Did computer games for Commodore 64 really take "25 minutes" to load "if everything went alright"?

Question

In the Swedish-language song "Nostalgi" by Markoolio from 2003, some of the lyrics go:

My first computer had 64 kB
you surfed on the water, and that's that
The computer games were loaded in from tapes
it took 25 minutes if everything went alright.

(Obviously, it doesn't rhyme when translated into English.)

He was likely referring to his Commodore 64. As far as I understand, that home computer/console came with a cartridge slot built in, but no floppy disk station or tape player. You had to buy the two latter separately. (But they had slots to connect them.)

I realize that piracy was absolutely rampant, so maybe people invested in separate tape players and then copied games easily onto cheap tapes and that's why it took so long to load them in? But even then, 25 minutes still sounds insane. Even if it was really just 1-2 minutes, that would still be significantly longer than the slow floppies that I remember impatiently waiting for. Imagine sitting down to play a game before lunch, only to realize that the game has not finished loading by the time the food is ready...

Did they actually sell commercial/legal software for Commodore 64 and other early home computers on tapes on shelves in stores? For full price? Surely those "genuine" copies must have come on the expensive-to-produce and impossible-to-reproduce-at-home cartridges with near-zero loading time? So if you had actually purchased a Commodore 64 game, it would not take "25 minutes" to load, but more like 25 milliseconds? You'd just slide in the cartridge and press "power on" and it's running right away?

Is the only reason that people talk about games taking so long to load in those days that they only experienced pirated/warez copies of games? They never bought any for money? It makes me wonder how anyone could make a living in those days...

Also, what's that part about "if everything went right"? It could also fail to load, even at that slow speed?

Is there something I'm fundamentally missing?

Answer 1

Theoretically it could take 25 minutes (or more), in practice it never did.

Theoretically it could, because the C-64's built-in tape handling routines had a data rate of about 300 bit/s. That's 37.5 bytes per second, or almost 30 minutes for a full 64K.

In practice, it never did, because the tape handling / decoding was done almost entirely in software, Commodore's own tape handling routines were horribly inefficient*, and third party speeders managed data rates close to 5 kbit/s, reducing load times dramatically. Almost every bit of commercial software, and every quality crack released on tape ;-), had a two part loader, first loading a fast loader at normal speed, then loading the rest of the game at higher speed. I can't recall waiting more than a few minutes for any game to load.

As for cartridges: Hardly any game or application software for the C-64 was sold on cartridges. Tapes and disks were way cheaper to produce.

* One source claims that the inefficiency was a response to the quality of audio tapes typically used at the time the PET was designed (mid-70s), and that only improved tape quality in the mid-80s made fast loaders possible. I'm not convinced.

Answer 2

Yes, cassettes were common, they took ages, and they were error prone.

In Europe, disk drives for the ZX Spectrum and Commodore 64 were uncommon. It's the same for cartridge games for the C64. The market for the 8-bit micros in the 80s in the UK was driven by price, so the cassette interfaces were the majority here. (BBC Micros with floppy drives were common for education.)

As for 'did they actually sell' cassettes - yes, in fact They Sold A Million.

There's at least 3 of these compilations - each compilation's games were chosen as between them they sold at least a million copies.

In this answer, the C64 tape data rate is calculated at 55 bytes per second, giving approximately 15 minutes to load 48 kilobytes. On the Spectrum, a title screen alone (6912 bytes) would take just under a minute (like in this video of Jetpac) to load from tape. (Though to a kid it would feel like forever :) .)

The typical C64 tape interface was a first party Datasette with little user tweaking required. On the Spectrum using an off the shelf tape recorder you'd need to adjust the volume level using trial and error. On both, especially with pirate cassettes there might be an extra bit of calibration required - cassettes recorded on one recorder might not 'like' another.

A View To A Kill is a multiload C64 game (more like a sequence of minigames) and each section has a very lengthy loading time. I haven't measured it but I could accept AVTAK having a combined loading time of 25 minutes.

I suggest you read a magazine like Zzap64 to get an impression of what was available at the time.

Answer 3

TL;DR: Yes, it could take up to 25 minutes (from tape) and it could fail due to many reasons, like record dropout or bad adjusted head.

Long story: It was pretty common to distribute software and games on tape cassettes. Cartridges were rare, much rarer than the cassettes. And it was not only for Commodore - almost every home computer of the early 80s' has the tape deck to load and save software; floppy drives were a little bit uncommon, at least in Europe, especially behind the Iron Curtain.

Cartridges are pretty fast but remember: it is a solid piece of hardware, some ICs on the PCB, so they cannot beat the price of the more versatile cassettes and their manufacturing was much expensive. They were common for game consoles, because of its instant loading time. But in the home computer world (don't forget that all those Commodores, Ataris, Spectrums, etc. were computers, not only game consoles) you need a memory medium, not only the "game load medium". On the other side: it was slow and you have to face "Tape Loading Error" from time to time...

