Do you think is worth recapping the ECS K7S5A? I only need to get the caps and a proper soldering iron

Well why not? 😀

I would perhaps not use the most expensive Japanese Nichicon caps though.

Read up on caps and get some cheap Taiwanese caps that other people have found working well. It's a good motherboard to practice recapping on! 😀

Have to agree with Skyscraper on that, especially if it's your only SiS 735 board.

Skyscraper wrote:

Well why not? 😀

I would perhaps not use the most expensive Japanese Nichicon caps though.

Read up on caps and get some cheap Taiwanese caps that other people have found working well. It's a good motherboard to practice recapping on! 😀

Yes, I'll do that, also i'm really curious about how would perform that board with SDRAM and DDR, and what difference will do

Carlos S. M. wrote:

i got two old PCs which one of these came with an ECS K7S5A and an Duron 1000, it have the HDD, videocard and RAM missing. I plugged the computer and got it turning on and POST sucessfully (after adding RAM and a videocard to test it), but the mobo needs a recap, the ECS K7S5A is based around the SiS 735 chipset and supports both PC100/133 and DDR200/266, Universal AGP 4x slot, ATA-100, USB 1.1, 5 PCI slots and 1 AMR slot, board has only 1 20 pin power as power source

the Duron 1000 is a Morgan core which is derivated from Palomino, although it still uses ceramic PGA instead of organic PGA in Palomino, the Duron model is DHD1000AMT1B

do someone have plans to try SiS chipset based Socket A systems? this is my first SiS based chipset Socket A mobo i get in my hands

At that time I had a K7S5A with Cheepobios, with a XP1700+ JIUHB and FSB to 150Mhz it reached 2250Mhz @1.8V using pinmod.

Today I have here a K7S5A Pro(newer version) that is almost the same, but have a problem on CMOS(lost config even with a new battery) and I used their VRM caps on a A7N8X-E Deluxe...

But I remember a few tests I did on my first board, SDR and DDR makes a difference but not too big, especially because on that time we needed to buy less DDR or more SDR 🤣 on benchs you can see the difference

On general it performs like a KT266/KT266A, I think was one of the better SiS chipsets on their times.

The unfortunate fact was that ECS was highly variable in terms of how stable they ran, but the SIS 735 chipset did have some impressive features.

I'm testing an Asus A7V333 REV. 1.01 motherboard this weekend and I'm not finding it as bad as old forum threads seemed to indicate.

I thought that this was a good opportunity to see if AMDs Athlon XP performance rating (Thunderbird equivalent) makes sense when comparing Thoroughbred to Barton.

The Thoroughbred B 2600+ with 166 MHz FSB runs at an actual speed of 12.5 x 166 = 2083MHz, the 2700+ runs at 13 x 166 = 2166MHz and the 2800+ runs at 13.5 x 166 = 2250 MHz. It would seem logical that a Thoroughbred B running at 14 x 166 = 2333 MHz would have a PR rating of 2900+. My Thoroughbred B 2400+ can run at this theoretical Athlon XP 2900+ setting at stock voltage so I ran some benchmarks to see how it would perform.

The Barton 3000+, the next step up in the PR rating scheme runs at 13 x 166 = 2166 MHz but does it deserve a higher PR rating than the Thoroughbred B "Athlon XP 2900+" running at 2333 MHz? Lets find out!

I planned to use my Geforce4 Ti 4600+ video card but it seems to be in a system somewhere so I used a Geforce Ti 4200+ overclocked to 315/580 which is fast enough for this purpose. The rest of the system spec is Asus A7V333 REV 1.01, 1x1GB memory running at 166 MHz CL 2-2-2-5 T1, an Audigy and some random HDD.

Thoroughbred B 2900+ (14 x 166 = 2333 MHz). Super PI 1M: 52 Seconds

Thoroughbred B 2900+ (14 x 166 = 2333 MHz). 7-Zip 32M: 1597 MIPS

Thoroughbred B 2900+ (14 x 166 = 2333 MHz). PCMark 2002: CPU Score: 6713 Memory Score: 3954

Thoroughbred B 2900+ (14 x 166 = 2333 MHz). 3dmark 2000: 3dmark score: 12765 CPU score: 704

Thoroughbred B 2900+ (14 x 166 = 2333 MHz). 3dmark 2001: 3dmark score: 10802

What do you think? will the Barton 3000+ beat the Thoroughbred B "2900+"? 😁

Skyscraper wrote:

What do you think? will the Barton 3000+ beat the Thoroughbred B "2900+"? 😁

Interesting comparisons.

According to tests back in the day, we have seen that AMD has set the 3000+ (for the 166FSB version) too 'optimistic'. They have tried to release a counterpart to intel's 3.06GHz CPU.

