• Please review our updated Terms and Rules here

C65 in eBay...OMG!

I believe I saw estimates that the Tiger Learning Computer was expected to sell a few hundred thousand units in 1997 if the license agreement had continued. That doesn't seem like a bad result for an Apple IIe turned toy. Other 8-bit systems might have been able to capture the same bottom rung secondary system for children market but prices would need to be severely limited and I doubt anyone would buy a $500 fancy monitor to use the advanced graphics capability of a $100 computer.
 
The Tiger Learning Computer might've worked in 1992, when the Apple II line was nearing its death but was still ubiquitous in schools. But just like the C65, it was five years too late.

Also as I recall one of the problems with it is that it included AppleWorks which ran in 80-column mode and thus was nearly unreadable on most TVs.
 
I don't think there would have been much of a market for the C-65 by the early 90's. For games playing, there were surely already better alternatives in dedicated consoles. On the other hand for productivity, PCs with more capable processors were already the norm.

The situation with dedicated consoles having their lunch eaten by the likes of the C-64 had reversed by this time. When the C-64 came out, the general purpose home computers were on par price-wise with a dedicated games console. However, anything past the NES era the prices were considerably cheaper than a personal computer.
 
Chuck(G) said:
The contention that an 8 bit CPU can run rings around what amounts to a 16-or-32-bit CPU on a 16-bit bus for general applications, I think, is misguided. Going back to my example, computations in my friend's application were done in double-precision floating point (64 bit FP words). Surely you're not going to tell me that a 6502 can beat a 68K or even an 80286 in that setting?

If so, I'd like to see some solid benchmark data.
Click to expand...

This isn't that setting. My AMC Eagle can run rings around your Corvette, but not on a dry race track, obviously.

There are few situations where floating point maths are actually required in the real world. The most precise things in the world are manufactured with fixed precision maths.

If I can find it, I'll post it. But in 2017 I can't get Google to find anything useful. A few years back someone put a number of processors through their paces and comparing a 68000 to a 6502, the '02 was generally faster. And that didn't factor in interrupts, which are very critical on the kind of machines we're discussing here. There isn't a mass produced microprocessor in the world* that can come close to the interrupt processing speed of a 65c02. A 65EC02 is significantly faster in most operations than a 65c02. So is the 65c816.

*Mind you my knowledge of microprocessors is about 20 years out of date. I have no idea what current chips are like.
 
krebizfan said:
The Amstrad PCW line continued having respectable volume until the mid-90s. Not exactly great systems, even by the standards of CP/M word processors, but just enough functionality to do the assigned task while being inexpensive.
Click to expand...

In the UK and Europe, this "word processor" sold around 8 million units by the time it was finally retired (in 1998 according to Wikipedia). "Outstanding", rather than "respectable", and I would challenge your "not exactly great" assessment, as Locoscript was much easier to use than WordStar and it had a sort of bitmapped screen, so could do graphics and even play games like Head Over Heels and Batman, which were two of the class leading isometric games of the time. Coming with everything you needed - including a printer, for less than the cost of a typewriter was quite an achievement, and there are anecdotal reports of many still in use today (by users, not retro computing nuts). The CP/M Plus implementation is pretty good too.

Of course, much depends on what you wanted to use it for. If games, look elsewhere, but if you wanted a cheap machine for writing letters, documents, etc then it was a bargain. Had you priced up, say, an Atari ST with a printer and decent WP software I suspect you would have paid an awful lot more (although the ST is obviously superior).

Anyway, sorry, that was off topic of me... how much has the C65 been bid up to? (Checks eBay... OMG 17,550 Euros!!!)
 
Last edited:
KC9UDX said:
There are few situations where floating point maths are actually required in the real world. The most precise things in the world are manufactured with fixed precision maths.
Click to expand...

Golly, most scientific work, which one can argue has been the driving force behind improving processor performance relies heavily on floating point. Seymour Cray, according to your statement, was wasting his time on getting his system's floating-point performance optimized. Fast integers would have been adequate.

Did you see your weather forecast today? What do you think was used for that? 16-bit integers?

Both Intel and Motorola must have been wasting their time rushing numeric data coprocessors (which happen to be significantly more complex that the MPUs of the time) out the door.

