Was anyone else disappointed with the A1200?

[Bruce Abbott]

Originally Posted by Promilus

As it happens WDC was the only entity which actually did develop 6502 processors further. Commodore just made few variants for their own machines (like 7501 and 8502) with just mild adjustments and hardly any improvements.

MOS Technology 6502

The main change in terms of chip size was the elimination of the tri-state drivers from the address bus outputs. A three-state bus has states for "1", "0" and "high impedance". The last state is used to allow other devices to access the bus, and is typically used for multiprocessing, or more commonly in these roles, for direct memory access (DMA). While useful, this feature is very expensive in terms of on-chip circuitry. The 6502 simply removed this feature, in keeping with its design as an inexpensive controller...

The 6510, a direct successor of the 6502 with a digital I/O port and a tri-state address bus, was the CPU utilized in the best-selling Commodore 64 home computer.

CSG 65CE02

The CSG 65CE02 is an 8/16-bit microprocessor developed by Commodore Semiconductor Group in 1988. It is a member of the MOS Technology 6502 family, developed from the CMOS WDC 65C02 released by the Western Design Center in 1983.

Like the 65C02, the 65CE02 was built on a 2 µm CMOS process instead of the original 6502's 8 µm NMOS technology, making the chip smaller (and thus less expensive) as well as using much less power. In addition to changes made in the 65C02, the 65CE02 also included improvements to the processor pipeline to allow one-byte instructions to complete in 1 cycle, rather than the 6502's (and most variants) minimum of 2 cycles. It also removed 1 cycle delays when crossing page boundaries. These changes improved performance as much as 25% at the same clock speed.

Other changes included the addition of a third index register, Z, along with the addition and modification of a number of instructions to use this register. The zero-page, the first 256 bytes of memory that were used as pseudo-registers, could now be moved to any page in main memory using the B(ase page) register. The stack register was widened from 8 to 16-bits using a similar page register, SPH (stack pointer high), allowing the stack to be moved out of page one and to grow to larger sizes.

The 65CE02 was the basis for the system on a chip CSG 4510 that was developed for the unreleased Commodore 65. The 65CE02 was later used for the A2232 serial port card for the Amiga computer.

[Bruce Abbott]

Originally Posted by desiv

[S]C2P- 020 Blitter: 5.275

Interesting.

Cammy at AmigaWorld got 5.8 fps on a CD32 with stock 14MHz 020 and 8MB FastRAM using Akiko C2P. In low detail mode she got 11.4 fps, 15% faster than my A1200 with 50MHz 030 in high detail mode. That means all you needed to run Doom at a playable frame rate on the CD32 was the essential FastRAM.

[Thomas Richter]

Originally Posted by grond

Nope? Isn't that exactly what I was saying? MOS could have created updated variants of the 6502 and kept on selling that enhanced microprocessor to other companies producing their own computers. The 6502 did sell well and could have continued to sell well if kept competitive.

Not sure about that. WDC provided multiple extended versions of the 6502, but none of them were ready for the rising demands - they were mostly "glue and gum" fixes without really addressing the deficiencies of the design. While there was a 24-bit variant with a larger address space, it was a quite unorthogonal design that was still badly equipped for any higher language.


Consider all complexity you had to go through for "stack based" parameter passing necessary for higher languages or generating a proper higher-precision arithmetic with the condition codes of the 6502. All these issues were solved by better designs like the 68K, and there were reasons why 6502-based designs died out - not only due to the small stack and the small address space.

[grond]

Originally Posted by Thomas Richter

All these issues were solved by better designs like the 68K, and there were reasons why 6502-based designs died out - not only due to the small stack and the small address space.

And then an 8080-based design took over. How was that any better suited? The 6502 addressed the low-cost market much the same as the 8080 did and its binary incompatible 16 bit variant 8086. I don't see why the 6502 couldn't have been turned into a 16 bit variant in a very similar way.

And let's not forget that microprocessors weren't only about computers, microcontrollers were an important market.

Originally Posted by Thomas Richter

not only due to the small stack and the small address space.

It seems like the small stack was fixed for the CMOS variant of the processor mentioned above.

[Promilus]

Originally Posted by Bruce Abbott

MOS Technology 6502

CSG 65CE02

5 years behind small fabless business was a long time back then. In 88 WDC had 8/16 bit design for several years and contract with Apple. So... the only machine "enhanced" 6502 from Commodore ever appeared was basically Commodore 65, never released and directly competing with a couple of years older A500. I'd say that's yet another example of brilliant moves by Commodore. AFAIK SNES is based on customized 65C816 and all licensing fees went to WDC as well (through Ricoh). Not a penny for Commodore. So much for being innovative...

