Intellivision development, back in the day

[Walter Ives]

On 11/29/2016 at 10:59 AM, intvnut said:

The 6502 does seem to have a performance edge over the CP-1610 at the same clock rate for general purpose code because its instruction cycles are just so lean

In comparing the microprocessors of the day it is more useful to compare memory cycle times and instruction times than clock rates. For example, if you look at their respective instruction times, you'll see that a 2 MHz 8080 effectively runs at the same speed as a 1 MHz 6800 or 6502 despite its higher clock rate—among other things the faster clock was used to pipeline nibbles through the ALU so as to use that portion of the silicon more effectively, but it was of no help whatsoever in moving data in and out of the chip. Similarly, although the CP-1600's designed clock rate was 5 MHz (CP1600 Microprocessor User Manual, May 1975, Section 1.0), the resulting instruction and memory cycle times were about the same as those of a 2 MHz 8080 or a 1 MHz 6800 or 6502. Providing a faster clock for a given process doesn't make the transistors switch any faster, doesn't decrease the read/write times of the on-board registers and doesn't change the propagation time through the ALU. What it does do is give the designer more clock phases with which to be clever. As a relevant example, the fact that the Keyboard Component was clocked at 3.58 MHz provided many clock phases with which to implement the dual port memory system, even though the 6502 was only clocked at 895 kHz.

 

You can't fault the CP-1600 for Maine and Harrower's decision to design a chip set that called for the CPU to be clocked as slowly as it was. The fully compatible STIC 1A could easily have allowed the clock rate of the CP-1600 to be doubled, but didn't because there was no point: it was a pure cost-reduction effort required to be 100% backwards compatible. STIC 1B, which extended the STIC 1A architecture, did allow the CP-1600 to be clocked at double speed.

 

WJI

[Walter Ives]

On 12/1/2016 at 2:47 AM, intvnut said:

Now, if we were to look at the original CP-1600 (of which the CP-1610 is a detuned variant), it ran at 2x the clock rate.

The CP-1610 used exactly the same die as the CP-1600. The ONLY physical difference was that it was packaged in plastic.

 

WJI

[Walter Ives]

On 5/10/2017 at 5:59 AM, mr_me said:

APh might have had an actual DEC PDP-11 or two.

Or twenty or thirty. One per programmer that wasn't using the PDP-20 or the VAX. Except Minkoff and Levine had to share one for Bowling.

 

WJI

[Walter Ives]

On 6/5/2017 at 6:18 AM, mr_me said:

APh, where Kimo Yap and Tom Loughry worked, had hard disks, they seemed to get the bleeding edge technology. In those days hard disks might have been removable platters.

Disk cartridge not included.

 

WJI

[Walter Ives]

On 6/6/2017 at 5:07 AM, decle said:

OK, the "Mattel tried to do things on the cheap" thing is something I have heard before, but again I've not found a quote for it. Do you have one?

Here you go: "Mattel tried to do things on the cheap."

 

WJI

[Walter Ives]

On 6/5/2017 at 6:18 AM, mr_me said:

I always wondered what was "modified" in the Keyboard Components development systems aka black whales.

There were blue whales, but the only black ones were in Keith Robinson's imagination. The blue whales used for development needed a ROM cartridge and a Rockwell 6551 ACIA cartridge. For large cartridges, or if you were unclear about relocating code, you could add a T-card populated with RAM. A bank-switching "T-card" allowed emulation of paged ROM. A Master Component jumper was required if you wanted breakpoints.

 

WJI

[Walter Ives]

On 9/24/2017 at 9:39 AM, decle said:

A DECSYSTEM-20 was a big beast of a machine requiring a three phase electric supply, properly air-conditioned computer room and technicians to tend it.

Yup. Except APh anarchistically dispensed with the technicians and allowed anyone into the machine room.

 

WJI

[Walter Ives]

On 9/24/2017 at 9:39 AM, decle said:

Therefore, APh could practically develop using a remote machine

Over one acoustically coupled 300 baud payphone line? Dream on.

 

WJI

[+DZ-Jay]

20 hours ago, Walter Ives said:

(For those unfamiliar with the work of Jan Łukasiewicz, you don't "pull" something off a stack, you "pop" it off. Using the term "pull" in this context is like squeaking chalk on a chalkboard. It marks the microprocessor as having been designed by an unsophisticated rube from [insert name of your favorite backward village here] and robs it of credibility.)

