Half a croissant, on a plate, with a sign in front of it saying '50c'
h a l f b a k e r y
Resident parking only.

idea: add, search, annotate, link, view, overview, recent, by name, random

meta: news, help, about, links, report a problem

account: browse anonymously, or get an account and write.



All-VRAM computer

See everything working
  (+9, -1)(+9, -1)
(+9, -1)
  [vote for,

A computer whose memory is entirely video RAM, visible to all.
Twenty-four bit CPU, registers in memory, firmware copied to RAM, each memory location displayed as a true colour pixel. Around a megabyte of memory, forming a display equivalent in resolution to a DVD display. Special instructions allowing quick decoding and display of video and MP3 files, but mainly pretty primitive, useful for learning programming on a fairly low level, playing DVDs (some screen garbage), CDs and music.
Great for Core Wars.
nineteenthly, Dec 08 2008

CP-642B General Purpose Computer http://www.dluper.com/CP642B.html
This is what I worked on in the Navy, 1984-1990. I'm not the guy in the picture. [phoenix, Dec 09 2008]

Fortran purity test - quite old http://www.netfunny.../92q1/fortquiz.html
There´s probably more [nineteenthly, Dec 12 2008]

Manchester 'Baby' http://news.bbc.co....hnology/7465115.stm
The memory of this machine was human-readable - the CRT _was_ the memory (I think) [monojohnny, Dec 13 2008]

Megaprocessor http://www.megaprocessor.com/index.html
LEDs not only for the ram, but for the inputs and outputs of all of the logic gates! [notexactly, May 24 2017]

Williams(–Kilburn) tube https://en.wikipedi.../wiki/Williams_tube
The first random-access digital data storage device was a CRT modified to store and retrieve data. However, the pickup plate blocked viewing, so "if a visible output was needed, a second tube connected in parallel with the storage tube, with a phosphor coating, but without a pickup plate, was used as a display device." [notexactly, Dec 18 2018]


       Baked, 1986-88, on a prototype development system for 8-bit microcontrollers in a specialised embedded application. Only 16K of main memory, but when the machine fell over, it was possible from the video image to determine whether it was a software, firmware or hardware problem.   

       Sadly it was a one-off, but it worked really well.   

       Not widely known to exist.
8th of 7, Dec 08 2008

       Multiple computers in the 1940s used CRTs for memory. I quite like the EDSAC.
Spacecoyote, Dec 08 2008

       Oh yeah, EDSAC and blinkenlights generally. Tell me more, (can´t find square brackets on Icelandic keyboard)8th(same again).
nineteenthly, Dec 08 2008

       BT,DT. Won the prize. Thirty two 32 bit CPUs, each with a portion of a distributed frame buffer. The only bit that wasn't visible was the cache (but that's not meant to be seen anyway - that's why it's called "cache")
coprocephalous, Dec 09 2008

       You'd need a big screen these days.
wagster, Dec 09 2008

       Size isn't important; it's resolution that matters.
8th of 7, Dec 09 2008

       I'm sure the BBC 'B' computer (early 80's) had no distinction between RAM and VRAM - i.e the screen could be mapped to any section of RAM you liked.
hippo, Dec 09 2008

       //Size isn't important; it's resolution that matters//
I'll resolve to tell my wife that.

Yes [hippo] you're correct.
coprocephalous, Dec 09 2008

       Concerning the BBC, i have page five hundred of the User Guide in front of me and it says graphics RAM ends half way up the address space but the start is defined by HIMEM. I remember it used to scroll the screen by changing the beginning of the VRAM, which was really weird, so the memory map would end up starting halfway down the screen.
Presumably the 6845 could be reprogrammed to do that. Registers twelve and thirteen tell it where to start the display, so you could maybe bypass the O/S and do that. Which would presumably mean the likes of CGA and EGA are able to be programmed in the same way.
That leads me to a sixty thousand dollar question (for me): Do contemporary PC graphics cards have something that looks to the system like a 6845 in them?
nineteenthly, Dec 09 2008

       Modern BIOSes allow you to select how large a chunk of regular RAM to reserve for extra VRAM. In theory you could make it map it to wherever you want and implement this; in practice you would have to ditch the OS altogether and hit the hardware directly to be able to accomplish that.   

