VOGONS


First post, by matskatsaba

User metadata
Rank Newbie
Rank
Newbie

Hello!
Got a strange problem:
286 board with 25mhz harris cpu. Accepts chip ram or 4x simm modules.
I threw in 4x1mb (and threw an isa 4mb and an isa 2.5mb xms boards), and was working fine. Got the vibra16 pnp working too, with a little messing.

Now, I got my hands on 4x 4mb simms, so removed the xms cards, and tried the 4x4megs.
It is working, bios counts them allright, did over 100 passes with memtest, everything should be fine but it isn't.

Thing is, if I want to initialize the cdrom, the computer hangs. Without the cdrom driver, everything runs.
Strange thing is, that if I put back the 4x1meg modules, everything is cool, cdrom driver loads and it works.
Any suggestiongs?

cdrom is a creative 32x, but tried with a 8x mitsumi as well, same results.
using a random isa hdd controller card, the cdrom is a 40pin ide.
technically I could bruteforce a solution by putting in a scsi ultraplex40, but come on, it's a 286 😁
Also curious how I seem to have ended up with too much memory, feels a bit nonsense.

Reply 2 of 58, by Grzyb

User metadata
Rank Oldbie
Rank
Oldbie

I would say some bug in the CD-ROM driver...
Have you tried various drivers?

Żywotwór planetarny, jego gnijące błoto, jest świtem egzystencji, fazą wstępną, i wyłoni się z krwawych ciastomózgowych miedź miłująca...

Reply 3 of 58, by weedeewee

User metadata
Rank l33t
Rank
l33t

try with 2x4M +2x1M modules, just for fun 😀

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 5 of 58, by matskatsaba

User metadata
Rank Newbie
Rank
Newbie

some test results:
same creative32x cdrom

4x1meg rams:
oakcdrom.sys - not working (numlock and tcrl-alt-del is responsive)
vide-cdd.sys - works flawlessly
gscdrom.sys - not working: mostly hard freeze, once there was a restart.

2x1meg+2x4meg (2x1megs in lower address space, 2x4meg in higher address space):
oakcdrom.sys - not working (numlock and ctrl-alt-del responsive)
vide-cdd.sys - works
gscdrom.sys - hard freeze

2x1meg+2x4meg(1megs in upper, 4 megs in lower address space)
no POST

2x4 megs in lower address space:
oakcdrom.sys - soft freeze, as always
vide-cdd.sys - hard freeze
gscdrom.sys - hard freeze

2x4 megs in upper address space:
no POST

8x4megs:
oakcdrom.sys - soft freeze
vide-cdd.sys - hard freeze
gscdrom.sys - hard freeze

Reply 7 of 58, by weedeewee

User metadata
Rank l33t
Rank
l33t

I'm just guessing here... If I encountered this situation... which sadly I can't since the only two 286 mainboards I have are annoying when it comes to memory, one is limited to 4M and the other is an old ibm with 512K...

I'd try more cdrom drivers, try the freedos ones, or freedos itself. With 16MB i'd even try an old OS2... no idea if it has ide cdrom support 😀

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 8 of 58, by mkarcher

User metadata
Rank l33t
Rank
l33t

The "no POST" configurations are expected. Dual-bank 286 boards often do not tolerate the first bank being unoccupied, and require the second bank to not have modules that are smaller than the modules in the first bank. The CD driver issue looks like oakcdrom.sys and gscdrom.sys generally are incompatible with your system (for whatever reason), so the interesting question is why vide-cdd.sys only works in some configurations. I notice the pattern that vide-cdd.sys works as long as the low bank is not using the 4M simms, so it is likely that you indeed have problems with the 4M modules, in a way that memtest does not fail, but the CD driver does.

Memory problems that are not found by memtest could be refresh problems, but the refresh requirements for 9-chip 4M modules and 3-chip 1M modules are often the same (requiring a 10-bit counter, and refreshing one row per 16µs). Can you read the type number of the memory chips? What chipset does the mainboard use?

Reply 9 of 58, by Horun

User metadata
Rank l33t++
Rank
l33t++

Good points mkarcher !
What motherboard / chipset are you using is important for us to know. Did that motherboard come natively running at 25Mhz or is that just the 286 cpu rating ?
Also: What IDE controller are you using and What version of DOS are you running ?