 

AMEN!

 

20 hours ago, Walter Ives said:

Programmers, on the other hand, especially programmers of more capable machines, don't give a rat's a** about the details of the implementation but feel that the addressing mode is naturally associated with the argument rather than the operation.

HALLELUJAH!!!

 

The first time I saw a sample of APh code in Your Friend The Exec and noticed the addressing mode indicator on the arguments, I thought, why isn't as1600 doing that? (Just like I thought, PULR??? Really?!  Who pulls from the stack??)

 

It just seems so much more intuitive.

 

I guess @intvnut was basing the syntax on GI's data sheets.

 

 

20 hours ago, Walter Ives said:

The CP-1610 used exactly the same die as the CP-1600. The ONLY physical difference was that it was packaged in plastic.

 

WJI

You have made this point several times in the past few days, so I suppose it's a notable distinction.

 

However, for those of us who do not really know much of anything about hardware and electronics, would you (or someone else) mind explaining he implications of this difference?

 

Does it improve or impair heat dissipation or something like that?  Or is it merely that plastic is cheaper than ceramic?

Inquiring (and ignorant) minds want to know ....

 

    dZ.

[Walter Ives]

On 2/12/2023 at 8:12 AM, Walter Ives said:

you'll see that a 2 MHz 8080 effectively runs at the same speed as a 1 MHz 6800 or 6502 despite its higher clock rate—among other things the faster clock was used to pipeline nibbles through the ALU so as to use that portion of the silicon more effectively, but it was of no help whatsoever in moving data in and out of the chip.

    Ooops. Meant Z80, not 8080.

 

    WJI

[Walter Ives]

On 2/13/2023 at 4:55 AM, DZ-Jay said:

 

On 2/12/2023 at 8:13 AM, Walter Ives said:

The CP-1610 used exactly the same die as the CP-1600. The ONLY physical difference was that it was packaged in plastic.

 

WJI

You have made this point several times in the past few days, so I suppose it's a notable distinction.

    I'm just stamping out burning ducks. I respond to comments in threads and don't particularly keep track of what I've written from one thread to the next. This site appears high in google search results for Intellivision and is taken as somewhat of an authority: my comments as much if not more for people who stumble onto the site while googling as for current members of the forum.

 

    WJI

[Walter Ives]

On 2/13/2023 at 4:55 AM, DZ-Jay said:

 

On 2/12/2023 at 8:13 AM, Walter Ives said:

The CP-1610 used exactly the same die as the CP-1600. The ONLY physical difference was that it was packaged in plastic.

 

WJI

You have made this point several times in the past few days, so I suppose it's a notable distinction.

 

However, for those of us who do not really know much of anything about hardware and electronics, would you (or someone else) mind explaining he implications of this difference?

 

Does it improve or impair heat dissipation or something like that?  Or is it merely that plastic is cheaper than ceramic?

    The CP1600 was developed under contract with Honeywell's Process Control Division to enable Honeywell to build what were in essence low-cost PDP-11s. To Honeywell in 1975, which needed a moderate volume of parts with an easy-to-program instruction set, each device was worth a hundred dollars (this is ballpark, I don't know the actual price. Also, it dropped over the years.). Selling the same part into the consumer electronics market presented GI with a problem: how to keep selling the part to Honeywell for a respectable price while selling the same part to Unisonic and Mattel for three. The solution: change the part number and the published spec for maximum clock rate. Presto! Changeo! Sparks fly like electricity! Problem solved. Note that the data sheet is silent on the question of plastic vs. ceramic packaging and that the specified operating currents for the CP1610 are precisely the same as those specified for a 5MHz CP-1600. Running the part slower reduced the required operating current and thus the amount of heat that had to be dissipated, but the individuals who actually typed in the altered specification apparently weren't sophisticated enough to know that.

    It was GI, not Mattel, that designed the Master Component circuit board, including the RF shields and heat sinks. GI was contractually responsible for providing chipsets that worked in that operating environment. The earliest chipsets it delivered were packaged in a ceramic package that conducted heat a little better. After considerable experimentation it settled on plastic packages with attached heat sinks because that was the cheaper solution.

