• Please review our updated Terms and Rules here

Ibm ps/1

I think any XMS board would work if it fits inside, remember that these CPUs are small and most ISA ram cards are full length.
To disable the rom drive, just run customiz.exe and select 'Try diskette first, then try fixed disk'
 
I was thinking more so removing it from the memory space... Oh and it appears the 2121 does not have a standard IDE interface in the typical pc memory map... I tried Xenix and it couldn't even find the IDE controller .....

Was there ever an IBM techref for the PS/1 2121 2112?

Its kind of weird that IBM built a machine incapable of running OS/2, but that was the war between the OS people and the Hardware people....

Ux0WIl.jpg

I know its not a disk size thing, as Ive got the 80MB disk...
 
Well, in the 2011 model there are 4 27C1001 EPROMs. Two have only rom drive and the other two have part of the rom drive, basic, bios, vgabios, etc. In the 2121 there are only 2 EPROMs, so maybe removing one of them could work.
 
PS/1 without PSU?

PS/1 without PSU?

Hi members i would like to know the psu pinout standard for ps/1 any model (at least for 2121).
2nd question where can find technical data sheets of vintage ps/1... scanning and sharing docs should be a good way to get it.
 
Reviving the thread! :)
Does anyone know where in Europe I can find memory cards for 2121...
I am tired of editing my AUTOEXEC removing texts in order not to load mouse driver or other programs, just to spare some memory for a particular game that needs more RAM :)
 
Deciding to bump this, because I've got questions about how the RAM slot actually works (in a 2121 context).

So, there's three reserved pins according to the 2011 reference's pinout of the 512 kiB RAM expansion. One of the pins (30) is in a logical place to put a 10th multiplexed address pin, which would give 2 MiB (one RAS, one CAS).

What are the other two used for? The pinout using "-RAS1" is suspicious, so that tells me that there's probably a -RAS2 to strobe RAS on the other bank, most likely pin 23/25... but then, what's pin 25/23? (It's also worth noting that there's both low and high byte CAS, which I wouldn't think would actually be necessary, I'd think you could just strobe four chips off the same CAS strobe, unless they're trying to optimize some sort of 8-bit DMA thing or something...)

Edit 2016-08-25: Ah, didn't look elsewhere in the reference. There's presence detect pins to define a 2/4 MiB card, and a second bank presence detect. And there is a second RAS, presumably for that second bank, and that's how they handle selecting banks - just don't send RAS to the other bank I assume.
 
Last edited:
Alright, replying to myself, with stupid PS/1 tricks, at least on the 2011 and 2121 (which is really what this thread's about, as literally every other PS/1 model began as a 386 or newer).

I slapped a Xircom PE3-10BT on my 2121, and noticed that it came up in non-bidirectional mode. Wait a minute... the 2121 is new enough that it should have a bidirectional port, right? There's no setting for this in CONFIGUR.EXE, though...

So, well, time to look at the 2011 tech ref... and sure enough, the port supports bidirectional... but the Programmable Option Select register at 0102h controls the port configuration in bits 7-5 (having, after the fact, looked at other stuff, all of this is actually very, very consistent with the PS/2 line - the 2011 being a derivative of the 8525/8530 286, as I understand). Unlock the register by clearing bit 7 of 0094h (changing it from FFh to 7Fh), clear bit 7 of 0102h (changing it from 9Fh to 1Fh), set bit 7 of 0094h again, and bidirectional works... but that doesn't persist across reboots, of course.

The 2011 tech ref doesn't say where it's stored in the CMOS RAM, though... so time to dump it, and found 9Fh at CMOS address 31h. (Turns out, this is also consistent with PS/2s, having googled for this afterwards.) The tech ref does say that CMOS is checksummed across 10h to 31h, so I shoved that into a checksum generator that calculated a bunch of common checksums, turns out it's using CRC-CCITT with a starting value of 0xFFFF. So, I checksummed the new CMOS configuration with 31h set to 1Fh, rebooted, and had bidirectional parallel.

