Devices that contain 65C802

[Kyle22]

Yes, be careful of those soft pins.

[JamesD]

This is one of those great ideas that never went anywhere. By that I mean the 65802 pin compatible with the 6502. For example it wasn't compatible with the XL series or the C64. So the only thing it was a drop in replacement for was the Apple and those users stayed away from it in droves. I think Kyle found a simple 816 => 6502 hack in the comp.sys.atari.8bit newsgroup and built it. Still that was only good for 800 users with the earlier processor board that took a 6502B.

I followed the comp.sys.???? groups back then and a lot of people were talking about it.

The problem comes down to this:

Does it speed up existing software? No.

Does any existing software take advantage of it? No.

Is any new software being written for it? No.

So why get it?

Then the IIgs was introduced and it offered so much more and software that actually used the new features.

There were some die hards that had them, but there really wasn't much point for the most part.

[+slx]

There were some die hards that had them, but there really wasn't much point for the most part.

 

One could argue there isn't much point in upgrading 8-bits at all The 65c802/16 don't make a lot more practical sense then swapping in a 65C02, it's probably more of a "because it's there" thing.

 

Is the "sweet 16" something that could be replicated (with stiffer pins maybe) or does it contain some "secret" components?

[+David_P]

A fex pix: http://atariage.com/forums/topic/114268-whatever-became-of-turbo-816/page-3#entry1481110

[JamesD]

 

One could argue there isn't much point in upgrading 8-bits at all The 65c802/16 don't make a lot more practical sense then swapping in a 65C02, it's probably more of a "because it's there" thing.

 

Is the "sweet 16" something that could be replicated (with stiffer pins maybe) or does it contain some "secret" components?

Unlike the 65802, the IIgs supported extended memory, better sound, better graphics, more speed, new software, detachable keyboard, a mouse, and a gui as well as backwards compatibility..

For anyone serious about an upgrade the choice was obvious.

About the only reasons to get a 65802 was bragging rights or if you were writing your own software.

I'm sure somewhere out there, there must be a hand full of programs that require a 65802 or 65816 but were designed to run on a standard Apple II... but I've never seen any.

 

About the only use I can really see back in the day, was that some of the people that got them used them to learn how to program the 65816 before getting a IIgs.

 

[Geister]

I though the IIGS was the way forward for Atari and other 6502 based PC makers.

 

The IIGS had everything the next generation of Atari should have had. Many people think that Apple backed the wrong horse when they choose the Macintosh over the IIGS. It took a long time for the Macintosh to get many of the features the iiGS had out of the gate.and the IIGS was faster than the tombstone Mac and had broader software support with the backwards compatibility.

 

An Atari GS version would have given Atari users a compatible path forward instead of Tramiel's stealth next gen C64. Hell, even the Amiga was a better path forward for Atari but greed and dirty dealing killed that option.

 

If Apple hadn't followed that psycho Jobs and gone with the better machine, if Atari (and Warner communications) had a better understanding of the market and even once brought out a next gen computer that people actually wanted instead of recycling the same design in a new suit every couple of years, the market might be a lot different today. Western Design Center might have seen a future in their 65C832 chip or used the COP instruction to develop a math coprocessor that would have made a 16-bit Atari or Apple a serious number cruncher. Instead, lack of growth relegated WDC to packaging 6502 cores into embedded chips instead of developing a third serious PC architecture. And we got recycled 8-bits as long as Jack could wring another dollar out of them. At least Jack gave us 128K to play with. Atari (WC) didn't have that much foresight.

Edited January 24, 2018 by Geister

[phaeron]

and the IIGS was faster than the tombstone Mac

Are you sure about that? You're pitting an 8MHz 68000 against a 2.8MHz 65C816 that has to slow down to 1MHz when accessing video memory. My uncle used to be a IIgs user and I distinctly remember how slow GS/OS was to boot... and that was with "ExpressLoad."

 

That having been said, Apple did a lot with regard to evolving the Apple II series. They upgraded the CPU, firmware, memory, storage, graphics, and eventually sound... and also made DMA-capable SCSI and network adapters. Forget the Apple IIgs, it would have been nice if Atari had made the equivalent of the Apple IIe -- which used its auxiliary memory bank to provide 80 column text and double hi-res graphics. Then they used double CAS memory access to double memory bandwidth again for super hi-res in the IIgs. And ported pretty much the entire Macintosh toolbox and OS. The IIgs memory design also has nice features which haven't been replicated in 65C816 adapters people have been making for the Atari, such as having bank 0 be all fast RAM and moving the legacy address space to $E0/E1 with shadowing, so you can access both without hitting bank switch registers. Even the ill-fated Apple /// had some interesting features including character attributes, extended indirect addressing, and alternate stack/zp page.