       Lots of systems had a framebuffer pointer that could be set to anywhere within RAM and changed at any time. I remember that a lot of graphical tricks were accomplished that way on the Amiga and GP32.   

       You may remember the loaders (which were builtin to many games to decompress the binary to make up for slow tape and disk speed and limited space) on the Commodore 64. The memory was very limited, so during decompression it would use VRAM as a scratchpad. Pacman (the arcade unit) did the same thing on startup.
Spacecoyote, Dec 09 2008

       I don't know much about the 6845, the 6502 series of chips is more my bag (Commodore 64's 6510 specifically) - the graphics chip could be pointed at one of a number of address 'banks' - that specified the entirety of the graphics space (all 320 × 200 pixels of it) - There must be an analogue to the same thing today - define a space, and then use the information located there to paint the screen with the appropriate coloured dots - it would seem churlish for a chip manufacturer to make that address-space fixed.   

       And yes, often, with the limited memory available, you'd see 'working' memory spilling over into the VRAM - and watch it squiggling about on screen - alternately, you might accidentally POKE the VRAM into watching the zero-page (a memory area reserved for the internal workings of the chip itself) and get to watch it tick (sort of).   

       And then there's the "snowcrash" of old Macs - where, they would occasionally freeze and populate the screen with a snowstorm of memory garbage.
zen_tom, Dec 09 2008

       I´m more a Z80 person myself. On the ZX81, for some reason the character "font" was defined in a memory area pointed to by the I register, and changing that register to another area of ROM (but not RAM) would display the bytes there rather than the alphanumeric shapes. If you aimed it at the normal character ROM and did the hi-res trick, you could get the highly weird resolution of sixty-four by one hundred and ninety two (plus two hundred and fifty-six alternating stripes if you really wanted them). It meant you could display histograms.
nineteenthly, Dec 09 2008

       "Geeks .... the woods are full of 'em this Fall ......"
8th of 7, Dec 09 2008

       I've linked to some information on the CP-642 Bravo, a refrigerator-sized computer I worked on in the U.S. Navy. It had lots of blinkenlights, some of which were also buttons used to program the thing.   

       When we were bored, we would run a "52" test which would light alternate bits in the displayed registers (a "5" being 101 in binary and a "2" being 010).   

       It wasn't VRAM though - it used core memory.
phoenix, Dec 09 2008

       Brilliant, [phoenix], thanks. Clearly stripes are the answer to everything.
[8th], i've never been able to work out if i'm a geek or not. I've done the tests but i also have a lot of way-off things like the whole alternative medicine thing. Having said that, i do sort of see herbalism as a form of hacking.
nineteenthly, Dec 09 2008

       //Concerning the BBC, i have page five hundred of the User Guide in front of me //

Ahh, how I hanker for the days when computers came with proper manuals that told you about the machine's architecture and how the thing actually worked.
DrBob, Dec 09 2008

       [DrBob] The BBC user guide also came with a full circuit diagram. On a single sheet!
coprocephalous, Dec 09 2008

       Not quite. I thought that too though, til i looked at it. It's got, on consecutive pages:
Motherboard layout
External connections
Three pages of memory maps
Circuitry for User I/O and 1MHz Bus
Same for video and RS423 - i've used this page to work out why the composite video was greyscale, and later for when the UHF TV output also lost its saturation.
Analogue inputs and Disc interface
Cassette interface.
So, lots of circuit diagrams but not the whole machine. The MZ80A and probably a lot of other micros actually did have the whole thing.
I once heard it said that the BBC was the last computer which could be completely understood by one person, but i think that's an exaggeration. The original IBM PC is probably less sophisticated for a start, as it also has an eight-bit databus and no fancy graphics or sound, unlike the BBC, so all it's really got over the BBC is more RAM. The BBC with an 80186 second processor would've been better as a PC "clone" than the IBM PC - more like a mid- to late-'eighties PC really.
nineteenthly, Dec 09 2008

       //It's got, on consecutive pages//
You must have a later version than mine - it has the whole lot on a folded sheet of A3. My user guide is a loose-leaf binder.
coprocephalous, Dec 09 2008

       Last day of May, MCMLXXXIII, it says on the inside front page, and it's spring bound with an addendum about using the Welcome cassette with Econet.
nineteenthly, Dec 09 2008

