sebaz_ri wrote:

@DK
Is there a 386 mobo that have PCI bus, or this was brand new with the 486?

it was a Pentium thing that they adapted for 486'es. PCI is "usually" done via VLB -> PCI bridge, so the performance is no better than VLB. I think there was ONE chipset that had "native" pci support and don't quote me on it but I think it was the 8881/8886 chipset from UMC.

The biggest problem with the VLB-to-PCI bridge is that it seems to kill PCI's DMA performance. This means that one of PCI's biggest advantages (busmastering DMA) is gone. HDD DMA modes don't save you CPU time anymore, NICs hog the CPU, etc.

The UMC chipset might indeed have real PCI. Over in the "Ultimate 486 Thread" in Marvin you can see tests of it. Feipoa likes the chipset a lot.

Wow, we have very similar 386 machine. Possibly the same motherboard too.
I have AMD 386DX 40Mhz, Tseng Labs ET4000, 8MB Ram, difference is I have no 387 coproc and only 128kb cache. Also I have installed a Gravis Ultrasound, can't remember if it's classic or PNP or other though now. My norton SI result is 41.3, 3d bench result 16.3, also I usually test the output of most of all bunch of my gfx cards from the video bandwidth test from this demo (you don't need to run the demo, just click upper left). My ET4000 is making 85fps in this test. This is the fastest ISA I could get so far (I am wondering if someone makes more).

With this configuration I am able to run Second Reality fairly well compared to some older 386 configs with lame cards and no GUS.

p.s. Also, I had to tinker a bit with bios settings. Initially SI showed 31.2. But if I lowered cache read cycle from 3-1-1-1 to 2-1-1-1 I got up to 40. Hidden Refresh, if I enable it, gave me one more point. But mostly the cache read cycle. I honesty don't know much about what these values are, I was just trying and retesting (We have the same bios setup btw as I see from your screenshot)

Optimus wrote:

Wow, we have very similar 386 machine. Possibly the same motherboard too. I have AMD 386DX 40Mhz, Tseng Labs ET4000, 8MB Ram, di […]
Show full quote

Wow, we have very similar 386 machine. Possibly the same motherboard too.
I have AMD 386DX 40Mhz, Tseng Labs ET4000, 8MB Ram, difference is I have no 387 coproc and only 128kb cache. Also I have installed a Gravis Ultrasound, can't remember if it's classic or PNP or other though now. My norton SI result is 41.3, 3d bench result 16.3, also I usually test the output of most of all bunch of my gfx cards from the video bandwidth test from this demo (you don't need to run the demo, just click upper left). My ET4000 is making 85fps in this test. This is the fastest ISA I could get so far (I am wondering if someone makes more).

With this configuration I am able to run Second Reality fairly well compared to some older 386 configs with lame cards and no GUS.

p.s. Also, I had to tinker a bit with bios settings. Initially SI showed 31.2. But if I lowered cache read cycle from 3-1-1-1 to 2-1-1-1 I got up to 40. Hidden Refresh, if I enable it, gave me one more point. But mostly the cache read cycle. I honesty don't know much about what these values are, I was just trying and retesting (We have the same bios setup btw as I see from your screenshot)

You're not missing out on much not having a co-processor, anyway. They wouldn't become important to most people until Quake was released and required one. Prior to that, it was mainly professionals running CAD/CAM or other apps that required floating point who used them. Not much else out there needed it.

if you play a lot of simcity then a co-processor is helpful. by a lot actually. I think some flight-sims also make heavy use of it but past that.

yep, Sim city, Falcon 3.0, Scorched Earth, and a not particularly well known game called Smashing Pumpkins into Small Piles of Putrid Debris, are all games that I know of that make use of the maths copro.

I don't have it installed at the moment because I took it out to confirm that the speedsys score was affected by it, and its proven to be a lot harder to put back in without removing my optical and floppy drives - so I haven't bothered 😜

As for cache read, I remember trying 2-1-1-1 but it would hang on POST even though all my cache and TAG chips were 15ns. I haven't tried with less than 256kb though. I should give it a go one of these days as the lower latency would likely be better than the larger cache.

I remember back in 1996 I had a 386 and was running 3D Studio V4.0. It wasn't very fast but I was patient. I had rendered a little UFO and was also rendering some models from various 45 degree angles to use as sprite graphics in Doom (enemies needed a lot of different graphics 45 angles revolving). I loved that program.

