jakethompson1 wrote on 2021-09-29, 17:48:

Vipersan wrote on 2021-09-29, 17:38:

Finally got the system to boot from floppy... There is an option in bios options to disable CPU internal cache. set to enable by […]
Show full quote

Finally got the system to boot from floppy...
There is an option in bios options to disable CPU internal cache.
set to enable by default.
Disabling this allows to boot from floppy..
Faulty CPU ???
If it is the cpu at fault then I have 2 x DX4-100 cpus with exactly the same fault.
what are the chances ?

You have one of the jumpers on your board relating to a Write-Back vs. Write-Through CPU set wrong (or the board is missing one of said jumpers)

Thanx
I will closely examine the board Jake ..
A lot of jumper options on this one...and I'm working blind without a manual.

Now this is curious..
When I got this mobo the first 4 cache sockets were populated with 32 pin GLT751208 512k chips.
I naturally assumed the mobo jumpers were set up for this config.
But having got hold of a very rough scan/photocopy of the manual showing cache configurations...and relevant jumpers.
There isn't a config that shows 32 pin chips fitted.
All are shown with 28 pin chips installed with permutations all the way up to 1mb.
and the rightmost 4 pins in the socket always left empty.
So no config is shown that support 512k 32 pin chips at all.
I dont have any 28 pin chips but I'm going to try and get some.
which might take a while.
In the meantime I will pull all 4 of those 512k chips and the TAG ram again...and see if there is an option to run without external cache enabled in the bios.
I can't see that this would explain why disabling internal CPU cache would allow boot from floppy but then this is my first experience of external cache so anythings possible.
This motherboard must be able to run without external cache installed surely ...but who knows ...maybe it doesn't.
Answers on a postcard to .....
If anyone has one of these motherboards working ...please upload a photo so I can compare.
rgds
VS

you say you didn't find a manual... yet this was the first hit on my google search.

Is this the wrong manual ?

GA - 486VS / 486VF USER'S MANUAL

edit: yes there's some captcha you got to go through to see it, meh.

weedeewee wrote on 2021-09-29, 20:17:

you say you didn't find a manual... yet this was the first hit on my google search. […]
Show full quote

you say you didn't find a manual... yet this was the first hit on my google search.

Is this the wrong manual ?

GA - 486VS / 486VF USER'S MANUAL

edit: yes there's some captcha you got to go through to see it, meh.

I think there was a shift on page 3 to working on a Soyo 4SA2.

Vipersan, I'm assuming you've downloaded the bundle from here http://www.vogonsdrivers.com/getfile.php?fileid=731
I'd assume that the board can run without external cache by disabling it in the BIOS, at which point the external cache jumpers won't matter.
Seems curious that the internal cache setting makes it fail to boot from floppy, but not the hard drive.

Yep ..I now have that bundle Snufkin..
Very useful info...and a couple of bios dumps to try.
Yep ..we are on the Soyo now Weedeewee...
In fact it was in the downloaded bundle where I found it was suggested that 32 pin cache wasn't supported looking at the cache options that is.
I will be trying other bios dumps as soon as I have some chips flashed...but I still can't understand why cpu cache enabled stops floppy boot but allows HD boot..
I may get to the bottom of this mystery eventually...but I have some learning to do first I think.

I'm sure I have a DX2-66 somewhere ..and maybe a slower cpu might provide some pointers.

Grrrr :-p

any chance of getting the board out of the case and getting a good photograph without any other cards or cables plugged in?
and getting a bios dump
since the Soyo SY-4SA2 isn't listed on win3x/uh, though the 4SA5 does seem to be marked as also known as 4SA2 45A2 and 45A5
and the 4SA5 does have a jumperlisting.

edit: http://www.win3x.org/uh19/motherboards/4718

Even stranger behaviour with the DX2-66
Highly Odd ...Highly unstable and illogical.
I have had it boot from hard disk and from floppy ..
ctrl alt del ...never works...
The keyboard and system stops functioning ...random..
I have to power down and wait several minutes between each boot attempt...
It never freezes if and when I manage to get into bios.
This makes making bios changes difficult as you can imagine.
This _could_be timing issue/clock or some such ...and once an error ocurrs I then have to wait several minutes before trying anything again...
is the bios imaged/written into memory somewhere ?
Not that I have any guarantee this will have any effect ...but I'm seriously thinking about recapping the electrolytics on this motherboard...and there are many.
where exactly is the CMOS stored ...
I have had motherboards in the past that have to be removed from any source of power ...battery etc ...and left for days sometimes until they 'forget'
This is one weird fault.