@Thomas Richter - I think 6502 do access DRAM faster than 6800, especially zero page. The only issue with WDC enhancements is they came with same 8bit data bus and I think 24 bit addressing was done the same way as enhanced 8051 - through the use of latch. That makes most 16bit enhancements essentially useless but should there be "full 6502" kind of like 80C251 to 8051 I'm pretty sure it might've been fairly competitive against 68000 (that particular model, not entire 68k line).

[desiv]

Originally Posted by Bruce Abbott

Interesting.
Cammy at AmigaWorld[/URL] got 5.8 fps on a CD32 with stock 14MHz 020 and 8MB FastRAM using Akiko C2P. In low detail mode she got 11.4 fps, 15% faster than my A1200 with 50MHz 030 in high detail mode. That means all you needed to run Doom at a playable frame rate on the CD32 was the essential FastRAM.

Interesting...
Obviously without a CD32, I can't test that... But it's surprising that Akiko did so well on a slower machine.
As I mentioned, maybe the C2P routines in DoomAttack aren't that good?
And also, I was using sound and not changing any other settings from defaults other than the C2P routines... So that could be at play also... I'm thinking Low Detail isn't default, so I might try that to see how it performs.

I haven't tried any other Doom ports either (ADoom is one mentioned in a lot of threads too), so perhaps they have better performing routines...
A quick google seemed to show that DoomAttack on my 28Mhz 30 seemed to be running about as fast as it does on a 386 of comparable speed with a non-fancy VGA card.

[Thomas Richter]

Originally Posted by grond

And then an 8080-based design took over. How was that any better suited? The 6502 addressed the low-cost market much the same as the 8080 did and its binary incompatible 16 bit variant 8086. I don't see why the 6502 couldn't have been turned into a 16 bit variant in a very similar way.

Consider some very elementary operations in higher languages: Access an element of a structure. In the 8080 and variants, you load an index register with the base address of the structure, and use the displacement + register addressing mode. That already existed for the 8080. Now consider the 6502: There is no 16-bit wide index register. Rather, you need to place the base address in a pair of zero-page registers, load the Y register with the offset and then access the element.... Actually, you would need to turn the 6502 addressing inside out to make this happen.

Originally Posted by grond

And let's not forget that microprocessors weren't only about computers, microcontrollers were an important market.

Sure, and the 6502 is still used there, but that was not quite the point.

Originally Posted by grond

It seems like the small stack was fixed for the CMOS variant of the processor mentioned above.

Yes, it was - yet the 6502 lacked stack-indirect addressing. You first need to transfer the stack pointer to the X register, and then perform the addressing. The 6502 was really badly equipped for stack-based operations.

[huggy70]

>>Sure, and the 6502 is still used there, but that was not quite the point

Exactly, the point is:
>Was anyone else disappointed with the A1200?

[Bruce Abbott]

Originally Posted by Promilus

5 years behind small fabless business was a long time back then. In 88 WDC had 8/16 bit design for several years and contract with Apple. So... the only machine "enhanced" 6502 from Commodore ever appeared was basically Commodore 65, never released and directly competing with a couple of years older A500. I'd say that's yet another example of brilliant moves by Commodore.

You said "WDC was the only entity which actually did develop 6502 processors further. Commodore just made... hardly any improvements.". That was not correct. So now you change the goalposts and reveal your real agenda - yet another example of unfairly bashing Commodore.

This narrative that WDC had the right stuff while Commodore were useless is getting as tired as the 6502 was. Of course Commodore only developed enhanced 6502s for their own machines. Why would they do any different? Only one other home computer used the WDC 65816 - the Apple II GS, released in 1986. It was not a success.

When I found out about the 65816 I studied the datasheet and was not impressed. Not fully backward compatible with the 6502, and barely 16 bit. Apart from the 24 bit address extension its registers were similar to the 6809, which was introduced in 1978 (I bought one in 1981 bundled with the official programming manual, for US$15). The 6809 was used in several successful home computers including the Tandy Color Computer 2 and 3 (which I now have and must get into programming some day!).

[Bruce Abbott]

Originally Posted by desiv

A quick google seemed to show that DoomAttack on my 28Mhz 30 seemed to be running about as fast as it does on a 386 of comparable speed with a non-fancy VGA card.