    Most vendors supplying a part in a different package just change a letter appended to the part number, as in 74LS00N vs. 74LS00J. Had GI done that here it would have destroyed the illusion. Several members of this forum, relying on GI's smoke and shiny mirrors, have suggested that the CP1610 was somehow fundamentally different from the CP1600. For example, the post to which I was responding stated, "Now, if we were to look at the original CP-1600 (of which the CP-1610 is a detuned variant), it ran at 2x the clock rate." I wanted to make it clear that there was no process tweaking to "detune" the part and that the ONLY difference was the packaging; the two parts used exactly the same die.

 

    WJI

[Walter Ives]

On 11/26/2016 at 7:38 AM, mr_me said:

We know that in 1977, prior to APh involvement, Mattel was developing for the Intellivision in preparation for January CES. Don't know who that programmer was but I'm guessing they used a GI development system and assembler, likely on a GI cp1600 based computer.

Not sure what happened to the image associated with my previous response. Let's try again:

 

    APh's involvement began in December, 1975. You have undoubtedly read elsewhere that APh brought the GI chipset to Chang's attention. It was APh personnel who prepared the Winter CES demo in the secret windowless room at Mattel headquarters in 1977. GI had supplied a TTL emulator built into a chassis similar in size to the one shown in this EE Times article and mounted in a mini-rack.

    The APh engineer sent to survey the situation found a setup that had been unpacked but was otherwise "doornail dead." He plugged it in and spent a couple of days preparing a minimal development environment by hooking up a Teletype model ASR-33 to it, writing and hand-assembling a minimal version of APh's debugger and writing and hand-assembling a minimal interrupt driven program to establish that the system could display background and objects and that the hand-controllers could be read. After that was in place, APh dispatched two programmers to the same room to write and hand-assemble the rudimentary demonstration program shown at CES.

 

    WJI

[Walter Ives]

On 6/5/2017 at 6:18 AM, mr_me said:

APh, where Kimo Yap and Tom Loughry worked, had hard disks, they seemed to get the bleeding edge technology. In those days hard disks might have been removable platters.

    I'm not sure what happened to the image in my previous reply. Let's try again:

   

    Disk cartridge not included.

 

    WJI

[+DZ-Jay]

1 hour ago, Walter Ives said:

    I'm just stamping out burning ducks. I respond to comments in threads and don't particularly keep track of what I've written from one thread to the next. This site appears high in google search results for Intellivision and is taken as somewhat of an authority: my comments as much if not more for people who stumble onto the site while googling as for current members of the forum.

 

    WJI

That is fair and appreciated.

 

1 hour ago, Walter Ives said:

    The CP1600 was developed under contract with Honeywell's Process Control Division to enable Honeywell to build what were in essence low-cost PDP-11s. To Honeywell in 1975, which needed a moderate volume of parts with an easy-to-program instruction set, each device was worth a hundred dollars (this is ballpark, I don't know the actual price. Also, it dropped over the years.). Selling the same part into the consumer electronics market presented GI with a problem: how to keep selling the part to Honeywell for a respectable price while selling the same part to Unisonic and Mattel for three. The solution: change the part number and the published spec for maximum clock rate. Presto! Changeo! Sparks fly like electricity! Problem solved. Note that the data sheet is silent on the question of plastic vs. ceramic packaging and that the specified operating currents for the CP1610 are precisely the same as those specified for a 5MHz CP-1600. Running the part slower reduced the required operating current and thus the amount of heat that had to be dissipated, but the individuals who actually typed in the altered specification apparently weren't sophisticated enough to know that.

    It was GI, not Mattel, that designed the Master Component circuit board, including the RF shields and heat sinks. GI was contractually responsible for providing chipsets that worked in that operating environment. The earliest chipsets it delivered were packaged in a ceramic package that conducted heat a little better. After considerable experimentation it settled on plastic packages with attached heat sinks because that was the cheaper solution.

    Most vendors supplying a part in a different package just change a letter appended to the part number, as in 74LS00N vs. 74LS00J. Had GI done that here it would have destroyed the illusion. Several members of this forum, relying on GI's smoke and shiny mirrors, have suggested that the CP1610 was somehow fundamentally different from the CP1600. For example, the post to which I was responding stated, "Now, if we were to look at the original CP-1600 (of which the CP-1610 is a detuned variant), it ran at 2x the clock rate." I wanted to make it clear that there was no process tweaking to "detune" the part and that the ONLY difference was the packaging; the two parts used exactly the same die.

 

    WJI

I understand now.  Thanks for the context.  It is a rather interesting historical note.

