification – those documents seem to be lost or they have never been made to the public...

Let’s move on with the comparison:

Below on the left you can see the results from ECS P6S5AT (SiS635T) Vs. ECS K7S5A(SiS735) on the right – The conditions are almost always the same. The chipsets are very similar, the rest is as usual: Win 98SE, DX8.1, RAM ten sam GoodRam 1 GHZ 2,5-2-2-6. The only thing that is really different is the FSB speed. It was impossible for me to get that crappy ECS K7S5A working at 166 MHz...

On the Intel side we’ve got: PIII-S SL6BX: 166 x13,5 = 1577 MHz (for sale from July2001?)
on the AMD side: Athlon XP 2400 +: 133 x15 = 2000MHz (for sale from August 2002?)

Pentium has higher FSB and twice as much cache, AtlhonXP has lower FSB and half the cache, but is clocked higher by 423 MHz! I wanted to set this Athlon at 166 MHz so that the conditions were more similar, but this board somehow does not accept it. All the results presented here were tested with the GeForce Ti 4600 graphics card (44.03 drivers)

AMD's performance in Everest definitely outperforms Intel in almost every respect except for memory latency.

In SiSoft Sandra Athlon beats Pentium again but the difference is a bit smaller in some instances:

PCMark 2002 is already more favorable to Intel but In the CPU Score PIII losts totally. In the Memory Score the result is almost identical, while Intel is leading in the HDD score.

3D Mark 99 produced very similar GPU results - 22 point difference. But AMD's CPU score wiped away Intel ...

Intel dominated AMD slightly in the 3D Mark 2000 by 92 points 😉

In 3D Mark 2001se the difference in results is merely 100 points so it’s a small thing 😉

Now, let’s check which one is better In games 😀

In Quake III, both in low and high resolution, Intel wins….(in the low res Pentium gets 14 frames more and in the high res. It scores 6.1 frames more) In Serious Sam, however, Thoroughbred wins the fight by 1.6 frames in 640x480 and 3.7 frames in high resolution. What is more interesting is that with a more powerful graphics cards such as the GeForce 6800 GT, the results were better for Intel 😀

Intel with Gf6800 GT at default clocks - 13472 points, AMD - 13435 points

The conclusion is that the Intel PIII 1266 MHz overclocked to 166 MHz bus (CPU clock - 1577 MHz - "TualWarrior") is more or less the equivalent of the Athlon XP2400 + 😉

I dare to say that Tualking 1.4 GHz overclocked to 166 MHz (1743 MHz - "TualGOD") will be able to compete with Barton 2500+ and maybe even 2600+ on “fair” conditions ;]

In the attacked file, you can compare the results described above with many differnet platforms. There are lots of results but I won’t get into details just take a look and see yourself which one is better. Of course it is unfair to compare results on a 2001 motherboard with a 2003 mobo, so it will always be unclear which CPU is better. A lot depends on the application.
Nevertheless I think that Overclocked Tualatin on a SiS635T mothearboard has a huge potential 😀

Next step is to play with TUALGOD!!! (SL6BY at 166FSB = 1750 MHz) HOW FAST DO YOU THINK IT WILL BE? 😁

Wow, judging by all those comments I can see that the interest is huuuuge 😜 😁 So I’m not gonna keep you waiting for an update 😉

Here’s how the fastest single socket 370 board runs with the best Tualatin CPU at 166 MHz, for the first time in the history:

Booting + 2 games with GeForce 4600 and WIN98SE :
https://www.youtube.com/watch?v=L2K43KxSHsI

MemoryTest86 Raport + BIOS settings:
https://www.youtube.com/watch?v=Y0BbW4iA__8

Doom III benchamrk on Windows XP with 6800GT:
https://www.youtube.com/watch?v=02x8L9u2kLk

30 minutes of gameplay, 2003-2005 era games (winXP and 6800GT)
https://www.youtube.com/watch?v=NHhYhPDT70U

STALKER on very high setting:
https://youtu.be/VoE6Y37pmd8

Before we go on with DDR results you can check the results with SDR memory:

Finalny WINDOWS 98SE and DDR memory!!!:

GeForce 4 Ti 4600:

FX 5950 ULTRA:

GeForce 6800GT win98SE:

small OC:

Test results with GeForce 6800 GT on WindowsXP
















Some of my best records on this board with overclocked GeForce 6800GT:











I’m not gonna share you the secret how to run this board at 166 MHz with a Tualatin CPU at this point, let’s just wait some time and next year you gonna see this baby run at 200 MHz 😁


If you are interested you can take a look at this New project releted to this mobo, where I compare it with other chipsets at 166 MHz 😉
Project "TualGOD": Pentium III-S 1,4 GHz @ 166 FSB = 1750 MHz - 5 chipsets comparison

Just a small update!