 

An unaccelerated 65C802/65C816 isn't that interesting, frankly. You get 16-bit operations, but the advantage is reduced by the code size and speed penalty of having to pepper REP/SEP instructions everywhere to switch modes. Stack operations? Great, except you can only use stack addressing modes with A and setting up a stack frame with D as a frame pointer takes a dozen cycles. The yucky M/X mode switching design also destroys one of the major advantages of the 6502, low interrupt latency, since it takes so long to get to a uniform register size and safely save registers. This makes servicing DLIs at 1.79MHz in native mode challenging and you end up having to do things like use long and absolute mode instead of absolute and zero page addressing. If you have an accelerated CPU, then you can overcome all this with the faster clock speed. But just an unaccelerated 65C802? It's like the time I put a 68010 in an Amiga 2000... faster in a few cases, breaks compatibility with a couple programs, and is overall not worth it.

 

[jmccorm]

Are you sure about that? You're pitting an 8MHz 68000 against a 2.8MHz 65C816 that has to slow down to 1MHz when accessing video memory.

I remember when I was on the couch, watching this YouTube video when I was sick around Christmas time. It did a side-by-side comparison of the two. They stated in the video (and I've heard elsewhere) that Apple limited the clock rate of the IIGS (which was already more efficient per clock cycle) specifically so that it would not be competition with the Mac.

 

Oh! My AtariAge message toolbar items aren't working since FireFox updated. Well, I can't work the code to embed the YouTube video, but here is the URL (and maybe the forum will embed it anyhow):

 

http://youtu.be/7h4tepFbMso?t=1m19s

[ricortes]

I remember when I was on the couch, watching this YouTube video when I was sick around Christmas time. It did a side-by-side comparison of the two. They stated in the video (and I've heard elsewhere) that Apple limited the clock rate of the IIGS (which was already more efficient per clock cycle) specifically so that it would not be competition with the Mac.

 

Oh! My AtariAge message toolbar items aren't working since FireFox updated. Well, I can't work the code to embed the YouTube video, but here is the URL (and maybe the forum will embed it anyhow):

 

http://youtu.be/7h4tepFbMso?t=1m19s

IIRC: It was much worse then this. Remember the first Mac was only 128k! I've heard in order to save $.02/mother board Jobs had them take out the address lines on the MB. This made expansion of the system in house impossible w/o a redesign of the MB and expansion of the original Macs by third party vendors expensive. Then there was the fan or lack of one. No fan and failures due to overheating.

 

I dwell on systems often. The shortfalls of designs people have mentioned in this thread i.e. no speed increase, no new software that take advantage of the new hardware, all ring true to me. A lot of this is just plain due to the hardware design of the original system. Would'a could'a should'a. I find the Intel/IBM way of doing things was superior to everything else in several key aspects. What was originally a flaw, number of clock cycles used to execute an instruction, they cured by doing a pipeline architecture. They had a fast enough bus<~6 MHz by the AT> and display cards that ran independently of the processor. I've actually run a VGA card on an AT and it wasn't horrible.

 

Some of the other schemes like display/CPU interleaved access were OK, but limited just how fast you could go because of the memory speed needed for this to work at higher clock rates.

 

Then you have us, we are a weird lot of purist and hackers. You can only do so much before some say it isn't an Atari anymore and on the other extreme, you have people that can never get enough.

 

I think what could have worked for all masters is ~a 65C02C with FULL one instruction per clock cycle pipeline architecture. Drop in replacement that would execute code ~4X the speed of an existing 6502C with better noise immunity, lower power, less heat. If all interrupts first step was to put it in pipeline mode, 4x the VBI code could run and ditto for DLI. We would still have problems with lack of a great display but at least it could get price and compatibility issues put to rest.

[JamesD]

Are you sure about that? You're pitting an 8MHz 68000 against a 2.8MHz 65C816 that has to slow down to 1MHz when accessing video memory. My uncle used to be a IIgs user and I distinctly remember how slow GS/OS was to boot... and that was with "ExpressLoad."

...

An unaccelerated 65C802/65C816 isn't that interesting, frankly. You get 16-bit operations, but the advantage is reduced by the code size and speed penalty of having to pepper REP/SEP instructions everywhere to switch modes. Stack operations? Great, except you can only use stack addressing modes with A and setting up a stack frame with D as a frame pointer takes a dozen cycles. The yucky M/X mode switching design also destroys one of the major advantages of the 6502, low interrupt latency, since it takes so long to get to a uniform register size and safely save registers. This makes servicing DLIs at 1.79MHz in native mode challenging and you end up having to do things like use long and absolute mode instead of absolute and zero page addressing. If you have an accelerated CPU, then you can overcome all this with the faster clock speed. But just an unaccelerated 65C802? It's like the time I put a 68010 in an Amiga 2000... faster in a few cases, breaks compatibility with a couple programs, and is overall not worth it.

 

Western Design makes some claims for the speed of the 65816 as far as number of instructions per whatever, but they leave out the part where it takes more instructions to do the same thing. The 68000, is sort of a 32 bit CPU running on a 16 bit buss and offers full 32 bit buss versions where there is no upgrade path for the 65816. Opcodes take quite a few more cycles than on the 65816, but again, it takes fewer of them to do the same thing, and as you say, it's clocked faster than the IIgs. I think the biggest problem with the Mac, is that it was written in PASCAL. The larger number of registers on the 68000 actually supports compilers pretty well. The problem back then was that compilers didn't generate very optimized code. In just a couple years from the release of the Amiga, the code optimizers improved quite a bit. I have to think there was room for at least a 10% speedup on the original MacOS with a better compiler, but maybe I'm wrong.