More realistic rating for the barton would be the 2900+ ... 😊

melbar wrote:

Interesting comparisons. […]
Show full quote

Skyscraper wrote:

What do you think? will the Barton 3000+ beat the Thoroughbred B "2900+"? 😁

Interesting comparisons.

According to tests back in the day, we have seen that AMD has set the 3000+ (for the 166FSB version) too 'optimistic'. They have tried to release a counterpart to intel's 3.06GHz CPU.

More realistic rating for the barton would be the 2900+ ... 😊

Lets see! 😀

Barton 3000+ (13 x 166 = 2166 MHz). Super PI 1M: 53 Seconds. A bad start.

Barton 3000+ (13 x 166 = 2166 MHz). 7-Zip 32M: 1576 MIPS. Another loss.

Barton 3000+ (13 x 166 = 2166 MHz). PCMark 2002: CPU Score: 6418 Memory Score: 4525. Mixed results, the Bartons 512KB cache does help when it comes to memory performance.

Barton 3000+ (13 x 166 = 2166 MHz). 3mark 2000: 3dmark score: 13203 CPU score: 716. A narrow win for the Barton.

Barton 3000+ (13 x 166 = 2166 MHz). 3mark 2001: 3dmark score: 10948. Another narrow win for the Barton.

The conclusion must be that this two CPUs perform about the same and member melbar is right! The Barton Athlon XP 3000+ (FSB166) should have had a PR rating of 2900+ for the PR rating to be consistent with the PR rating of the Thoroughbred B Athlon XP with 166 MHz FSB.

Skyscraper wrote:

The conclusion must be that this two CPUs perform about the same and member melbar is right! The Barton Athlon XP 3000+ (FSB166) should have had a PR rating of 2900+ for the PR rating to be consistent with the PR rating of the Thoroughbred B Athlon XP with 166 MHz FSB.

What would happen if you compare the Barton and the Thoroughbred at the same clocks, i heard the difference would be mainly 0-1% due to the exclusive L2 design on AMD Athlon Thunderbird/XP

Carlos S. M. wrote:

Skyscraper wrote:

The conclusion must be that this two CPUs perform about the same and member melbar is right! The Barton Athlon XP 3000+ (FSB166) should have had a PR rating of 2900+ for the PR rating to be consistent with the PR rating of the Thoroughbred B Athlon XP with 166 MHz FSB.

What would happen if you compare the Barton and the Thoroughbred at the same clocks, i heard the difference would be mainly 0-1% due to the exclusive L2 design on AMD Athlon Thunderbird/XP

I'm trying to get a Barton to run at 14 x 166 at the moment but with this tiny little Akasa cooler and the Asus A7V333 motherboards less than stellar voltage regulation this 14x multiplier capable voltage hungry Athlon XP-M 2500+ is acting a bit flaky. I'm using it because the Barton 3000+ is locked and will only run at the 13x multiplier and I needed to test this CPU anyhow.

The core voltage is very unstable (+-0.1V), I guess this Asus A7V333 motherboard would need new caps and perhaps some mods to run decently with an overvolted and overclocked Barton. In any case it seems using "1.75V" was enough to get the system stable enough for benchmarking, I will upload the result later.

Skyscraper wrote:

God Of Gaming wrote:

Hello, I would like to build a high performance win98 machine for 1600x1200 100Hz gaming, if I can dig up one of those 35W 266FSB mobile bartons. For graphics I'll go with FX5950 Ultra or maybe 6600GT or 6800GT, they're all AGP8X I think. I also need enough PCI slots for Aureal Vortex 2 and gigabit network adapter if one is not built into the mobo. I'd like to run the CPU at no higher than 2ghz because of NFS Porsche bug, and undervolt it as low as it would go while still running stable. Which motherboard would be the best fit for this?

I have no idea if it's possible to control the XP-M CPU multiplier from within Windows 9x but if so any VIA chipset motherboard with or without multiplier control should work as long as it's able to POST and BOOT Windows with the XP-M.

If it's not possible to control the multiplier from within Windows 9x then you need a motherboard which both is able to run the XP-M and support multiplier control from the BIOS otherwise you could be stuck at 5x133 MHz.

Any particular motherboard models to look out for?

I really ought to do something with these at some point.

Best,
Chris

Skyscraper wrote:

I'm testing an Asus A7V333 REV. 1.01 motherboard this weekend and I'm not finding it as bad as old forum threads seemed to indicate.

For me the ASUS A7V333 was a joyful experience before I accidentally killed it 😊
I didn't push mine very hard though, using a R9600 and a 2000+

ASUS boards of the time (or at least the AMD ones) tended to have slightly increased voltages and this often resulted in increased stability.

God Of Gaming wrote:

Any particular motherboard models to look out for?

I will know in a month or so when I have tested more motherboards. 😀

The Asus A7V333 I'm testing now seems nice though, it has multiplier control in the BIOS and as long as you stay at default voltage it seems to run just fine with the Barton XP-M.