In the real world, precision is a notion, not a tangible thing, as few measurements of a physical quantity are exactly precise.
 
Chuck(G) said:
Golly, most scientific work, which one can argue has been the driving force behind improving processor performance relies heavily on floating point. Seymour Cray, according to your statement, was wasting his time on getting his system's floating-point performance optimized. Fast integers would have been adequate.

Did you see your weather forecast today? What do you think was used for that? 16-bit integers?

Both Intel and Motorola must have been wasting their time rushing numeric data coprocessors (which happen to be significantly more complex that the MPUs of the time) out the door.

In the real world, precision is a notion, not a tangible thing, as few measurements of a physical quantity are exactly precise.
Click to expand...

Seymour Cray was not designing desktop computers for home use. We don't need to calculate weather forecasts at home, especially in 1991. Integer math is not fixed point math. Do you know the kind of range and precision you can get with 24 bit fixed point? 16, 24, 32 bit fixed point calculations are pretty trivial on 8 bit processors, and still usually faster than FPU calls. Floating point only adds rounding error.

Sure due to the overabundance of cheap computing power people are using their home computers for things like solar system simulations and fractal calculation. But beyond whimsical entertainment, what legitimate uses do we have for floating point maths at home? On a mass market C65, A500, Apple ||gs or Macintosh? Even today with 64 bit processors everywhere; don't those do 64 bit integer maths? Do you know what kind of range and precision you can achieve with 64 bits?

Floating point is mostly used in large systems requiring varying precision. The solar system example is a good one. You need both large numbers and small numbers with the same number of digits of precision. Floating point is also unnecessarily used in many, many situations by programmers who just don't realise that they can get the job done without it, and faster.

All measurements of physical quantity have precision, and tolerance. You wouldn't use a gallon jug to measure a quarter teaspoon of salt. Floating point reduces your bit width by sacrificing precision on large numbers and incidentally increasing it on small numbers. In most systems, common precision is needed on the largest and smallest numbers in the system, and increased precision is not profitable.
 
Floating point is extensively used in digital audio production, since it enables clipping-less extreme headroom in processing plugins. FP DSP is pretty standard for professional audio production inside the DAW, prior to export to some integer audio representation, such as the common 16-bit WAV.
 
lowen said:
Floating point is extensively used in digital audio production, since it enables clipping-less extreme headroom in processing plugins. FP DSP is pretty standard for professional audio production inside the DAW, prior to export to some integer audio representation, such as the common 16-bit WAV.
Click to expand...

But is it necessary? What does it give besides rounding error that an extra 48 bits of fractional resolution couldn't accomplish? If you want clipping-less, you need fine resolution, not expanded range. Or go analogue..

Would the CPU in a machine like the C65 need it for that? Or wouldn't you add a 56000 if you needed DSP? There weren't competing machines doing floating point DSP at that time.
 
Floating point adds rounding error? Howzat? Truncation, perhaps.

Floating point is a scheme to preserve significance. I'm familiar with fixed-point computation (i.e. implied point) as opposed to integer. The problem is that fixed-point doesn't preserve significance. Let take a fixed-point example where the decimal is after the rightmost digit (i.e. integers). In 24 bits, if you have the number for 10, it has, at most 2 significant digits, where 12345678 has 8--there's no way to show that the 10 has the same significance as 12345678. Granted, FP doesn't track significance either--that's what error analysis is all about--but at least it gives one a tool to deal with the problem.

Lots of desktop computers in 1991 were applied to solving problems requiring floating point--consider, for example, statistics.

I don't play computer games--and I suspect that your idea of "1991 desktop" applications is very different from mine. Maybe I'm prejudiced--after all, I made some money writing floating point packages for 8- and 16-bit MPUs.
 
I don't play a lot of games either, but that's what the C65 was mainly intended to do.

The only thing I need FP for at that time was AutoCAD, and that shouldn't have needed it either. Unless people were designing things larger than 6 miles in any direction with precision better than .0001", which no one should do.