       Sorry, [19thly], back home now and checked the manual - mine's actually the Advanced User Guide (Bray, Dickens&Holmes), 3rd edition November 1983.
But yes, the whole lot's on a single side of A3 - floppy i/f, printer, serial, extension bus, Tube, Econet - the works.
coprocephalous, Dec 09 2008

       // never been able to work out if i'm a geek or not //   

       You're a geek. Trust us on this. We would be delighted to assimilate you into our Collective.
8th of 7, Dec 09 2008

       One technique I used in the early '80s is to use the address lines to drive a pair of D/A converters, (higher order bits to one DAC, lower order to the other) and take the analog DAC outputs to an X-Y oscilloscope.   

       This will show execution/access as an x/y map, with each unique address occupying a coordinate, and dwell time (how long the processor stays at a given address area) showing up as brightness of the trace.   

       Can also be used on the data bus to see what data is being used, or a combination.   

       Very helpful in debugging code and hardware (why is it spending so much time accessing $E048 ?)
csea, Dec 09 2008

       Did anyone else find the Easter Egg in the Electron´s ROM where Jim, Sheila or whatever they were called would be on the BBC?   

       [8th], well in that case:   

       -----BEGIN GEEK CODE BLOCK----- Version: 3.1
GPdxs:a41C++L+P+E+++ W-N++K-w---
OM-V+PS+++ PEYPGPt-5--X--R-!tv

       Plus, of course, the HB version:   


       Speaking of emacs (E+++, as you can see), [coprocephalus], i do remember the fold out sheet now you come to mention it. One of my friends at school had it and he later went on to write bits of emacs, for instance Tetris. I wonder if he ever comes on here.
nineteenthly, Dec 09 2008

       Lots of great nostalgia here.   

       The BBC/Electron game Exile used memory assigned to video to display the game. On the Electron, this couldn't be hidden so you would actually see the game's data as noise above and below the main display.   

       I think the original idea is good. Certainly possible for a 1MB computer. It'd be interesting to see for a modern OS as well, displaying the modulo sum of 1Kb on one pixel, but getting 1GB out at something approaching 50Hz would be really pushing memory and wouldn't leave a lot for the CPU.
Srimech, Dec 09 2008

       In a curmudgeonly sort of way, although i know nothing of how programming is done today, i feel strongly that it seems to waste enormous amounts of computing resources. Elite and Rogue for example, had pseudorandom features fitting in very little memory space, but a pseudorandom number generator amounts to extreme compression for a series of levels or other entities. Clearly a first person shooter needs most of the work doing beforehand, which takes up loads of RAM, but SURELY...well, you know.
What i´m working up to is this. A relatively small quantity of RAM would methinks be good discipline for a budding programmer and totally visible memory would enforce this requirement.
nineteenthly, Dec 10 2008

       <kicking myself that I sold my BBC 'B' in the late 80's for a pittance...>
hippo, Dec 10 2008

       <kicks [hippo] for not selling it to me>
coprocephalous, Dec 10 2008

       eBay is a magnificent source of ancient computers - there's a BBC B on there right now for £28 (+£32 p&p) - only 4 hours to go.
zen_tom, Dec 10 2008

       I removed the rubber case from my Spectrum and replaced it with a "good" keyboard. Of course it's only now that I realise I've ruined it.
wagster, Dec 10 2008

       I have a lurking Jupiter Ace, ZX81 and Tandy Color. The Ace is broken though, sadly. I don´t think the arrival of a BBC would enamour me of [grayure] because of her aversion to clutter. It´s a great big slab of a thing. I´ll just get a PDP-11 then.
nineteenthly, Dec 10 2008

       [bigsleep], that´s incredible. I always felt the ZX81 had an air of an early ´sixties mainframe about it, what with the capitals and the white background.
[Ian], though it may have been by Microsoft, i actually really liked the Dragon´s Basic. Didn´t it also have a Welsh version of Basic available? The 6809 would´ve been ideal for Forth though.
nineteenthly, Dec 11 2008

       // The 6809 would´ve been ideal for Forth //   

       It isn't bad, but around that time a number of specialised Forth engine chips were produced.   

