That this 8008 computer from 1972 should have 16KB of RAM sounds unlikely. I haven't been able to find any real info stating this - still, the headline on tomshardware says "16KB". That's very hard to believe. Static RAM of that size would have been extremely expensive at the time, the much much later Altair 8800 didn't get past 4KB without a dynamic RAM board (which was also very unreliable).
The Q1/Lite, a Z80-based computer from 1974 (correction - 1977!) - well, that I can kind of believe.
I wonder if the article author is confusing the theoretical max memory address space of the 8008 (which is 16KB) with actual RAM size of the 1972 model Q1?
> They use the Intel 8008 CPU, an 8-bit CPU capable of processing a whopping sixteen kilobytes of memory
Which is correct. As you say, highly unlikely that the computers had that much, though.
I'm not sure how Tom's Hardware does things, but it's well-known in the _newspaper_ industry that the headlines aren't written by the person who writes the article, and misinterpretation happens from time to time. I suspect whoever wrote the headline just got confused, here.
Of course, because of the plague of LLMs which now lies over the land, anyone trying to find out how much RAM such a machine had in 2050 will be confidently, incorrectly told 16kB by GPT47, even though it's obviously wrong, due to this headline...
Surely by 2050 they'll have figured out how to associate a credibility measure and attested provenance with every bit of knowledge to prevent that sort of thing. What a time to be alive.
The headline is probably wrong, the article text just says "capable of" 16 kilobyte. I also think it's unlikely it had 16 installed. On the other hand, doesn't the screen look like a flatscreen, not a CRT? So the screen itself was probably very expensive in itself. (Probably plasma screen?)
Edit:
Look at the image and the bank of 3 * 8 socketed chips:
Probably not a plasma screen in the 70s. It looks like a bunch of individual characters. They are probably vacuum fluorescent like what was on calculators and the Speak & Spell educational toy from that period.
From this link posted in another thread, the structure of the display is a little clearer: https://www.peel.dk/Q1/
In the shot of the display frame by itself without the tinted plastic in front of it, it almost looks like a miniature arrival/departure board at a train station.
Yes, it is pretty certain that no bitmap plasma displays existed at that time, because the drivers for them would have been more expensive than an entire desktop computer (due to the high number of rows and columns).
It is said that the display was an unusual plasma display and not a vacuum fluorescent display, which were much more frequent in the desktop calculators of that time, but it must have been an alphanumeric display with segments, perhaps with 16 segments per character.
What we think of as "plasma screens" were invented at UIUC for the PLATO project in (checks) 1964, by Donald Bitzer, H. Gene Slottow, and Robert Willson. They had a resolution of 512x512. They started cranking them out in bulk in 1972.
These were monochrome orange vector displays. They didn't use segments.
Wikipedia says that the 512x512 PLATO plasma display had a cost of $2500 at a time when that was a much greater value than today.
Moreover, I doubt that the cost includes the one thousand high voltage transistors that would have been needed to drive the display.
As I have already written, such a display would have been too expensive for a desktop computer with an 8-bit CPU.
In the early seventies, as an alternative to the red LED displays and green or blue vacuum fluorescent displays, there were the orange Panaplex plasma displays with segments for numeric (7 segments) or alphanumeric (14 or 16 segments) characters.
I believe that the Q1 computer must have had such a Panaplex display, which was cheap enough.
Vector displays are possible only where there is a drawing spot that can be moved in any programmable direction, i.e. only on CRTs or mechanical plotters. They are not possible with plasma displays.
Where did you get the 1000 high voltage transistors thing?
I used PLATO at UIUC in the late 80s briefly, because they were still around, but everything had been WYSE-50 or Sun already (or contemporaries) and the PLATO terminals weren't some monstrosities....just looked 20 years old.
In the simplest case you need one transistor per row and one per column for bitmap displays.
If any transistor in omitted, than that row or column must have a fixed voltage, which means that the corresponding column or row of pixels would no longer be controllable and they would display the same pixels as whatever other column or row is selected at a given instant.