Recently I tried to install it again on my new 386. I only now realized it requires a coprocessor. Gotta find one..

Markk wrote:

retro games 100 wrote:

BTW, I read Vogons user Markk's post here. He mentions using a diode, when using an external battery. Markk, should I use one for my 386 mobo? If you look at my post above, I have just attached the external battery to the mobo, without much thought! 😦

No you don't have to do that. I would only fit a diode if the motherboard doesn't have an external battery connector, and I have to connect it where the barrel type rechargeable battery was. I've only encountered this just once on a more "recent" motherboard. A soyo socket 7 one, that had barrel type battery soldered(strange for a board of that era) with no external battery connector at all.

Help!, i got a Soyo SY-5VC socket 7 mobo that's got one barrel battery and it already had leaked the BIOS chip, what value should be the diode, and what's the polarity, i'm going to use a coin cell holder with a CR2032 battery

Hey, I think that was the same board I had there. I don't know if there's a specific value. I just asked for it at a local electronics shop. The diode lets electric current pass one way. On the other end it blocks it. Check here.

Donutking does that Contaq board handles 386 SX processors?

No, 386DX and 386SX processors usually need different motherboards due to the different data bus widths. (I suppose there might be hybrid boards available but I don't think they'd be very common).

This is why a 386DX needs 4 30pin SIMMs to make a bank but a 386SX only needs two.

DonutKing wrote:

yep, Sim city, Falcon 3.0, Scorched Earth, and a not particularly well known game called Smashing Pumpkins into Small Piles of Putrid Debris, are all games that I know of that make use of the maths copro.

When you say SimCity, that is the original, not SimCity 2000 or SimCity classic? (Those games may support it too, but there were years later)

when I tested it, I had sim city classic. it was still the disk dos version. But if I recall correctly, the performance of having a co-pro was around 40-50% faster.

I see only a few benchmark programs used. Back in the days I used this benchmark programs:

Checkit (CPU, FPU, Graphics & HDD bench)
Pmips (CPU)
3dbench (my version seems different then the ones used here, it isn't 1.0c but it shows the same XXX.Y digits)
Winbech 3.11 (Windows benchmark)
Wintach (---"---)

I would like some Checkit & Pmips results, mainly Checkit FPU ones (Intel VS Cyrix VS IIT 387).

Are there some problems about licensing? Else I can upload this programs somewhere.

Isn't most of this already available here:
ftp://78.46.141.148/dos/benches/
and ftp://78.46.141.148/dos/sysinfo/ ?

They might just be. But it seems they are useless if noone is using them (posting results).

Though I don't see Checkit V3.0 there.

I also used F1GP game for benchmarking. It doesn't have a demo but you can get quite reliable speed reading on the starting grid and pressing O. I remember having 286/16 and some TI 256KB VGA card and F1GP run really slow with every detail to the minimum. Later I upgraded the mobo with 486slc/33 and using the same VGA card - I was really disappointed in finding out that F1GP run just a little bit faster - I really cursed that SLC processor 😦 Some months later I purelly by a chance replaced the VGA with Trident 8900 and the game was drasticly faster 😳 From then on I knew that the system should have balanced components 😊

question. Every time I see the mother board in the OP's machine it has the SIS chipset. I even just acquired a second board with the SIS chipset but recently I came across a machine with the exact same board except it has a UMC chipset instead. is this of any significance? unfortunately the board was damaged beyond repair due to acid damage.

It's not a UMC chipset. It's a Contaq chipset. The UMC 206 chip is just supplying the DMA controllers and a few other basic devices. The real guts of the chipset are in the two other black square chips. I think there are a few forum users that like the contaq boards. They're probably better than average.

Anonymous Coward wrote:

It's not a UMC chipset. It's a Contaq chipset. The UMC 206 chip is just supplying the DMA controllers and a few other basic devices. The real guts of the chipset are in the two other black square chips. I think there are a few forum users that like the contaq boards. They're probably better than average.

Ah, of course, silly me. thanks for the correction. So far I have to say using it it is a nice 386 board. I wonder how it stacks up to the other 386 board I have that does use the SIS "Rabbit" chipset. other then that Its pretty much identical to this one in features except its a good bit longer.

unfortunately I damaged the RAM sockets trying to get it into a smaller case and they need replaced 🙁

soviet conscript wrote:

unfortunately I damaged the RAM sockets trying to get it into a smaller case and they need replaced 🙁

that makes me VERY VERY SAD.