       In a bitter irony which is hard to comprehend, one of the best CPU's for Forth is the 8086; with its segmented adress space and dedicated deep stack pointer, it's ideal. As a general purpose CPU, it's rubbish.   

       Quite a few Forth robotic controllers were built around the 80186 (the one with the built-in I/O capability).
8th of 7, Dec 11 2008

       Yes, i suppose that´s true. I always thought the dedicated Forth CPUs were deeply cool. I´ve also wondered about the idea of extending them so that the instruction set had a byte like 01000000 for @ or 00100000 for ! .
I´ve also wondered what my ownership of a Jupiter Ace says about my character. I always seem to back the loser. Do Jupiter Ace owners tend to be Prefab Sprout fans?
nineteenthly, Dec 11 2008

       Did you drive home with it in your Sinclair C5?
wagster, Dec 11 2008

       What, you mean the one fitted with a Squarial, an eight track and a Betamax (they bloody well were better!) and a pair of fluffy Pyraminces?
nineteenthly, Dec 11 2008

       //I always seem to back the loser//

Heh! I went for the Spectrum (48K) and followed that up with the Amiga A600. I feel your pain.
DrBob, Dec 11 2008

       <Throws [Dr.Bob] a bottle of Bailey's Irish Electric Soup and Brain Ointment>
8th of 7, Dec 11 2008

       How was the Speccy a loser? It beat everything else at the time on price, popularity, software availability, you name it. It's the Dragon32 owners I pitied.
wagster, Dec 11 2008

       I think he meant the Amiga. At least the Dragon had bit-mapped graphics. The ZX81 hadn´t. On the Ace you could store characters in the right place and redefine them, so it had the same resolution as a Spectrum in a way. In a way.
nineteenthly, Dec 11 2008

       Well, i would say that was a natural progression because of the extended Microsoft Basic, but it doesn´t sound like you used it much on the Dragon. By then, i was trying to go cold turkey on computers because i felt my interest was unhealthy, but i was using the Vax at the Uni then as well. Clearly it didn´t work, because here i am on the HB going on about it, but it did result in a complete failure to grasp Object-Oriented Programming.
nineteenthly, Dec 11 2008

       //i was using the Vax at the Uni //
I used to use a Vax at a Uni, but it was mostly to vacuum up crisp crumbs around the common room.
AbsintheWithoutLeave, Dec 11 2008

       Did you ever have a job in a space station?
Forth is a total wonder. It seems to have gone a bit pear-shaped recently.
nineteenthly, Dec 11 2008

       //I think he meant the Amiga//

He did. It never had quite enough grunt in the processing department and was superceded by the A1200 a couple of months later. I still rather liked it though (well all the Amigas really). It was one of those infinitely customisable devices and one of the best copying machines ever invented. It is still sitting in a box in one of my cupboards. Fully functioning but unused. <wipes a tear>
DrBob, Dec 12 2008

       I think of it as the newest old computer, except maybe for the Sprint.
nineteenthly, Dec 12 2008

       I never really got any sort of confidence with Forth - much like Prolog, which always seemed a bit weird too. LISP, however, I loved. There's probably a personality type questionnaire somewhere based on your programming language preferences.
hippo, Dec 12 2008

       Yes, the X purity test or something. Lisp is of course what my emacs friend was into. One of the things that really surprised me about home micros was the fact that most of them had Basic in ROM rather than a plethora of other languages. I would´ve thought that Logo for a start would have been popular, and maybe Pascal as a teaching language, and various others, would have been in home micros when you bought them, but no, it was almost always Basic when you bought it and other languages as extras.
Then of course there´s APL, whereof this very computer has an implementation, partly to torture [eleventeenthly] but also because it´s fun (for suitably small values of that word, as they say).
I´ve got Prolog on here too, but i´m a Philosopher, so you can expect that. We used that at Uni, of course.
nineteenthly, Dec 12 2008

       Pascal is so far as I'm aware an exclusively compiled language. Interpreted languages are good for beginners because you can stop execution at any point and interrogate the current state with commands typed in the same language, or run individual parts of a program without worrying about how the compiler has treated internal states.   

