Rockwell instruction extension

[litwr]

I am curious to know which manufacturers have realised Rockwell bit instructions (RMB, etc.). It is definitely WDC. GTE and CSG 65CE02 did not realise them. What about the others, such as Synertek?
Interestingly, it seems that the BBC Model B+ uses the Rockwell 6512, which supports these instructions. However, the BBC Master uses the GTE 65SC02, which does not support them. Is this true?

[GARTHWILSON]

GTE (which later became California Micro Devices, CMD) and Synertek do not have them.

[hoglet]

litwr wrote:

Interestingly, it seems that the BBC Model B+ uses the Rockwell 6512, which supports these instructions.

The Rockwell 6512 in the B+ is an NMOS processor, so has the same instruction set as the 6502.

FYI, the datasheet is here:
http://archive.6502.org/datasheets/rock ... x_r651x.pd

Dave

[L0uis.m]

Hello Dave,

hoglet wrote:

FYI, the datasheet is here:
http://archive.6502.org/datasheets/rock ... x_r651x.pd

I was interested, tried to download the 6512 datasheet.
To my surprise, the link returned a "404 Not Found".

[GARTHWILSON]

Maybe it was supposed to be http://6502.org/documents/datasheets/ro ... _r651x.pdf ?

[L0uis.m]

Hello Gart,

GARTHWILSON wrote:

Maybe it was supposed to be [··]

Ah, jea, I'm "a bit" dyslexic, didn't notice there was missing an F.

Thanks

[litwr]

Thank you very much. So it seems that only Rockwell documented the use of these instructions. The WDC documentation doesn't mention them but all (?) the WDC chips support them. There were also Japanese and other companies that were second-source...

[BigDumbDinosaur]

litwr wrote:

Thank you very much. So it seems that only Rockwell documented the use of these instructions. The WDC documentation doesn't mention them but all (?) the WDC chips support them. There were also Japanese and other companies that were second-source...

The Rockwell extensions are described in the WDC data sheet.  They are also covered in Appendix C of the Eyes & Lichty programming manual.

That said, if you want your programs to be portable, you should avoid using the Rockwell extensions.  They don’t exist at all in the 65C802 and 65C816, despite the former being intended as a drop-in replacement for the 65C02.

[hoglet]

L0uis.m wrote:

Hello Dave,
To my surprise, the link returned a "404 Not Found".

Sorry, my bad. The last character got chopped.

Thanks to Garth for posting a working link.

Dave

[L0uis.m]

hoglet wrote:

The last character got chopped.

Duke Nukem says "Typos happen".

[GARTHWILSON]

I also removed the "archive." before "6502.org", and inserted "documents/" before "datasheets".  Actually, I just went to 6502.org, went down and clicked on "Documents Archive", then clicked on "Datasheets" on the next page, then "Rockwell", then right-clicked on the second entry and "Copy link" to get the URL.

[litwr]

IMHO all matters relating to this instruction extension remain rather controversial. It used to be said that all the undocumented instructions of the NMOS 6502 were replaced by new typical/standard instructions or NOPs. However, it seems that all 65C02 chips manufactured by WDC have this extension. Furthermore, some people even claim that all 65C02 chips have the extension!
https://www.masswerk.at/6502/6502_instruction_set.html
WDC created the 65C02. What was the reason that some second-source manufacturers got the extension and some didn't?

[BigDumbDinosaur]

litwr wrote:

IMHO all matters relating to this instruction extension remain rather controversial...WDC created the 65C02. What was the reason that some second-source manufacturers got the extension and some didn't?

As is customary in the world of proprietary chip designs, most purchasers of the 65C02, e.g., Apple, required that there be at least two separate sources for the product to avoid a supply-chain disruption.  As WDC itself has no foundry, they licensed production of the 65C02 to Rockwell, Synertek, NCR and others, thus meeting the second-source requirement.

Rockwell’s interest in the 65C02 was for use in embedded applications, especially modems.  Since manipulating I/O devices often involves bit testing and twiddling, Rockwell devised new instructions specifically for expediting such operations and added them to their production parts.  Also, they mapped their I/O into page zero to improve performance, which is why the extensions only have zero page addressing.

However, in doing so, Rockwell had forked the development of the C02, creating a significant compatibility issue for designers who required at least two sources for the MPU.  As the extensions were Rockwell’s intellectual property, other 65C02 producers could not use them—only WDC was legally able to license the 65C02 design to anyone.  So the result was two different 65C02s were being produced: the WDC base design and the Rockwell enhanced design.

WDC eventually remedied this problem by incorporating the Rockwell extensions into the base 65C02 design and updating the license given to the other producers.  Later on, WDC, having developed an interest in licensing the 65C02 design to producers of implantable medical devices, added the STP and WAI instructions.  With Sanyo’s assistance, WDC converted the 65C02 (and 65C816) to a static core and shrunk the geometry from 1.2µ to 0.8µ and then to 0.6µ, leading to the current W65C02S.  For a long time, Sanyo was the foundry producing the 65C02s sold directly by WDC.  Since around 2005, TSMC has been the foundry, designated with a T in the part number.

[litwr]

BigDumbDinosaur wrote:

litwr wrote:

IMHO all matters relating to this instruction extension remain rather controversial...WDC created the 65C02. What was the reason that some second-source manufacturers got the extension and some didn't?

As is customary in the world of proprietary chip designs, most purchasers of the 65C02, e.g., Apple, required that there be at least two separate sources for the product to avoid a supply-chain disruption.  As WDC itself has no foundry, they licensed production of the 65C02 to Rockwell, Synertek, NCR and others, thus meeting the second-source requirement.

Rockwell’s interest in the 65C02 was for use in embedded applications, especially modems.  Since manipulating I/O devices often involves bit testing and twiddling, Rockwell devised new instructions specifically for expediting such operations and added them to their production parts.  Also, they mapped their I/O into page zero to improve performance, which is why the extensions only have zero page addressing.

However, in doing so, Rockwell had forked the development of the C02, creating a significant compatibility issue for designers who required at least two sources for the MPU.  As the extensions were Rockwell’s intellectual property, other 65C02 producers could not use them—only WDC was legally able to license the 65C02 design to anyone.  So the result was two different 65C02s were being produced: the WDC base design and the Rockwell enhanced design.

WDC eventually remedied this problem by incorporating the Rockwell extensions into the base 65C02 design and updating the license given to the other producers.  Later on, WDC, having developed an interest in licensing the 65C02 design to producers of implantable medical devices, added the STP and WAI instructions.  With Sanyo’s assistance, WDC converted the 65C02 (and 65C816) to a static core and shrunk the geometry from 1.2µ to 0.8µ and then to 0.6µ, leading to the current W65C02S.  For a long time, Sanyo was the foundry producing the 65C02s sold directly by WDC.  Since around 2005, TSMC has been the foundry, designated with a T in the part number.

Thanks a lot for this detailed answer.

[litwr]

It would be interesting to know if this extension could help existing computers from the '80s. I can think of only one system that could benefit: the Commodore +4. The Commodore +4 uses a port at 1 for communicating with serial devices, such as disk drives, so the extension could provide a faster method of working with this port and potentially speed up the turbo loader.
Does anyone have any similar examples?