Answer 4

As other answers note, the basic tape and disk routines on the C64 were notoriously slow. Commercial game releases usually were on tape, but tended to use a "fastloader" routine to bring the loading times down to something vaguely sane. The loader routine also served as an excuse to display a loading screen and play music while you waited - and some of those loader tunes were works of art in themselves. Many games didn't need to load a full 64KB into RAM before starting. This also tended to speed up loading.

For concrete examples, I timed loading Cybernoid into an emulator, with the "warp load" feature of the emulator switched off so that it ran in simulated realtime. It took about 5m15s until the title screen and music started. This is a game that doesn't have a distinct loading screen and music; it takes about a minute to bootstrap the fastloader, and the rest of the time is spent loading the game proper.

Arkanoid is another famous example that does have a loading screen. The fastloader bootstraps in about 40 seconds, the loader music starts at the one minute mark, and about the next 45 seconds are spent loading the loader screen. In total it takes less than 5 minutes before the loader music changes to the title music, at which point loading is complete, despite the extra time taken by the loading screen.

The Last Ninja actually uses both sides of the tape - it is a comparatively big game for the machine. However, it is effectively several distinct games chained together, so there is a relatively normal loading time at the start, and subsequently after completing each level, each with its own set of loader music. The loader music begins at about 1m10s, and the first level completes loading in 7m30s total.

In short, there's a genuine reason why short 10 and 15 minute tapes - this rating counting both sides - were commonly produced for use with home computers.

Answer 5

Did they actually sell commercial/legal software for Commodore 64 and other early home computers on tapes on shelves in stores? For full price?

I remember that the Oric 1/Atmos had a fast and a slow mode.

Slow mode was 300 bauds. But it wasn't the default. Default was "fast" (2400 bauds). Both modes were built-in (there were slightly faster custom formats but I never saw them in any production)

However, some commercial programs (by Tansoft, mostly) put the fast loading version of the game on one side and the slow loading version on the other side. So it probably decreased the risk of not loading the tape with a bad tape player (there were a ton of them)

But it was dropped gradually. For instance there are 2 versions of Harrier Attack

The Oric-1 version had a tape with 2 fast recordings on side A, and a slow recording on side B, and was labelled "Oric".

The Oric-1/Atmos version only had one fast recording on both sides, and was labelled "Atmos / Oric 1".

I remember that Harrier Attack loaded in around 3 minutes in fast load mode, so more than 20 minutes on slow mode probably (who used that ? :))

Of course if the tape player chewed on the tape, it destroyed both sides so it was unlikely to be safer. The only proper safety was to record software on both sides (fast was okay) just if one side was erased by mistake (because someone disabled the read-only tape safety?)

Cracked copies were always spread in fast mode. And the Oric had no commerical software on disk or cartridge (note: there's an almost-cartridge system now with Erebus card which is able to read tape data from SD card and transfert data through parallel port at lightning speed)

Answer 6

Loading from cassette was very, very common and the default option for most 8-bit micro owners. The early micros like the C64 were reasonably affordable, and the fact you could hook them up to a regular television set as a monitor and an ordinary cassette deck to load data added to that affordability. Floppy drives were quite expensive and could double the cost of owning a system.

Games took various different lengths of time to load from cassette depending on a number of factors, but each individual game always took the same length of time to load because the cassette ran at a specific speed. Games loaded from the cassette into the computer's RAM so, broadly speaking, the more memory the game took up when loaded, the longer it took to load.

In the very early days of all 8-bit micros, games tended to be quite simple in concept and graphically. Some were designed to fit into far less RAM than your computer had because many micros came with memory options (for example, the Commodore had 16K and 64K models, and the ZX Spectrum 16K and 48K models). Games designed for 16K naturally loaded faster than later, more complex titles that took full advantage of the enhanced machines.

Part of the tape-loading experience was a 'loading screen' - usually a graphical display that you could look at while the game loaded. This had to load and show on the screen before the main game started loading, so this added to the loading time. A few games even had loading music that played while the main game loaded. These things loaded in a sequence, so the loading screen/music would load first and be cleared from RAM by the time the main program had loaded.

As development on 8-bit micros progressed, programmers found ways to make games load faster by coding their own tape loading systems which could receive data faster than the system's in-built method. To make use of this though, the fast-loading system had to load itself first using the system method, and then the rest of the data on the tape would be in the faster format. These systems often had a timer on the screen which counted down so you could see how long the game had left to load. As this later era of software coincided with bigger titles, games could still take some considerable time to load.

Finally, it is worth mentioning 'multi-load' games. These were games that were too big to fit into the memory in one go. Often these were arcade-conversions which had not been designed with small micros in mind. These would usually load the game engine first, then you were asked to stop the tape. Each level of the game would load separately and you were prompted to stop and start the tape in-between levels.