I've managed to run a different batch of that ECS P6S5AT at 166 MHz and Tualatin at CL2! (2-2-2-6 exactly) but there are some bad side effects:
https://www.youtube.com/watch?v=zRKR7M4D408
You can see lots of artifacts. When I switch to CL 2,5 the problem is gone...
I think that CL2 is a chipset limitation that cannot be solved. Bear in mind that we are talking here about a motherboard from 2001 and the DDR333 became a standard in the middle of 2002 I think - correct me If I am wrong....

So.... perhaps CL 2,5-2-2-6/5 is the maximum this mobo/chipset can take. I will try to expermient with different RAMs and see if there is any improvement.

Nice 3dmark 2001 results.
My best socket 370 score was 12105 : epox 3pta / celeron 1.1a sl6ca @ 1793 (163fsb) / radeon 9800 pro.

Could always try a high quality ddr400, it’s possible the chipset / memory controller itself has latency

overdrive333 wrote on 2022-01-04, 01:44:

Nice 3dmark 2001 results.
My best socket 370 score was 12105 : epox 3pta / celeron 1.1a sl6ca @ 1793 (163fsb) / radeon 9800 pro.

Thank you 😀 This mobo can score a lot on almost every graphics card 😀 I think the best option would be to run this configuration with GeForce 6800 ULTRA 😀 - then, even Crysis would be playable, I guess 😁

rmay635703 wrote on 2022-01-04, 04:51:

Could always try a high quality ddr400, it’s possible the chipset / memory controller itself has latency

NO NO NO! Not with this motherboard! The better sticks you use the worse it's going to be! The best so far were cheap GOODRAM and some Infineon RAMs. But still it's a lottery - you can try two IDENTICAL sticks and one is gonna work better and one worse. In some cases it doesn't work at all...

Now that I have two WORKING boards I can compare how they behave at 166 MHz with tualatins, and I can say that they DO NOT behave in the same way... So I can assume that it's also a matter of luck to find the RIGHT BATCH of rev 1.6 to be fully able to run it at 166 MHz... What is interesting is that on both batches there are absolutely NO PROBLEMS to run them with Tualatins at 166 MHz and proper SDR memory (qimonda chips)!!! 😀

If that’s the case should test over clocking
Some antique Pc200 and pc266 ddr

I'm recapping some of my old motherboards, with polymer caps. This is called "polymod". Right now, I'm thinking about recap my P6S5AT. Every cap was made by G-Luxon.

The bigger caps is 1500uf/6.3v/10mm (x21 or x22)

There are x4 1000uf/10v/8mm

And x4 470uf/16v/8mm

About the smaller ones, if I really recap, I'll leave the way it is or I'll replace, at least, the 100uf/16v (x13 or x14).

In the past, do you saw problems with this capacitors mobo ?

VooDooMan wrote on 2021-04-08, 07:39:

Sorry, I haven't checked in here for a while...

Kami27 wrote on 2020-08-21, 17:32:

You talked about a software called CMOSTool to overclock the FSB but I really can't find it anywhere, could you help me please ?

If you are still interested please write me a PM.

Hi, I'm also interested but I can't compose PM's for now, can you write me one ?
Thank you.

The worst thing happened to me, I have a Jetway J-635CS, 2x 1gb DDR 400 Kingston, Radeon HD 3650, Pentium III 1.1 GHz SL5QW Coppermine aka fastest coppermine, and pretty much the best 370 setup.
I switched PSU's and suddenly no post, switched back to the old one still no post, the CPU fan spins for a second then nothing, the cpu gets warm, but the chipset is cold, I checked everything seems the mosfet shorted so i can't get correct reading on the caps, and the GPU's caps blew. I pray that I'll manage to fix the motherboard, even if I have to re-solder the whole thing from ground up, hope the chipset is fine because that would be a real pain in the ass to source and solder even then it might not be compatible.

Just had to write this down here as a 635 user in pain 😒 If anyone has any advice I'll be super thankful!
Also I've got a dead Intel D815 board, it worked 10 years ago and the whole PC was just stashed away, the caps were dead so I recaped it and it still wont post,
why are they selling those 2 ancient boards for huge amounts of money? I can't find anything under a couple hundred dollars.

