Haven't tried this particular board, but the connector looks pretty standard and was widely used in the socket 7 era and I've had a lot of boards from that era. Needless to say, mice worked fine on those.

Was PS/2 not as common during the last years of the 486 platform?

Yes, works fine for me. You must use a keyboard controller IC which supports PS/2 mice. Look up the spec sheet for yours. Perhaps the previous owner swapped keyboard controller chips.

feipoa wrote:

Yes, works fine for me. You must use a keyboard controller IC which supports PS/2 mice. Look up the spec sheet for yours. Perhaps the previous owner swapped keyboard controller chips.

The keyboard controller on my board is a HOLTEK HT6542B and is not removable (soldered onto board)... It's data sheet states that it does support PS/2 devices.

What puzzles me a bit is the fact that the header is labeled +5v (pin 5) nc (pin 4) GND (pin 3) Signal Out (Pin 2) and Signal In (Pin 1), instead of pins 1 and 2 labeled as Mouse Clk and Mouse Data. I tried reversing these two pins, still no go. I tested the header with my Biostar MB8433 and it works fine.

Also, the board seems to be outpointing 5V trough both pin 1 and pin 5...

Frankly, this board is appealing because of the PS/2 header, since my Biostar MB-8433-UUD is an early revision and it stops receiving both mouse and KB input after a few minutes of use when using an optical mouse (works fine with regular PS/2 ball mice tough) - this and the undocumented 3.6V voltage make it a great board for a 586 machine, if it will work properly.

I have this exact same Board and revision, and Yes, my boards PS2 mouse works perfectly.
The only downside is that it doesn't support EDO memory, but You can't have it all I guess.

Here is the latest bios, also the one mine has installed.

I'm using the same PS2 header as my Epox EP-58MVP3 mobo, and from that manual, the pin configuration is
1 Data 2 Clk 3 Gnd 4 NC 5 Vcc

Imperious wrote:

I have this exact same Board and revision, and Yes, my boards PS2 mouse works perfectly. The only downside is that it doesn't s […]
Show full quote

I have this exact same Board and revision, and Yes, my boards PS2 mouse works perfectly.
The only downside is that it doesn't support EDO memory, but You can't have it all I guess.

Here is the latest bios, also the one mine has installed.

I'm using the same PS2 header as my Epox EP-58MVP3 mobo, and from that manual, the pin configuration is
1 Data 2 Clk 3 Gnd 4 NC 5 Vcc

Thank you. I tried that bios - PS/2 mouse still won't work... The header I use is a regular PS/2 header that works fine on both VIA MVP3 boards and my Biostar UM8433, so the header and pinout is ok. PS/2 mouse is inabled in bios, I checked with modbin. There is no visible damage to the motherboard, neither traces or broken parts - somehow, this is a software issue.

looking around at pictures, I noticed some boards come with a Holtek KB controller like mine, while others come with an AMIKEY controller. Does your board use a holtek or amikey KB controller?

Also, I noticed that in bios (using modbin) the KB controller has several electable "types", like KB-200, KB-200x and so on. This board's bios has "KB-200" selected, while the BIOSTAR has KB-200x (and PS/2 works on it, but I think PS/2 is handled by the chipset and not a dedicated KB controller).

UPDATE: I changed the KB controller with modbin from KB-200x to KB-200 and re-flashed - still no mouse...

The mouse header pins on the MB-8433UUD and the 4DPS are very different.

MB-8433UUD
PIN Assignment
1 Data
2 N/C
3 GND
4 Vcc
5 CLK

4DPS
PIN Assignment
1 CLK
2 Data
3 GND
4 N/C
5 Vcc

On the MB-8433UUD, the chipset controls the PS/2 mouse (the chipset contains a keyboard controller chip). On the 4DPS, the keyboard controller controls the PS/2 mouse.

You shouldn't need to alter modbin. Both AMIKEY and the HOLTEK should work.

feipoa wrote:

The mouse header pins on the MB-8433UUD and the 4DPS are very different. […]
Show full quote

The mouse header pins on the MB-8433UUD and the 4DPS are very different.

MB-8433UUD
PIN Assignment
1 Data
2 N/C
3 GND
4 Vcc
5 CLK

4DPS
PIN Assignment
1 CLK
2 Data
3 GND
4 N/C
5 Vcc

On the MB-8433UUD, the chipset controls the PS/2 mouse (the chipset contains a keyboard controller chip). On the 4DPS, the keyboard controller controls the PS/2 mouse.

You shouldn't need to alter modbin. Both AMIKEY and the HOLTEK should work.