I'd say that 5-10 minutes is probably a more realistic average loading time for a decent game at the height of the 8-bit micro era, but to a child that wait could seem a lot longer which may account for the exaggeration in the song you quoted.

Having games on cassette did make piracy quite easy, but it also kept production costs quite low so arguably there was less temptation to pirate them when you could buy them at an affordable price. Because there was no internet, piracy was limited to copying games from other people you knew which was far more limiting than file-sharing today. The time it would take for a game to copy tape-to-tape didn't put us off. I'd buy 90-minute blank cassettes which I would append newly copied games to. If a friend brought round a new game we'd load it, and then then I'd copy it while we played it.

The line from your song which says "if everything went alright" seems to refer to the fact that tape-loading could sometimes fail. Tapes could get damaged or just degrade over time. Poor quality cassette players could "eat" cassettes - causing the tape to come off the spool and get tangled up inside the deck. Cassette player heads could also be slightly misaligned, and users with persistent loading difficulties were often advised to adjust the azimuth alignment screw which could improve the tone of playback in a way that would be condusive to successful loading. Copied (pirated) cassettes could also be of poorer quality. Some twin-cassette decks available at the time had a feature called 'high-speed dubbing' that copied cassettes at double-speed. I found their games copied this way didn't always load so reliably, possibly because the process left a slight high-pitched whine on the dubbed copy which could interfere with loading.

Answer 7

Can confirm, at least as a ballbark figure: I remember Impossible Mission taking 15 minutes to load. We just went out in the back yard and played in the meantime, being about 12 at the time.

How did we put up with it? That was just how long it took. We didn't know how fast it would load 30-odd years in the future, so we didn't have that point of reference. I don't remember knowing that cartridges were even available for the C64, though the Spectrum definitely had. Not everything made it all the way out to NZ, back then.

Answer 8

The other answers cover the historic reality, but I also think it’s worthwhile to illustrate what were the technical reasons for tapes to be “slow”. Audio tape as a medium is not “slow” and is quite data-dense, compared to what one might think based on what the home micros did back then. Compared to C64’s default rate, the tape can do 4 orders of magnitude better at least!

Primarily, a lot of the information from signal recorded on tape was lost. It was passed to the CPU not as if you had a sound card, there wasn’t a parallel-output A/D converter that would sample the signal from the tape and feed 8-bit values to the CPU. The most popular tape interfaces had a simple analog signal shaper and a comparator for threshold detection. The CPU saw a serial bitstream on some input port. All the information you had left were zero crossings of the audio signal. Some micros probably did things a bit better, e.g. using a low resolution sigma-delta converter that would provide the equivalent of 2-4 bit sampling. It had to be cheap, even 8-bit A/D converters that could sample at kilohertz rates were expensive in the early 80s as well as power hungry, and usually were considered “glam” peripherals rather than something you could waste on tape duty.

8-bit microprocessors used in home micros had no multiplier, and to do the digital signal processing needed to demodulate high density tape data you definitely need lots of multiplications to implement the DSP algorithms. You could definitely hook an 8-bit ADC to a micro and implement a state-of-the-art demodulator system that could offer quite high data rates with error correction - think 1-5kbytes/s on a mono audio tape. Said demodulator would probably run at least 1000x slower than real time, though :( It couldn’t keep up with real-time tape data, and even if a large RAM buffer was available to sample the data, store, and process offline, you’d still get throughput much slower than the “naive” methods allow.

As far as what tapes are actually capable of: with a 16-bit sound card and processing throughput of a 486DX2, you could load 64 kilobytes from stereo tape in 15 seconds or so, with state of the art error correction and so on - that tape could be pretty well mangled, and the head azimuth a bit off, and it still worked. But this is the sort of stuff that even today requires specialist knowledge, and most people who designed I/O for micros of the era did not have that sort of a background. I did a class project very long ago on this and it was lots of work, especially that the access to literature was much harder in the late 90s, so collecting all the necessary bits of wisdom wasn’t easy - and I forgot like 99% of it since then, too :) This problem is harder than what the voice line modems had to cope with: they have a very limited adaptive ability for channel estimation, so if the channel changes too much, they must retrain to re-estimate the channel model, and while they do that, the data transmission is suspended. They also don’t tolerate any sort of timing slips - those usually force a full retraining. With tape you have a time-variable channel whose model you have to continuously estimate and adapt. Then the timing is not constant either, since the tape speed regulation is orders of magnitude worse than the phase stability needed for demodulation of dense data, so you have to do clock recovery and resample the data. All of this is computationally heavy, even in fixed point integer arithmetic. On a “fancy” CISC like the 486, you could run integer processing in parallel with floating point, but this had to be hand-coded in assembly at the time. Tedious stuff.