One thousand transistors do not need a huge space (they would fit easily on a PCB of a size compatible with a PLATO terminal, based on the photo from Wikipedia), but they were quite expensive. Suitable high-voltage integrated circuits have appeared later. In 1972, one could have used Nixie decoders like 74141 for the negative voltage, replacing 512 transistors with 64 IC packages, but 512 discrete transistors would still have been needed for the positive voltage.
For segmented alphanumeric displays you need only 14 or 16 transistors for the segments plus one transistor per character.
> As I have already written, such a display would have been too expensive for a desktop computer with an 8-bit CPU.
They were subsidized to some extent, but they were nonetheless cranked out by the hundreds.
> Vector displays are possible only where there is a drawing spot that can be moved in any programmable direction, i.e. only on CRTs or mechanical plotters. They are not possible with plasma displays.
I'm sorry, you are simply wrong here. The Plato display did have vector graphics.
Your own Wikipedia citation disagrees with you:
"The display was a 512×512 bitmap, with both character and vector plotting done by hardwired logic. It included fast vector line drawing capability, and ran at 1260 baud, rendering 60 lines or 180 characters per second."
>"The display was a 512×512 bitmap, with both character and vector plotting done by hardwired logic. It included fast vector line drawing capability, and ran at 1260 baud, rendering 60 lines or 180 characters per second."
Sounds more like it has Bresenham's algorithm implemented in hardware. It would still plot individual pixels on a 512x512 grid, which is different from how an actual vector display draws lines.
Maybe we can agree to disagree about the terminology here. To me if it takes vector inputs and displays them, it doesn't matter if it's pixels on a grid or tiny elves holding up colored signs. It's a vector display.
In any case, the PLATO display absolutely did not use "segments", as the OP claimed that it must have. He also claimed that bit-mapped plasma displays "didn't exist at the time".
A vector display is a non-pixel-based screen like a radar screen or oscilloscope. I used a vector computer terminal in 1979. I also used a Plato, which did have vector graphics but was not a vector display.
No, a vector display is one that takes vectors as input and displays them. What happens inside the display is irrelevant.
PLATO used this type of display because it was running over very slow dialup modems, and vectors, while having their limitations, are much more compact than rasters.
Example: suppose you want to draw a line on the 512x512 PLATO display. With a raster method, you have to send 512x512x(number of brightness levels) down the pipe. With a vector display, you only have to send DRAW(X1,Y1,X2,Y2) or (equivalently) DRAW(X,Y,length, angle) 5 numbers versus hundreds of thousands.
The PLATO display had direct hardware support for line-drawing commands. That made it a vector display.
A vector display is not the same thing as a vector format for storing/transmitting graphics. On a raster display (what we all have today), a line must be made up of discrete pixels, whereas on a vector display, there is an analog integrator that moves the X/Y deflection smoothly between the endpoints.
This might not matter if you have a high resolution, multiple brightness levels, and a CPU/GPU powerful enough to do antialiasing, but for 512x512x1bpp the pixels making up each line will be very visible as "stairsteps".
You are thinking of digital vector GRAPHICS displays with hardware line-drawing support. Analog vector monitors and digital vector graphics raster displays are two different things. Vector monitors can draw nonpixellated curved lines as well as straight lines. Vector monitors can draw lines at any angle without staircasing. If a screen has pixels, it is not a vector monitor.
The 1970s Asteroids arcade game had an analog vector monitor without pixels. Ditto the PDP-1 Spacewar and PDP-11 Lunar Lander, both of which I saw in the 70s. Ditto the Star Wars arcade game. Some vector monitors had rapid-fade phosphors and had to be constantly redrawn, others were bistable storage tubes that retained an image for several minutes without being redrawn. The Plato terminal did not have an analog vector monitor.
And.. the original article's picture is also a Q1 Lite, as far as I can tell. Introduced 1977. That one used NEC 4116 dynamic RAM. And did not use the 8008.
It's very difficult to find anything about the original Q1. Even the "First Q1 Brochure" from TheByteAttic is actually a Q1 LMC, which was the second model.
The 2200 type 2 (1972) was indeed expandable to 16k. Yes it cost $14,000, but you could buy it. Remember: this was an expensive piece of business equipment, the Altair was a cheap (for the time) kit for hobbyists. Companies buying these machines were choosing between them and a minicomputer like the PDP/11 which cost $20,000.
