What was the very first computer you owned or used ?

[ebastler]

Not a good time to visit museums, maybe, and the vintage computer festivals either go virtual or get cancelled. But on the other hand, plenty of time to spend at home and tinker! 

I built this 100 MHz 65C02 over the past months, to give my Apple II and various old chess computers a boost:
http://e-basteln.de/computing/65f02/65f02/

[brucehoult]

Not a good time to visit museums, maybe, and the vintage computer festivals either go virtual or get cancelled. But on the other hand, plenty of time to spend at home and tinker! 

I built this 100 MHz 65C02 over the past months, to give my Apple II and various old chess computers a boost:
http://e-basteln.de/computing/65f02/65f02/

Mate! That's fully sick!

(this is an Aussie BBS after all)

I've been playing with 15 MHz 65C02 recently, but that's something else. Cool idea cloning the system ROM and RAM internally.

[MK14]

Not a good time to visit museums, maybe, and the vintage computer festivals either go virtual or get cancelled. But on the other hand, plenty of time to spend at home and tinker! 

I built this 100 MHz 65C02 over the past months, to give my Apple II and various old chess computers a boost:
http://e-basteln.de/computing/65f02/65f02/

You're right, there are balancing pros and cons, about the current (virus) situation. I had plenty of opportunity to visit such places, in the past. But now I can't (in some cases, i.e. meetings cancelled, some are shut down at the moment, etc), it is sort of annoying/frustrating.

WOW!, that is amazing (A 100 MHz 6502 (65C02)). As brucehoult said. It is clever/inventive to use the very fast FPGA ram/stuff, rather than the relatively much slower, external RAM/ROM/stuff.

It will be good, when you have the public release version of the PCB, as it seems a neat and interesting project. That particular FPGA, seems to be relatively affordable/cheap, at something like £15 a time.
Although BGAs can be a pain to put on to PCBs.

The old Chess computers, had interesting (vintage) playing styles. But could take a long time, when set to a high setting, in order to give a somewhat good move. E.g. 2 .. 15 minutes. Call it 5 minutes (assuming 1 MHz 8 bit cpu, such as a 6502, or equivalent).

So, that 100 MHz 65C02, would allow playing with the Chess computer thinking it was set to 10 minutes a move, but it would only take, typically, 6 seconds a move. Allowing investigations, into how old vintage Chess computers, can analyse interesting positions/games, at long (equivalent) time settings, and yet the results would come out very quickly.

As your notes say. You do need some kind of 'Turbo' button, or normal (x1), medium (x10) and full turbo (x100), speed settings. Ideally external, without needing to open the devices case. But I suppose a simple turbo mode on/off, would suffice.

[ebastler]

WOW!, that is amazing (A 100 MHz 6502 (65C02)). As brucehoult said. It is clever/inventive to use the very fast FPGA ram/stuff, rather than the relatively much slower, external RAM/ROM/stuff.

Thanks for the kind words! The concept was actually borrowed from a 1980's accelerator for chess computers, the TK20 from Schaetzle & Bsteh. The TK20 also plugged into the CPU socket and copied the ROM contents into fast on-board RAM. It filled an external box of electronics back then, and would run at 18 MHz max., but was conceptually the same as my little FPGA board: https://www.schach-computer.info/wiki/index.php/TurboKit

It will be good, when you have the public release version of the PCB, as it seems a neat and interesting project. That particular FPGA, seems to be relatively affordable/cheap, at something like £15 a time.
Although BGAs can be a pain to put on to PCBs.

I do intend to publish the PCB layout and VHDL code as open source; still need to do a bit of cleanup. Soldering the BGA and the little leadless packages for the level converters was a first for me too; I got some good advice on this forum. I am using a $50 pizza oven plus a temperature probe connected to a multimeter. No closed-loop control; I just switch the heating on and off while watching the thermometer to get reasonably close to the specified ramp, soak time, and peak temperature.

The old Chess computers, had interesting (vintage) playing styles. But could take a long time, when set to a high setting, in order to give a somewhat good move. E.g. 2 .. 15 minutes. Call it 5 minutes (assuming 1 MHz 8 bit cpu, such as a 6502, or equivalent).

So, that 100 MHz 65C02, would allow playing with the Chess computer thinking it was set to 10 minutes a move, but it would only take, typically, 6 seconds a move. Allowing investigations, into how old vintage Chess computers, can analyse interesting positions/games, at long (equivalent) time settings, and yet the results would come out very quickly.