If you have the old jumper settings for the dx-100 that's a nice overclock for that chip 😀

I mean that it's too much high and fails. Looks like a too far OC.

Vipersan wrote on 2021-09-29, 21:17:

but I still can't understand why cpu cache enabled stops floppy boot but allows HD boot..

When write-back L1 cache was introduced to 486 CPUs, the pinout changed. With the prior write-through L1 cache, the only cache coherency issue that arises is if a device performs a DMA transfer to RAM (bypassing the CPU), the affected addresses must be invalidated in the L1 cache. With write-back, it is more complicated, because DMA devices must also be blocked from reading from RAM until it is ensured there is not newer data sitting in L1 that needs to be flushed to RAM. Thus, the pinout change.

The floppy controller is the only stock ISA DMA device on an AT-compatible system. The hard drive uses programmed I/O, where data transfers go through the CPU and the caching issue doesn't crop up as soon.

This page has a list of the pins to check. http://web.inter.nl.net/hcc/J.Steunebrink/amd5x86.htm
I'm not sure why the page says you have to use a DOS 5/6 floppy (as opposed to Win9x) to test write-back L1.

Vipersan wrote on 2021-09-29, 22:57:

is the bios imaged/written into memory somewhere ?

If Shadow ram is on for the system bios, then yes.

Vipersan wrote on 2021-09-29, 22:57:

where exactly is the CMOS stored ...

It's in one of the big SIS chips.

jakethompson1 wrote on 2021-09-30, 03:36:

When write-back L1 cache was introduced to 486 CPUs, the pinout changed. With the prior write-through L1 cache, the only cache coherency issue that arises is if a device performs a DMA transfer to RAM (bypassing the CPU), the affected addresses must be invalidated in the L1 cache. With write-back, it is more complicated, because DMA devices must also be blocked from reading from RAM until it is ensured there is not newer data sitting in L1 that needs to be flushed to RAM. Thus, the pinout change.

I was thinking in the same lines, though it's the why does the floppy access work ok once booted from hard drive that has got me stomped.

Latest update..
I have had the system booting with cpu cache enabled using my Intel 486DX2-66 fitted..
Once I realised that the instability issues were down to the fact I was still running at 3.4 core voltage.
I changed the JP40 jumpers to the alternative 5V position and all booted ok both from floppy and HD.
Silly me ..
But it must prove something ?
The DX2-66 certainly requires some cooling now.
??

So, DX2 boots from floppy with internal cache enabled but DX4 doesn't (tested with 2 different DX4s)? Both work if internal cache is disabled?

Can you post a photo showing the whole board with the jumper settings you used for the DX4 (with the CPU fitted)?

Which dos version are you trying on the floppy to boot from ?

I'll grab a photo later with the DX4 reinstalled.
The DOS disk was made on this system from Win 3.11
So it is in fact a win 3.1 system disk
Also tried and failed with a windows 95 installation disk ...and a win98se startup disk..
These also failed to boot
As did disk 1 of SOS 6.22 installation disk
rgds

Windows 3.x doesn't have system files. those come from the underlying ms/pc/whatever-dos system it is running on. the fact you formatted it in windows is rather meaningless.
Good to know you tested with w95 and 98Se disks as well.
Still leaves the same question jakethompson1 has. 😀

cheers.

I'll correct myself ..
You are quite right of course Weedeewee
The DOS disk made whilst booted to the hard drive containing dos 6.22 and win3.11..requiring date and time setting each time it boots...from that floppy.
So effectivelly a dos boot disk with basic minimum startup capabilities..

Vipersan wrote on 2021-09-30, 13:58:

The DOS disk made whilst booted to the hard drive containing dos 6.22 and win3.11..requiring date and time setting each time it boots...from that floppy.

copy con a:autoexec.bat
@echo off
^Z <- ctrl-Z

Since I'd already taken these images ..thought I might as well post them
All taken using the DX2-66 installed...
booted from 4 x floppies of various flavours.
plus cache settings in the bios...
plus a photo of the S0yo set for DX2

and the 5th/6th image

Thanks for the photos. If you can when taking a picture of the motherboard don't have any cards or cable in the way so people can see the whole thing. I had a look at the jumpers and manual and it looks to me like there are a couple out of place for a DX2 (&E5VIX, non-write back internal cache, which I think it says on the CPU fitted). The manual doesn't show JP39 as having any jumpers and it looks like there's a green one fitted on the bottom pair. Also JP34 has a yellow jumper fitted and the manual doesn't show it. No idea what they actually set, and it boots with the current settings so they don't sound like an immediate problem.

[Ah, it looks like they were a hold-over from the setting for the AMD DX4 V8T jumpers]