 

The 65816 has a lot of shortcomings and you certainly only touched the surface. Even compared to the 6809, there is a significant difference in code size. If it weren't for the 65816's instruction pre-fetch, it would be no contest.

But the 65802/65816 has significant improvements over the 6502. It has a 16 bit stack pointer, you can have 64K pages for code or data that can be accessed with smaller code than the 6502 could for even 1K, it has stack relative addressing (though limited), and even though the 16 bit accumulator and memory move instructions aren't ideal, they are still a significant improvement over the 6502. One of the most important things about those changes is that they support compilers better. The larger stack and stack relative addressing alone should make a significant difference in code size and speed. You can use the hardware stack instead of one maintained on Page 0.

 

Motorola analyzed tens of thousands of lines of code to come up with the 6809 design, I have a hunch that wasn't the case with the 65802/65816.

Where the 6809 was designed from scratch to support high level languages and broke backwards compatibility to do so, the 65816 appears to be designed to look more like the 6809 while still being able to run 6502 code.

The mode switching choice was probably due to the number of available opcodes and simplicity of decoding them.

They didn't want to use any 2 byte opcodes like the 6809.

A few 2 byte opcodes on lesser used instructions require fewer clock cycles than mode switching, but it may have a significant impact on the hardware required to decode them.

 

[JamesD]

...

well. The problem back then was that compilers didn't generate very optimized code. In just a couple years from the release of the Amiga, the code optimizers improved quite a bit. I have to think there was room for at least a 10% speedup on the original MacOS with a better compiler, but maybe I'm wrong.

...

Keep in mind that also applies to applications running on the Mac.

[+slx]

Yes, I know, unfortunately...

 

I am looking for a 40 pin DIP to put in my Atari 800 (with standard 6502, not C014806).

 

A "fake" 65C802 made from a small form factor 65C816 with some auxiliary chips and jumpers is sold here for 35$ including shipping. Ordered one although it could be a tight fit on the 800 with the extra height of the board.

[Kyle22]

 

A "fake" 65C802 made from a small form factor 65C816 with some auxiliary chips and jumpers is sold here for 35$ including shipping. Ordered one although it could be a tight fit on the 800 with the extra height of the board.

 

Thanks for the info. It may help someone else. I already have a real WDC 65c802 in my 800.

 

Edit: If they could put some linear RAM on it addressed @ 10000 and up, it would be perfect. As long as they prevent access to $1D302, etc. from showing up on the external address bus.

Edited December 1, 2018 by Kyle22

[+slx]

 

Thanks for the info. It may help someone else. I already have a real WDC 65c802 in my 800.

 

Edit: If they could put some linear RAM on it addressed @ 10000 and up, it would be perfect. As long as they prevent access to $1D302, etc. from showing up on the external address bus.

Interesting idea. Still far from real need territory here but Ill ask the guy. I assume that ideally any access above $FFFF should not show on the external Bus and the $1D302 you mentioned is required to keep hardware registers safe?

 

Redirecting everything above 64K to the onboard RAM would allow using it on all 6502 based systems regardless of their hardware register location.

[+slx]

Thanks for the info. It may help someone else. I already have a real WDC 65c802 in my 800.

P.S.: where did you get yours?

[Kyle22]

P.S.: where did you get yours?

I bought it from someone here on AA a number of years ago. Can't remember who now.

[Kyle22]

Interesting idea. Still far from real need territory here but Ill ask the guy. I assume that ideally any access above $FFFF should not show on the external Bus and the $1D302 you mentioned is required to keep hardware registers safe?

 

Redirecting everything above 64K to the onboard RAM would allow using it on all 6502 based systems regardless of their hardware register location.

Yes, that's what I was thinking. Address onboard RAM above 64K and keep access off the bus so as not to interfere with hardware registers that can only decode 16 bit addresses. If you only see the lower 16 bits then $1Dxxx, $2Dxxx etc. looks like $0Dxxx to them and they will react when they shouldn't.

[Wyatt Wong]

How many W65C802 do you require ?

[Wyatt Wong]

On 11/29/2013 at 7:42 AM, Kyle22 said:

I'm looking for info on the 65C802 chip. I have used search engines, but gotten no real results. I would like to know what devices use these chips. I know they were somewhat popular in Apple & Franklins. I can't find them for sale, and I am looking for one for my 800.

 

If I know what they are used in, then I know what to look for at junk sales, flea markets, etc.

 

Thanks!

 

-Kyle

 

How many W65C802 are you looking for ?

[Kyle22]

On 11/10/2020 at 8:07 PM, Wyatt Wong said:

How many W65C802 are you looking for ?

PM Me with cost per piece / cost per 10, etc.

Thanks.

 

[Wyatt Wong]

6 minutes ago, Kyle22 said:

PM Me with cost per piece / cost per 10, etc.

Thanks.

 

PMed

[Wyatt Wong]

I have some W65C802-P chips for sale.