Answer 9

Commercial software was certainly released on cassette tapes. This was a very common practise in the days of early micro-computers since floppy disk drives were very expensive, and some micro-computers did not have floppy disk drive interfaces. For these computers, audio cassette was the only means of data storage.

I can remember having to wait for many minutes for games to load on my BBC Microcomputer (which only had 32 kilobytes of memory). However, I cannot remember the data format or the data speeds, so I cannot estimate the maximum time that you might have to wait for a game to load.

It is possible that on commercial tapes, each block of data may have been recorded more than once for redundancy purposes, and this would slow things down considerably. Also, to keep users amused, a smaller program might be loaded to play music while the main game was loading, and this would extend the overall loading time even more.

Answer 10

I don't know if the Commodore 64 used the same tape loader as the Commodore PET. The Commodore PET saved the program twice on tape. On loading, it would load the first image into memory and then read the second image and verify it against what had been loaded into memory. If there was a discrepancy, then you would get a verify error.

So a 25 minute load actually took 12.5 minutes. The rest of it was verifying the image. If there was a verify error, the program would still be in memory with possibly at least one byte being corrupted or a read error or a corrupted verification image.

It was a lot of fun watching engineers look for the eject button for the cassette drives. One guy, possibly an apple user kept typing eject. It was just this look of amazement when you lifted the cover to take the tape out.

We used to use 30 or 60 minute tapes, never anything longer. The cassette motor used to struggle with 90 minute tapes.

Edit Sometimes, these tapes had built in security so they couldn't be copied on the Commodore machines. They would load in two parts - first the security bit and then the rest of the program. If you saved the image in memory, it only saved the second part. If you then tried to load it on another machine and run it, it would fail because the second part would look for the first part, which was missing.

A common way of getting round the security was to do a tape to tape copy on your hi-fi unit with the volume turned right down.

Answer 11

Yes, although the disk releases here in North America weren't much faster. However with that said, awesome loader music made the waiting time far less painful. Cartridges may have been rare in the UK and most of Europe, however they were quite common in the early days of the computer here in North America. Most of the early Activision games for example were released in cartridge AND disk format. For the record cassette tapes kind of died out here in North America around late 1983-early 1984. I hardly know anyone who used a Datassete or a third party cassette deck here after the VIC-20 days.

Answer 12

The loading times were initially really long with the VC-20/VIC-20 and C-64, as described. The manufacturers also used special cassettes for this, which were intended for around 15 minutes audio (C15). From the mid-1980s, fast chargers were developed, which were available as hardware versions (chipsets and plug-in modules) or as software versions. A distinction must be made between programs that manipulate the hardware of the cassette and floppy devices (yes, there too!) and the programs that packed the data stream; in addition, several programs increased the data transfer rate. The basics of compressing data also come from this time, which was then called crunching/crushing or packing, called compressing. Many of these methods have been adopted by software vendors, making load times more tolerable, by using a pre-loader

The waiting time was really long at the beginning, but that changed quickly.

Answer 13

in the very early days of owning commodore 64 (1981) there were hardly any games available and any games had to be typed in from a book and were written in BASIC. Next came games on tape. These were mainly written in the mysterious machine code. Longer running tapes (90 mins or more) could stretch a bit, for example we had a tape with lots of games on and the ones at the beginning of the tape usually loaded whereas the ones later on usually didn't load. Tapes of 60 min length were usually safer. Load screens sometimes with music were sort of slightly later in the tape game loading experience. We had one cartridge (a maze game with a mouse and some cheese) which loaded instantly, presumably it was expensive. Later on in the owning experience came the external disk drive which was of course far quicker and more reliable (mid to late 80s?). As to '25 minutes' I wouldn't say that was not the norm. The longest loading game for tape we had was entirely text based adventure game The Hobbit. This took 15 minutes to load and 'usually' loaded but not always. According to wikipedia, Hobbit was released for C64 in 1984.

Answer 14

Common tape formats in those days were c-30 c-60 c-90, where the number is the time in minutes it takes to play the tape. Most games fit on 1 side. Some games used up to 4 c-60 tapes. (The lord of the rings text adventure?) c-90 tapes were rarely used as the thinner tape was more likely to get caught on dust and grease in the mechanism and end up being "chewed" by the playback machine.

Answer 15

Telengard was the one that did this for us.

We used to load Telegard on Fridays, and leave the machine powered on all weekend. This game sucked up every byte of RAM.

Answer 16

I recall working out my Vic 20 (and C64 I think) actually loaded the program twice, then ran it. I would load it properly, divide the counter by 2, and write that down on the cassette case so I would load just over half, stop the load, run the program (usually a game) and it always worked a treat. So the load time was cut in half. B.

Answer 17

Some games, Gauntlet on Atari 800, Lord of the rings on BBC, for example where multiple 90 minute tapes.