Folks on the German schachcomputer.info forum have done quite a bit of testing already. Typical 65(C)02 chess computers in the 1980s had 3 to 5 MHz clock rates, so you don't get an acceleration factor of 100. But even a factor of 20, which applies to the later 65C02 chess computers with smarter programs, provides a big improvement:

The established rule of thumb seems to be that doubling the CPU speed provides about 80 ELO points of improvement in playing strength. (Although that tapers off with further doubling -- slowly or quicker, depending on the program and its approach to pruning the search tree.) The better 6502 chess computers can now hold their own against the later generations of 68020 and 68030 machines!

As your notes say. You do need some kind of 'Turbo' button, or normal (x1), medium (x10) and full turbo (x100), speed settings. Ideally external, without needing to open the device's case. But I suppose a simple turbo mode on/off, would suffice.

The idea was to connect a reed contact to the 65F02's speed selector input, so that the acceleration can be disabled from outside with a magnet, without the need to drill a hole into the case for a toggle switch. None of the handful of chess computer users has opted to install that, I believe -- they all keep the acceleration active permanently. (Being collectors, they probably have a second system at original speed anyway...)

For the Apple II and Commodore PETs one would definitely want the switch, either to compare original vs. accelerated speed, or since acceleration is simply undesirable for some software -- say action games.

Anyway -- this project has been good fun. But I would enjoy showing it off "live" at a vintage computer festival at some point, when these get back on the map!

[MK14]

The TK20 is a clever idea. Using 100 fast logic chips of that time era, to make a much faster 6502. Copying the relatively slow EPROM, into faster (presumably) static ram, is good.

I'm glad you intend to publish the PCB layout and VHDL source code. Because 40 pin DIL outlines, are relatively small, already. Using BGAs, is sort of forced, if you want the device to have many pins as well, which that FPGA application needs.

When I said x100 faster, I meant when it was the original 1 MHz 6502 applications/hardware. Such as most Commodore Pets/64's, early home computers, and (presumably) some of the early standalone dedicated Chess computers.

E.g. The KIM1 1 MHz 6502, running Microchess.
http://www.benlo.com/microchess/index.html

Amazingly, the above link, has the actual Microchess program, which you can type in, in 6502 Hexadecimal format.

But, you are right. The later Chess computers, with faster cpus, such as 5 MHz 65C02's, would **ONLY** be x20 times faster. But that would still be a substantial speed improvement.
E.g. 5 minutes a move, would then be a much more palatable, 15 seconds a move. So, well worth it.

Anyway, speed up the old commodore Pet computer by around x100 times, sounds amazing. That would work, not just for any chess programs, but everything!
It would be fun, to see/play with such a turbo charged pet computer.

On the other hand, there may be some technical hurdles which still need to be solved/fixed. I know you mentioned about 'slowing down', when I/O activities (via memory map), are detected. But some things are timed in other ways, let me explain.
I've played around with simulators/emulators, of computers of that era, but the huge speedups, can cause it to malfunction. E.g. Keyboard denouncing failure, on a Jupiter Ace (Forth) machine emulator.

E.g. The Commodore Pet, probably does keyboard denouncing in software. So if you speed it up by x100 times, it might cause the keys to bounce (unwanted extra repeat keystrokes, when typing), in a very annoying way.
Because the (guesstimate) 10 msec denounce, would then only be 100 usec in duration, which would probably not be long enough.
Similarly, that (and maybe other) related timing issues, may apply to other computers and things.

EDIT: I'm probably being too pessimistic here. Hopefully/probably the Pet uses a timing reference, such as the screen refresh rate (50 or 60 Hz), to act as a reliable timing signal, for debouncing the keyboard and/or uses ICs, which automatically read/control the keyboard, and send the results to the 6502.
With it being such an old computer design, I wouldn't be surprised, if it reads in the keyboard matrix manually, but as long as the timing is still exact (i.e. independent of the 1 MHz 6502 clock signal), you should be fine.