A Barton XP-M 2500+ running at 11 x 166 = 1833 MHz would only need about 1.4V - 1.5V but you will need to set the voltage using the jumpers or you will get 1.675V with an XP-M intended to run at 1.5V.

AMD was optimistic on Barton's PR rating, I never seen that difference. A good board + Tbred shows very better performance than a Barton on a slow chipset/board.

Here is the results for the Barton at 2333 MHz.

Still using the overclocked Geforce4 Ti 4200 @ 315/580.

Barton 3200+ (14 x 166 = 2333 version). Super PI 1M: 51 seconds.

Barton 3200+ (14 x 166 = 2333 version). 7-Zip 32M: 1668 MIPS

Barton 3200+ (14 x 166 = 2333 version).: PCMark 2002: CPU Score: 6851 Memory Score: 4694

Barton 3200+ (14 x 166 = 2333 version). 3dmark 2000: 3dmark Score: 13757 CPU Score: 731

Barton 3200+ (14 x 166 = 2333 version). 3dmark 2001: 3dmark Score: 11275

A quick recap.

Super PI 1M

Thoroughbred B 2333 MHz : 52
Barton 2166 MHz: 53
Barton 2333 MHz: 51

7-Zip 32M

Thoroughbred B 2333 MHz : 1597
Barton 2166 MHz: 1576
Barton 2333 MHz: 1668

PCMark 2002

Thoroughbred B 2333 MHz : 6713/3954
Barton 2166 MHz: 6418/4525
Barton 2333 MHz: 6851/4694

3dmark 2000

Thoroughbred B 2333 MHz : 12765/704
Barton 2166 MHz: 13203/716
Barton 2333 MHz: 13757/731

3dmark 2001

Thoroughbred B 2333 MHz : 10802
Barton 2166 MHz: 10948
Barton 2333 MHz: 11275

The Barton is a fair bit faster clock for clock in comparison to the Thoroughbred.

résumé, (with your results): Tbred @2333 vs. Barton @2333

Super Pi 1M: 1.9%

7-Zip 32M: 4.4%

PCMark 2002 - CPU: 2.1%
PCMark 2002 - memory: 18.7%

3dmark 2000 - score: 7.8%
3dmark 2000 - CPU: 3.8%

3dmark 2001: 4.4%

Considering that the exlusive L2 cache working on AMD's XP... it's not 100% more cache for the Barton, it's 66% (384kb cache vs. 640kb cache - overall).

melbar wrote:

résumé, (with your results): Tbred @2333 vs. Barton @2333 […]
Show full quote

résumé, (with your results): Tbred @2333 vs. Barton @2333

Super Pi 1M: 1.9%

7-Zip 32M: 4.4%

PCMark 2002 - CPU: 2.1%
PCMark 2002 - memory: 18.7%

3dmark 2000 - score: 7.8%
3dmark 2000 - CPU: 3.8%

3dmark 2001: 4.4%

Considering that the exlusive L2 cache working on AMD's XP... it's not 100% more cache for the Barton, it's 66% (384kb cache vs. 640kb cache - overall).

It's possible that the difference in 3dmark 2001 would be more in line with the difference in 3dmark 2000 if I had used a faster video card. Other than that I think the results give a fair representation of the performance difference between these cores when running in a single channel VIA motherboard. I think the difference might be even less when running on a 200 MHz FSB and a lower multiplier and probably somewhat larger if running at an even higher multiplier using 133 MHz FSB.

Today I did a quick AGP performance comparsion between the Asus A7V333 and the Asus A7V8X, basicly the same motherboard with VIA KT333 and VIA KT400.

The BIOS versions for both motherboards have pretty much exactly the same settings so it's easy to compare apples for apples. The only difference is that the A7V8X wont go lower than "tRAS 6" while I tested the A7V333 using "tRAS 5", this do not affect the performance in any measurable way. I'm using the GF4 Ti 4200 128MB @ stock, AGP subsystem performance differences should still affect the performance even if the system is bottlenecked by the video card.

I had to install the VIA AGP driver to persuade the A7V8X to not run the AGP @ PCI while the A7V333 is using what ever driver XP felt like using as it worked, I never install the VIA driver packages unless I have to.

Asus A7V333, Thoroughbred B 2333 MHz (2900+), Geforce 4 Ti 4200 @ stock: 3dmark 2000

Asus A7V333, Thoroughbred B 2333 MHz (2900+), Geforce 4 Ti 4200 @ stock: 3dmark 2001

Asus A7V8X, Thoroughbred B 2333 MHz (2900+), Geforce 4 Ti 4200 @ stock: 3dmark 2000

Asus A7V8X, Thoroughbred B 2333 MHz (2900+), Geforce 4 Ti 4200 @ stock: 3dmark 2001

Not much difference at all, the A7V333 KT333 motherboard is faster than the A7V8X KT400 motherboard but the difference is not worth mentioning.