VOGONS

Is there any hope for this crap motherboard?

Topic actions

First post, by Fish3r

Posted on 
User metadata
Rank Newbie
Rank
Newbie

I've had this PC for a while, it's a slot 1 build with the following specs:

CPU: Pentium 3 Katmai 550Mhz
Motherboard: Chaintech CT-6SSA2
RAM: 256MB PC100
Video Card: S3 Savage 4
Sound: Onboard ESS Solo 1

Out of the box, performance is absolutely terrible. I'm not sure why chaintech did this but basically every feature of this board's SIS 5600 chipset that would make it perform well is turned off, using this PCChips M747 / SiS 600 Tuning: The Anti Thesis Thread. NOT the bane of MY existence. very helpful writeup from Oerg866 dealing with a similarly awful board I was able to bring performance in windows games up to a level that I'd consider to be acceptable.

Memory bandwidth goes from:

The attachment before.png is no longer available

To:

The attachment after.png is no longer available

Which puts it within ~20% of the 440bx in games from the testing I've done. With that I thought I was done with the thing, as it seems to be performing OK.

One lingering issue though is that DOS game performance is all over the place and I can't really figure out why, specifically with DOS games running in VESA modes. Duke 3d at 320x200 is north of 100fps which I think is what I should be getting on a Pentium 3 at this speed, however, once I bump the resolution up to 640x480 I'm now hovering around 40fps. If I look up performance for this tier of CPU, I should be seeing at least 60 at this resolution. So something is bottlenecking, but I'm not exactly sure what it is. One other game I've tried that this system should handle just fine is Screamer Rally, which like Duke3d is very smooth in the low resolution mode but becomes very choppy when trying 640x480.

I know the solution to this problem is to probably just quit while I'm ahead and get a 440bx board, but I've *almost* got this thing to where I want it performance-wise, it's just DOS games that seem to be giving it trouble so I'm not actually sure if I should be blaming the board despite its many issues. Also, I'm not running these games in DOS, just windows 98 which (I think?) bypasses the problems P6 CPUs have with DOS game performance. The S3 card shouldn't be causing any issues either as far as I'm aware.

I'd be interested to know if anyone here has any insight on this, maybe I'm missing something really simple?

Reply 1 of 5, by rasz_pl

Posted on 
User metadata
Rank l33t
Rank
l33t

might be the case of Sis 5600 not having MTRR registers or defaulting to only enabling magic window over standard VGA ram space while high resolution VESA modes are mapped in 15-16MB window or somewhere else MTRRLFBE and AGP/PCIe cards in DOS

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 2 of 5, by Fish3r

Posted on 
User metadata
Rank Newbie
Rank
Newbie

So, I've spent way more time than I should have faffing with this today but I've arrived at a solution, and to answer the thread title, yes there is hope for this crap motherboard.

rasz_pl wrote on 2026-08-15, 00:47:

might be the case of Sis 5600 not having MTRR registers or defaulting to only enabling magic window over standard VGA ram space while high resolution VESA modes are mapped in 15-16MB window or somewhere else MTRRLFBE and AGP/PCIe cards in DOS

I had a read through the linked thread, tried a few different things but only saw a speedup by just disabling the LFB using S3VBEFIX, that helped somewhat but performance was still probably lower than it should be due to relying on banked display modes. This also only worked within DOS.

I decided to go and check the datasheet for the chipset, so far I'd only made the register tweaks suggested in the post I linked above and hadn't actually read into anything myself. I found this section:

The attachment 5600sheet.PNG is no longer available

All of these were unset by default on my motherboard. I'm not going to pretend I know exactly what these all mean, but since they're related to accessing the memory of an AGP device I just turned all of the options on using WPCREDIT.

Here's what I was getting:

The attachment dukebefore.png is no longer available

And here's the result of enabling those options:

The attachment dukeafter.png is no longer available

(Both screenshots are Duke running within windows and the system had the performance tweaks applied that improved memory bandwidth from the original post)

I think I may have to investigate the datasheet further and see if I can get more performance out of this, it's actually insane how badly configured the chipset is by default.

Last edited by Fish3r on 2026-08-16, 00:50. Edited 3 times in total.

Reply 3 of 5, by Ozzuneoj

Posted on 
User metadata
Rank l33t
Rank
l33t

Wow, that is a HUGE difference. It is truly amazing how badly some devices were configured back in the day. I wonder if Chaintech was so desperate to put out a product that wouldn't be an unstable nightmare (which could cost them money in returns\support) that they just made them as slow as possible from the factory, to be on the safe side.

These days we do some tweaks and get a slight reduction in power consumption\heat and maybe a tiny bit of extra performance (single digit percentages). Back in those days you could change two or three BIOS settings from "off" to "on" and double your gaming performance.

Nice work!

Now for some blitting from the back buffer.

Reply 4 of 5, by Fish3r

Posted on 
User metadata
Rank Newbie
Rank
Newbie
Ozzuneoj wrote on 2026-08-15, 17:05:

I wonder if Chaintech was so desperate to put out a product that wouldn't be an unstable nightmare (which could cost them money in returns\support) that they just made them as slow as possible from the factory

I think this is likely what happened, these were from what I can tell cheap boards sold in bulk to system builders that were never expected to see anything more than a low end Celeron.

Reply 5 of 5, by rasz_pl

Posted on 
User metadata
Rank l33t
Rank
l33t

Pcchips loved this chipset. My favorite is this board and list of expected problems: https://theretroweb.com/motherboards/s/pcchips-m666

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Go to top of page Go to top of page
Back to General Old Hardware
Close window
Close window