Vintage Computing

[Slim]

Found this in my February 1980 copy of Personal Computer World:

Highslide Image Viewer

Adjusting for inflation that £10 would be more than £40 now. Probably the Star Trek that I remember playing on a PET myself. But look at that keyboard .. you wouldn't want to use it for word processing.

[Slim]

From the same magazine:

Highslide Image Viewer


A 16K memory upgrade, costing more than £400 in today's money. The best value ever to be offered to UK 101 users.

[The Picnic Wasp]

The thing about translating old prices into today's rough equivalent value doesn't really impart that unattainable feeling of wanting something so badly, but being young and not earning, or earning very little.

My only source of income in those days was my parents and I was aware that the amount of money I was looking for would impact them quite significantly given their commitments.

I think the algorithm for teenagers in the 70s to give a similar difficulty for my wallet today would need to be about times ten. Just my gut feeling when I see or hear of these comparisons.

[Slim]

From the back page of my slightly distressed-looking February 1980 copy of Personal Computer World:

Highslide Image Viewer

A dual floppy drive taking 1.2MB disks for about £10,100 in today's money. Exclusive of VAT - but that was only 15% in 1980.

Or: an Apple II word processing system complete with a printer for about £21,000 in 2025 terms. Today you could buy a Raspberry Pi with a monitor and a laser printer then put some free word processing software on it for about £180. It would do a far better job.

[Slim]

Had a play with the online ZX81 emulator at https://www.zx81stuff.org.uk/zx81/jtyone.html this morning. I entered the following short program to perform an arithmetic calculation 10,000 times.

Highslide Image Viewer

I used "FAST" mode as you can see (this allows the CPU to ignore its display maintenance duties until the program completes). It took 98 seconds to complete. I believe the emulator does match the original hardware speed, so I'm confident that the old 3.25 MHz processor in my ZX81 in the loft somewhere would give a very similar result. If I used "SLOW" mode as I usually did it would have taken a lot longer than that.

Then for fun I wrote what's in essence the same program, if I can use the term to describe three terse lines of Bash, and ran it on my desktop PC. Took a bit less than 0.02 seconds. The "echo $EPOCHREALTIME" lines are there for timing purposes.

Code Select Expand

#!/bin/bash
echo $EPOCHREALTIME
for ((i=1; i<=10000; i++)); do
  ((x = i*5 - i))
done
echo $EPOCHREALTIME


[Slim]

A reminder from the June 1985 Personal Computer World that although AI didn't have the prominence in the public consciousness that it does now, it was still a thing 40 years ago. Not only that, but you could "dabble with AI" on a BBC Micro in 32K. Quite a far cry from an LLM humming away in a few terabytes of VRAM in a data centre.

Highslide Image Viewer

The magazine printed programs written by users every month. Often they were short, so it was reasonably practical to type one in if you fancied it. But the following 10 pages are dedicated to this one, not counting ads running in the margins. I wouldn't be surprised if exactly 0 people took the trouble to type the code into a BBC Micro. And if anyone did I'd be amazed if they didn't make a few mistakes.

Still - fascinating. Some sort of rudimentary natural language processing function.

[Slim]

Personal Computer World occasionally published benchmarks - brief programs intended to be timed in their execution, to assess the performance of various hardware being tested or reviewed.

In June 1985 the magazine ran a piece on BCPL, a language I was never familiar with but which was a precursor to C. A BCPL compiler had just been released for the Sinclair QL so they published some benchmarks for the language and ran them on the QL.

I thought it would be fun, in a too-much-time-on-my-hands sort of way, to run one of the benchmarks on a modern PC. I chose this one, because it was easy enough to work out what it did, and looked like it would be easy enough to write in C. I wasn't going to attempt to install a Linux BCPL compiler, if indeed one exists.

Highslide Image Viewer


This is the C code. It's equivalent, it does the same job.

Code Select Expand

#include <stdio.h>

int main(void) {
   int a = 0;

   putchar('S');

   for (int cnt = 1; cnt <= 10000; cnt++) {
      a = cnt * 2 / 3 + 4 - 5;
   }

   putchar('E');
   putchar('\n');

   return 0;
}

So I compiled it and ran it. On the Sinclair QL back in 1985, according to PCW, the benchmark took 5.4 seconds to run. On my own PC the same task took 0.002 seconds, or 2 milliseconds.

