So, I've been watching a YouTube channel of a local guy named Adrian
Black. His channel is called Adrian's Digital Basement. He's into retro
computing and does board repair to get old, broken, abandoned computers
working again, generally from the Apple II, Commodore 64 era, up to and
including IBM AT era machines. Recently, he's mentioned the Gravis
UltraSound audio card a few times, which rang a bell with me. I knew
that I once had one, and that I probably still had it. Some versions of
it, I learned, are quite sought after in the retro computer community.
Or, at least sellers have the perception that they are sought after. If
you search on e-bay, you will find sellers asking for $500-$1000 for
(again) some version of them. They were unique at the time in that they
did mixing on the card, rather than in the CPU, or something vaguely
like that. They were producing really good quality sound on PC-style
computers before the CPUs were really fast enough to be able to do the
mixing. I remember having it in the mid-late 1990s, and playing MOD
files on it in a linux box. So, in a moment of inspiration, I went down
in the basement and looked for the card. And, at first, I only found it
visible from the outside of a computer that was wedged behind some other
things. So, a day goes by and I'm thinking about it, and then last night
I go back down in the basement and extract the box from the corner it is
tucked in and pop it open. Sure enough, I have a Gravis UltraSound Plug
& Play Version 1.0. It is one of the later versions. The date codes on
the chips suggest it was manufactured in the 2nd quarter of 1996 or
maybe a little later.
Well, so here in front of me is the box it is sitting in. I don't
remember very much about the machine it is in, except from the markings
it is: an ASUS P/I-P55TP4N motherboard, with ISA and PCI slots, half the
memory slots are filled, and it has a Fujitsu IDE hard disk. Not really
any idea how long it has been wedged in the corner, except that it's
probably been there for "a while". No idea if the power supply even
works. So, not wanting to risk the power supply blowing up and killing
the maybe-potentially-worth-something GUS (as they were known, there is
still support in modern kernels for these boards), I pull the card out,
find a PCI video card, a VGA monitor to plug in, also an adapter for an
AT-style chonky keyboard connector, and with a little trepidation and
sideways thoughts about the location of the nearest fire extinguisher, I
turn the power on.
And it whirs to life!
It became rapidly apparent that this had been a server of some kind,
because I had configured it with a serial console, so I find a
null-modem cable I can plug into a laptop to get a 9600 baud console
shell. Then I remember that I don't remember what the password is or
was. So, reboot again, and from the GRUB shell, I supply an
init=/bin/bash option to the kernel commandline, which boots me to a
semi-functional root shell. I remount the root filesystem read-write,
find the password hash in the /etc/passwd file (this was before i'd
caved into the modernity of /etc/shadow to hide the password hashes),
remove the hash so i can log in temporarily with no password, and
reboot. Now I can log in. Maybe or maybe not, I was too dumb to do the
same thing for root the first time, but maybe there are a few more
reboots, I'm not saying. Anyway, I am in. The machine is a 133MHz
Pentium-S, whatever that was. It has 64MB of RAM. The IDE hard disk is
1GB. It is running a 2.4.20 kernel. From looking at the log files, it
was shutdown sometime in May 2003, and it's running fine. I notice that
the system clock is in October 2023, which is a little surprising,
because the network configuration is wrong for what it is plugged into
(it had a static config, and the upstream network is different than it
once was). It gradually dawns on me that it has a 2023 date, not because
of NTP, but because the realtime clock somehow survived, unplugged and
ticking, for >2 decades.
This evening I went downstairs to take a closer look at the battery, and
discovered that it's a Dallas DS12B887. And the clock is running about
25 minutes slow. After 20 years and 5 months.
Anyway, I just wanted to share that.
--
Russell Senior
[email protected]