[m k]

First used '82 or '83, any of
Z80 proto board (1st ASM coding)
ABC 80
MikroMikko 1

First owned '82 or '83
Commodore Vic-20 (2nd ASM coding)
- brain damaged by BASIC
2. DEC Rainbow 100 (ST412 upgrade)
3. Commodore C-16 (hardly ever used)
- brains partially healed by Pascal
4. Amiga 500 (later heavily upgraded)
- USA high-tech embarco softens
- Xenix was also around
5. Unisys PC
- found a mess called simply C

Around '90 I wondered what I'm gonna do with
VAX 11/750 with RL disks, RM disks, TS11 tape and silky LA120 console.
The machine had a 20A mains plug.
Luckily it wasn't finally my problem.

[ebastler]

When I said x100 faster, I meant when it was the original 1 MHz 6502 applications/hardware. Such as most Commodore Pets/64's, early home computers, and (presumably) some of the early standalone dedicated Chess computers.

E.g. The KIM1 1 MHz 6502, running Microchess.
http://www.benlo.com/microchess/index.html

For the Apple II and the various Commodore PET generations, we do indeed get very close to a factor of 100. (With a bit of loss for e.g. keybord polling, and some more loss if printing or plotting to video RAM.)

The minimalistic chess programs like Microchess might in fact benefit nicely from the acceleration -- relatively speaking, starting fro a low level -- since they probably don't use pruning of the search tree much, if at all. But then, they might run into memory limitations soon when the recursion goes too deep?

E.g. The Commodore Pet, probably does keyboard denouncing in software. So if you speed it up by x100 times, it might cause the keys to bounce (unwanted extra repeat keystrokes, when typing), in a very annoying way.
Because the (guesstimate) 10 msec denounce, would then only be 100 usec in duration, which would probably not be long enough.
Similarly, that (and maybe other) related timing issues, may apply to other computers and things.

In most cases this should work with the 65F02 accelerator. As soon as an I/O address (which is known to be time critical) is accessed, the 65F02 falls back to the original host speed. And it stays in that mode as long as code on the same subroutine level, or in deeper subroutine calls, is being executed. Only when the code returns "upwards" from that level via an RTS or RTI, the 65F02 switches back to accelerated mode.

While this is not waterproof, and it is easy to show coding patterns where it fails, this works well in practice. The Apple II runs Apple DOS nicely, and the Woz disk controller and RWTS routine are notorious for relying heavily on software-defined timing! Keyboard polling, speaker beeps, and timing the RC delays on the analog paddle inputs also work.

[MK14]

this works well in practice. The Apple II runs Apple DOS nicely, and the Woz disk controller and RWTS routine are notorious for relying heavily on software-defined timing! Keyboard polling, speaker beeps, and timing the RC delays on the analog paddle inputs also work.

You're making me too jealous, being the first (or one of the first), to try that 'new' 6502/65C02 100 MHz cpu, on the Apple and things. 

I like the way you are automatically slowing down (back to the original base speed), when the slightest hint of I/O activities, are taking place. Using the RTS/RTI is clever as well, to help detect when to do this.

As you seem to be saying, software delay loops, can be problematic. Tricky to easily/reliably/robustly detect, yet potentially a pain, if run too quickly. E.g. Things shown on screen, way too briefly, as the delay loops are running too fast.

Even on the original vintage machines (not 6502, but DOS X86), running a game on, say a 80286 at 12 MHz, when it was designed for an 8088 at 4.77 MHz. Would sometimes result in a completely unplayable game (way too fast), even in those 'old' days.
Hence some had turbo buttons, to help solve that problem. Which is why I like the option of an external magnet, to activate a reed switch (or hall effect device), and hence switch the speed of the FPGAs emulation of the 6502/65C02.

As you said,m some early chess programs, weren't designed for long computation times and/or didn't have access to enough memory. Which may mean they can't usefully use time periods, greater than some amount, such as 15 minutes.

I remember playing with an ancient (now) Chess program (I think it was Psion Chess, by Richard Lang, for Dos), on an ancient 8086 computer, or later. When it was current and freshly available, around the 1980's. (Possibly later in time, when I revisited that program).
I seemed to find that the maximum useful search time (despite infinite time settings), was around 15 minutes. Any longer, and it never seemed to update the move or do anything useful, after that time period.
That seemed to correspond with something like 8 moves (ply), which was the most, it was happy to display (advanced thinking, moves line).

I.e. you leave it overnight, and the results seem to be the same as they were, after the first 15 minutes, on a fairly powerful computer.