That's pretty dramatic. Possibly as much to do with the way modern CPUs do arithmetic in hardware as the basic difference in clock speed.

Fun fact: on my PC, a single

a = cnt * 2 / 3 + 4 - 5;

.. happens in the time it takes light to travel about 60 metres.

[Slim]

Wearable tech, 1980s style. From the June 1985 Personal Computer World, a brief news item about the Seiko RC-1000, marketed as a wristwatch computer terminal. You hooked it up to your Commodore 64, or Apple II, or Spectrum, where you ran the "wrist terminal data manager".

Highslide Image Viewer

But it was superseded quite quickly: this is an ad from the November issue for the 'Memo Diary', with a four line screen. I don't remember the RC-1000, but I do remember these. A standalone keyboard was used - I suppose the watch functioned as the screen while you were typing in your data.

A sort of prehistoric smartwatch then, but I can't imagine many people found them useful.

Highslide Image Viewer

[Slim]

I've been going through my old files and I came across the very first video I ever downloaded, or accessed in any manner, from the Internet. September 1997.

This is a full-size screenshot.

Highslide Image Viewer

It's what you might call amateur, domestic soft-porn. She basically performs a sort of erotic dance and partially disrobes.

Browsers didn't play video then, you had to download them and play them separately in your local media player.

As you can see it's extremely crude by modern standards - 160x120 resolution, 12fps and it uses an antique codec called Cinepak. 8-bit mono audio,unsigned PCM. It's 28 seconds long.

Because the audio is uncompressed it takes roughly the same space, despite being very low quality, that 28 seconds of decent quality MP3-encoded soundtrack would use up, even today.

It's a 2.6MB file which would have been a lengthy download back then over a modem, but (of course) I downloaded it at work. Investment banks had big fat Internet pipes but even so, it wasn't instant.

As rubbish as it is - it was a "wow" moment. Not only because it was a video accessible over the Internet, a medium pretty much dedicated to text, static images and (much less ubiquitously) low-quality audio streaming, but because it was an amateur production. I realised instantly that the Internet was going to be transformative.

[Slim]

From the February, 1990 edition of Personal Computer World, a piece by Tim Bajarin, a technology writer who's still active today. Here, he makes predictions for the new decade in computing.

Inevitably he gets a lot wrong but there's some real prescience here, if not necessarily in the detail or the timing.

His idea that the PC will become an "information and entertainment appliance" is spot on, though the idea that the PC and the TV will become "one and the same in the home" is not quite right. I think there were TV tuner cards in the late '90s but very few people had one. I had a DVB PCMCIA card about 20 years ago myself. It's not until the '00s that telly on a PC goes mainstream with Netflix, the iPlayer etc. But even now, the standalone telly in a living room is still very much a thing, and it doesn't normally have a mouse or a keyboard. However his prediction that the PC would become "an important part of the home" was undeniably correct.

The huge 1990s phenomenon that comes to dominate personal computing is of course the Internet, which understandably doesn't get a mention. At the beginning of the decade it does exist, but not in everyday life. The World Wide Web does not yet exist. And yet Bajarin foresees a future in which computers fetch information into a "personal newspaper", something that pretty much did come to pass albeit not in the '90s, with online news aggregators.

His prediction about CISC chips didn't happen, the Intel architecture clung on and continued to dominate.

The voice control he envisages didn't happen in the '90s either. But it did happen, albeit more typically used to direct a smart speaker or phone than a PC.

Highslide Image Viewer

[Slim]

This made me smile. I picked up the April 1990 edition of Personal Computer World and found a letter from a Mr (or Ms) RJ Wilson.

Invoking Godel's Theorem to underline his (or her) point, RJ argues that Artificial Intelligence is a pipedream, a mirage - and concludes that its only credible objectives are to "attract research funds from the ingenuous, or to provide an illusory glimmer of hope".

What's he (or she) on about? Well there were two branches of AI research back then, the most commonly understood of which was "symbolic AI" - based on rules and logic. It's this to which his (or her) comments seem to apply.

But it's the other branch of research, based on neural networks, that eventually evolved into Gemini, ChatGPT, Deepseek et al. I'd studied both in the final year of my degree but perhaps neural networks were the preserve of academia back then. Out in the real world, AI usually meant expert systems.

I'd love to think that Mr (or Ms) Wilson was still around when ChatGPT turned up in 2022.

Highslide Image Viewer