Yep. The typical 386SX-25 that most people were buying in 1992 was much slower, even with a fancy ET4000 graphics card.

[freehand]

I had a 486 did not know anyone with a 386, what was your typical amiga user using in 1992 ?

[lmimmfn]

Originally Posted by freehand

I had a 486 did not know anyone with a 386, what was your typical amiga user using in 1992 ?

In 92 I had an A500 with 2.5Meg RAM, loan of an A2000 with I think 2Meg RAM, 12Meg Harddrive, 286 card, multisync monitor.
Late 92 I replaced my A500 with a stock A1200, had external dusk drive though.

[grelbfarlk]

I had an A4000/040 in December of 1992. Also an A2000 with 4MB and and IDE HDD and a 286 bridgeboard.

[desiv]

Originally Posted by freehand

I had a 486 did not know anyone with a 386, what was your typical amiga user using in 1992 ?

I wouldn't call myself typical, but... ;-)
I did get an A1200, but I'm sure it wasn't in 92. It would have been 93 or 94...
So in 92, I still have my A500 and 1084S monitor.
That had 1M RAM and a Dataflyer 500 SCSI. (Can't remember the hard drive size... Maybe 40M?)
I also had an external floppy, Aprotek modem (1200 baud????), a digiview and a HAM-E external graphics board (I won that in a Compuserve event with Black Belt Systems)... Oh, and a Panasonic dot matrix printer. ;-)

Now, I also used to help people with upgrades and I got to keep the old parts. So PC wise, I was able to piece together an XT. Originally an 8088. Then I got a 286 Inboard for it. Then a 386SX. I think I had that one till I got a 486...
I think I was still using the 386 when I got my 1200...
My original Doom experience was on that 386, so sub 10 FPS was my original Doom experience... ;-)

[freehand]

My dx2 66 cpu cost me 10 floppy disks to a local company when I was naughty boy back in the day

Making money was easy from PC users coming from the amiga most had problems with just configuring there pc's

[Promilus]

Originally Posted by Bruce Abbott

You said "WDC was the only entity which actually did develop 6502 processors further. Commodore just made... hardly any improvements.". That was not correct. So now you change the goalposts and reveal your real agenda - yet another example of unfairly bashing Commodore.

BS. EVERY CPU has DIGITAL I/O and 6510 additional lines were directly related to C64 memory mapping. It's not actual development. It's just those mild adjustments to target platform. So don't give me that wiki nonsense to support your silly arguments.

This narrative that WDC had the right stuff while Commodore were useless is getting as tired as the 6502 was.

Of course WDC did develop 6502. They implemented CMOS first and then made 16bit version used largely in Apple II GS (~1.2M), SNES (~62M) ... They produced (and still do!) microcontrollers based on 6502 used in variety of applications including medical and telecommunication. So when you sum things up it is pretty clear WDC made bigger impact and on wider market than Commodore with their CPUs made for their own stuff only. Now you get my point or you'll stubbornly continue to deny the facts?

[sandruzzo]

After all Our loved AGA wasn't so bad! If only had a better chip mem access management..

[CCCP alert]

After losing interest in all things 'retro' 'vintage' now I can look back with an unbiased view of my time with the A1200 in 1992 and remember it did some things better than my 250% of the price soundcard devoid 486 PC bought weeks earlier. Did some things worse too. But it was a solid build quality great value for money computer in 1992. The keyboard was horrible, apart from that never had any complaints for the machine. The AGA games library on the other hand.....

[Promilus]

Originally Posted by sandruzzo

After all Our loved AGA wasn't so bad! If only had a better chip mem access management..

It wasn't so bad but it still was bad (in terms - OCS was cheap and powerful solution only most professional PCs could match when it has been released, AGA on the other hand was already lagging behind new commercial PC solutions). Obviously you still couldn't get anything quite like AGA on PC for the same price as A1200... but it was clear Amiga was no longer in the lead. And if indeed PC was then ... what's the actual point of developing games for new platform which was neither all that popular nor would've become more powerful easily? That's something Commodore didn't anticipate. After all AGA is just placeholder for mismanaged and underdeveloped AAA. Which still wouldn't be able to win the crown but was not something so easily dismissed

[sandruzzo]

@Promilus

Aga was too late. what if it is was full 32-bit chips set at 14 mhz with one cicle ram access...