Thanks for the response and for pointing out what I missed... mosue works fine now.

I'm really hoping to find a suitable replacement for my MB-8433UUD, since it's been running unstable for a while now, even at stock clocks for any CPU used. Turns out there's really bad corrosion on the CPU socket, and I don't think I can fix it - but I'll give replacing the whole socket a go at one point. The corrosion is probably caused by the mainboard being left outside in the elements at the recycling center I got it from.

If I can get the tomato going with a PS/2 mouse, and it can run 160MHZ for an AMD and 120MHz for a cyrix stable @ 3.6v, I'm happy.

The issue I have with the 4DPS is that it can only accept single-banked cache, so it is unlikely that you will be able to get the L2 cache timings stable at the fastest settings when using a 40 MHz FSB, this is especially true when using a Cyrix 5x86-120. You might have better luck with an AMD X5-160 with 256 KB though.

This could be some incompatibility triangle between the KBC, Mouse and Keyboard... and sometimes those go away if you turn off typematic in Advanced CMOS Setup.... sometimes you just have to give up on an older keyboard, or confine it to slower machines.

feipoa wrote:

The issue I have with the 4DPS is that it can only accept single-banked cache, so it is unlikely that you will be able to get the L2 cache timings stable at the fastest settings when using a 40 MHz FSB, this is especially true when using a Cyrix 5x86-120. You might have better luck with an AMD X5-160 with 256 KB though.

Yeah, I noticed the single banked cache. It does take a performance hit compared to the biostar - but it's perfectly stable at 2-1-2 / 40MHz, even with enhancements enabled (at least in DOS). All benchmarks run fine, it scores 61.24 in speedsys with enhancements enabled and 54.47 w/o enhancements. I'm using 15ns UMC 256k cache chips and 60ns FPM ram. Quake runs fine, it gets 15.8 fps with fast fpu on and a cirrus logic GD5446, 13.4 with sound. This is quite a bit slower then the 17.1 fps the Biostar could get - guess the single banked cache is to blame. Red Alert for dos also runs fine - played a couple of hours since I got the PS/2 port to work (thanks for that).

Thing is, I have a hunch I'm not running the fastest bios - I think 4DSP400A is faster then 4DPS2013 - and it has more settings to mess with.

To sum up, even if it's slower then the biostar, what I'm looking for right now is some semblance of reliability. My LS486e board is very reliable (I use it for my pentium overdrive rig), and faster then the 4DPS, but no PS2 port (and I'm kinda tired of cleaning mouse balls every couple of days and relying old clumsy serial mice).

I'm planning to hunt for another Biostar MB8433-UUD as soon as I can, but they are getting really rare, and finding one is a bit of a pain in the ass (or wallet). Hopefully I can trade something from my soundcard collection for a working board at one point.

All this was to finally complete my "childhood PC", since I got most of the components in some form or another, including the case, witch, except for the 133MHz Cyrix 4x (witch I don't have), was the hardest to find so far. Turns out it originally had some version of the ECS Elite UM8810PAIO mainboard - the one with Award BIOS and a PS/2 port on top of the AT port - and I got this from a reliable source - the shop where I used to go get it fixed / upgraded back in the day. They indeed confirmed the CPU was a Cyrix 133Mhz running at 4x33, as well as the rest of the parts - 800MB Quantum Fireball (I'm using a 3GB Quantum Fireball for this build) - a Cirrus Logic CL-GD5446 made by Eagle computer, a Genius Yamaha YMF714 chipset card, and it ran 4x8MB FPM ram at one point. The machine was made by a romanian company who today only builds point of sale PCs, ATMs and cash registers. They confirmed the original configuration of the machine, as being a mid end model sold in october 1996.

BitWrangler wrote:

This could be some incompatibility triangle between the KBC, Mouse and Keyboard... and sometimes those go away if you turn off typematic in Advanced CMOS Setup.... sometimes you just have to give up on an older keyboard, or confine it to slower machines.

I've encountered the compatibility thing on the biostar - it will outright refuse to detect some PS/2 mice, while others work fine. In fact, classic ball ps/2 mice work great, while optical mice - some are not detected, others cause the system to crash the KB controller after 10-20 mins of use (no kb or mouse input, tough the game / program still runs).

P.S. @ feipoa - I'm having trouble getting my POD to perform as expected in GL quake... in dos quake, it gets 19-21 fps with sound (19 with FPM, 21 with EDO witch also allow for tighter timings on the LS486e for some reason) - but it only gets 23 fps and quite a few slowdowns in gl_quake (voodoo 2 used). I also tested on my biostar (when it was working), and got even slower results... I got 25 fps w/o sound on the 120mhz cyrix (enhancements enabled), so I expected the POD83 to be faster... what gives?