Upshot is, here's what you need to do to enable bidirectional parallel on a 2011/2121 (should work for most PS/2s, too, really, but most PS/2s will let you change this without resorting to this):

Dump CMOS from 10h to 31h
Subtract 80h from the value currently at 31h and store it there
Checksum CMOS from 10h to 31h using the CRC-CCITT function with a 0xFFFF starting value
Store the new checksum at 32h and 33h (high order byte first)
Reboot
 
Very interesting! It would be wonderful if someone makes a little program to perform this steps automatically :rolleyes:
 
Also, fun fact of the day for the PS/1: the 2011 and 2121 don't have a VGA controller.

They have an SVGA controller - same one as used by the 8525SX. It's a pretty lame one, 640x480 256 colors is all she wrote, 800x600 isn't happening at all. There's a VESA VBE TSR for it, and there's apparently an OS/2 driver floating around (not that OS/2 works on a 2011 or 2121) thanks to the controller being used on some servers that ran OS/2 (as a cheap alternative to the XGA-2), but no Windows 3.x support was ever done for it.

The trick is that IBM only populated half of the VRAM on the PS/1... but note that there's solder pads for more.

In any case, with the stock VRAM, and a VESA TSR intended for the Model 25SX, you can get 640x400 at 256 colors (which works quite well), and 132x25 text mode is also enabled (but I'm not yet sure if it works).
 
Well, that is only correct for the 2121 model. I populated the VRAM pads in the 2121 and after loading the VESA TSR it correctly uses VESA 640x480x256 mode.
The 2011 has the same old and slow VGA chipset found in the first PS/2 series like the 8580. I think it uses an 8 bit interface because of it slowness.
 
That's what I get for trusting Google Images for a 2011 motherboard image... and it gave me a 2121 (duh, that's obviously a 386SX...)
 
bhtooefr said:
Also, fun fact of the day for the PS/1: the 2011 and 2121 don't have a VGA controller.

They have an SVGA controller - same one as used by the 8525SX. It's a pretty lame one, 640x480 256 colors is all she wrote, 800x600 isn't happening at all. There's a VESA VBE TSR for it, and there's apparently an OS/2 driver floating around (not that OS/2 works on a 2011 or 2121) thanks to the controller being used on some servers that ran OS/2 (as a cheap alternative to the XGA-2), but no Windows 3.x support was ever done for it.

The trick is that IBM only populated half of the VRAM on the PS/1... but note that there's solder pads for more.

In any case, with the stock VRAM, and a VESA TSR intended for the Model 25SX, you can get 640x400 at 256 colors (which works quite well), and 132x25 text mode is also enabled (but I'm not yet sure if it works).
Click to expand...

OS/2 2.0 does work on the 2121 BTW, they have special code for it.
 
nestor said:
Very interesting! It would be wonderful if someone makes a little program to perform this steps automatically :rolleyes:
Click to expand...

I made an attempt at this but it remains to be seen just how wonderful it is. :D Since I don't have any of these machines the program hasn't really undergone any testing worth a damn. Run it at your own risk.

Two things though;
bhtooefr said:
Subtract 80h from the value currently at 31h and store it there
Click to expand...
Doing this would really only toggle the most significant bit. My code is made with the assumption that clearing that bit enables bidirectional mode and setting it disables bidirectional mode.
Checksum CMOS from 10h to 31h using the CRC-CCITT function with a 0xFFFF starting value
Click to expand...
Apparently there's at least two different ways to do the CRC-CCITT function. My implementation is based on the C code at here.

I hope this is useful. Source is included of course. It's a bit bloated at 819 bytes but that's because I unrolled the CRC calculation loops. ;)
View attachment 32877
 
bhtooefr said:
Also, fun fact of the day for the PS/1: the 2011 and 2121 don't have a VGA controller.

They have an SVGA controller - same one as used by the 8525SX. It's a pretty lame one, 640x480 256 colors is all she wrote, 800x600 isn't happening at all.
The trick is that IBM only populated half of the VRAM on the PS/1... but note that there's solder pads for more.
Click to expand...
That's weird because usually SVGA means it supports 800x600 while VGA is 640x480.
 
You must log in or register to reply here.
Back
Top