But here's the thing. If the designers of the C65 thought they needed FP, the solution would be to add a Weitek coprocessor, not jump ship to 68k or x86. If anyone bought one and then panicked that they didn't have hardware FP, putting a coprocessor on a cartridge would be easy enough.

The lack of hardware FP does not make a C65 inferior to an A500. It didn't make the ||gs inferior to the Mac.
 
By the way, that's only 32 bits, some 6½ miles at .0001" resolution. How many bits do you need to represent that in floating point? I don't even know, but it's more than 32 bits. How long does 32 bit integer calculation take versus greater than 32 bit floating point calculation?

The whole reason I stopped using FP was that I found I could use larger values without losing precision. I wasn't even concerned with speed.
 
I think the real issue is learning where you should be using integer math and where you should be
using floating point.
I've seen too many times where floating point was chosen as a mistake ( In one case, I know of, people died ).
I think the real problem is the general lack of understanding what the limits are.
Thing are easier to do with floating point. No one seems to ever really look to see where it fails and why.
Dwight
 
The argument is senseless.

Floating point math is a semi-logarithmic (or exponential; two sides of the same coin) way of looking at the world. Like it or not, that's the way most of our natural world works. Audio signal processing is a great example of this--the ear is not a linear device--it's basically logarithmic. This is a good thing--it gives us the ability to hear things both very lout and very soft and is not a linear relationship with power. Continuous compounding of interest; standard deviation...

I've seen proposals over the years that suggest that our floating-point math should be fully logarithmic.

It's an alternate (and often useful) notation system. Like phasors in AC circuits. Sure, you could compute everything in rectangular coordinates, but why?

P.S. There is the issue of what kind of floating point--implicit or maintained?
 
Last edited:
vwestlife said:
The motherboards came from the C64GS after that braindead idea of turning the C64 into a keyboardless game console in 1990 failed miserably (which maybe was the catalyst to pull the plug on the C65).
Click to expand...
According to Brian Bagnall's book, "Commodore: The Amiga Years", CBM had limited financial resources. They could either release the Amiga 600 or release the C65. They chose to release the A600.

Truly,
Robert Bernardo
Fresno Commodore User Group
http://www.dickestel.com/fcug.htm
 
RobertB said:
1992
Click to expand...
To quote from Wikipedia:
"In March 1994, at CeBIT in Hanover, Germany, Commodore announced that the C64 would be finally discontinued in 1995,[54] noting that the Commodore 1541 cost more than the C64 itself.[54] However, only one month later, in April 1994, the company filed for bankruptcy."
[54] - https://archive.org/stream/AmigaFor...ure_Publishing__GB__300dpi_#page/n21/mode/2up

Amstrad CPC was discontinued in 1990, Apple IIGS, Atari 8-bit, and SAM Coupe in 1992, Panasonic MSX TurboR in 1993.
Yes, C64 was The Last Mohican of 8-bit - not counting word processors and other specialized devices.
 
If you count the ||gs as 8 bit, you include the Mensch Computer, no? Or is that too specialised?
 
Xacalite said:
To quote from Wikipedia:
"In March 1994, at CeBIT in Hanover, Germany, Commodore announced that the C64 would be finally discontinued in 1995,[54] noting that the Commodore 1541 cost more than the C64 itself.[54] However, only one month later, in April 1994, the company filed for bankruptcy."
[54] - https://archive.org/stream/AmigaFor...ure_Publishing__GB__300dpi_#page/n21/mode/2up

Amstrad CPC was discontinued in 1990, Apple IIGS, Atari 8-bit, and SAM Coupe in 1992, Panasonic MSX TurboR in 1993.
Yes, C64 was The Last Mohican of 8-bit - not counting word processors and other specialized devices.
Click to expand...

IIRC, Commodore had been planning to discontinue the C64 as far back as 1986, and reportedly even stopped production of it at some point, but restarted production due to continued popular demand for it.

As for 8-bit computers in the '90s, there was the Brother Super PowerNote series, circa 1996-1997, using the Z80-based Hitachi HD64180. Not just a word processor -- these could run applications from disk, including two games that were released for them (Tetris and Turnabout, a.k.a. Reversi).

6978858519_cb8beea1a4_b.jpg
 
You must log in or register to reply here.
Back
Top