[MK14]

First used '82 or '83, any of
Z80 proto board (1st ASM coding)
ABC 80
MikroMikko 1

First owned '82 or '83
Commodore Vic-20 (2nd ASM coding)
- brain damaged by BASIC
2. DEC Rainbow 100 (ST412 upgrade)
3. Commodore C-16 (hardly ever used)
- brains partially healed by Pascal
4. Amiga 500 (later heavily upgraded)
- USA high-tech embarco softens
- Xenix was also around
5. Unisys PC
- found a mess called simply C

Around '90 I wondered what I'm gonna do with
VAX 11/750 with RL disks, RM disks, TS11 tape and silky LA120 console.
The machine had a 20A mains plug.
Luckily it wasn't finally my problem.

That's a nice lot of interesting computers, to learn from. But I'm not sure I'd fully agree with your opinions, on the various programming languages.
Pascal is nice, we seem to agree!
The old/original C, (in my experience), was marred in the early days, by rather/very buggy and arguably poor quality C compilers, of the time. But it could of been, that I didn't have access to, or obtain, the best C compilers, available, all those years ago.

Some people, seem to dislike Basic, but I think it is quite a good language, within limits, as regards the original/old versions of Basic.
E.g. To write a quick/short program in, to do some useful things with. It just seems quicker and easier, than some of the other languages.
But as the program, becomes bigger and more complicated, the short comings (at least of older Basic dialects), seems to become an increasing hindrance. I.e. not so good (Basic), for bigger and/or more complicated programs.

[SilverSolder]

First used '82 or '83, any of
Z80 proto board (1st ASM coding)
ABC 80
MikroMikko 1

First owned '82 or '83
Commodore Vic-20 (2nd ASM coding)
- brain damaged by BASIC
2. DEC Rainbow 100 (ST412 upgrade)
3. Commodore C-16 (hardly ever used)
- brains partially healed by Pascal
4. Amiga 500 (later heavily upgraded)
- USA high-tech embarco softens
- Xenix was also around
5. Unisys PC
- found a mess called simply C

Around '90 I wondered what I'm gonna do with
VAX 11/750 with RL disks, RM disks, TS11 tape and silky LA120 console.
The machine had a 20A mains plug.
Luckily it wasn't finally my problem.

That's a nice lot of interesting computers, to learn from. But I'm not sure I'd fully agree with your opinions, on the various programming languages.
Pascal is nice, we seem to agree!
The old/original C, (in my experience), was marred in the early days, by rather/very buggy and arguably poor quality C compilers, of the time. But it could of been, that I didn't have access to, or obtain, the best C compilers, available, all those years ago.

Some people, seem to dislike Basic, but I think it is quite a good language, within limits, as regards the original/old versions of Basic.
E.g. To write a quick/short program in, to do some useful things with. It just seems quicker and easier, than some of the other languages.
But as the program, becomes bigger and more complicated, the short comings (at least of older Basic dialects), seems to become an increasing hindrance. I.e. not so good (Basic), for bigger and/or more complicated programs.

There is nothing inherently wrong with Basic...  even VBA in Office applications can be pressed into service and produce useful results. 

[bd139]

Depend which variety. I used a fair number of derivatives and the only one that stuck was BBC BASIC on ARM. That was pretty good.

[SilverSolder]

Depend which variety. I used a fair number of derivatives and the only one that stuck was BBC BASIC on ARM. That was pretty good.

I recall doing Spice network analysis with an Acorn Atom,  that would have been a 6502, running a very terse version of Basic.  It didn't actually perform too bad, thinking back! 

[m k]

I'm still picking Basic for LibreOffice macros.

That Rainbow was a sophisticated machine, 3 in 1 and toolless case.
Forgot that I also got a bargain deal for a color monitor that was not available without a service agreement.
And of course the monitor was RGBC or SOG, so not compatible there eighter.
The motherboard had no general bus but it had "second floor" for add-on boards.
Memory seems to be mixing things with PC350 but I'm quite sure they both were no tools needed style.

Amiga then, it had external add-ons, finally so that without tweakings it was sort of a dinner table machine and original part being just a peripheral pass through controller.
It had a straight through board side extension connector system, like a backplane that can't be separated.
For that I also installed a piggy pack ALF? ST01 SCSI disk controller with a ribbon cable bus, it was finally quite stable, after adding so much groundings that twisted pair alternative would have been easier.
Luckily that machine had RGBC available but a small pain still, the connector was Cannon D-sub but 23pin, nothing normal 25pin and a saw can't handle though.
On the other hand, all of my Commodore manuals have been very good ones.