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. Stuff: https://archive.org/details/@horun

Reply 10 of 58, by Grzyb

User metadata
Rank Oldbie
Rank
Oldbie

What do you mean by "memtest"?
I guess it's neither MemTest86 nor Memtest86+, as I never could get them to work on a 386, leave alone a 286...
My preferred tools for memory test on a 286 are:
- Checkit 3.0, with "quick memory test only" set to NO
- Checkit Pro 1.11, also with some "comprehensive" option

Anyway, if there's only one driver (VIDE-CDD.SYS) that works depending on the memory amount, there's still the option that the driver is buggy - so try more drivers.

If you can't find a working driver for ATAPI drives, there's still the option that's more appropriate for a 286: one of those proprietary-interface drives, like Misumi, Panasonic, Sony...

Żywotwór planetarny, jego gnijące błoto, jest świtem egzystencji, fazą wstępną, i wyłoni się z krwawych ciastomózgowych miedź miłująca...

Reply 11 of 58, by maxtherabbit

User metadata
Rank l33t
Rank
l33t

My suspicion is that the chipset on this board does not actually support 4MB SIMMs. But why it POSTs and counts all the memory, I can't resolve with my theory

Reply 12 of 58, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie

You need to make sure (on a 286) with 16mb

That the bios does not try to relocate or use the 384kb of memory as extended

The drivers you are using aren’t trying to go up to the very top of memory

My best guess is the driver is trying to use a memory location that is mapped to something else.

Most 286’s need to put a memory hole at the top of memory for 16 bit ISA cards that use the top of the memory space,
many do not

The 384kb area also needs to be a memory hole.

Most 286’s handle 12mb much more successfully.

Reply 13 of 58, by Disruptor

User metadata
Rank Oldbie
Rank
Oldbie
Grzyb wrote on 2021-05-08, 00:02:

What do you mean by "memtest"?
I guess it's neither MemTest86 nor Memtest86+, as I never could get them to work on a 386, leave alone a 286...

Do not try Memtest86+ past version 4.10

Reply 14 of 58, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

What was the behaviour with 8mb?

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 15 of 58, by Jo22

User metadata
Rank l33t++
Rank
l33t++
maxtherabbit wrote on 2021-05-08, 01:56:

My suspicion is that the chipset on this board does not actually support 4MB SIMMs. But why it POSTs and counts all the memory, I can't resolve with my theory

That's a possibility. Though I've never seen a 286 myself that couldn't handle 4MB mimimum (1MB SIMMs).

Not sue about 4MB SIMMs, though.
Occupying the full 16MB address space isn't trouble free perhaps.
Some ISA VGA's used the 15-16MB region for the frame buffer.
That region is ATs' counterpart to the 640KB-1024KB range of the PC/XT or the 3,5GB-4GB range of Pentium era PCs.

Btw, by definition, a "chipset" was something that the more sophisticated 286 PCs had (NEAT and successors).

And such sophisticated PCs surely were meant to be able to run a real OS, like OS/2 or Xenix,
or Netware, which in turn required lots of RAM.
Okay, enough nitpicking. I'm quite now. 😇

"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel

//My video channel//

Reply 16 of 58, by maxtherabbit

User metadata
Rank l33t
Rank
l33t

There were very few true AT-clones brought to market that used discreet logic for everything. The vast majority of clone boards used a chipset of some kind, most of which did little more than provide basic AT functionality.

OP's system does work fine with 4x1MB SIMMs. OAKCDROM does NOT work on a 286 period, regardless of memory - it requires a 386. I can't speak to the GS whatever one because I've never tried it, but I suspect the issue is the same.

Reply 17 of 58, by Jo22

User metadata
Rank l33t++
Rank
l33t++

Ah, I see. I've used a PAS16 and the Trantor driver on my 286 back then.

Edit: Does OAKCDROM use 386 Real-Mode instructions? If so, EMU386 could help.

"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel

//My video channel//

Reply 18 of 58, by maxtherabbit

User metadata
Rank l33t
Rank
l33t
Jo22 wrote on 2021-05-08, 18:43:

Ah, I see. I've used a PAS16 and the Trantor driver on my 286 back then.

Edit: Does OAKCDROM use 386 Real-Mode instructions? If so, EMU386 could help.

Does EMU386 work for other installable device drivers? That would be an interesting solution but I'm a bit skeptical that it would work

Reply 19 of 58, by Caluser2000

User metadata
Rank l33t
Rank
l33t
Jo22 wrote on 2021-05-08, 04:01:

That region is ATs' counterpart to the 640KB-1024KB range of the PC/XT or the 3,5GB-4GB range of Pentium era PCs.

That is just an NT issue and not a hardware issue I believe.

There's a glitch in the matrix.
A founding member of the 286 appreciation society.
Apparently 32-bit is dead and nobody likes P4s.
Of course, as always, I'm open to correction...😉