14 year old me in 1978 had a whopping 4K of RAM in my S-100 system (a whole card full of 2102s) and I couldn't have had that much without my parents helping me out.
A full 16K six years before seems unlikely, especially given the size of programs back then.
You could do quite a lot with a small amount of RAM. And if you were using the front panel switches (not on the acorn I suppose) you didn't want a large program!
I worked for Daniel Alroy the founder of Q1. https://www.peel.dk/Q1/
I have never seen one in real life yet. He told me lot of stories, especially how he got into a fight with Intel. He had set out to prove a point that a cheaper simpler machine can be made.
omg hey vachi it's Danielle, your former post-Q1 coworker :)
I've also been trying to track down a Q1 FOR YEARS and a friend sent me this article.
Daniel Alroy would be so thrilled about this coverage. To this day, I still haven't visited the Computer History Museum in San Jose, out of respect for his grudge with Intel. (They don't include Q1 in its history.)
the neon-orange plasma displays invented for the plato system in the late 01960s had an enormous, unique advantage for this kind of thing
they were bistable and switchable! so you could have a 512×512 display which had, in the physical display itself, 262144 bits of memory, equivalent to 32768 bytes. by contrast (heh), a lcd or normal crt fades in a second or two if you don't redraw it, which means you need a separate memory somewhere to hold the data to redraw, memory you have to run through many times a second in order to generate a steady display
the tektronix 4010 and 4014 graphics terminals (derived from the tektronix type 564 and the hughes aircraft company model 104d) had a similarly bistable crt, making it possible to do detailed high-resolution graphics at an affordable cost, but you couldn't erase pixels; you could only erase the whole screen
so normal plato users enjoyed multiple fonts, equations with superscripts and subscripts, and full-screen high-resolution graphics that users of other systems could only dream of, unless they had enough money to devote tens of kilobytes of ram to a framebuffer—which, of course, is precisely what the gas-plasma display inspired alan kay to lead a group at xerox parc to start doing, a few years later
(why didn't people use the plasma panels for computer ram? because they were slooow. updating a pixel took on the order of a millisecond. same reason they didn't use dekatrons or ne-3 lamps)
brian dear wrote a very interesting book about the history of the plato project
Still pretty cool. I wish I could find an 80x24 one (or two 40x24) for a sensible price. Would be lovely for a text-only terminal thingie.
> brian dear wrote a very interesting book about the history of the plato project
It's an interesting book, full of interesting stories. It is a bit confusing though, as it goes back and forth in time quite a lot - it has a very Primer-like timeline.
As someone who collects and restores old computers, I can tell you this is how they are preserved. Museums are only the final resting place of these machines, but, in the meantime, they are preserved by ordinary people in their own collections.
Sometimes it's the other way around. I saved a Cromemco Z2D from the trash when the company where my father worked cleaned out their basement prior to moving to a different building.
You could also go back to the 1965 Olivetti Programma 101 (https://en.m.wikipedia.org/wiki/Programma_101). It had labels and memory cells, and could be programmed in a kind of assembly.
So many articles have terrible / gibberish writing. I can't help thinking, is this AI?
Eg. In this article, right out of the gate there are two...
..
"They are now on display at Kingston University in Surrey, England, but only for the rest of February 17th."
Are they saying that the machines are available on one day only? February 17th, The day the article was published? If so, that's a very strange way to say that.
..
"As far as what to expect, don't expect too much— these are the first desktop PCs created with a fully integrated single-chip microprocessor, including the CPU."
I am just guessing what they mean here. Do they mean... "the first desktop PCs created with a fully integrated single-chip microprocessor, which includes the CPU" ??
..
ps. I was using Miro recently and a popup appeared asking should the AI rewrite the text I wrote "for clarity" .. I just laughed, clarity is certainly not an LLM's strong suit.
But I reckon these tools are everywhere and they are being used everywhere.
I wonder if the article author is confusing the theoretical max memory address space of the 8008 (which is 16KB) with actual RAM size of the 1972 model Q1?