Much later that RGBC came back.
That time it was Playstation and regular H/V-sync PC monitor.
Luckily time passed and I didn't need those on purpose ordered, sync separator and video driver chips, with what I missed their package types and got msops.
I thought it was too much, even for trusted Magnastat.

[RJSV]

My first, owned was a Commodore 64.  I was ambitious, early 1980's and wanted (more) success, than hanging out... waiting for my tech company employer to get bought out or worse...

   Guys in office said "C-64 is the happnin thing, this year ('83), so off to TOYS R US I went, including purchase of an assembly language enabling CART called HESMON.
  Funny, all that software, (we sold Z-80 controllers with a mid level main frame), all thoses smart people, and they crowded around my C-64 like they never saw microcode before.  Hesmon could do some real-time tricks, such as running at one thousandth the usual rate, which made screen graphics functions entertaining and informative.
Commodore rocked, in 1983

[SilverSolder]

I had the C-64, and also the Vic-20.  Fun and well thought out,  well made mechanically too.

[m k]

Were you Elite?

[DrG]

Were you Elite?

That's 1337 (Leet) https://en.wikipedia.org/wiki/Leet 

[SilverSolder]

Aren't we all? 

[m k]

Nope.

[gooligumelec]

First computer?

Commodore CBM 4016 with 12" display (aka the "fat 40").
It was the family's, but I upgraded it to 32k by soldering RAMs (4116s, I assume) to the motherboard.  Luckily it still worked!

Stayed loyal to C= after that: C64, then Amiga 1000, Amiga 2000 (later upgrade to 68030), Amiga 4000 (PowerPC card added later).  Fun times.

[m.drachov]

Guess I'm a bit younger than most here.

My first computer was a Packard Bell (don't remember the specific model), looked exactly like this one:
" alt="" class="bbc_img" />

[Smokey]

...
2. DEC Rainbow 100 (ST412 upgrade)
...

I was at a swap meet about 20 years ago and bought a fully functional Digital Rainbow for $15.  I took it home, plugged it in, confirmed it worked, then stuck it in a closet where it has sat ever since.  One day when I have a ranch or something I'll dedicate a room to obsolete stuff like the Rainbow and Tek 7000 series mainframes and such.

[Phil_G]

My very first wouldnt really count as a computer, an intel 4004 based on a practical electronics (practical wireless?) article.  Then came a homebrew switch-&-leds SC/MP, then my mainstay for many years, a Nascom-1 built from a kit.  It was expanded to the point it was unreliable but with ZEAP assembler and MS Basic both in ROM it was a great development machine.  It helped me persuade work to buy a Nascom-2 which was perfectly reliable apart from one day when a tantalum suddenly burnt like a match-head, it had been reversed since its build.
Eventually I got fed up with my own Nascoms problems and bought an Eaca Genie which was great for a few years using ZEN assembler.  Next was a CP/M box, the Micronix MX800 which was a badged Ferguson Bigboard and took me to the PC era and an Opus XT-Turbo (8Mhz!). The rest is of no real interest to a retrobate, except to say that the story has gone full circle back to CP/M Z80s
Cheers
Phil

[YurkshireLad]

I don't know what it was, but my friend's dad brought a computer home from work. It must have been somewhere in the late 70s to '81. All I remember is that it had a green screen and I used it to play and ascii based golf game for hours. Addiction baby!

[DrG]

My very first wouldnt really count as a computer, an intel 4004 based on a practical electronics (practical wireless?) article.  Then came a homebrew switch-&-leds SC/MP, then my mainstay for many years, a Nascom-1 built from a kit.  It was expanded to the point it was unreliable but with ZEAP assembler and MS Basic both in ROM it was a great development machine.  It helped me persuade work to buy a Nascom-2 which was perfectly reliable apart from one day when a tantalum suddenly burnt like a match-head, it had been reversed since its build.
Eventually I got fed up with my own Nascoms problems and bought an Eaca Genie which was great for a few years using ZEN assembler.  Next was a CP/M box, the Micronix MX800 which was a badged Ferguson Bigboard and took me to the PC era and an Opus XT-Turbo (8Mhz!). The rest is of no real interest to a retrobate, except to say that the story has gone full circle back to CP/M Z80s
Cheers
Phil

whoa real old-school stuff - do you have any of the 4004 gear left?