With identical hardware, the SiS 496/497 shouldn't really perform worse than boards with the UMC 8881/8886. In fact, it should be marginally faster. You may have the system configured a little differently, see for example, Help with SiS 496/497 Tomato 4DPS 486 motherboard

I would suggest running the 4DPS at 33 MHz with the Am5x86-133 and comparing the performance to the MB-8433UUD at 33 MHz to establish a baseline. From my experience, SiS 496 boards need some BIOS setting set which cripples performance when a 40 MHz FSB is used, which I think is due to the single banked cache. My SiS 496 boards with double-banked cache did better with 40 MHz and fast timings. I did not extensively test 256 K, single-banked cache on SiS 496 boards, and dropping the cache quantity down helps a bit with faster timings. Single-banked cache stability with SiS 496 boards can be elusive to test for as the systems often appear initially stable. Try running it long-term and run some Windows 9x games. Anyway, I found the Cyrix 5x86-120 to work better with UMC 8881 boards compared to SiS 496 boards when a 40 MHz FSB is used.

From memory, I recall 4DPS400A had stability issues with my 4DPS boards with certain hardware configurations, so I created 4DPS2013 from 4DPS172f.

What revision of MB-8433 are you using? I recall reading that the earlier revisions have issues with the PS/2 mouse.

Did this shop have actual records of the bill of sale to you for the system, or are they recalling from their own personal memory? If they were realy selling Cyrix 5x86-133-containing systems, were they in Evergreen upgrade modules? The reports online suggest that all the Cyrix 5x86-133/4x chips were sold to CPU upgrade companies.

For POD speeds in GLQuake, take a look at this thread, A brief comparison of Voodoo-Quake results on a 486
I get 36.1 FPS with a POD100 and a GF2. I haven't done much POD83 tests with a Voodoo2.

The company has records of all systems sold. My system was part of a lot sold to the company my dad worked at - 42 machines in total, all identical, all with 133MHz cyrix CPUs, 8MB of ram, 800MB Quantum fireball drives, and CL-GD5446 video cards made by Eagle. It seems they got the CPUs + ECS motherboards as a combo, and they were pretty heavily discounted. Source - a chinese company responsible for selling ECS boards right out of the factory. While true that your average end-user could only but the 133Mhz part as an upgrade kit, OEMs and large resellers had access to the part. The CPU was ceramic, branded Cyrx 586 133GP/4x. On top was plain heatsink held on with plastic retention clips, and a 40mm fan. The motherboards lacked a 4v option, instead had 3.65v required to run these chips, just like the Zida 4DPS. All boards had PS/2, but the PCs came with a serial mouse (and a nice mechanical keyboard made by chicony) Some kits came with pentium overdrive CPUs instead of the cyrix 586, and the company sold those with 16mb of ram, a 1,2GB western digital HDD and a 2MB ATi Mach64 video card. The cyrix machines were shipped with 14" crt monitors made by either Logix or ADi (budged chinese companies) as well as a OKI microline dot matrix printer. The POD models shipped with a 15" AOC monitor and an Epson dot-matrix printer, as well as a serial bar code scanner.

After further testing, I can confirm that 4DPS400A has some issues with certain hardware - artefacting with S3 virge and trio64 cards when fsb 40 is selected, IDE controller issues when loading CD-ROM drivers in dos, and so on. These seem to occur when using an AMD cpu, and sometimes with the pentium overdrive. It will hang on boot with the AMD @ 160MHz unless I set cache timings to 3-2-3 and memory speed to slower, and the IDE controller will "missfire" when using the POD, even at 33MHz fsb. I had no issues with the cyrix @ 120MHz - got it stable with 2-1-2 cache timings, 0ws for memory, memory CL set to 2-1-1-2, and speed to faster.

4DPS2013 on the other hand runs great with the AMD, stable at 160MHz 2-1-2, memory set to faster, CL 2-1-2-2. GL_Quake gets 17.8 fps.

The POD83 is SLOW on this machine. 15.4 to 16 fps regardless of what I try. The Cyrix scored a whopping 18.6 fps with enhancements enabled @ 3.6v (JP27 set to 2-3, JP28 empty, JP30 empty), almost as much as the POD does on the Biostar UMC board.