The family is getting biggger 😁

Here are all the possible revs and all mobos I have collected with the date of production:

Rev 1.3 - 01.04.24

Rev 1.5 - 01.05.22

Rev 1.5 - 01.05.23

Rev 1.5 - 01.06.28

Rev 1.6 - 01.08.29

Rev 1.6 - 01.09.04

Rev 1.6 - 02.02.14

Rev 1.6 - 02.02.19

Rev 1.6 - 02.07.11

In case you have this board and you want to participate in this project please provide your rev and date of production 😉 The higher rev with production date the bigger OC potential 😀

1) Rev 1.3 - 01.04.24 - NO OC potential (Tualatins work only at 66 FSB)

2) Rev 1.5 - 01.05.22
3) Rev 1.5 - 01.05.23 - Tualatin Support, Coppermines work good at 166 MHz, Tualatins only at 150 MHz.
4) Rev 1.5 - 01.06.28

5) Rev 1.6 - 01.08.29 - Tualatin Support, Coppermines work good at 166 MHz, Tualatins have some problems at 166 MHz.
6) Rev 1.6 - 01.09.04

7) Rev 1.6 - 02.02.14
😎 Rev 1.6 - 02.02.19 - Tualatin Support, Coppermines work good at 166 MHz, Tualatins can work better at 166 MHz.
9) Rev 1.6 - 02.07.11

All revs support 2 GB of Ram (not only 1,5 GB as suggested by the chipset producer)

When it comes to HDD the maximum transfer rate is around 82,4 MB/sec (ATA UDMA100 onboard controller)
Here's how it scales with FSB:

66FSB - 78,8 MB/sec

100FSB - 81,1 MB/sec

133FSB - 81,7 MB/sec

150FSB - 82,6 MB/sec


166FSB - 82,4 MB/sec

Here's how long it takes to boot into windows on 166 FSB:
https://www.youtube.com/watch?v=SEiI4V9jP1g - Tualatin
https://www.youtube.com/watch?v=VSTXBHm6F2k - Coppermine

gabimor wrote on 2022-01-06, 22:22:

In the past, do you saw problems with this capacitors mobo ?

Those caps were of low quality just like the vast majority from that period of time 😉

JustJulião wrote on 2022-02-08, 19:36:

Hi, I'm also interested but I can't compose PM's for now, can you write me one ?

I attach everything you need.

Mazuto wrote on 2022-10-07, 22:36:

Just had to write this down here as a 635 user in pain 😒

I feel for you 🙁 I just hope you will be able to get this mobo back to life! In case you need help just PM me 😀

I would also like you to make so HQ photos of your mobo and make them public - THERE ARE NO GOOD PHOTOS IN THE WHOLE INTERNET 😉 You are the lucky owner of Jetway J-635CS 😁 Please share 😉

Hi, I am also tinkering around with a P6s5AT. At the moment I am using a PIII-S 1400 MHz and there is absolutly no OC potential. Even at 138 MHz FSB it didn't boot into Windows.
I purchased a 1333 MHz PIII with 256k in China to ckeck if the 512k cache are the limiting problem. But I guess it will take some weeks to be shipped.

In the meantime, I would kindly ask if there is something worth cheking out with the 512k Tualatin. Are there maybe any chipset registers to set or anything else that could be tweaked?
Or is it all just about having the right batch of the Mainboard to get the 512k Tualatins running at higer FSB frequencies? My Revision is 1.5A.
And where is the manufacturing date written?

Thanks

3faltigkeit wrote on 2023-12-13, 18:44:

Hi, I am also tinkering around with a P6s5AT. At the moment I am using a PIII-S 1400 MHz and there is absolutly no OC potential. […]
Show full quote

Hi, I am also tinkering around with a P6s5AT. At the moment I am using a PIII-S 1400 MHz and there is absolutly no OC potential. Even at 138 MHz FSB it didn't boot into Windows.
I purchased a 1333 MHz PIII with 256k in China to ckeck if the 512k cache are the limiting problem. But I guess it will take some weeks to be shipped.

In the meantime, I would kindly ask if there is something worth cheking out with the 512k Tualatin. Are there maybe any chipset registers to set or anything else that could be tweaked?
Or is it all just about having the right batch of the Mainboard to get the 512k Tualatins running at higer FSB frequencies? My Revision is 1.5A.
And where is the manufacturing date written?

Thanks

The 512k of cache are very unlikely to be the limiting factor here
I would investigate board stability (caps?), RAM, Power supply. I don't remember if that board comes with voltage settings, have you tweaked it? 138 seems low even at stock voltage though.

Mhm, the Power Supply should by fine. The caps have at least no visible issues (I know this does not mean that they are fine).
I tried different memory sticks. DDR and SDR. OCZ, Kingston, Infineon, MDT, Now I am using cheap Tiquila Sticks.
And there is nothing tweaked so far. Voltages are at default because they can not be changed. I also tried a SATA PCI Card if maybe the PCI BUS could be the culprit.
I am using a modded BIOS (based on latest official BIOS) which makes the supported FSB Settings available in the BIOS (100,133,138,150,166).
I also have aa Celeron 1200 MHz this one runs at 138 MHz. For 150 MHz it needs a volt mod to 1,65 V to be stable (1800 MHz). Until now, I didn't had time to do the mod.
But I will give it a try maybe next week. But what I get from VooDooMans postings is that there are issues with Tualatin CPUs with 512k with older revisions of this Board.

