I like the onboard ps/2 mouse input!

Markk wrote:

I like the onboard ps/2 mouse input!

So do I. 😀

This is unfortunate - I can't successfully boot up the mobo, because the CMOS battery inside the Odin RTC block is dead. When the mobo displays its BIOS POST messages, it says that the CMOS battery is dead. I did a search for OEC12C887A (seen on the RTC in the photo) on ebay, and I can find one seller who has them. The listing is bad, because there's no photo and no description. It's here. In the description, it says that "Littlediode [the seller] will provide as much information as possible about the item", when there is nothing at all about the item.

Does anyone know if I can substitute this ODIN RTC thing, for a DALLAS RTC? The reason I ask is that DALLAS RTCs are more commonly found on ebay. Thanks a lot for any info.

After some googling I've come to the conclusion that this ODIN chip is the same as the DALLAS DS12887

I recommend buying them from www.futurlec.com, have bought some of these evil chips in the past and their service has been excellent

OK thanks a lot! I'll get one or two, and then continue my testing on this board...

I had a look at the BIOS options, and there is something called "Host clock / PCI clock". Is that something which is useful? The mobo's BIOS datestamp on the BIOS POST screen says March '96, so it's quite "modern". 😉

Before buying a dallas chip, why don't you try to hack yours? I've done it twice, and both chips worked fine. Here's the RTC chip I fixed yesterday on a 386 motherboard, which was useless before.

Uploaded with ImageShack.us

Markk wrote:

Before buying a dallas chip, why don't you try to hack yours? I've done it twice, and both chips worked fine. Here's the RTC ch […]
Show full quote

Before buying a dallas chip, why don't you try to hack yours? I've done it twice, and both chips worked fine. Here's the RTC chip I fixed yesterday on a 386 motherboard, which was useless before.

Uploaded with ImageShack.us

Nice! 😁 Unfortunately, I don't have the skills you have. I'm no good at stuff like that. Also, I don't have any tools. Perhaps I could pay you to hack a few of my RTCs? 😀

Anyway, I have had some success! 😀 I had a look in some of my boxes of retro junk, and I found another 486 mobo with a DS12887A RTC. I removed this RTC, and tried it in the mobo as featured in the original post. It worked! The DS12887A RTC can replace an ODIN RTC. Thanks a lot for that tip, Amigaz. (Also, I tried a DS12887 and an RTC by BenchMarq, but they did not work.)

Believe me, the hardest part is to "cut" the plastic, where you need to find the connectors...
Most motherboards I've seen use the DS12887. But that particular 386, when I desoldered the chip, it had DS1287 written underneath. I don't know what difference a number(8) can make, but 2 spare 12887 I have wouldn't work 🙁

I have got this mobo working well. Tweaking its BIOS settings to the max, I get 83.3 in 3DBench, using a 1MB PCI Trio64V+ card. That's using a 5x86 P75 chip, running at its native speed of 133 MHz. I'm using 1 stick of 16MB EDO RAM.

I ran Doom shareware 1.9s, and set the screen to full size. I then ran doom -timedemo demo3, and I get 2134 / 1667. I ran Quake shareware 1.06, and set the screen to default, which is "2 notches" below full size. I then ran quake timedemo at the command console, and I get 14.2 FPS.

I wonder if my Quake timedemo score would improve, if I used more RAM? Currently, I am using 16 MB. Quake locks 14 MB of data, just before it begins. Also, I suppose I could use a slightly faster video card. I'm sure I can find something faster than a Trio64V+, but then again, this card did give me 83.3 on 3DBench, so perhaps it doesn't make much difference?

That's quite good performance actually. Perhaps you would see 1 or 2 more frames you used something like a Matrox, but it really doesn't make much of a difference does it?
If you add more RAM you'll get less swapping, because the system deletes bytes from RAM much less frequently.

Very pretty motherboard. Did you check bios settings stability under Win95? I love 486 boards based on UMC chipset. I think those are the best! (:

retro games 100 wrote:

I had a look at the BIOS options, and there is something called "Host clock / PCI clock". Is that something which is useful?

My guess is that's the 'magic' jumper 😉

Have you tried it yet?

That is really magic jumper, especially during finding undocumented FSB settings. (;

Thanks a lot for the info, people. I haven't got a W95 HDD at the moment, but I have got W98 and also Wme. I'll try them as soon as possble. Also, I will try to find some undocumented bus speed settings. I'll see if I can get all of this done today. I've already increased the RAM from 16MB to 32MB, and I'm using EDO.

I tried a few different PCI video cards. I noticed that a Voodoo3 did not POST. The other 2 video cards I tried were Diamond Viper V330, and Dell S3 Nitro GX. All video cards tested so far perform about the same. The V330 gave me 14.3 FPS in Quake, which was a 0.10 point improvement. The Nitro gave me 1664 in Doom, a 2 point improvement.

Quick update. I set the mobo's bus jumper to 40 MHz, and the P75 chip now runs at 160 MHz. I tried another video card, this time a 1MB Number 9 Virge 325 card.

Doom = 2134 / 1388, quake = 17.1, 3dbench = 0.00, 3dbench v1.0c = 93.9

RG100, have you tried the magic jumper yet?
If that's what it is, theres basically no limit except for what the cpu can handle 😉

Tetrium wrote:

RG100, have you tried the magic jumper yet?
If that's what it is, theres basically no limit except for what the cpu can handle 😉

Sorry, I've been insanely busy today. I've got some time now, and so I will set the BIOS' "magic jumper" setting to "1/2", and then set the mobo's bus jumpers to undocumented settings. If I hit one of the jackpots (60 or 66 MHz bus speed), then the BIOS setting of "1/2" will hopefully allow the PCI video card to still function correctly, rather than me getting no POST display signal, because the PCI bus is being overclocked too much. Actually, what I've just said - is that correct? Is that technically accurate? I'll post back later on today with some results...

Success! I got the mobo running at 180 MHz, using an overclocked P75 (133 MHz) CPU. Speedsys is below, and I get 106.7 in 3DBench 1.0c, and 2134/1260 in Doom timedemo demo3. Quake won't complete its timedemo demo1 test because of instability.

I could not get the cool ADZ P75 variant to overclock to 180 MHz, but I could get another hotter P75 variant to overclock OK. I can't remember its code name, because it's currently sitting snuggly underneath a socket 370 (!) heatsink at the moment. I am using a CoolJag copper heatsink, and although the heatsink clips do not fit the socket 3 heatsink "lugs", the heatsink itself sits perfectly OK on top of the CPU.

I set the mobo's CPU voltage to 5V. I am using an S3 325 PCI video card. The "magic jumper" BIOS setting is set to "2/3", which sets the PCI bus to 2/3 of the system bus speed. 2/3 of 60 is 40, which the PCI S3 video card appears to be able to cope with.

Also, the only maxed out BIOS setting I had to alter to get the mobo running at 180 MHz, was the DRAM read wait state. I had to increase this value from 0 to 1. All other maxed out BIOS settings were left unaltered.

To find the undocumented 60 MHz bus speed setting on the motherboard was very easy. I reckon that if you are using a UMC based PCI 486 mobo, the bus jumper settings are all the same. You will see 3 jumpers on this type of motherboard to control the bus speed. If you want 60 MHz, simply leave the middle jumper closed, and the two outer jumpers (to the left, and to the right of the middle jumper) open. Also, I tried to set the "magic jumper" BIOS setting to "1:1", meaning that the PCI bus was overclocked to a crazy 60 MHz setting, but it would not boot up DOS. Interestingly, it POSTed OK.