The POD83 seems slow on anything I try to use it on. On the 4DPS it's particularly slow since the sistem will hang if I set cache tag to 7 bits. If it didn't, I think it would get over 19 fps w/o sound.

I seem to have diagnosed my MB-8433UUD-A (after wasting 2 hours replacing the dirty corroded CPU socket) - the culprit is the bios - not the chip, but the ROM itself. I stuck in the boards original eprom, and all issues are gone (it's slow tough, and optical PS2 mice still hang the KB controller). My board is an early revision made in '95. Can't tell you what revision since it's not printed on the PCB (possibly rev 1.0?). It's possible I accidentally flashed a bios that was meant for the non-A version, or a later revision, witch would explain the weird behavior.

Interesting about the Cyrix 5x86-133/4x. That's the first instance I've heard of an OEM using these CPUs.

It is possible that your motherboard doesn't properly support the POD in write-back mode. Many 486 boards do not.

The early revision of your MB-8433 may be why you are having mouse issues.

You should get an EEPROM and flash to the latest Biostar BIOS.

I'm not sure what the 'A' represents on this board.

feipoa wrote:

Interesting about the Cyrix 5x86-133/4x. That's the first instance I've heard of an OEM using these CPUs.

OEMs in east europe loved the 120 and 133MHz cyrix as well as the 133MHz AMD 586 - they were extremely popular both here and in russia, as well as parts of asia. The reason is simple - PCs were EXPENSIVE, and income was low (still is in most parts). People didn't have the financial power to buy a shiny new pentium, so the old 386 had to be pushed a little bit further - both performance and usage time-wise, or they had to settle for a brand new (old stock) 486 class machine. Pentiums were very rare here, and I'm not talking about the original socket 4 pentium - I'm talking about most pentium 1 chips. Intel only really started selling decent volumes of chips here in late 1998 / early 1999 with the 3rd gen celerons, and that's because of cheap an cheerful chinese companies like lucky star (lucky tech), Matsonic, PCChips and so on, who sold boards for half the usual price. Even so, last gen intel processors were too expensive for most individuals, so most opted for alternatives like Cyrix/IBM and AMD. In fact, well over half of all socket 7 motherboards sold in 1997 were sold with either a Cyrix/IBM or an AMD chip - and these were considered high end systems. And why not? You could either buy a 133MHz pentium 1, or for the same money (~320 USD) in 1997 you could buy a Cyrix PR233 or a 200MHz AMD K6, witch performed better in productivity apps, and performed about the same in games. In 1999, most machines built here were either super socket 7 machines with AMD K6-2 CPUs ranging from anywhere between 300 and 450MHz or older stock Slot 1 i440bx boards with FPGA celeron processors (plus slot 1 to socket 370 adapters), ranging from 266 to 500MHz. All this despite the intel / AMD GHZ race that was going on. Those machines were like flying saucers - we read about them, but we merely dreamed of seeing one ourselves, let alone owning one. Personally, in 1999 I was looking over the AMD K6-II, deciding what chip/board combination I should get as an upgrade for my aging 586, and I remember thinking about compromising about ram amount so my parents could afford to buy me a faster chip. I think the fastest CPU sold here back then was a 700MHz Slot 1 Athlon, and one of those cost an arm and a leg (but hey a kid can dream).

My point is, these chips were sold in eastern europe up to 1999-2000. In 1999 you could still buy a brand new 486 here, just like you could get a 386 in 1995 or a 8086/8088 in 1986-1988, even tough much newer and faster machines were available. The western world loved to flog yesterday's hardware here at good prices, because they sold well. Hell, in 1990, local romanian PC companies were building custom Z80 based machines, even tough the 386 was the norm, and 486 chips just launched.

feipoa wrote:

It is possible that your motherboard doesn't properly support the POD in write-back mode. Many 486 boards do not.

I think I'm going to try running an LS486e bios in the Zida board, since the POD seems to work fine in WB mode on that. Both use the same chipset and have a very similar layout.

feipoa wrote:

The early revision of your MB-8433 may be why you are having mouse issues.

yeah. Just wish there was a way to fix it.

feipoa wrote:

You should get an EEPROM and flash to the latest Biostar BIOS.

I have quite a few eeproms, the issue is finding the right BIOS. The latest bios for the MB-8433 causes all kinds of stability issues, even at stock clocks.

It seems that a fast stable 486 is my "white whale". I have loads of parts, but nothing works properly, and fast parts like a 586GP-120/4x still elude me.

486 boards were still sold new in 1999 in N. America as well. I recall seeing the PC Chips M919 being sold without the cache for $99.