By the way, is there maybe some progress running the p6s5at at 200 MHz FSB? I wonder If the chipset is able to handle such high frequencies. I am searching for an alternative to the TUSL2. The TUSL2 has no issues with 200 MHz FSB. But the only Graphics Card which (which I found so Far) handles 100 MHz AGP clock is the GF3 Ti200. A working faster one would be great. The 6OXET has great deviders but using these is unstable. So the P6s5at ist the only other candidate that maybe could achieve this.

3faltigkeit wrote on 2023-12-22, 15:19:

By the way, is there maybe some progress running the p6s5at at 200 MHz FSB? I wonder If the chipset is able to handle such high frequencies. I am searching for an alternative to the TUSL2. The TUSL2 has no issues with 200 MHz FSB. But the only Graphics Card which (which I found so Far) handles 100 MHz AGP clock is the GF3 Ti200. A working faster one would be great. The 6OXET has great deviders but using these is unstable. So the P6s5at ist the only other candidate that maybe could achieve this.

Check you private messages please - you got all the answers there 😉

VooDooMan wrote on 2022-10-08, 17:51:

The family is getting biggger :D https://u.cubeupload.com/roksana/allboards.jpg […]
Show full quote

The family is getting biggger 😁

Here are all the possible revs and all mobos I have collected with the date of production:

Rev 1.3 - 01.04.24

Rev 1.5 - 01.05.22

Rev 1.5 - 01.05.23

Rev 1.5 - 01.06.28

Rev 1.6 - 01.08.29

Rev 1.6 - 01.09.04

Rev 1.6 - 02.02.14

Rev 1.6 - 02.02.19

Rev 1.6 - 02.07.11

In case you have this board and you want to participate in this project please provide your rev and date of production 😉 The higher rev with production date the bigger OC potential 😀

1) Rev 1.3 - 01.04.24 - NO OC potential (Tualatins work only at 66 FSB)

2) Rev 1.5 - 01.05.22
3) Rev 1.5 - 01.05.23 - Tualatin Support, Coppermines work good at 166 MHz, Tualatins only at 150 MHz.
4) Rev 1.5 - 01.06.28

5) Rev 1.6 - 01.08.29 - Tualatin Support, Coppermines work good at 166 MHz, Tualatins have some problems at 166 MHz.
6) Rev 1.6 - 01.09.04

7) Rev 1.6 - 02.02.14
😎 Rev 1.6 - 02.02.19 - Tualatin Support, Coppermines work good at 166 MHz, Tualatins can work better at 166 MHz.
9) Rev 1.6 - 02.07.11

All revs support 2 GB of Ram (not only 1,5 GB as suggested by the chipset producer)

When it comes to HDD the maximum transfer rate is around 82,4 MB/sec (ATA UDMA100 onboard controller)
Here's how it scales with FSB:

66FSB - 78,8 MB/sec

100FSB - 81,1 MB/sec

133FSB - 81,7 MB/sec

150FSB - 82,6 MB/sec


166FSB - 82,4 MB/sec

Here's how long it takes to boot into windows on 166 FSB:
https://www.youtube.com/watch?v=SEiI4V9jP1g - Tualatin
https://www.youtube.com/watch?v=VSTXBHm6F2k - Coppermine

gabimor wrote on 2022-01-06, 22:22:

In the past, do you saw problems with this capacitors mobo ?

Those caps were of low quality just like the vast majority from that period of time 😉

JustJulião wrote on 2022-02-08, 19:36:

Hi, I'm also interested but I can't compose PM's for now, can you write me one ?

I attach everything you need.

Mazuto wrote on 2022-10-07, 22:36:

Just had to write this down here as a 635 user in pain 😒

I feel for you 🙁 I just hope you will be able to get this mobo back to life! In case you need help just PM me 😀

I would also like you to make so HQ photos of your mobo and make them public - THERE ARE NO GOOD PHOTOS IN THE WHOLE INTERNET 😉 You are the lucky owner of Jetway J-635CS 😁 Please share 😉

How to know exact date of production?

P.S. I would not confuse people with overclocking potential by production date , revision as it's alot of lottery .

Hi, yes, I read the PM - Thank you very much for that. I am still not able to write back. It seems I did not enough Postings, yet.

In the meantime I also tried different memory Sticks. Goodram, Nanya, Quimonda, OCZ...
And on my Board the first 1GB Kit that I tried, gave me the best result: 2,5-NORMAL...
This cheap Tequila M&S modules with fancy cactus Sticker:
https://www.tiger-technik.de/Arbeitsspeicher- … r672::8302.html

Maybe this is helpful for somebody.