 

   dZ.

[mr_me]

4 hours ago, Walter Ives said:

Not sure what happened to the image associated with my previous response. Let's try again:

 

    APh's involvement began in December, 1975. You have undoubtedly read elsewhere that APh brought the GI chipset to Chang's attention. It was APh personnel who prepared the Winter CES demo in the secret windowless room at Mattel headquarters in 1977. GI had supplied a TTL emulator built into a chassis similar in size to the one shown in this EE Times article and mounted in a mini-rack.

    The APh engineer sent to survey the situation found a setup that had been unpacked but was otherwise "doornail dead." He plugged it in and spent a couple of days preparing a minimal development environment by hooking up a Teletype model ASR-33 to it, writing and hand-assembling a minimal version of APh's debugger and writing and hand-assembling a minimal interrupt driven program to establish that the system could display background and objects and that the hand-controllers could be read. After that was in place, APh dispatched two programmers to the same room to write and hand-assemble the rudimentary demonstration program shown at CES.

 

    WJI

Dave Rolfe spoke of being sent to Mattel in late 1977 to program a demo for CES.  He mentioned someone at Mattel previously attempting it but was unable.  GI had assemblers and cross-assemblers, did they not provide Mattel some sort of development system to go along with that asr-33?

[Walter Ives]

On 4/25/2023 at 7:51 AM, mr_me said:

Dave Rolfe spoke of being sent to Mattel in late 1977 to program a demo for CES.  He mentioned someone at Mattel previously attempting it but was unable.

Mattel didn't receive the emulator from GI until December of 1977, and the demo (including the fragile emulator!) had to be in Chicago by January 3 for CES, Christmas notwithstanding. In order to make that happen, Chang arranged for APh to program the demo the month before the emulator arrived.

 

The program that was attempted by a Mattel programmer and later taken over by APh was NFL Football. The Mattel programmer didn't start work on that project until shortly after the students started, about July of 1978, when the EXEC was more-or-less stable.

 

WJI

[Walter Ives]

On 4/25/2023 at 7:51 AM, mr_me said:

GI had assemblers and cross-assemblers, did they not provide Mattel some sort of development system to go along with that asr-33?

Nope. It didn't even provide the ASR-33. All the system had was an Addmaster 601 paper tape reader and a ROM loader that could use that reader to load a program from paper tape and begin executing it.

 

WJI

[decle]

Here's the latest update of the Intellivision development tools description:

 

intellivisionDevelopmentBackInTheDay-20230801.pdf

 

The main additions are:

• A new section 5.5 on the development tools and processes used by Roklan / Coleco / Parker Bros (p43)
• Additional information on SOUNDT, Bill Goodrich's sound development tool (page 21)
• A summary of the differences between v2.0 of Rick Koenig's debugger (page 23) and the version embedded in World Championship Baseball
• A brief piece in Section 4.8.2 on the various versions of Killer Bees and their uses (page 39)

Get it while it's hot!

[mr_me]

19 hours ago, decle said:

Here's the latest update of the Intellivision development tools description:

 

intellivisionDevelopmentBackInTheDay-20230801.pdf 4.81 MB · 4 downloads

 

The main additions are:

• A new section 5.5 on the development tools and processes used by Roklan / Coleco / Parker Bros (p43)
• Additional information on SOUNDT, Bill Goodrich's sound development tool (page 21)
• A summary of the differences between v2.0 of Rick Koenig's debugger (page 23) and the version embedded in World Championship Baseball
• A brief piece in Section 4.8.2 on the various versions of Killer Bees and their uses (page 39)

Get it while it's hot!

Thanks for keeping it updated, it's amazing.  A couple of questions.

 

They had CP/M for 8088/86 computers by that time.  The CompuPro 8/16 systems supported both 8-bit and 16-bit processors.  Could Roklan have made a cross-assembler on an 8-bit computer, would that work?

 

And what is the Nuvatec cross-assembler syntax INTV was using, is that not the GI assembly format?

 

edit:

By the way, the guy from Roklan in the interview said he has all their games source code on floppies or hard disk, plus the original computers to run them.

Edited September 20, 2023 by mr_me

[+Lathe26]

Cool!

 

Coincidentally, I was at a used electronics store yesterday and saw the same terminal with multiple loose keyboards as shown in Figure 40.  They also appear to have some CompuPro cards for sale on their eBay site.