How did the LS486e BIOS workout on the 4DPS? You might need to modify the LS486e BIOS using modbin to allow for the PS/2 mouse. There is a checkbox you need to tick.

It is unlikely that a Cyrix 5x86-120/4x will work stable at 133 MHz. Mine does not.

feipoa wrote:

486 boards were still sold new in 1999 in N. America as well. I recall seeing the PC Chips M919 being sold without the cache for $99.

I apologize. What I meant to say is that about half of all PCs sold in 1998-1999 were new old stock 486 machines - I know first hand due to working part time after school for one company - loads of 486's rolled out the door - some new, some second hand. Some re-sellers were using refurbished or second hand mainboards for "new" machines, while others were using new old stock 486 parts and calling these "budget computers". Cyrix had it right with the mediaGX - most people were perfectly happy with these 486's back in '99 and even up to 2001. If cyrix would have developed a cheap mediaGX mainboard with AGP, L2 cache and a faster clocked 586 CPU, it would have sold here very well.

feipoa wrote:

How did the LS486e BIOS workout on the 4DPS? You might need to modify the LS486e BIOS using modbin to allow for the PS/2 mouse. There is a checkbox you need to tick.

Haven't had the chance to try yet.

feipoa wrote:

It is unlikely that a Cyrix 5x86-120/4x will work stable at 133 MHz. Mine does not.

My 5x86-120GP/2X works fine at 66x2, even with fast FP on @ 3.7 volts - BUT - the mainboard I used it in died. It was a Pine PT-430A from a PoS machine. I guess running it at 66MHz killed it - either that or something else. Now my MB-8433 is acting up with the latest bios, and with the stock bios it's stable but way to slow, and the 120 is not stable at 2x66 on it, regardless of voltage or bios settings. I've invested way to much time in this board, so I'm giving up on it for now.

A 4X chip would be a lot easier to run in most boards, but I can't seem to be able to get a hold of one. Meh - I'll probably come across one at one point - even if I have to trade my voodoo 5 for it.

Well, the latest Biostar BIOS worked on my original version MB-8433UUD. Try running your board at 33 MHz and adjust the CMOS settings.

I'm pretty sure that chips marked as Cyrix 5x86-120/4x were originally meant to be Cyrix 5x86-133/4x chips, but they failed to qualify at 133 MHz. So the 120/4x chips were specifically tested and failed at 133 MHz. The Cyrix 5x86-120/3x chips would not have been tested at 133 MHz, so there is a greater chance of getting the 5x86-120/3x chips to run at 133/2x.

Cyrix 5x86-133/4x chips are a lot rarer Voodoo 5 boards, so I don't think anyone would make that trade.

kanecvr wrote:

feipoa wrote:

It is unlikely that a Cyrix 5x86-120/4x will work stable at 133 MHz. Mine does not.

My 5x86-120GP/2X works fine at 66x2, even with fast FP on @ 3.7 volts - BUT - the mainboard I used it in died. It was a Pine PT-430A from a PoS machine. I guess running it at 66MHz killed it - either that or something else. Now my MB-8433 is acting up with the latest bios, and with the stock bios it's stable but way to slow, and the 120 is not stable at 2x66 on it, regardless of voltage or bios settings. I've invested way to much time in this board, so I'm giving up on it for now.

A 4X chip would be a lot easier to run in most boards, but I can't seem to be able to get a hold of one. Meh - I'll probably come across one at one point - even if I have to trade my voodoo 5 for it.

I had an Amptron/Elpina UMC chipset board and a 5x86-100... Now after a few months of owning it, sometime in 96, I stumbled across the higher clock settings and ran it at 2x60, I also found 66Mhz, I think, thing was, it didn't actually seem quite as fast at theoretical 133 as 120, benchmarked slightly lower, I guess it was extra wait states or something starting to kick out recoverable errors. Apart from that it did seem to run okay if you kept it cool. I had a clip on fan jammed on bonded on heatsink.

(Thanks very much for documenting the high clock speeds on here, just about everyone I ever tried to tell about them called me a n00b, liar or crazy.)

kanecvr wrote:

How did the LS486e BIOS workout on the 4DPS? You might need to modify the LS486e BIOS using modbin to allow for the PS/2 mouse. There is a checkbox you need to tick.

What an incredible coincidence, as I have been doing exactly that. You need to set the Keyboard controller to Award in Modbin otherwise it won't work.
Other than that You lose 40Gig hdd support and gain the Y2K bug.
That was with the LS486E-D bios.

I am playing around with other bioses so will update this when I have some success.