       Logo is a functional language. It is most remembered for its graphical primitives which would have been difficult to implement on many early microcomputers. Even disregarding that, functional languages are even now of mainly academic interest, and even those who know the difference between functional and imperative languages and realise the potential of both tend to use imperative languages for practical purposes. Functional languages certainly have a place, but early microcomputers were designed for practical purposes, not as learning tools.
Srimech, Dec 12 2008

       Logo was implemented on the TI-99/4A, the 380Z and the Apple to my knowledge. Those are pretty early. With a turtle, it could have been done earlier (and yes, i understand it´s more than just turtle graphics, but there was list processing on those interpreters too.) There was also a ROM replacement for the ZX81 which ran APL, and of course the IBM 5100, which is over three decades old. The Apple II could do UCSD-p, so it had Pascal.
The capability was there, but it wasn´t exploited, and if they weren´t primarily learning tools, why would they be running a language for Beginners?
nineteenthly, Dec 13 2008

       From the (very) little I know about Logo, I've always thought it would be fun to introduce it as a scripting language for a proper vector graphics package such as Illustrator.
wagster, Dec 13 2008

       How's the new arrival getting on anyway?
wagster, Dec 13 2008

       If you hooked a terminal up to a Postscript printer, wouldn´t you have a useable computer?
nineteenthly, Dec 14 2008

       <distant memories>...there was someone I was at college with who wrote a Fresnel Lens generator in Postscript...</dm>
hippo, Dec 14 2008

       How? Did it produce actual Fresnel lenses and if so, what hardware was involved?
nineteenthly, Dec 14 2008

       I have a laser printer, but it doesn´t do Postscript, sadly.
A Psion organiser! That´s deeply groovy. I wonder if it could be done with a Microwriter.
nineteenthly, Dec 14 2008

       The Fresnel lens patterns were printed on paper and then photocopied onto sheets of plastic (the kinds used for overhead projectors). They worked, too!
hippo, Dec 14 2008

       I see. I´m just thinking about the possibilities of magnetic ink on a laser printer, with Postscript. Circuits? Could you maybe make an actual set of logic gates by drawing them using Postscript to parse propositional calculus sequents? Or could you not combine magnetic ink and a laser printer? Graphite maybe?
nineteenthly, Dec 14 2008

       Right, i'll tango down to the smoky lobby and take a look. I thought those were to do with CDs. Ta!
nineteenthly, Dec 14 2008

       I think the lower end Macintosh laptops currently have only one type of RAM, shared between video and other. I imagine it would take a pretty outlandish hack to do what you want, though.   

       I also used to point my C64's video pointer to various parts of the RAM. I even tried using it for debugging purposes - including for self-modifying code! Astonishingly primitive and crude in hindsight, but a lot of fun.   

       Even the program counter and CPU registers were in the first 256 bytes of RAM. Self-executing programs would load into zero page, and write themselves over the program counter, setting it to the start of the code. Ah. Nostalgia isn't what it used to be.
spidermother, Dec 17 2008

       I miss my Beeb.
BunsenHoneydew, Dec 18 2008

       Everyone likes this visible memory mode so much(note), perhaps there is a plausible reason to bring it back   

       on a tablet PC, even with many GB or TB a video display of memory could let another PC or person actually see that the software was working as it was supposed to, its a "all memory is bus" diagnostic as well as security feature. Just show the compressed view of all memory on the display, then any other computer can see that its working properly "noncompromised", it might possibly even permit the viewing PC to ask zooms if the memory content looked askance thus visual security is an "excuse" to bring back the see it all feature of computer memory   

       (note) popular .5b idea was putting a VRAM display on flash ram   

       mystery narrative doing random peeks at a TRS80III the screen resolution suddenly changed so that there were bright nonperiod pixels on. If I had only written what the random peeks were to tape Id have found an unknown assembler graphics mode. I started using the Tandy color computer around 1981 as a HS sophmore   

       The .5b could use a wider demographic to support this idea. where is the observation that an nvidia Gforce card programmed with assembler could emulate some other CPU while literally being another computers graphics memory, that would be like a thirtysomething response. Or how about a current college response where a person says, yeah, like I programmed an ipad to do feedback from its own video screen (via software or mirror) such that the stable whirly structures on display are usable data objects.
beanangel, Nov 30 2011


back: main index

business  computer  culture  fashion  food  halfbakery  home  other  product  public  science  sport  vehicle